What ref is yours? Mine are rev4 (or possibly rev5), and according to Soeren the transistor mod is no longer necessary for the latter revisions.
Reference DAC Module - Discrete R-2R Sign Magnitude 24 bit 384 Khz
I hope he releases a new FW where there is a switch between movie and ludicrousmode.
//
I think currently the clock is updated at a rate of 400 Hz. Maybe this could be customized to define a threshold of abberation to trigger an adjust, or make it possible to lower the rate to an arbitrarily low number.
According to Soren no mods were ever needed.
The addition of transistors lowers the output impedance, which makes it easier to obtain low wide band impedance while adding less capacitance.
Many here ended up supplying Vref from batteries which require even less additional capacitance.
I had a look at what he's done and at the design of the 1941 in general.
It seems to follow the design guidelines I was referring to: Bigger electrolytics closest to the regulator and furthest from the parts supplied with the power.
Success with decoupling has a lot to do with combining caps that result in an even response, especially no impedance spikes. We might inadvertantly create them with the wrong selection of caps here.
I just tried Nichicon UKA, a cap designated for professional audio, 220uf 16V instead of the four 1.800uf Panasonic FR caps. ESR of the former measures almost an order of magnitude higher than the latter.
It sounded horrible. Abrasive high end, compared to either no cap or the Panasonic.
It would be great if someone with theoretical knowdledge and measuring equipment could analyze the situation. There surely is an optimized version.
The transformer mod looks really messy on a V4/5 board. I'd rather not do this with 17 DAM1021 boards...
The wishful thinking of the young
Sadly, theoretical simplifications just do not translate into better sound.
Empiricism still rules in higher end audio.
Do you allow the caps you try out a sufficient break in period? How many capacitor types have you listened to extensively?
Are you using the active buffer stage on the board?
As for subjective dynamics, there are topology limits. Listen to a 1121/1941 for a comparison. Or to a Totaldac if you enjoy feeling depressed.
The wishful thinking of the young
Sadly, theoretical simplifications just do not translate into better sound.
Empiricism still rules in higher end audio.
Do you allow the caps you try out a sufficient break in period? How many capacitor types have you listened to extensively?
Are you using the active buffer stage on the board?
As for subjective dynamics, there are topology limits. Listen to a 1121/1941 for a comparison. Or to a Totaldac if you enjoy feeling depressed.
I wish I were that young.
I practice empiricism all the time, a lot more than you might think. In the studio the difference is not only in listening - every change you make influences an outcome. It's an artistic decision making process as much as it is a technical process.
But with a converter all I want is transparency. I have plenty of gear to add a "sound".
I'm sure there are differences to be measured and technical details to be optimized. What is important psychoacoustically just might not be what we are measuring. But with decoupling there tends to be a technically correct way that works best. I have been there and done that, albeit only with purely analog circuitry.
As for the output stage, I am using one I build myself. It is based on the most transparent sounding op amp I know, a certain brand 5532 from a certain batch (unfortunately, they all sound different). To my ears it is a massive improvement over the DAM1021s build in output stage and it fares better in a null test, despite it having a DC blocking cap on the output.
The topology limits vs. the DAM1121 I don't get. From what I understand they major difference is the clocking?
Crossover/Delay possibilities
Hi there, i'm a happy user of the DAM1021 voor the last 2 years.
Also making good use of the 15 biqaud filters for room correction
I would like to make my audio system 2way active by adding a second DAM1021.
Because there are 15 biquads per board available there is a good way to actively filter the speakers.
The only issue is there is no possibility to add a delay for "time alignement purposes". I allready asked Soren but stubborn as I am, i'm also asking you guys if you see any possibility to add a delay.
Hi there, i'm a happy user of the DAM1021 voor the last 2 years.
Also making good use of the 15 biqaud filters for room correction
I would like to make my audio system 2way active by adding a second DAM1021.
Because there are 15 biquads per board available there is a good way to actively filter the speakers.
The only issue is there is no possibility to add a delay for "time alignement purposes". I allready asked Soren but stubborn as I am, i'm also asking you guys if you see any possibility to add a delay.
Today I received some polymer caps. I replaced the 4x Panasonic FR 1800uf caps directly on top of the 4 ceramic caps to the right of the ladder with Murata 750 1500uf.
Big difference! There is now the engaging bass and effortless high end I associate with quality analog recording.
The difference signal (original file vs inverted recording) is a few db quieter than with the previous best (no added caps) and has is a lot less high end in it, matching the listening experience.
There must be something going on with antiresonance wrt to the other caps. The difference in ESR my tester shows is minimal, the FR are only slightly worse there than the Polymer caps. The later have significantly less Vloss, but I don't think this should matter here.
The final test tomorrow will be 12 caps directly at the ladder vs the current configuration. A little more transient attack/midrange punch would great, but the DAC now has a much improved and very useable (in a studio environment) sound.
I'll also try soldering one of the caps to the clock at the J9 header (danny_66's mod). Has anyone found this to make an audible difference with the S/PDIF input?
Big difference! There is now the engaging bass and effortless high end I associate with quality analog recording.
The difference signal (original file vs inverted recording) is a few db quieter than with the previous best (no added caps) and has is a lot less high end in it, matching the listening experience.
There must be something going on with antiresonance wrt to the other caps. The difference in ESR my tester shows is minimal, the FR are only slightly worse there than the Polymer caps. The later have significantly less Vloss, but I don't think this should matter here.
The final test tomorrow will be 12 caps directly at the ladder vs the current configuration. A little more transient attack/midrange punch would great, but the DAC now has a much improved and very useable (in a studio environment) sound.
I'll also try soldering one of the caps to the clock at the J9 header (danny_66's mod). Has anyone found this to make an audible difference with the S/PDIF input?
Last edited:
On my rev5 (I think) boards there are two bigger ceramic caps each to the right of the 4 ladders, in line with the 3 bigger ceramic caps between the ladders. Put a cap on top of the one closer to the ladder, following the polarity marked on the ceramics between the ladders.
From photos I've found it appears rev4 PCBs doesn't have these ceramic caps installed.
If you can wait a little longer, I will report back my results from installing the 12 caps soon. It might well prove to be a superior mod.
16 1500uf polymer caps don't sound good.
After comparing recordings of the various alternatives I have concluded, that the stock board sounds closest to the original recording.
You get more bass with the added caps, but fidelity suffers. Attacks just sound better and more natural without them. High end clarity stays intact with in the stock board. The more caps, the more muffled the small details get.
Maybe the transistor mod can help? But it seems noone has figured out how to do it on a latter rev. board.
Also, the recordings of the board with the brand new polymer caps on sounded distinctly different after some hours, I have the recordings to prove it. No idea if it was the DAM, the caps burn in (I'm skeptical this exists with polymer caps) or the AD converter.
After comparing recordings of the various alternatives I have concluded, that the stock board sounds closest to the original recording.
You get more bass with the added caps, but fidelity suffers. Attacks just sound better and more natural without them. High end clarity stays intact with in the stock board. The more caps, the more muffled the small details get.
Maybe the transistor mod can help? But it seems noone has figured out how to do it on a latter rev. board.
Also, the recordings of the board with the brand new polymer caps on sounded distinctly different after some hours, I have the recordings to prove it. No idea if it was the DAM, the caps burn in (I'm skeptical this exists with polymer caps) or the AD converter.
Everyone thinking about adding the 16 caps be warned that removing them from the vias is no fun, even with professional (de)soldering equipment.
I added a 1500 uf cap for the clock at J9, but couldn't find any difference from it.
The stock board sounds great. I compared the DAM1021 to some widely used studio multichannel DACs - no contest. The latter all sound various degrees of muffled dynamically constrained, with the strangely swishy/blurred high end I have come to associate with delta sigma.
Soeren, please give us a different re-clocking option! With the clarity of the stock board and TNTs filter the differences between clocks have become painfully obvious.
I added a 1500 uf cap for the clock at J9, but couldn't find any difference from it.
The stock board sounds great. I compared the DAM1021 to some widely used studio multichannel DACs - no contest. The latter all sound various degrees of muffled dynamically constrained, with the strangely swishy/blurred high end I have come to associate with delta sigma.
Soeren, please give us a different re-clocking option! With the clarity of the stock board and TNTs filter the differences between clocks have become painfully obvious.
Last edited: