Indeed nice work, Richard!
Hmmm... may I ask a silly question?
Now that you have developed / improved nearly everything around the DAC chip; most work being compatible -or easily transposable- to other DAC chips... is there a particular reason why all your DACs are buillt around TDA1387s?
Is is because it is non expensive, or sounding good to your ears... or is there anything particular with this chip?
I ask because you have mentioned a couple of times other DAC chips in your threads and when looking at their spec sheet (which doesn't say how they sound admitely), they all seem good or better performer. And perhaps make sense if chips become the bottleneck and you have to stack many to even this out
Just asking for my curiosity, not detracting of course, as I wasn't there when all this fascinating work started.
Enjoy music
Claude
Hmmm... may I ask a silly question?
Now that you have developed / improved nearly everything around the DAC chip; most work being compatible -or easily transposable- to other DAC chips... is there a particular reason why all your DACs are buillt around TDA1387s?
Is is because it is non expensive, or sounding good to your ears... or is there anything particular with this chip?
I ask because you have mentioned a couple of times other DAC chips in your threads and when looking at their spec sheet (which doesn't say how they sound admitely), they all seem good or better performer. And perhaps make sense if chips become the bottleneck and you have to stack many to even this out
Just asking for my curiosity, not detracting of course, as I wasn't there when all this fascinating work started.
Enjoy music
Claude
Is there a reason for all the designs to use TDA1387? There are several, and I went into some detail on this on my Hackaday project. I won't link again to that here as I've linked it several times already on DIYA. The TL;DR version of that is - its cheap, its widely available, its simple to implement (single PSU), low power, no bells and whistles included, has a convincing technical story (continuous calibration) and reasonable performance for a 16bit part. Added to that nobody else in the DIY space seemed to be using it so little danger of reinventing the wheel and - yes it sounds uncoloured to my ears.
When I started out, I took onboard the advice of seasoned DAC designers that mostly agreed that the SQ of a DAC is 90% in the power supplies and analog support circuits and only 10% in the chip. So I figured the chip itself didn't matter that much so why not choose a cheap one? That wisdom has held up fairly well over the years. Now though, as a result of experiments over what seems like a very long time, I have figured out the particular limitations of the 1387 I'm ready to select alternative DACs which exhibit less of those. Stacking is the way to get around the noise issue which is the major limitation of the chip.
When I started out, I took onboard the advice of seasoned DAC designers that mostly agreed that the SQ of a DAC is 90% in the power supplies and analog support circuits and only 10% in the chip. So I figured the chip itself didn't matter that much so why not choose a cheap one? That wisdom has held up fairly well over the years. Now though, as a result of experiments over what seems like a very long time, I have figured out the particular limitations of the 1387 I'm ready to select alternative DACs which exhibit less of those. Stacking is the way to get around the noise issue which is the major limitation of the chip.
I've been unusually quiet on this thread for a rather long time. Partly that's because the Chinese wave of 'the virus' struck me down just before Xmas. Partly though its because I've used the time to reflect on whether my latest 'Stack DAC' design really makes sense as a project. One new datapoint injected into the mix over this period was a comparison of a stack of six of my old 'DenaDAC' boards (each has 36 TDA1387s) against the new PCM56 design.
192 paralleled DACs is a bit beyond where I was intending Stack DAC to go but I still preferred the sound of the single PCM56 overall. My conclusion from this is - I think I've exhausted the potential of TDA1387 and while its an excellent choice for low through to upper-mid level DACs, the very highest echelons are not suited to it, not even massively paralleled. I now don't think going through the work-up of building a stack of 5 boards is worth the effort when better results can be achieved with somewhat less effort when building on the foundation of PCM56.
Rather than ditch 'The Stack' entirely what I'm currently planning to do is reposition and re-package the design to more properly target mid level DAC SQ, and do so with less 'hair shirt' (e.g. the fan, two power supplies) compared to the original design.
Thoughts / comments / disappointments / questions? I'm interested if anyone has any feedback.
192 paralleled DACs is a bit beyond where I was intending Stack DAC to go but I still preferred the sound of the single PCM56 overall. My conclusion from this is - I think I've exhausted the potential of TDA1387 and while its an excellent choice for low through to upper-mid level DACs, the very highest echelons are not suited to it, not even massively paralleled. I now don't think going through the work-up of building a stack of 5 boards is worth the effort when better results can be achieved with somewhat less effort when building on the foundation of PCM56.
Rather than ditch 'The Stack' entirely what I'm currently planning to do is reposition and re-package the design to more properly target mid level DAC SQ, and do so with less 'hair shirt' (e.g. the fan, two power supplies) compared to the original design.
Thoughts / comments / disappointments / questions? I'm interested if anyone has any feedback.
That sounds like it makes sense.
So you would have a range of qualities of DAC which is reflected in the price right?
I think it comes down to how much work you are willing to put in for our benefit really 🙂
What are you currently offering again? Would the effort on this make it fit into the range or would it be more fun just working on the PCM56?
With prices?
Dorati (replacement for Kubelik, right?) 168RMB
LingDAC - this thread. What sort of RMB do you reckon it would be?
Celibidache (upgrade of DecaDac) 780RMB
Where do you reckon the PCM56 would sit in SQ and price?
Actually, while we are at it, what else do you do?
Regulator for Dorati ?RMB
Your amp 520RMB
Shunt regs for the amp 100RMB
What else?
Might it be worth putting a list of things you are offering at the bottom of the first post of a new item.
With links to their threads?
So you would have a range of qualities of DAC which is reflected in the price right?
I think it comes down to how much work you are willing to put in for our benefit really 🙂
What are you currently offering again? Would the effort on this make it fit into the range or would it be more fun just working on the PCM56?
With prices?
Dorati (replacement for Kubelik, right?) 168RMB
LingDAC - this thread. What sort of RMB do you reckon it would be?
Celibidache (upgrade of DecaDac) 780RMB
Where do you reckon the PCM56 would sit in SQ and price?
Actually, while we are at it, what else do you do?
Regulator for Dorati ?RMB
Your amp 520RMB
Shunt regs for the amp 100RMB
What else?
Might it be worth putting a list of things you are offering at the bottom of the first post of a new item.
With links to their threads?
Hi Richard, sounds like a case of diminishing returns with the 1387 chip. You are up against limitations of the chip design and need greater and greater complexity to get a small gain. I think that your on the right track considering moving to another chip. Miro's ad1862 design is a case in point, fairly simple design, easy to build, with great SQ due to the strength of the chip design.
Makes completely sense to me!
And yes, taking a breath to look at the entire range, assessing sound vs price... perhaps some iterations are redundant or too close etc., especialy as you improved a lot of things recently...
With the stacking experiment, you were creating "an admittely strong monster-building... that could barely walk"...
Hey, let's face it, it was getting so complex that even you had problems counting the numerous DAC chips LOL! (eg. I presume your Dena was rather 32 than 36 🙂
At the end I see your expriment, as all others, with a lot of positiveness: you found a complex path that reveals being very complicated and less rewarding at the end than a simplier solutions, so PCM56 is the way forward IMHO
Now the question is: is there a reason to move the entire line to this chip (depending on the oncost) or just the upper range?
That's a difficult one to answer, as chip oncost might not be the only oncost to make it run. I can also see that the market is agressive, especialy as I consider that DIY should enable to access a sonic quality (at the cost of effort, risk, looks, flexibility vs commercial products) of say 1 to 10 (eg 100E parts should be on a par from a pure sound POV to a 1000E commercial product. OK, it is always a fun DIY project to learn from building a DAC, spend some time chatting with nice folks etc., so can be a bit lower ratio, but not much more as you lose a lot of functions and ready product features... while though being able to tune the sound.
I believe in a kind of bipolar market. For me there would be a non expensive but qualitative DAC to enable to get a proper taste of R2R and in that regard I consider the latest DecaDAC say below 100E with its PS (no interface and no casing) to be the right price range. Both because it is (for me!) OK money to spend for fun/ a try, while also nowadays some commercial (chinese) DACs at 800E can offer all the sound I need. For a upper range model I would defo not exceed 200E parts for a complete working unit, but that should really being able to compete with the best VFM sub 2000E DACs, so be giant killer somewhere. Finaly, an other interesting price point is the simple DAC at 20E, as that competes with low cost Amazon offers and if sounding reasonably good, why not.
I don't know if there is more interest for the upper range or the lower range, nor if something around PCM56 replacing the DecaDAC wouldn't just fill all bills being in the right middle... you to know better based on number of kits you sold in the past in the various price ranges...
Last but not least, all these comments because you asked for our "thoughts", at the end whatever I applaud your effort to try to offer something different, qualitative with great support, at an affordable price!
Long live your DACs, whatever their evolutions, as that's more what I see than a complete change of directions eventhough a new chip is inside
Claude
And yes, taking a breath to look at the entire range, assessing sound vs price... perhaps some iterations are redundant or too close etc., especialy as you improved a lot of things recently...
With the stacking experiment, you were creating "an admittely strong monster-building... that could barely walk"...
Hey, let's face it, it was getting so complex that even you had problems counting the numerous DAC chips LOL! (eg. I presume your Dena was rather 32 than 36 🙂
At the end I see your expriment, as all others, with a lot of positiveness: you found a complex path that reveals being very complicated and less rewarding at the end than a simplier solutions, so PCM56 is the way forward IMHO
Now the question is: is there a reason to move the entire line to this chip (depending on the oncost) or just the upper range?
That's a difficult one to answer, as chip oncost might not be the only oncost to make it run. I can also see that the market is agressive, especialy as I consider that DIY should enable to access a sonic quality (at the cost of effort, risk, looks, flexibility vs commercial products) of say 1 to 10 (eg 100E parts should be on a par from a pure sound POV to a 1000E commercial product. OK, it is always a fun DIY project to learn from building a DAC, spend some time chatting with nice folks etc., so can be a bit lower ratio, but not much more as you lose a lot of functions and ready product features... while though being able to tune the sound.
I believe in a kind of bipolar market. For me there would be a non expensive but qualitative DAC to enable to get a proper taste of R2R and in that regard I consider the latest DecaDAC say below 100E with its PS (no interface and no casing) to be the right price range. Both because it is (for me!) OK money to spend for fun/ a try, while also nowadays some commercial (chinese) DACs at 800E can offer all the sound I need. For a upper range model I would defo not exceed 200E parts for a complete working unit, but that should really being able to compete with the best VFM sub 2000E DACs, so be giant killer somewhere. Finaly, an other interesting price point is the simple DAC at 20E, as that competes with low cost Amazon offers and if sounding reasonably good, why not.
I don't know if there is more interest for the upper range or the lower range, nor if something around PCM56 replacing the DecaDAC wouldn't just fill all bills being in the right middle... you to know better based on number of kits you sold in the past in the various price ranges...
Last but not least, all these comments because you asked for our "thoughts", at the end whatever I applaud your effort to try to offer something different, qualitative with great support, at an affordable price!
Long live your DACs, whatever their evolutions, as that's more what I see than a complete change of directions eventhough a new chip is inside
Claude
Last edited:
Agreed it makes total sense, I have wondered the long windy threads on your dac projects and thoroughly enjoyed the insights, trials & tribulations and product decisions you have made, with reasoning behind them, which is quite rare, it has helped me and I am sure many others understand more about how dacs work and what makes them work well.
If there are better low cost NOS chips out there, please do chase them! Seeing what you have achieved with the basic TDA1387, I am excited at what will come with the PCM chip range.
Huge respect and admiration as ever Richard, put me down for a PCM56!
Jules
If there are better low cost NOS chips out there, please do chase them! Seeing what you have achieved with the basic TDA1387, I am excited at what will come with the PCM chip range.
Huge respect and admiration as ever Richard, put me down for a PCM56!
Jules
Richard, I think the comments that I see so far from Grarea, PJN, ClaudeG, and Shertzy are very good. I only want to add that I think most importantly, whatever you do, it should be enjoyable and rewarding for you and your wife (very important that your wife stays happy), and you should be profitable enough that you can make a living. It is no fun if you cannot pay your bills. You seem to be extremely adept at researching and finding excellent solutions that are very cost-effective. Expensive solutions are easy to find, but not always a reasonable solution. For example, how many of us have spent a lot of money on a component that had a very good review, just to find out we do not really like it so much in our system? I would have wished there were a low-cost way to dabble. Having an entry level solution is helpful to many people. Maybe they can add to that board or go a different route and not be out much money. When writing a technical paper, you consider your audience, an engineer, a newbie, or an artist. Each would need something different and you cannot please everyone with one solution. However, few people have the means to build every product they would like to build. So you will pick and choose.
And there is merit to an elegant solution (that term is not always well defined). You already know that I prefer elegant and simple over brute force, but sometimes other factors take precedence over elegant.
Oh, I am especially eager to hear your development with a PCxxx or a miro AD1862 board and one of your new I/V boards with a filter.
So whichever way you decide to go, I (and many others, I am sure) appreciate your explanations, products, and other contributions. I thank you for that and look forward to trying out some more of your offerings.
And there is merit to an elegant solution (that term is not always well defined). You already know that I prefer elegant and simple over brute force, but sometimes other factors take precedence over elegant.
Oh, I am especially eager to hear your development with a PCxxx or a miro AD1862 board and one of your new I/V boards with a filter.
So whichever way you decide to go, I (and many others, I am sure) appreciate your explanations, products, and other contributions. I thank you for that and look forward to trying out some more of your offerings.
Hi Richard
What has been written so far from these good men above about your presence here and your stance as a true experimenter and developer, reflect my thoughts. Kudos to you and your wife.
Now what I would like to see as a future DAC from you.
Not a huge number of chips and stacks of boards.
I would like to see another chip-not a delta sigma one- on a flexible format.
One version as one board that would have everything that your previous boards have DAC-filter-I/V-buffer and the proper PSU section.
Another DAC board version with pick-up points for connecting external PSUs optimized for each section (par the DAC’s, which has to be onboard really close to the chip) and/or different boards for the I/V, buffer with a choise of designs.
The experimental board with the opamp-less Abbado has given me ground to support the “less is more” approach.
You already provide cheap DACs and you can’t beat ebay simple board modules or Topping all-inclusive units.
I would be OK with your decent pricing for a good basic DAC that has the provision for changes/upgrades, so it will allow me to keep the ball rolling with my soldering iron while learning.
A healthy New Chinese Year and a productive one.
George
What has been written so far from these good men above about your presence here and your stance as a true experimenter and developer, reflect my thoughts. Kudos to you and your wife.
Now what I would like to see as a future DAC from you.
Not a huge number of chips and stacks of boards.
I would like to see another chip-not a delta sigma one- on a flexible format.
One version as one board that would have everything that your previous boards have DAC-filter-I/V-buffer and the proper PSU section.
Another DAC board version with pick-up points for connecting external PSUs optimized for each section (par the DAC’s, which has to be onboard really close to the chip) and/or different boards for the I/V, buffer with a choise of designs.
The experimental board with the opamp-less Abbado has given me ground to support the “less is more” approach.
You already provide cheap DACs and you can’t beat ebay simple board modules or Topping all-inclusive units.
I would be OK with your decent pricing for a good basic DAC that has the provision for changes/upgrades, so it will allow me to keep the ball rolling with my soldering iron while learning.
A healthy New Chinese Year and a productive one.
George
I would choose to use a chip that is still in production and capable of a balanced sound. Attention, I did not say neutral sound. A neutral sound can be bad.
A modern available DAC chip could indeed enable future developments without worrying about supply and while possibly offering better performance... if produced in high numbers to be readibly available.
Does such part exist at a competitive price?
Does such part exist at a competitive price?
AD1860R-J is 3 times cheaper than 1862-N . It can be purchased new from reliable sites.
It must be tested if 2 or 3 such chips in parallel can be heard as well or better than a single Ad1862-N.
I have seen a lot of tests that are done in stereo mode. From my point of view it is a mistake. All tests must be done in mono mode.
The stereo sound deceives the hearing.
But if something sounds good in mono mode, it will sound even better and balanced in stereo mode. Nothing like natural sound or sound delivered by an instrument is stereo , but mono. When I buy an amplifier, I try it in mono mode, if I don't like how it sounds in this mode, then I don't buy it.
It must be tested if 2 or 3 such chips in parallel can be heard as well or better than a single Ad1862-N.
I have seen a lot of tests that are done in stereo mode. From my point of view it is a mistake. All tests must be done in mono mode.
The stereo sound deceives the hearing.
But if something sounds good in mono mode, it will sound even better and balanced in stereo mode. Nothing like natural sound or sound delivered by an instrument is stereo , but mono. When I buy an amplifier, I try it in mono mode, if I don't like how it sounds in this mode, then I don't buy it.
Last edited:
Difficult to hear 3D soundstage out of mono only, for me soundstage size in all dimensions and accurate placement of instrument is also important, but then that's just me...
And yes, I do use the mono function yet sometimes, like when placing LS in a room etc.
And yes, I do use the mono function yet sometimes, like when placing LS in a room etc.
If a design made of a maximum of 3 chips per channel does not sound good, I don't think it's worth the effort to make a design with more chips per channel. Simply switch to another chip that meets these conditions. In this way, you will get time to think about other ways of implementation (other approaches, maybe innovative) of that chip. For example working of that chip with other programmable chips like FPGA.
That's the general idea yes. The complication is I'm always trying to improve on designs, so newer designs incorporate newer knowledge to get better sound.
I am hoping (and aiming for) its going to be one step above Celibidache. I have the third generation prototype on my desk now and listening to it but so far no direct comparison with Celibidache. We'll do that over the next few days hopefully.
The repackaged Stack DAC is planned to slot in the gap between the Dorati and Celibidache with the potential of moving past Celi if multiple boards are used. The first extra board ideally gets used with a splitter to run balanced, additional boards after that may (to be verified in experiments) be paralleled for even lower noise.
The TDA8932 amp yes and various PSUs to support the DACs. And just this week, a standalone I/V stage to use with existing DACs.
I'm not sure if this idea violates the forum policy of keeping commercial stuff in the 'Vendor's Bazaar' area. We (well wifey really) have started a website to catalogue the offerings but with new stuff coming along it needs updating for 2023 (or Year of the Rabbit).
Hi Claude - actually although my memory is definitely on the decline, its not gotten so bad as to get technical details wrong yet! You can see a 'Dena DAC' board here : https://www.diyaudio.com/community/...l-multibit-dac-adventures.343834/post-5944711
I'm of the view its just the upper range that needs to make the transition to different DAC chips.
Yes, that seems quite reasonable. We should be able to do better in SQ for less money (but more time) invested over commercial gear. But also not pay for features we don't really need, to me a key strength of DIY is having flexibility (granularity of features) to match the design to the need.
DecaDAC is going back over two years now, I would definitely hope the PCM56 (Marriner) would trounce it, especially as its at a higher price point.
Thanks to you (all) for chipping in in response to my request. Appreciated!
Couldn't agree more. The whole point of building DACs for me is its fun, more fun than other jobs that I could engage in. And definitely keeping wifey happy is a high priority too. Making it into a living takes time in building networks of customers, its much slower when we don't actively advertise what we're doing (other than posting here and occasionally on Hackaday which don't seem to me like advertising).
That is indeed a significant challenge to overcome. Hopefully the DIY route mitigates that in various ways - first its not dependent on professional reviewers. Reviewers that I've read seem to mostly shy away from direct comparisons between components for VFM preferring woolly phrases like 'competes with DACs at twice the price' and 'one of the best DACs I've ever heard' / 'amongst the best digital I've had in my system'. The other aspect is personal taste, whether it aligns with the reviewers' or not. Not everyone's looking for the same kind of sound I'm after which is why my designs definitely don't suit everyone.
Thanks for the encouragement Dave. I'm hoping you'll get enough free time in between new job commitments to get the I/V board up and running by this weekend, keen to hear your feedback!
Thank you George.
The huge numbers of chips are a thing of the past now, but I felt curious to know how far a particular chip could go. Seeing as I know the answer now with the result from 192 chips, my curiosity is assuaged.
That sounds like it would be 'Marriner' - attached a pic of the board (3rd generation) being assembled.
Wifey's been working away on a couple of PCBs - there's a very simple and cheap TDA1387 board mainly useful to test out the I/V stage and serve as a reference point in SQ upon which to build and then there's a PCM58 board coming along which has higher SQ aspirations. Both these boards are Iout only (no filter, no buffer).
That's the kind of experimentation it would be good to see more of. Especially the sharing of the output waveforms and getting Claude's comments back, I'd not seen that kind of interaction on a forum before around a DAC. Kudos for stepping out into something novel.
The learning aspect for me was the primary motivation behind Stack DAC so its great to see others getting into the same spirit.
