#513 – Audio DSP with Shannon Parks

Shannon Parks: This is The Amp Hour Podcast. Released October 18th, 2020. Episode 513. Audio DSP with Shannon Parks.

Chris Gammell: Welcome to the Amp Hour. I'm Chris Gammell of Contextual Electronics. And I'm Shannon Parks of Parks Audio. Hey Shannon, how are you? Doing fantastic, Chris. So we're going to be talking today about a little bit of audio stuff and your manufacturing business and your design. And like we were talking about a little bit before the show, we've had some people on that have done audio, but audio is your full-time thing. And there's a specific crowd that pays for high-end. In your case, it's affordable, but it's still high-end and it's very specific to people's tastes and what they want to do with that sort of thing. So what is the audio industry like these days? In 2020, what is it like to sell an audio product to not the mass market, but the public?

Dave Jones: Well, I mean, there's all different types of audio products now. And it's a bigger product base than ever. I mean, you have just some people wanting to convert audio off of their TV set to go into their stereo. You have people that want to have like a Sonos wireless setup so that everything's connected to their phone or whatever wireless streaming service that they're using. And then you have people that are using traditional stereo equipment, things that they've picked up from Goodwill or that they've got inherited from their folks. And that vintage market where people are listening to records and now even reel-to-reel tapes and cassette tapes, just people get a lot of fun out of that, a lot of entertainment.

Chris Gammell: Yeah. Yeah, it's interesting. The reel-to-reel one too. I know there's like special sound about it, like all this, you know, all this very subjective stuff. But at the same time, I'm just like, man, I hate rewinding things. I'm just amazed that people are like, oh yeah, that's the experience I want is like to rewind stuff. And sometimes the tape breaks too. That's, you know, that's great. I love my media getting destroyed.

Dave Jones: I've never been a tape guy. And the ritual is part of it in all these different medias and formats. Yep. Just how you allot your time to do what you're going to do and relax and listen to music. And definitely the reel-to-reel, it's not my cup of tea, but I mean, I have a friend down the road that's really big into it. And I've heard his setup and it's amazing and it's a great experience. Yeah. But I'm a record guy.

Chris Gammell: Yeah. Well, even records for me, I'm just like, I don't like listening to whole albums. And yet some people are like, Chris, I do not want to listen to the whole album. It's like the experience. I'm like, I got stuff to do, man. But like some people like just sit down and listen to the whole album as it was designed. And like there is like the whole like start to finish. You know, there was someone who put a lot of thought into what song goes into what song and how the whole thing feels. I'm just like, yeah, all right, great. I'm going to put on my Spotify playlist.

Dave Jones: Well, that's the beauty of these older records that, I mean, you go back to a Beach Boys record in the early 60s. And those album sides may be 12 minutes long. There may be, you know, six two minute songs. So just listening to one album side only takes 12 minutes. And I like I like that format better than, you know, getting back into the 90s and the aughts when CDs were taken over. I mean, they put every bit of those 72 minutes. They put they put songs on their songs that probably maybe shouldn't have been on there.

Chris Gammell: Right. And then they had to sell it. Right.

Dave Jones: Yeah. I mean, it's like people want more. And and then sitting down for 72 minutes. That's yeah. I can't do that. I can't sit down for a movie. I don't like sitting down for an hour. But a record that's that's a good format for me. I mean, it's like the length of a sitcom. Yeah. Twenty three minutes or whatever that is. Yeah. That's about the length of one side of a record.

Chris Gammell: Yeah.

Dave Jones: OK. Yeah, that's that's a great that's a great. I guess that's that's that's where our attention at least my attention span is, is that Gen Xer.

Chris Gammell: Mm hmm. So let's talk about records, too. So records obviously have had this crazy resurgence. I saw that there was a tour. I think Adam Savage did a tour with Jack White in his pressing factory in Detroit, I think. And it's like it is very interesting that it's like as a format is coming back. What is the status in 2020 of new records, old records? Like I just actually rewatched High Fidelity. So, of course, now I'm thinking about like record collections, too.

Dave Jones: I love that movie. I mean, if you like records, you absolutely have to love High Fidelity. Biographical. I mean, I mean, the Smiths and he's talking about the Smiths in there. He's talking about Zappa and things that I love Zappa. I love the Smiths. So, I mean, everything in that that movie, it's. Yeah.

Chris Gammell: It's all Chicago based, too. So I actually like I used to live pretty close to where the, you know, the fake store was on Honoré and Milwaukee. So like I used to go by there. It's like a, you know, high end. That whole area is like very high end now. But, yeah, it's fun to it's fun to spot that stuff.

Dave Jones: But what the status is now. I mean, I don't I mean, I know it's it's huge. And, you know, I see those articles that that somebody, you know, like an aunt posts on Facebook and saying that that records are going to outsell CDs this year and things like that. And, of course, you know, CDs who's buying CDs. This is right. Right. And but I think it's it's it's always I think it's got at least another 10 years on it. That's that's always my horizon. Of course, I would have said that 10 years ago.

Chris Gammell: Sure. Sure.

Dave Jones: But I remember going to, you know, Record Store Day, the once a year they have a big release in spring in April. Usually, I think it's been bumped back to I think it still hasn't happened yet because of because of COVID. Who's they? They is. Well, I don't know who's behind Record Store Day other than who started originally. But all these different record shops are behind it. And the record companies. And, of course, there's a lot of small businesses like like the one you mentioned. What's what's his little business? It's up in Detroit, right?

Chris Gammell: Yeah. The Jack White label. Yeah. Yeah. It's like three, three something. It's got three in the name. I forget what it is.

Dave Jones: But they do all kinds of special Record Store Day releases. RSD. Got it. Okay. And it's a big thing.

Chris Gammell: But it kind of sounds like a support your small business kind of like I think American Express sponsors it. But it's like, yeah, okay. So it's like support local businesses. And it's like a specific day that everybody kind of gets together.

Dave Jones: They do all kinds of special. Hey, here's something that was recorded live 18 years ago. It's never been out. And it's only going to be out on a thousand copies of vinyl. I see. Okay. And when I was going to record, well, just early on Record Store Day, it was just a bunch of old guys like me. Gen Xers and boomers. Yeah. And then like in 2014 Record Store Day, I went to, it was in Bloomington, Illinois, which is a college town. And the line was around the block. And it was all millennials. And I was like, oh my gosh. You're like, what is happening here? I was so thankful and so happy about it. Because I mean, hey, this is what I do for a living is make thonal preamps for people that listen to records.

Chris Gammell: Yes, we haven't said that yet, but we'll get there. Yeah.

Dave Jones: To see that there's going to be another generation.

Chris Gammell: Yeah.

Dave Jones: And I'm getting, you know, I get emails from Zoomers all the time too. So that's when I knew it was getting big back just six years ago due to the next generation taking over. And being, I mean, actually, you know, here the boomers and me, we were a bit lazy eating breakfast and showing up at opening time. Whereas all the millennials had been in line for hours. That's right. To get these limited special releases.

Chris Gammell: Huh. Yeah. So what I'm getting out of all this is that it's more experiential than it used to be. Well, not even than it used to be, but it's a hearkening back to a more experiential time. And people are really longing for that in an age of digitalization and stuff like that of, hey, I want to have a physical thing. One of my friends always talks about like tokenization. Not tokenization. That's a different term. But like tokens, like taking the actual thing with you, buying a record box. It has all the notes. It has everything with it. There's sometimes like, I remember like Radiohead would have like special sets that they sell as a record, that kind of thing. And it's an experience instead of just listening to the song like I do on Spotify or similar.

Dave Jones: Well, yeah. I mean, Spotify and all the streaming, that plays a part in all our lives and we all use it. But it's not something tangible that we own or have that religious experience of mellowing out with a glass of wine or a wine of our favorite homebrew or whatever. It's, they're two different. I mean, I listen to when I stream. I mean, I listen to a lot of YouTube when I'm doing work, when I'm laying out a board or doing something. And that's, that will make a lot of hi-fi guys wince. But it's like, it's, it, that's kind of acting like the AM radio of the old. Or music acts. Well, not, not music. But I mean, AM radio, you just got this scratchy little thing that, that you bring wherever you are to, to fill a need. Yeah. But then to have a, something where you are sitting down and, and, and, and relaxing and listening to music and the music is part of the experience. Like if we're watching Netflix or whatever, it's relax, uh, then having it in a, yeah, a real media that you can hold is important. And, and CD has, uh, it's, it's, it's, that's not the media. People don't, people don't say, Hey, here's a CD. I really want to share with you.

Shannon Parks: Listen to the depth of these bits. Yeah.

Dave Jones: But, uh. 96 kilohertz. Ooh.

Shannon Parks: Yeah.

Dave Jones: But, uh, I mean, the beauty of, uh, of vinyl is you buy the record and then you get the record, but you also get this piece of paper that's inside the record that says, here's all the digital, here's the digital copies too. So when, uh, when a zoomer buys a record now, they're buying things.

Chris Gammell: You're saying zoomer, is that a combining of Gen X and boomer?

Dave Jones: No, that's the, that's the, after the millennials, you millennials are, you guys are going to hit 40 this year. So, yeah, that's true. So, uh, I, so, uh, my daughter's a zoomer and I, so I, the zoomers are on my, uh, zoomers.

Chris Gammell: So that's like Gen Z. Yeah. All these knees names, you know, I think zoomer has to be the zoomers. I think it's great actually. Cause it allows you to be disparaging, but also, you know, respectful.

Dave Jones: Well, it's, it's, it's what, you know, it's what their experience is. I mean, uh, it's Gen Z that kind of means nothing, but are they living through zoom right now? Because they have to, are they all going to remember zoom? Yeah.

Chris Gammell: Yeah. Yeah. Yeah. Yeah. Yeah. Okay. That makes a lot more sense.

Dave Jones: Oh yeah. They're the, they're the zoom generation. I mean, I mean, I can't, my daughter's in fourth grade, so she can kind of use a computer fairly well, but, uh, for kindergarteners, kindergartners that are, uh, the zoomers that are having to use all this. I, I can't imagine what their experience is, but Hey, they're resilient and that's who they're going to be. Yeah.

Chris Gammell: Yeah. No, that's great. All right. So let's talk a little bit. Okay. So we've set the stage now for the fact that it is very experiential. There is a lot of, uh, you know, interesting stuff here. Like you mentioned, what we're talking about here is the puffin. It is a, uh, phono preamp. The first thing that I think is you have a lot of potentially grumpy customers, true or false?

Dave Jones: I mean, I think anybody that's selling something is, is trying to provide a solution to somebody's problem, right? I mean, that's, that's kind of a business one-on-one is somebody has a problem and you need to fix it. And, uh, definitely with, uh, so I don't know if you meant just car curmudgeoning. Customers. I guess touchy.

Chris Gammell: I guess I would expect, so like I've read the Amazon reviews and first off, people love this thing. Uh, there are some really great reviews on Amazon. Uh, and we're going to talk about Amazon a little bit as well, but, uh, I guess you're selling to the public, right? You're selling to actual people that are, that have this, you know, this very specific experience. They love their sound. They want it to sound good. You're dealing with a lot of different setups. And I would imagine that it's, I guess touchy is probably the better, I don't know if I said, I said grumpy, probably touchy, but like, what is it like to deal with people? Because it is a subjective experience, but also revered, you know? Like, so you're making something for, for public consumption that potentially could go very wrong. And then how do you, how do you make something that goes very well? Like, like it does.

Dave Jones: Well, I mean, you got to just, I've been listening. I've been emailing with these people. When I say these people, I mean, people like me, so I don't mean that, uh, but audio people, I mean, I've been conversing and I mean, I've been an audio person my whole life too. So, I mean, I've had a lot of these same experiences. Uh, so if I was not part of that community, I certainly couldn't do this. I mean, I wouldn't have the, uh, uh, I wouldn't have the sympathy or, uh, you know, putting in the time to help solve some of these problems to get where they're wanting to go. But I think, and here's an interesting point, especially for engineers. I'm sure a lot of engineers think that a lot of these audio guys are, are crazy with some of the things, uh, some of the terms they use, uh, on some of these snake oil devices. But I found that, you know, most of these things that they're saying, you know, the words may not be, uh, the correct or the, their implement, their, their solutions may not be correct, but they're talking about real problems that they have, uh, with how the, how it sounds. I mean, these are systems, uh, problems, right? I mean, speakers and rooms and the media. Yeah.

Chris Gammell: And, uh, give us an example of, of like a phrase they might say that outset sounds weird or wrong, but actually is describing a real problem.

Dave Jones: Uh, I, well, here's a couple, you know, one, uh, I, I, and I use this term that it sounds thin. The music sounds that's a great example.

Shannon Parks: Yeah.

Dave Jones: And, uh, and, and a thin sound is just, uh, and I've noticed this with, uh, certain types of cartridges. I mean, all these different cartridges, which are part of the turntable and actually are the transducer. They're vibrating on the record.

Chris Gammell: They all have their own cartridge. Cartridge is the thing that for people that don't, some people may have not actually experienced, uh, you know, phono players in the past, but the cartridge is the thing that actually touch the needle that touches the LP, right. Or the record.

Dave Jones: Right. And every, uh, manufacturer kind of has their own characteristic sound. I mean, these aren't just, uh, flat frequency responses.

Chris Gammell: Yeah.

Dave Jones: Uh, their data sheets may show a flat frequency response, but there's definitely a tone that comes, that comes out. And who knows what type of load. And I may just be into some a hundred kilohertz resistive, a hundred, uh, a hundred, uh, kiloohm resistive load or something. And, and, and not the cable with the capacitance and then what the, uh, the input to the phono preamp works because capacitance really wreaks havoc on these. Uh, these, these cartridges and actually is used as a tool to make them sound correct or, or was in the past. And, uh, so a customer's maybe sound, Hey, everything sounds thin. It just doesn't, it doesn't, I want more warmth. Now that's, that's a real term that I'm sure a lot of engineers are like, Oh, warmth tube sound warm. Now that's, that's a real, uh, I'm sure Dave Jones is, uh, warmth. He's squirming even though he's not here.

Chris Gammell: He heard, he heard you say it somehow.

Dave Jones: But my gosh, warmth is one of the key parts of, of good sound. And that's, I think.

Chris Gammell: What would you, what would you classify warmth as? Like from, so you know, you're like, you're, you're the, you're the, you're the encyclopedia or not the encyclopedia. You're the, uh, you're the dictionary between, you know, audio customer and engineer. What is warmth?

Dave Jones: I mean, I think it's just simply a balance of the lower frequencies versus the higher frequencies. And this is, these are problems that sound engineers that are doing final mastering of, of, of records or albums. Uh, this is the stuff that they deal with, uh, every time that they're doing their production with pro tools, they're trying to balance, uh, how things are in the low frequencies versus the highs and just to make it sound all weighted properly. Got it. And, uh, in a warmth, uh, you know, think about how they maybe produced some of these records back in the seventies. They wanted to dial back some of the bass frequencies so that they could get those grooves closer together. You know, this is all out of it. This is, this is, this is a, well, you know, when you get these grooves closer together, then you can put more minutes on a record.

Chris Gammell: Yeah.

Dave Jones: And not only can you get more minutes on a record, you have a record that plays better on more turntables because you want it to play on everybody's record player. Right. Even something from Montgomery ward. Uh, if, if people can remember companies like Montgomery ward.

Chris Gammell: Yep. So one thing that I would, I would do a callback to as well as Ben Krasno has been on the show before, but he has that amazing animation of a needle in the groove of an LP. That should be in the Smithsonian.

Dave Jones: I, I, I, I, I've seen, I've seen that reposted so many times and I look at it every single time it gets reposted online. And, uh, and that's the only one I really know of that anybody has done something like that. Yeah. Ben, he deserves an award for that.

Chris Gammell: Yeah. Yeah, exactly. The, uh, the Gipheys or something like that. But the thing I like about that and the thing that was, that I always talk about is when I think about like RF and translating stuff and thinking about the term bandwidth, right? I think about like, Oh, like looking at a spectrum analyzer and seeing like how wide the band is. I feel like that's a very visual representation of that term, but then also for something, something about that image specifically as well. It's like seeing the needle in the groove and being like, Oh, it's going up and down left and right. Like, like an audio signal again. And so it's also the bandwidth there, right? It's like a visual bandwidth kind of thing. And what you're saying is sometimes that bandwidth is less to be able to make those grooves closer together. Is that a safe assumption?

Dave Jones: Yeah. I mean, uh, the whole idea of a lower frequencies, you know, being more difficult to physically get impressed on that PVC platter there. I mean, that goes back to the whole reason why I, I have a business in the first place doing, making these puffins, making these phono preamps because they have this, uh, this, this EQ that where they lower the base frequencies, uh, reference to one kilohertz, the base frequencies are down 20 DB. Wow. Wow. And then the upper frequencies going up to 20 kilohertz, those are boosted at 20 DB. So you've got this, you got this slope and you can imagine it hinged, uh, from, from one kilohertz. And then what these preamps do like my puffin, then we just reverse that. We go ahead and boost, uh, the base frequencies, 20, uh, 40 DB. Wow. And, uh, and just, well, we switch everything back. So it's, so it's flat, but if we didn't have that records would be, well, you'd have to go back and look at what they had with 78s because some of these 78s were flat, pressed flat. And, uh, also with these EQs, it gives it better noise floor because then you're, you're doing this de-emphasis and pre-emphasis, uh, roll off. And they were actually going to do that with CDs too. When, when, uh, DACs were 14 bits in the early eighties. And I think they should have, they should have done it. I think that it would have been a better product if they had until it, before they got better DACs down the road.

Chris Gammell: Well, okay. So let's, let's take a quick step back here. So when you say the compressed stuff below one kilohertz, does that mean that the, so now, uh, Shannon, uh, named perhaps after Mr. Information Theory, Shannon, uh, is that information gone? So like, is it compressed below one kilohertz or is it, is it a cut below one?

Dave Jones: Well, uh, and I kind of hate using the word compressed because that makes us think of digital compression. I mean, we're talking about, uh, yeah, physical shrinkage just because it's a 20 dB quieter signal. So, so, uh, a 20 dB, uh, 20 dB quieter, uh, base signal on a record. It's just going to be a much smaller, uh, you're not going to need, you know, physically, whatever that would be three millimeters per groove or something like that. Uh, cause then you might only be able to do three minutes of music or something. And that's why indeed the 78s going back pre LPs. Yeah. They were very short. They're only like 70, you know, just a few minutes because they didn't have that, uh, equalization.

Chris Gammell: Okay. So it's basically like a low roll off then, but, but this is all because of the actual physical, the physical compression, the physical space constraint. Right. Is it that, so like, like a hundred Hertz signal that might've been otherwise, is it like, that would have been like a huge component. Again, I'm thinking of the, the animated gif of the, the, uh, the needle on the record. Would that have been a large component of that track width? Is that the idea or the groove rather the groove size? Yeah.

Dave Jones: Yeah. Yeah. I mean, at a hundred Hertz, that would have been maybe down, uh, minus 10 dB or something. They would have, they would have dropped that down. So, uh, you know, 3.16 times smaller. So you, so you gain that your, your groove there is 3.16 times smaller. So just for a hundred Hertz.

Chris Gammell: Huh? Okay. Yeah. That that's, that's the kind of like, so it is, it's all the physical constraint and then that plays into the actual compression or sorry, the filtering rather. And the, the mastering that happens that to make it fit on a record. Okay. So that makes sense.

Dave Jones: Yeah. I mean, that, that, that brought about great, uh, high. I mean, that was the whole high fidelity thing in the late fifties and why LPs just blew away 78s and how they took over because they, uh, they were able to just, uh, well, long players, what LP stands for. Right. So they were able to get more, more music on there and also at better fidelity.

Chris Gammell: And can you walk us through the, the different record types real quick? So 78s are the, or the large format, right? That's like, uh, is it a larger size?

Dave Jones: Well, 78s are, what are they? 10 inches, uh, primarily. Okay. And the 78s, I mean, that is really kind of a catch-all term going back to any disc. What? Back to the 1880s.

Chris Gammell: Okay. Yeah. Like after, after they went from the, uh, when they were actually flat pressed instead of those coils or whatever the Edison style thing.

Dave Jones: Well, I mean, Edison competed, uh, with, uh, what, Berlinier or Berliner. He, he was going for the discs and, uh, Edison, uh, he liked those, uh, you know, those cylinders. Canisters, right? Yeah. Yeah. Like what?

Chris Gammell: Yeah.

Dave Jones: They, yeah. But, uh, yeah, so there was about 70 years of, of 78s and they may not all be actually 78 and they had different equalizations. I mean, uh, the amount of equalizations that different countries used and different manufacturers used, it was, uh, it was a, a format war really. Right.

Chris Gammell: Yeah.

Dave Jones: And there was, I mean, there's hundreds of those. And now that was another reason why I thought DSP would be cool. Cause I could then do any of those and then, uh, hopefully get some of these, these folks that are really into some of these old records. And if you get people that are huge collectors of 78s, then you get a little bit of bona fides with people that are, that are in a regular records. Yeah. Okay. Newer stuff.

Chris Gammell: I think we have sufficiently set the stage that there are a lot of formats. There's a lot of emotion and there's a lot of interest in getting these records to sound good of all these things. So let's talk about the actual hardware now. So the Puffin is your main product, the only product. And we'll talk about that, I think too, but what, what is it and what's inside of it?

Dave Jones: Well, uh, it's, it's just a little box, uh, what four inches by six, three by six. And, uh, what's in it, I have a Cortex M4. That's the, that's the main brains of it. It has a, that controls the interface screen. I got a little encoder and a menu and, uh, it has a ADC for the analog coming in from the, from the record that gets processed and then gets sent back out through the DAC. And there is a, an adjustable gain stage, uh, before the ADC, just so you, you get in a sweet spot, uh, for your ADC. But it's just, it's standard, uh, standard components that you get from Mauser that people have used and you can get eval boards for any of those things, even on, on eBay, uh, TI parts. But, but the DSP, the Cortex M4s without it and without that single float, uh, my, the FPU on those M4s are just, it's just wonderful. It makes life a lot easier than dealing with fixed, uh, with fixed numbers.

Chris Gammell: I tell you. Okay. I'm trying to find the, the comment that I saw on there. Ah, where is it? Shoot. I can't find it now, but it was like, why would I want to put? An ADC or sorry, a DSP in front of all of my other stuff. It was basically that. And, you know, basically they're saying, Hey, Hey, we understand that this has, you know, digital signal processing internally on it. And that's like poison for my high end audio setup. Yeah.

Dave Jones: Well, I mean, he could, he could restate that question. Why would I want my, my system to be turned into MP3 music? You know, that's, that's kind of, that's kind of the question he has. And that's a good question. I mean, because digital is a, I mean, it's, it's a format. It's a conversion. It's, there's all different types. There's, it's the processing. Uh, they're just, and the quality can be all over the place depending on what you're trying to do. And that's exactly the same way for analog. I mean, it's, what are you trying to do? What quality? The beautiful thing is, you know, now it's easy for us to do 24 bits, 96 kilohertz. We can easily do that, uh, and get good quality, uh, sampling and, uh, aliasing is not an issue. Yeah.

Chris Gammell: Yeah. I feel like a lot of the disparaging remarks around DSP and digitalization and things like that. It's like when, when people are saying that they're saying like, oh, it's bad digital, digitalization or bad processing, you know, codec artifacting that types, types of stuff that happens. But like, sure. If people actually looked at the, the, the actual output of things that like they're, yes, they're not raw. They're not lost. It's not all the other things, but they are, you know, assuming you have everything done right, which I'm going to assume here, like. Everything. It is a perfect, it's almost perfect replication of the actual analog sound. And I try and like talk to people about that, but it's, it's tough.

Dave Jones: You know, it depends on who's doing it because I mean, every day, you know, if we listen, if you listen to YouTube, you can hear some crappy compressed audio on YouTube and, and, and you can also turn on your TV set and see some little compression, uh, artifact, uh, depending on what you're streaming from Netflix. So we see, we see digital artifacts and hear digital artifacts all the time in our lives. So we know they exist. And, uh, so some of these devices, they have to prove it to us that they aren't, uh, going to do that to us because yeah, if, if I listen to a record, am I going to put up with what I put up with from YouTube and Netflix with some of this, uh, some of these compression digital artifacts? No, I certainly don't want that. So yeah, you have to make a product that doesn't do that.

Chris Gammell: Yeah. Yeah. Right. So then you said you had, you had made this choice obviously to make it to DSP front end and things like that. So what does that process look like? So you, you're basically, so you, you described kind of the signal chain here. You bring it in, you clean it up a little bit with, you know, some filtering and some, uh, amplification preamping rather, uh, you do an ADC, uh, conversion. And then what's happening internally in this Cortex M4?

Dave Jones: Well, I mean, let me, and let me get back to the analog real quick before I answer that. I mean, because still, uh, gain and, and low noise is still, that's, that's the bread and butter of everything. If you can't have a gain with low noise, then your, your product's going to suck in audio. And that's, that's, that's one of the biggest problems people have. So yeah, getting that, getting in, uh, and that's all analog. That's just a, you know, a BJT type op amp, very low noise. So you get this good quality stuff in and then, then once you got it in a DSP, I mean, you can have so much fun and this is, this is where I'm excited because I'd love to do this, you know, for the next 20 years, just manipulating it with the DSP. And the key is once you get it in your digital realm, then you can do any number of preamps or filtering or whatever you want to do with that and just have a blast. I mean, so, and you can do it without adding massive amounts of, of noise by having all these discrete op amp, you know, uh, filtering with pots. I mean, there's always these digital, there's always these pots involved in traditional preamps that get noisy and scratchy and, and usually they're high impedance. So they're just noise magnets. So being able to do that in DSP is just, uh, fun, but.

Chris Gammell: That's a good, I like that as, I like that as a comparison as well though, too, because assuming it wasn't all, you know, like someone's like a purist quote unquote, and they're like, no, I want an all analog front end, but yeah, you have five stages. Each stage is amplifying the next. Any noise gets carried all the way through. So.

Dave Jones: So, I mean, a lot of audio purists, uh, have tried to do minimal systems and, and not use, uh, even preamps with tone controls, tone controls, defeatable tone controls have been important to a lot of hi-fi folks over the last, uh, you know, well, for a long time because. You said defeatable? Yeah. Defeatable as in you could switch it. So it would like bypass. Ah, okay. Bypass it all together.

Chris Gammell: Got it. So like for comparison, so you could be like, you know, like a, like an eye doctor of like version A, version B, version A, version B like that.

Dave Jones: Well, I mean, I think that's what they maybe were intentional, you know, the original, uh, how they were engineered so that people could do that. But I mean, I think that's been a, uh, especially with this vintage gear, if people can defeat and bypass some of this tone control, especially on, on something that's maybe 50 years old now, that's getting a little bit noisy and very, you know, all these aging components, then, then, uh, people like to do that. Uh, it's about these pots, these old pots are just.

Chris Gammell: I mean, there's dust in there and they start to get shorted out and yeah, it's, I mean, they, they have wear because of the physical mechanisms that's happening. Like they're carbon, carbon resistors effectively. Right. With like a turn. Exactly. Yeah. It's.

Dave Jones: So, uh, you know, what do I do in the DSP? Well, the first thing I thought of, uh, other than being able to do these different, uh, these different EQs for different records, uh, I was like, Hey, this, I want to be able to just do a simple warmth control because that's, that's something that every it's intuitive for everybody. It does this sound, uh, more, more detail and thin or, or does it, is it warm? And, uh, once again, the warmth is just the balance of more, more weighted, lower frequencies and a little attenuation on the higher frequencies, just so it balances just right to your ear. There's not a right or wrong.

Chris Gammell: Yeah.

Dave Jones: And of course, of course the rooms and the speakers, all that influences that, that, uh, that warmth. So I just wanted to have a knob that I could turn and say, Hey, turn it to where it sounds just right to you. Turn up the good, turn down the bad. Come on.

Shannon Parks: Yeah.

Dave Jones: Yeah. I think it's, I think it's intuitive. I mean, uh, these graphic, graphic equalizers, I think are completely unintuitive.

Chris Gammell: Yeah. Oh, I remember I used, my, my parents used to have like a console record or like a, you know, audio player thing. And it would have like 18 band EQ and I used to just like mess with everything, you know, it had the actual sliders on each one. And, oh man, little, little Chris really messed with their sound.

Dave Jones: Oh, those things are, I mean, if, if somebody is good at controlling those, they, they could probably become a, a famous audio engineer, but that changes from even record to record. Unfortunately, one record may be warm sounding and then the next one could be thin. And, and that's, and that's true. Like right now, a lot of the new pressings have been remastered to be, uh, to be more warm. And it's obvious. It's very obvious. And then the seventies pressings are thinning, more thin sounding just because a, they were trying to make it so it would play on more gear and they could put more minutes, uh, on that side. So just having a simple, simple knob. I like the idea of having a knob that a pleasure knob that you turn it in the direction to make it sound better. Very simple and intuitive.

Chris Gammell: Yeah, that's great. Actually. What would happen? So, so give us a little, like, I guess it's just going to be an explanation here, but like if you had a remastered album from 2020, that then you put on a system that had EQ kind of set for a 1970s thin sounding record, what would that sound like? So like newly remastered, but adjusted for a seventies record. What does that sound like in these days?

Dave Jones: Well, I mean, if you had something set up to warm up a thin record and then you put on a warm record, then you would have a very dull sounding, overly warm, uh, record. And, and, and that happens. I mean, that's, that's, it's nice that you can switch back and forth, uh, easily because, uh, yeah, you, you do go from playing, you know, modern pressings to, to 50 year old pressings. Sure, sure.

Chris Gammell: And so what, what does, what does that sound like? Is it like really heavy mid tones though, or high tones or like what, where is, when it's like, you know, warm on a warm, what is that actually, what's being saturated in that case?

Dave Jones: Well, it's just, uh, yeah, the, the mids and the lows are just, uh, you have less detail popping out and it just sounds, uh, you know, just the overly warm. I just, it's, it's funny that I can't think of a better term.

Chris Gammell: Yeah. No, it's tough. It's tough. It's, and it's like we said, it's subjective. There's, you know, these words that are used for it, but at the end of the day, some people, you know, you know, I can imagine like people that don't even listen to that much music, they wouldn't hear this. They wouldn't understand the difference between the two. They'd be like, yeah, they both sound like music. What's the problem? You know, so it's also, there's also that it's just like human ear and how we are able to, you know, describe it in words.

Dave Jones: I don't think so. I think it is. I think it's very clear.

Chris Gammell: I think that you could demo this to anybody. Here's the thing. Young Chris used to listen to, uh, you know, stuff downloaded off of a DC plus plus and, you know, LimeWire and, you know, 60, 64 kilohertz or less, you know, super compressed. And I was fine with it because I was listening through a, you know, an early 2000 cell phone too. Like, there are many different levels of how crappy people will listen to music and not care. Just saying.

Dave Jones: I mean, and I do that too. I mean, man, some of those Napster songs, my gosh, we're even going back. It's so compressed. Oh my goodness. But, uh, you know, once again, when you have, uh, a knob and you can say, Hey, is this, it's kind of like when you go to the optometrist, you know, uh, uh, one or two, you know, good or, you know, which, which one is better when you do that. But that is, uh, you don't even know, you know, what's he doing. You just know that one's a little more clear. And that's, that's the beauty of just having a knob or a switch versus having to really configure something and say a or B. And it's like, well, no, Hey, I hear more highs. I hear, you know, some, I hear the hi-hat a little bit better that that was kind of completely missing before. And, and, and the room doesn't sound as boomy too much bass in the room. You get, you get all this content from 200 Hertz to 400 Hertz that starts just really making a boomy bass sound instead of a tight bass. And that, and that's, once again, that's the studio engineers are fighting that and dealing with that every day. Sure. And things that, that, that we need to deal with, with our music when we, when we care.

Chris Gammell: Sure. I think another analogy that I've been thinking about as you're saying this stuff too, is it's like, I look at a photo that I take of my dog and I then go into the editing settings. And these days it's not like, Chris, would you like to pull up, you know, the, the greens or the blues or the, you know, the whites, the blacks, whatever, I guess whites and blacks are kind of like included these days, but like a lot of the settings and, and pre filters that are happening on like Instagram type of thing. It's not, you know, Chris, do you want to adjust all the, the color curves and the saturation stuff? It's just like slide left or right. What looks better to you? And then I, then I, you know, so it's kind of like that. It feels like.

Dave Jones: Yeah. I mean, and, and, and let's, uh, let's put this more into something that people use, see all the time. I think about the, the four, three TV format and all the video that's four, three. And of course nobody has, well, most of us don't have four, three TVs anymore. And looking at that format and how, uh, the cable provider, satellite, Netflix, how they decide to show us that old content can be, that can be a real mixed bag. And, and, and most of the time we don't even have a choice in how they do it. I mean, when they like just do a zoom in and, and half the faces. Clip off the top. And it's really blurry and it's like, oh my gosh. And if we just had a, you know, back with the DVDs, we would be able to switch the aspect ratio and all that. But being able to do that, it's, uh, fixing those types of problems, uh, is what I want to do with, with audio. Uh, so, so that you can get the, the aspect ratio and, and the saturation and all that easily and quickly, the way that you find pleasing and not obnoxious.

Chris Gammell: Yeah. Okay. So let's go back to the front end of this thing again. So you mentioned there's some, you know, preamplification and stuff on the front end. What about like the power supplies? How, I mean, that could ruin anyone's day in a signal integrity kind of standpoint. What does it have to look like for this, for this device?

Dave Jones: Yeah. I like simple power supplies and I don't like batteries. So I, I just use a simple, uh, 12 volt off the shelf, uh, switch mode power supply. Oh, really? And, uh, and of course, you know, their, their frequency content is, is up there. I mean, I think, uh, 30 to 50 kilohertz. So filtering is filtering is pretty easy and their, and their leakage currents are small. And the idea of having a, making something that plugged into AC mains, that, that would just terrify me too.

Chris Gammell: Mm-hmm.

Dave Jones: So, uh, yeah, I just, filtering is, is no big deal. I, I just, uh, the way I design things is to have very low noise references and I worry less about regulation so that, uh, things are folk, uh, things, uh, the denoise floor comes from the references as opposed to, uh, whatever's coming in from the power supply.

Chris Gammell: So the power, so what is, what is the actual back of the box look like? What is, what, uh, what is a user plugging in? Back of the box? It's like an office, it's like the office, off the shelf.

Dave Jones: Like, uh, like, uh, I don't, I don't have a back of the box. It's a, because it's, it's the top, it's all top construction, but, uh, uh, yeah, no, I, yeah, not wall warts. Uh, I just, I use the mini desktop supplies. I like the mini desktop that way that they had the detachable C eight connectors. Ah, okay. Or C seven, C seven connectors. I forget which is, uh, you know, the one of them, C seven, one of them, C eight, uh, but yeah, that way I can, it can easily ship overseas to different countries by swapping in a different cord and the wall warts, uh, you're more limited in options with, with the goofy connectors. Uh, so that, that's what I like to do.

Chris Gammell: Got it. Okay. So, so basically it's a DC jack that goes to a brick. Yeah. The brick has the switch mode. Right. And then the input to it's like, it's not, it's the, the, whatever you were saying, the C, whatever the C eight was or C eight. Yeah. Yeah. And that's like, uh, then you can plug into an IEC or whatever your, uh, or whatever your localized thing is. Okay. So then it's an auto transformer on the front side.

Dave Jones: Yeah. You use, you know, their universal supply. So they work anywhere.

Chris Gammell: Mm hmm.

Dave Jones: So.

Chris Gammell: Yeah, that's cool. That's cool.

Dave Jones: Cause, cause yeah, you got, you gotta be able to ship this anywhere. I mean, that's, that's, that's, yeah, yeah, of course.

Chris Gammell: Yeah. And so, okay. So noise you're able to, that's an interesting idea too, that, you know, you just take it up above the audio range and, you know, low pass filter everything on the, on the, uh, power and you're good to go.

Dave Jones: Yeah. Linear supplies, uh, may be great for power amps. It's pretty simple to just throw some huge, uh, capacitors and, and, and, and set up your, for like a power amp, a class B power amp, but for, you still get some residual, uh, low frequency noise, you know, your 120 Hertz, but that's, uh, in a balanced setup, you can get rid of that, but I'm doing everything single ended. Mm hmm. So, uh, I definitely want to stay away from linear supplies. I don't want to get, when you start amplifying, I mean, I can do 72 dB a gain. So when you're doing 72 dB a gain and you have rectified noise in the audio range, you're going to hear it. Yeah. So you gotta, cause that's, yeah, that's, that's, that's tough to do.

Chris Gammell: Yeah. Okay, cool. Okay. So now what does the, so now we've got the power, we've got the, the amplification, but what does the actual signal look like coming in? And like, what about the impedance off the cartridge? So like, it seems like you have to kind of deal with that incoming impedance.

Dave Jones: There's a fantastic, uh, webpage by Jim Hagerman who also makes phono preamps and you can just, uh, type into Google cartridge loading or Hagerman cartridge loading. But if you type in cartridge loading, it'll take you there. And he has got a wonderful document with little calculators even on what that signal looks like and how that cartridge behaves with these different loads. And this is, uh, in actuality, I, I, I read that page eight years ago and I was like, wow, all these different, uh, uh, changing this 47 K is kind of the standard load for phono preamps. And then you have maybe some different capacitive loading, some from your internet interconnect from your turntable and then others from switches that you do on your phono preamp. But with a moving magnet cart, which is the most standard turntable cartridge, it really has a huge effect in this 10 kilohertz to 15 kilohertz range because you have the inductance of the cartridge in parallel with the, the capacitance of, of the cable and the, uh, the capacitance at the phono preamp and the, and the loading resistance at the preamp. Um, so you add all this together and you do not get a nice flat line out to 20 kilohertz and beyond. You, you've got a nice little tank circuit, nice little peak resonance right there, maybe at 11 kilohertz, which is obviously within, uh, yeah, that's hearing range. Yeah. So, so the whole, uh, I know once again, a lot of engineers may be like, how can you hear all these differences? Well, with cartridges and records, it is, it is night and day. You will hear huge differences just due to this loading because you're having things happen at 10, 11, 12 kilohertz. Hmm. And, and, and going on his website really, I mean, it'll show you a little, it'll even, uh, emulate the, the resonance and everything. It's really cool.

Chris Gammell: Yeah.

Dave Jones: And, uh, so being able to adjust that to where you like it, where it sounds good to you, where it's not peaking too much and sounding too bright, where it's not overly loaded. So it's, it's damp where it's rolling off too much and not peaking enough at the highs. That's all about, it's all, you know, personal taste. So once again, a lot easier with DSP to adjust that.

Chris Gammell: Okay. So you do that in, so you basically then back calculate what that cartridge might look like in the frequency domain. And then you adjust for it, you're saying. Right.

Dave Jones: And, and, uh, it really all to adjust that resonance really only happens with high inductance, uh, moving magnet cartridges, uh, classic cartridges like Shure and then, uh, Ortophone. Uh, have a lot of, uh, they have a lot of, uh, inductance. So they really do have kind of a critical peak there. And, uh, the Japanese cartridges and other cartridges may have less inductance where they don't have that, uh, that peak down closer to 10 kilohertz. So they're flatter, but you don't have as much control over the sound. So yeah. In the DSP, I'm kind of emulating this, how you can change, uh, this like a Shure or, or a phone cartridge, but you can do that on any type of cartridge or even just a CD player that you hook up. Uh, so I think that adds a lot of, uh, it can add a lot of vitality and sound nice.

Chris Gammell: Yeah. Okay. What, so let's dive into the DSP side of things. So how do you actually work in that realm? I mean, are you writing basically, you know, code in C and then just kind of working through, like, how do you, how do you prototype all this DSP stuff as well? Cause you mentioned the FPU in, in the Cortex M4, that sounds great, but like, are you writing individual modules and testing things out or, or how do you set this up?

Dave Jones: I did like my, just my original filter work in, in MATLAB. And, and nowadays, you know, I would recommend folks to, to get, uh, to do that all on Python because the tools in Python now with NumPy and what is it? SciPy or, or Skippy. Uh, all these different filtering tools.

Chris Gammell: I'm not a Python person.

Dave Jones: Are you kidding me, Chris? You don't use Python? You know, I I've used it a little bit, but it's just, yeah. How do you use, how do you use KiCad and not use Python?

Chris Gammell: Well, what I do is I go and I find other people's, uh, scripts, then I implement those. And by implement, I mean, I mean, use them and, and I'm very, very grateful for them.

Dave Jones: I, I hand edit. I just use Notepad plus plus to hand edit my, uh, my, my KiCad files, but. Well, I'm, I'm shocked that you don't, that you don't use Python. Uh, yeah, definitely. You need some of these, uh, some of these filtering tools either on, in MATLAB or, uh, in Python to, to design some of these filters and decide what you want to do. And then you're just getting these, uh, standard, uh, uh, these standard digital filters. And then, uh, geez, yeah, I write it all in C. Yeah. I write it all in C. And, and, and later on I changed, uh, I changed it all to, to assembly just because I needed every single cycle.

Chris Gammell: Nice. Uh, okay. And I want to hear about that too, but like the, the, I guess the, so do you have like, uh, a test file that you put into Python or MATLAB? And then you like look at the frequency roll off that gets generated from that? Like what is, what does that process look like?

Dave Jones: Uh, I just use the general, uh, what in MATLAB FV tool parentheses B comma A. I mean, you're just, uh, so the, I use FV tool primarily in MATLAB where you're just, and then you, and you point it to your coefficients.

Chris Gammell: Is that frequency, uh, visualization tool or something? I don't, I don't know that one. Yeah, yeah, yeah. I don't know any of these actually.

Dave Jones: Yeah, yeah. Uh, FV, FV tool is really easy to use, uh, in that you don't have to set up anything. You don't like, uh, like with Python, you have to start doing, you know, using MAT plot lib and to do graphing and things. And it gets a bit complex, but yeah. Uh, and then there's also a filter designer tool with, uh, with, uh, MATLAB. And, uh, there's, uh, I don't use that as much as just looking at FV tool. Okay. Yeah. It just, it just, it just shows it from, uh, it just shows it from, you know, DC to, to Nyquist, uh, is all, is all it does. Cause it, cause you, you, you, you have to punch in what your, uh, what your sampling rate and all that is. So that's, that's what it shows you.

Chris Gammell: Okay. Okay. Cool. And so when you're doing, so then you're translating from Python or MATLAB to C then. And so what does that algorithm development look like then? So, and again, like I'm trying to kind of visualize, like, I haven't seen the menu structure, all this stuff. I assume there's different menus. You can adjust different things, but like you're creating these filters and then you're implementing them. I assume it's functions internally in C or later in assembly. So like, what does that process look like?

Dave Jones: I mean, originally and still, I don't know, maybe half of it still is just a pre-calculated coefficients and put in a table and there that, uh, then that the C, uh, program then references those coefficients and plugs them in. And so it's not, uh, there might be, so like I have a tilt function, which is, has 13 different settings and that'll just point to 13 different sets of coefficients that I've pre-calculated that are in memory. And then, uh, but, and then later on I did make it where, uh, my EQ was even more flexible for the different 78 types where people wanted to be able to do hundreds of different combinations. And I just, I couldn't pre-calculated, well, I couldn't even figure out exactly what they wanted to do.

Chris Gammell: Right.

Dave Jones: You just do it yourselves. Yeah. So I would use Python and MATLAB then to, to, to actually create those functions writing in, in those languages. And then after I got it doing what I wanted it to do, then I would port that over. Uh, I mean, I would, I would actually normally do it in both MATLAB and then Python and then rewrite it again and see, I tell you, getting the whole going from, from pound sign to a percentage sign to all the different commenting and different.

Chris Gammell: Oh yeah. Yeah. Your brain's all messed up trying to figure out which one's which. Yeah.

Dave Jones: Oh wow. Trying to communicate with somebody after doing that all day. But we don't have to communicate with anybody anymore since we're all sitting behind our computers.

Chris Gammell: That's right. Yeah. Right. Yeah. No, no typing necessary when we're, uh, when we're all remote. Right.

Speaker ?: Yeah.

Dave Jones: But I mean, I, I mean, I was, I'm not a, I mean, this is the first time I ever did it. So I was just learning and, and, uh, learning those tools and, uh, it was a lot of fun. Uh, and then, uh, yeah, writing it and see, uh, you just, you do want to write for speed. Uh, so, so, uh, getting out your logic analyzer and timing your cycles and all that, that's, that's a big part of it.

Chris Gammell: Okay. So is that, that's part of taking it into assembly as well? Is that like part of the motivation was basically the, the signal chain internally, like all the blocks of DSP that you're implementing could cause enough delay that it's problematic or what's the reasoning there?

Dave Jones: Yeah. I mean, uh, latency is that's latency is always something you worry about with, uh, with audio DSP. But I mean, in my case, it was more less of a latency issue than just running out of, uh, running out of resources. I mean, I have to do everything with 64 K bytes. So, I mean, I have as much memory as a Commodore 64 that I'm dealing with. And that's, and, and, and if you have, you know, when you're doing digital audio on a, on a notebook, I mean, you have gigs and you have, you have a lot of memory for storage and to do manipulation. Right.

Chris Gammell: You're not running MATLAB on 64 K, huh?

Dave Jones: Oh my gosh.

Chris Gammell: Or Python, to be honest, you know, like, yeah.

Dave Jones: Yeah. Well, yeah. Yeah. Python can chunk away. Both, both will chunk, chunk away. But, uh, yeah, to, to get your cycles down, uh, just so that you don't, uh, just so you have plenty of time for housekeeping. And so you're not spending a hundred percent of your time doing a DSP, you know, have to make sure you interact with, uh, the menu and that. Sure.

Chris Gammell: Yeah. You don't want your menu to take three seconds to respond or anything like that.

Dave Jones: Right. Exactly.

Chris Gammell: So are you doing like, do you do a real-time operating system to deal with that and like have priority or you just do interrupt based type things?

Dave Jones: No, I, I just, uh, it's, it's just interrupt based. Mm-hmm. And, and, and the, uh, the, uh, the, uh, the ADC and the, and the DAC, all that, that's what an, I, uh, I, uh, I2S setup. That's all, uh, that's all DMA based. Okay. So that, once again, to have, have it be DMA and have the, the FPU on the Cortex M4, just things that you just couldn't do on, on earlier microcontrollers. I mean, the fact that you can do this on a microcontroller at all is, is amazing. Mm-hmm. Uh, and, and what you can do is, is, is a lot of fun.

Chris Gammell: Yeah. I mean, so, so, so, some of this probably leads back to like the consumer level thing, right? So this is a product that currently is $489 on Amazon. Uh, so, you know, not cheap, but also, you know, not a ton of headroom in terms of a small batch assembly type of thing. Right. So I, I assume that the, the micro is kind of part of that, but I was going to ask like, why not upgrade the micro and get more than 64K?

Dave Jones: Well, that is, I plan to do that down the road. Got it. I mean, uh.

Chris Gammell: So time investment and time cost.

Dave Jones: Yeah. Well, it's, uh, the Cortex M7 is, is what I'm going to hit, is next.

Chris Gammell: Singing, singing your name.

Dave Jones: Oh, the Cortex M7, that, uh, you know, looking at, uh, the TNC 4.1, my gosh. Oh, yeah. Yeah. And, uh, does, does, uh, Adafruit have a M7 board out yet? I'm sure they do. Yeah.

Chris Gammell: They're, they're pretty on top of things.

Dave Jones: These, these, these Cortex M7s, uh, the amount of, I mean, what, 400, I mean, they're up around 500 megahertz, they have tons of RAM. Yeah. And they have double precision, which is something I want to do. You know, I'm curious how, what their cycles will be like on that double precision, you know, because you still gain a lot by doing things fast, but, uh, they say they support it. So I imagine that, uh, are they doing these calculations in single cycles? I don't know.

Chris Gammell: Right. I mean, I, when you do your processing on this stuff too, is it, so are you, what is, what is your unit? I mean, you're not using doubles, are you? I mean, like in turn in, in the actual code. So is it all fixed point math? No, no, no, no, no.

Dave Jones: It's all, it's, I'm, I'm, I'm doing everything in a single floats.

Chris Gammell: Okay. So it's all floats, but it's, yeah. Okay. So because, because that targets the FPU as well, you're saying?

Dave Jones: Yeah. Because, uh, the M4 has all these, uh, special DSP instructions that are single cycle. So, you know, when you're doing these single cycle, uh, floating point multiplies, I mean, what, what you can get done is, is a lot.

Chris Gammell: Yeah.

Dave Jones: And actually, uh, back when I, when I rewrote things a year ago, when I went and started to do things in assembly to get more cycles, I did do some, uh, some of the things I switched over to fixed point. And actually my education on that was Eli Hughes, you know, your guest from a couple of weeks ago. He's, uh, Eli has some great videos on YouTube, uh, documenting how to, how to do a fixed point, uh, DSP and these, uh, Q number formats. And, uh, if anybody's interested in that, uh, especially if you're not using an M4, if you're using one of the earlier M0 or whatever to do DSP, then, uh, check out those videos. Very informative. Yeah.

Chris Gammell: We'll link those in. That's great. That's great. I remember I used to do this as many, many, God, this is probably 2005 now, but we were doing like FPGA stuff and it was before they had a lot of floating point things internally there. And so we were doing DSP stuff inside an FPGA. It had, you know, custom math, Mac blocks and things like that. So you can make filters, but it was still floating point math. And that was part of the conversion stuff you had to go over. But I assume that things are moving towards that, like everything floating point, every, you know, like floating point units everywhere kind of thing. And so that, that is also helping to speed things up, make things cheaper. What, what are the main benefits of that kind of thing?

Dave Jones: I mean, floating point is just much more natural than fixed point in my brain. Uh, I guess if somebody has a RPN calculator, maybe fixed point is, is more of their style, but when I rewrote my code for, I mean, in particular, I rewrote this section of code that was dealing with, uh, the, the high pass filter, which was working down at, uh, 20, 20 Hertz and below. So this was, this was working, you know, real close down to DC and you really had to have super precision to get the absolute best performance. So I rewrote this in fixed point. And after I, I, I did all my tests, I was still a few cycles slower than, uh, than my original floating point code. And then when, and then when I test the distortion tests, they were, there was no change either. So, and also I ran into some errors that I never debugged because at that point I decided to scrap it there, I ran into some strange bugs. Um, actually it's all pretty hazy now and maybe I did get them figured out, but I think it was polarity.

Chris Gammell: This is the, uh, the Python, the Python C assembly, uh, MATLAB fugue state. Oh my gosh. That's what they call this.

Dave Jones: But, uh, but, uh, but polarity, you know, polarity is a lot easier to, uh, to never screw up, uh, with, uh, with, with floats versus, uh, fixed point. I mean, fixed points just not natural for me. And yeah, I had to write all kinds of conversion programs in, uh, in Python to get my, you know, pre-calculated coefficients. Yeah.

Chris Gammell: Yeah. Yeah. And I guess if you're, if you're like watching, if you're watching the, the coefficient or if you're watching all the output then too, you have to like either put it through a, a translator that puts it back to something you'd understand. Right. Or you have to try and like look at it natively and like understand the, what the floating point values represent in an actual. Yeah.

Dave Jones: When you're debugging your code, debugging your code with, uh, with a floating point is much, much easier than, uh, then, then, yeah, you're getting this fixed point data. I'm like, uh, what's going on here?

Chris Gammell: Yeah. And so when you look at the, the numbers and the code and stuff like that, is it, is it like, uh, SPLs or like DBs or what are you, what are the actual numbers representing, uh, like a single instant in time of a sample? G's. G's? Just, it's just, I mean. Or volts, I guess. Yeah. That's what I was. Yeah. Yeah.

Dave Jones: I mean, you're just looking at, uh, I mean. What does G stand for? G. G, that was a G E E Z. Like, like, like, like, what are you asking? Got it. Yeah. Yeah. If I just, uh, see a number in this string, uh, if, if it's in floating point, uh, uh, yeah, if it's in floating point, it's, uh, going between what plus one and minus one. I, I don't even, uh, since it's not. Yeah. Yeah. Yeah. And then of course, uh, with, uh, when it's regular, uh, uh, just numbers, 24 bit would be like eight point, you know, 8.3 million to plus or minus, you know, plus 8.3 million to minus 8.3 million. It's whatever, you know, 23, uh, 23 bits minus one is.

Chris Gammell: Got it. Okay. All right. So it's not like, it's not, uh, referenced any like hard, it's not an absolute value because it's unknown coming in, but you're saying you have this maximum range and then you can like amplify to try and hit like 70% of the maximum input range so that you get every sound in there.

Dave Jones: Like my clipping code, you know? Yeah. I was gonna ask about that too. Yeah. My clipping code would be like, Hey, is this ever more than a 24 bit number? Right. Or, or more than a 23 bit number, uh, sign 23, whatever sign 24 bit number. So yeah, it's, it's easy to just say, Hey, uh, if greater than eight, three, eight, eight, whatever, if, if it's greater than that, then, uh, yeah, you've, you know, you've clipped high or, or if it's minus that, you know, you've, you've clipped low and then you go ahead and process it to, to, to roll it off. Well, you just, you just make it that number actually. Or you just, do you just ever drop the packet or no? No, yeah. You just, you just do a hard clip. Uh, and so, so yeah, if you go beyond that number, if you go beyond 24 bits, you are limited at a signed 24 bit number.

Chris Gammell: Hmm. Okay. And then you also have like, like, uh, so then people that have listened to records before, they would probably hear the, the pops and the scratches and all that other stuff. And it's, you have a thing to try and get rid of that as well, right?

Dave Jones: Yep. That's, that was my big project for, for, well, actually last year and this year. And at the beginning of, uh, you know, back in, in March when things look like, oh man, this year is going to be a mess. I decided to go ahead and, you know, nobody was buying and everybody was bummed out. And I was like, Hey, this would be a great time for me to go ahead and finish this up. I intended to launch it in August, uh, just a couple of months ago. So, but I, I was like, Hey, if I go ahead and get it kind of working, working as good as I can, and then launching it as like a, a beta firmware for people to download, then they can go ahead and give me feedback and start using it and see if they like it. And, and if I totally lay an egg on it, you know, people are preoccupied with other more important things. It's not like I'm going to get too much egg on my face on this one, but yeah. So yeah, I've made this, uh, I'd spent a year working on something I call magic, which is a pop and pop and click suppressor attenuator. And, uh, it filters it out and, and, uh, the whole idea is you don't want to overly, you know, you don't want to overly process your music. And, uh, so it has to be very gentle how it does it. And, uh, people love it. I mean, it's, it's been a huge hit and, uh, and it's kind of one of those sticky features that you want in your product.

Chris Gammell: I'll never go anywhere else. Cause I need this one thing, right? I think actually I saw that in the Amazon reviews as well.

Dave Jones: I cannot listen to records without it now. So, uh, yeah, it's, uh, it's, it is, uh, it works great.

Chris Gammell: In that case, right. So a pop, I would imagine, you know, just thinking about, so I look at a lot of waveforms, mostly when I'm editing the amp hour or looking at other audio things and audacity. When I think about a pop, I would think like a high impulse event. And for, you know, a couple samples, when you really zoom in and, you know, however many samples it is. And then I think about trying to do that same thing, like DSP. Okay. I could probably come up with some logical tests that maybe start to operate on that sort of thing. But thinking about that in analog, that would be really, really tough to do because you need to have detection and reaction to it. And you don't want to make it actually impact other things by like, you know, you put an RC filters in front of something. Yeah. It rolls off everything. But like, if you do these high impulse events, you just end up taking out every, every, you know, T sound or every P sound, right? Every high frequency sound. So like, what does that then look like in, in the DSP realm? Is it like you're just testing and removing?

Dave Jones: Well, I do call it magic for a reason.

Chris Gammell: Oh, here's the secret sauce, folks. Sean is going to clam up. He's like, I can't, I can't divulge my secrets. You know, this is the, uh, this is the, uh, well, you're not part of the club.

Dave Jones: I mean, as far as like detection, I mean, there's a PDF out there from the early nineties that people can, can find, uh, just search, uh, record rest restoration or, or, or, or something like that. Uh, records impulse, something like that. It's from like 91, 92, a guy was going to use a PC to, uh, to, uh, look at these. Pops and clicks. And, and, uh, he, he covers how to find those events. Uh, it's, it's a pretty good document. The actual correction, uh, that method doesn't really work very well, but yeah, just looking for sharp edges. It's, it's pretty easy to, uh, uh, to use what he's got in that, that paper to find things, but then, then acting, you know, responding quickly and doing it all in just, you know, a few milliseconds with 64 K. Now that's, that's just as difficult as, as doing it tastefully. Yeah. But finding the impulse events, that's, that's not, that's not too difficult. I mean, but yeah, as you say, there's a lot of impulse events like high hats, pops.

Chris Gammell: Yeah. Pete. Well, you don't know the next thing that's coming, right? You see like, okay, you see this huge spike. I can imagine like you have like an acceleration factor, right? You say like, oh, well it went up, it jumped a whole lot more than I expected. I expect, you know, some certain range. It's going to jump way more than that. Okay. But you don't know the next packet. I guess you could look at, are you looking at multiple packets or multiple data points in order to, to characterize this sort of thing?

Dave Jones: Yeah. I mean, I spent a year monkeying around once again, back to Python. If I did not have Python to do this, if I was just a writing code and then taking it to my shop and testing it with, with records, it would have never happened. So, I mean, I had to, I had to sample, you know, tons of wave data that then I processed for months and months and months with Python. Python and then, uh, and then go back and see how aggressive it was. Uh, was it, uh, was it doing damage on things that it shouldn't? And then, uh, yeah, without fast iteration on Python, I wouldn't have been able to do it. And it still took a year of pulling my hair out and driving my wife crazy. It might, uh, when, when, when you just talk about something all the time. Yeah.

Chris Gammell: Now that's a good lead in too, because we talked about this a little bit before the show. You dropped your previous product to work just on this product. And so you have a philosophy, it seems like of single product, single focus on this. And it seems like it's paying off in this way. How did, how did that come about?

Dave Jones: Well, I mean, part of it goes to the whole, uh, I mean, all this theory on modern companies and products and focusing on not just living on, on what you've been making for years, but looking at what the next wave is going to be and then jump on that. And, and that's, that's difficult for big companies. And it was difficult for me too, because you have to, you know, go without money for a while while you're switching between products. But, uh, I mean, my previous product was, uh, the budgie and it was a tube phono preamp and it, uh, I had sold that for six years and I had had, you know, six years of emails, uh, with customers talking to people who used it and, and getting an idea of what they wished the budgie could do. So once you have a list of, Hey, what this product needs to be, and then, uh, being able to adjust the, the loading to that warmth on, on anything, and then having a wide range of, of gain because the budgie was just fixed gain because it was just the tubes, the tubes. Yeah. Right. They were, they can only do what, like 38 dB a gain. And then of course they couldn't do low, low noise amplification either. They're tubes. Right. Well, I mean, to a certain crowd, is there anything tubes can't do? Oh, I tell you, I, I get emails all the time saying, eh, my budgie's still working pretty good. Maybe I'll get a puff in one day, but I, I love my, I mean the budgie fans.

Chris Gammell: What you gotta do, Shannon, is you have to make a special cable to plug the puff in into the budgie and then it makes it better. People do that. And then you do both. You do both.

Dave Jones: People do that, Chris. But, uh, yeah, I mean, I, to do what I do, it being a, like a, a one person factory, everything has to be focused towards the product. Uh, so I have to, like with, with the budgie, I had these, uh, Italian real thingamajigs. They look like, uh, Oh, you would put your actual parts in it and it would cut all the leads off and bend the leads. Oh, wow. And then I would, uh, take all these buckets of bent parts and have my wife stuff them into these boards and then I would take these boards and then use a, a big solder, uh, bath and, and, and lay these boards on top of the solder bath to, to solder them all. And everything that I did was, was a process. And I, I just don't, I can't do two products at the same time.

Chris Gammell: That's right. You know, I, I have to call you out on this, that you did say it was a one person factory, but then you said your wife's, you know, stuff in the boards for you.

Dave Jones: So, you know, yeah, she, she was happy about the puffins design, uh, probably as much as anybody.

Chris Gammell: Right. So, so all surface mount, all, uh, outsourced assembly or are you doing the actual assembly now too?

Dave Jones: Uh, the surface mount stuff is done by Macrofab. Oh, nice. So they're Macrofab is, uh, I tip my hat to them. They, uh, are a fantastic company. I recommend them to anyone.

Chris Gammell: Yeah. And great podcast too. We, we love those guys.

Dave Jones: So, yep. You got it. Parker Dillman. I love seeing what he's doing with his, uh, Jeep or, or, or whatever on, uh, all right.

Chris Gammell: I will say I've heard enough about that goddamn Jeep. Look, can we talk about it? Buy another one, buy something, you know, he keeps buying them too. So that's the other problem.

Dave Jones: So can we talk about electric vehicles instead? Uh, yeah, that's true.

Chris Gammell: I guess, I guess the people in their cars, you know, Davis is a little crazy about that stuff now too. So yeah. Yeah. All right.

Dave Jones: So you, so you, yeah, everything's focused. Everything's focused on a product. And, uh, I mean, it really does save, you have to operate it at peak efficiency to be able to do this, um, to, to make a product. And, and I think a lot of people do have too many products and that's even a kind of a Steve Jobs, uh, type of thing is, is if you have too many products, you can't focus. And, uh, I think I can only do one.

Chris Gammell: Yeah. I, I, that's a really important thing to understand about yourself. I think, you know, like that it's, uh, I think one of it, one of the things is making sure you actually can make a living from it. Right. So like, sure. It seems like you were confident that you would, like you said, you had the six years of emails. That is some great product research right there. And you understood the process and it seems like this is a huge upgrade in terms of manufacturing and, you know, just iteration and all this other stuff. So yeah, it's really, it's a great, great transformation, I think.

Dave Jones: Well, I mean, and at the, at the very beginning of it, I realized, Hey, I can monkey around with DSP the rest of my life. I mean, I don't, I want to move away from hardware. I mean, I, I don't want, I mean, I still want to produce hardware. Right. But then I want to, uh, you know, hopefully I won't have to change the puff and hardware wise for a decade.

Chris Gammell: Until you put the M7 in, right?

Dave Jones: Well, yeah. And, and I do have a socket in there so, so that people with the current puff and hopefully I'll have a little module with the M7 so they can just plug in a new one.

Chris Gammell: Nice. Yep.

Dave Jones: But, uh, yeah, to be able to, yeah, you, you got me on that one. Hardware wise, I am going to be monkeying around with an M7 in the near future.

Chris Gammell: Well, I mean, you, you spoke about it like it was the Holy grail a couple of, couple, you know, 20 minutes ago. So like, I'm not, it's not a surprise. It is. It is.

Dave Jones: I mean, I think if any, uh, the Cortex M7 for, for a small business, for makers, for anybody, Cortex M7 is just, uh, unbelievable. Yeah.

Chris Gammell: I mean, if you, I think it's, it really comes down to like power really. And you have tons of power. So great. You know, like that's tons of resources.

Dave Jones: But, but I, I mean, I'd be happy just to, other than when I'm making puffins and shipping off boxes of puffins to Amazon, I'd be happy to just sit on, sit on my couch and, and monkey around with, uh, with Python. Yeah. And, and, and, and do new, uh, DSP ideas, improve magic. Magic is just, uh, it's still kind of rudimentary in my opinion. I think, I mean, what I've, the feedback I've got is, is, is amazing and awesome. And, and, and it made me feel that's been a real bright spot in this year of 2020 for me.

Chris Gammell: Mm-hmm.

Dave Jones: But, uh, I could, I could monkey around with magic for years just doing small improvements.

Chris Gammell: Yeah. I bet. I feel like almost the, you know, like the, so the Amazon reviewer says, oh, I don't want to put DSP at the front of my signal chain. And like, I think the argument against that every time is, yeah, but you get upgrades, you know, good luck upgrading your tube amp without a solder iron, you know, like, it's just like not going to happen. And so.

Dave Jones: Well. Yeah. Actually that's tube amp guys can swap out the tubes. That's, that's kind of, that's kind of the big fun thing.

Chris Gammell: Sure. And, and they, and they do, they do love it, but, but that's not really upgrading. That's more side grading, you know, like.

Dave Jones: I think all the people that are kind of hesitant, it's before they've actually used, uh, what I make before they use the puff and before they hear it in their own system. And then once, once they get to use it, uh, they're a believer. But yeah, I mean, could I make a, could I make a crappy product that, that does sound bad? Sure. It's, it's, it's easy to, it's easy to do that too. So, uh, getting this. And that's another reason why Amazon is so successful because people, people can easily get a product and they can easily return a product. So that's, that's another part of their, their success is the ease of return. Yeah.

Chris Gammell: Yeah. And so tell us a little bit about that because we talked about that before the show as well of like, you know, kind of getting into the FBA, the fulfilled by Amazon thing. You don't have to deal with logistics as a one man shop, but like that has other downsides like returns, you know?

Dave Jones: Yeah. Uh, Amazon, I mean, it's, I, I, I tell my wife that, uh, Amazon's kind of like a bad marriage. I mean, one you can't get out of, I think of, uh, Steve Ballmer back, uh, in like revenge, not revenge of the nerds, but nerds 2.0, uh, a documentary back in the nineties. Steve Ballmer talked about when, uh, Microsoft start first started working with, with IBM and they knew they had to work with IBM cause they were the only show in town and he, they referred, Microsoft referred to, uh, IBM as riding the bear. And the whole idea is, the whole idea is get on the back of the bear and hold on for dear life. That's right.

Chris Gammell: Because it doesn't turn around and eat you. Yeah.

Dave Jones: Yeah. And if you fall off, you're toast. And, and Amazon is kind of like that. Uh, Amazon, you got to ride the bear and, uh, uh, definitely, you know, inner trying to converse with them. They don't have a phone support anymore for Amazon FBA. So you know, you could have a hundred thousand dollars for the product missing someplace and wouldn't have a phone number to talk to somebody. Wow. Yeah. But you got to, that's what people use to buy things.

Chris Gammell: And, uh, yeah, exactly. Right. It is the marketplace. It is the place to go. And if you, if it changes in the future, you'll change too. Right. I mean, like that's, you go where people are, unfortunately.

Dave Jones: Yeah. And I mean, and the Amazon review system, that's, that's what people use. And it's, it's what I use. Right. I mean, it's, uh, if, and if something isn't, I mean, if there's two products on Amazon, one of them's seller fulfilled and one of them's prime, I'm going to go for prime. Uh, and I'm going to go with the one with the good reviews and it's just, it's just what people are doing nowadays. Uh, and I don't blame them.

Chris Gammell: So what's interesting about it is I'm, so I'm on the page here and I'm looking at the, you know, it says the compare with similar items and it's got three comparisons with you. Two of them are from pile, which I assume is made in China knockoff type thing. Cause it's one cost $14. Another cost $18. And then yours is sitting here at $490. And it's like, those are not comparisons, but it is what is Amazon showing. Right.

Dave Jones: And so, and I'm, but, uh, go ahead.

Chris Gammell: No, no, no. I mean, that, that's just, I guess that's it. It is a tough comparison, but it's, that's what the marketplace, that's what the algorithm shows. Right. And it's like, Oh, okay.

Dave Jones: Yeah. I mean, uh, I'm happy to be, uh, you know, a lot of people do go from a $20 phono preamp, uh, to, uh, to the puffin and, uh, yep. I mean, it's a phono D I named it a phono DSP. And, and when I first named it that, uh, my engineering friend was like, Oh man, that's kind of DSPs. He's a hi-fi guy too. But I was like, but this is what it is. And it also, you know, it, it differentiates itself.

Chris Gammell: Differentiate. Yeah, totally. It is.

Dave Jones: It is. It's the only phono DSP on Amazon. So, well, I don't know. There, I can't say that a hundred percent. There's a lot of things on Amazon, but, uh, yeah, it's, it's, it's a good marketplace. Yeah. If I didn't have any reviews, uh, then yeah, I don't think it would be that successful. So, so have still selling from your website.

Chris Gammell: Yeah. Did you sell to your previous customers? Like how did you convince people to get started with it? Cause obviously you need to have it. Reviews are, you know, proxies for word of mouth. Right. I think actually like the audio folks seem to like they have a lot of forum posts too, but how do you get people to start talking about this sort of thing?

Dave Jones: Forums are everything. Well, I mean, I, I, I, when I changed over, I kind of changed over right at the apex of, of my, of my budgie preamp. Mm-hmm . Got it. So I was, uh, getting in, in a lot of my interaction with, uh, my customers come while the main. Interaction other than emails is, uh, my Facebook, uh, parks audio account.

Chris Gammell: Mm-hmm .

Dave Jones: And, uh, so, you know, announced it there gets into the audio forums, you know, you do your kind of limited launch to get it in people's hands. And then it just went from there, but being able to sell from your website along with, uh, you know, Amazon, that's, that's key. And just having as much interaction with, with folks and, and having people that believe in, you know, I did, I have been doing this since the early two thousands. So having that bit of, uh, these fans and people that have used your products that once you get that going, it's something to nurture and we'll take care of you.

Chris Gammell: Yeah. Yeah. Kevin Kelly's a thousand true fans is a, that, that document has held up over the years, you know, talking about like how a thousand true fans will support basically, you know, small business, you know, and it sounds like you have a lot more than that. So that's, that's a good sign.

Dave Jones: I, I think thousand true fans, I think that's a good, uh, that's a good number. I think cause, uh, you want, I mean, I don't get on the audio forums and, uh, and like look for my net, you know, look for Shannon parks or look for parks audio stuff. Cause that would drive me crazy. That would, yeah.

Chris Gammell: You'd get a lot of, you'd get a lot of, uh, unintended glancing blows, you know?

Dave Jones: Yeah. And, and, and then, and then what do you do? Do you hop on there and, and, uh, talk back to these people and look like a complete fool?

Chris Gammell: I think, yeah, acting really defensive and, uh, swearing is, is the right way to do it on the internet, Shannon.

Dave Jones: Well, that's a lot of people that, that, you know, a lot of people do that. Yes.

Chris Gammell: As people can hear here, you sound very high strung and, uh, you know, you sound very chill and, and you're doing things right. So I'm drinking my green tea.

Dave Jones: Yeah.

Chris Gammell: Right.

Dave Jones: But, uh, yeah, yeah, definitely interacting with folks and not being a jerk and taking people's emails, uh, seriously and appreciating them. I mean, that's just appreciating your customer and trying to meet their problems. I mean, that's, yeah, that's, that's the key.

Chris Gammell: Yeah. And, and I think having, uh, having a focus on that too, right? I mean, like even the budgie, which was the previous product was in the phono market as well, right? So you knew you are serving a very specific audience and you're serving a very specific need and you're focused on it versus another company that might have 40 audio products that are trying to do all these different things. So there is that too.

Dave Jones: Right. Yeah. I mean, and I get emails from people that, you know, aren't going to buy any of my stuff, but just have a problem. But they heard that I actually respond to emails. I mean, and I'm, I'm happy to do that because I learn about problems people are having and, uh, that's, that's really insightful. And then, Hey, and you get good credit out there. I mean, uh, good karma. That's, that's, that's good. That's important.

Chris Gammell: Yeah, totally. Where can people, uh, find out more about you and see the product, maybe see the inside of the product.

Dave Jones: If there's any photos of that, uh, inside of the product, probably, I mean, there's actually some, a lot of photos, even on the Amazon reviews where people are showing them updating the firmware and, and things like that. But they can go to my website, which is parksaudiollc.com. They can go to, uh, parks audio on Facebook. They can just, uh, Google me. Oh, I guess I have a couple of videos on YouTube.

Chris Gammell: Yeah. You did a Q and a recently-ish or it's on your website. At least that was a nice, answering people's questions.

Dave Jones: That was geez. I think that's two years old now. I, I just purchased, uh, some new gear to do a second video, which, uh, uh, I haven't done yet. Uh, what's interesting in that video I did two years ago, I talked about possibly in the future doing, uh, SP diff digital outputs and, uh, doing, doing the magic type things. And I just, I wasn't sure I was going to be able to do any of that, but it's interesting looking back that I was able to talk about that even then. Um, yeah, Facebook and YouTube and, uh, and my website and just Googling me and seeing what people are saying in forums here and there, stevehoffmantv.forums. Things like that.

Chris Gammell: What is, what is that? I saw that one pop up, but I didn't know what it was.

Dave Jones: Yeah, the Steve Hoffman forums. That's real, uh, that's some real hardcore vinyl guys. Uh, and, uh, and I, I got some fans there and I really appreciate, uh, really appreciate them.

Chris Gammell: Hmm. Okay. That's great. Well, Shannon, thanks so much for being on the show and telling us about this stuff. I mean, I think it's, uh, I'm sure I didn't ask as many, you know, audio foolery questions as Dave would have wanted me to ask, but, uh, I still, I think it's an interesting thing because it is, it's this, you know, very human, you know, music is such a like visceral human experience that like people have very, very strong feelings about. And you're obviously dedicated to improving it and, you know, listening to your customers and working on interesting stuff. And at the end of the day, then you do interesting DSP things with it too, which is, you know, hits the nerd crowd here. So that's great.

Dave Jones: Hey, being able to work in music, uh, I just love it. I think it's a, always a net good in people's lives. And, uh, to be able to do something like that, uh, it's just fantastic. I feel really lucky. That's great. All right. Well, thanks a lot. We'll talk to you soon.

Chris Gammell: Thanks so much, Chris. Thanks to Suzanne for editing.

Shannon Parks: Our theme music is Chlor from the Amps by Paul Stevenson. Join our other Amphour patron sponsors on patreon.com slash the Amphour and chat directly with Dave, Chris, and the rest of the patrons on Discord. Special thanks today to our corporate sponsor, Binho. That's B-I-N-H-O and Binho is their name-o. We'll be right back.