Modding the DirectStream DAC MKI

@Anton_ss please can I use these diodes in 230V (EU) version? I was just going to order from Mouser, but noticed 100V in the specs

The primary voltage of the unit (220V or 110V) doesn’t matter. We have the same secondary windings on the tansformer that these rectifying diodes are connected to. So we all should have the same rectifying diodes.
Schottky diodes are the best option.


thank you Anton!

Regarding the labels.
ES2D is the type.
You have diodes from Farnell (thus “F” and AE12 just a manufacturer’s code), while @Stimpy7777 has diodes from ON semiconductors (ON).

1 Like

Hello guys. It’s been awhile. Still around and still digging for the best audio we can create. Was working on an amazing preamp the last several months. But also kicking some ideas around meanwhile on the DS Mk1 DAC. Fast fwd to now and I have implemented it and yes, its finally what the DS promises to be. And wow, is my response. Felt we needed to share here because I know this is a hard core group of audio folks who can appreciate it :slight_smile:

Basically we have taken the cork out of the DAC finally, ie ditched the industrial diff amps output stage altogether and replaced with a new super high performance discrete output stage. This is VERY high performance, discrete “error correcting” circuits if you will. Near the pinnacle of analog gain circuit design basically. They essentially have no distortion and no noise. Even the better industrial diff amps, the AD 8139 and other brands (that I prefer) aren’t even close to this level- I know, because I have personally tried them. So now we can finally hear DSD without the cork and it’s glorious! So much music, detail and needed information, soundstaging, no distortion, no noise, etc that was covered up all this time finally is allowed out. All the nuance, tone, presented. Everything sounds accurate and correct. Drums, bass, vocals, …

We also had to apply some creative discrete pre-filtering and configuration to get the BW down from the pure DSD128 range to not overwhelm and to properly feed the analog circuits. Its working splendid. The output level is also boosted a bit vs a stocker, I have 2.6V rms out with a 0dB digital input; that’ll work. The output signal is very well filtered and it has perfect square wave response as well, as much as one could hope for. The passives are generally Zfoil level, etc. Wiring is OCC Ag and Cu, Audience MORRE treated even.

-Also powered by a very high performance bi-polar PSU with 2 layers of DC/DC, ending with local very high perf discrete linear regs near each circuit.
AND because of the bi-polar PSU and the way things are architected NO output blocking devices are needed either (trafos or caps), even with the ~ 3.6V DC output on the re-clocker that feeds the output stage. Ie it is DC coupled for the win(!)

Note, beware- one big issue with trying to do this in the DS DAC Mk1 is a power on crowbar condition that can/will occur with an external PSU to the analog output stage. And it can smoke things, even the main transformer depending on how you’ve implemented things. Don’t guess how I know (ack :wink: BTW those transformers were a special PN to PS Audio and you cant get them anymore :wink: But that power on issue has been solved now thankfully with some inspired thinking :slight_smile:

I call it the “studio” version now because it is so much more like being in the studio, rather than listening at the end of the chain.

So next Thurs the 7th is a big deal, it will be the world premier of this new “studio” version in a friend’s very high performance, very good sounding system in Dallas. Even at higher than club volume levels it sounds amazing. I know from personal experience, debuting the new preamp (he bought it).
So we’ll see how it goes in that system and then proceed accordingly. Even thinking I might try and acquire a small lot of the Mk1’s to allow others to be able to get their hands on this level of performance. I expect great things on the 7th, and will likely return home without the “studio” DAC proto (lol) but we’ll see…



What changes did you hear from using the silver transformers? Was it worth the cost?

The silver version belongs to a friend of mine. It will stay with me for a few months. Mine is the lundhall copper version. We also changed the opamp and are feeding it with 3 separate LPS. It’s like listening to two different DACs. The silver version has a bigger and more spacious sound. Also embroidered details such as lace. I think it’s worth the money. Now it’s time for silver wbt exits.

I think I roughly understand the effect, as the molecular tuning I have for my equipment, among other effects, makes the whole signal path inside (not only the transformers) conduct much better than silver. It’s a level of effect one can’t get with any kind of equipment or circuit upgrades.

I just due to suspect rebooted my DS after some time again and bingo…more 3D and fine dynamics was back.

This is really annoying, but I recommend anyone to do it regularly as Ted suggested once. I wonder if the Mk II suffers from this, too.

Another weird effect I sometimes have is and not since long is, that after switching out the unit, the complete display flashes for undefined time, even without power connection. Anyone else?

1 Like

Yes silver in general is quieter and more transparent sounding than copper. So the transformer versions also ~ pretty much follow. But, it’s still a transformer in the signal path. The signal still has to traverse a lossy magnetic pathway and that leaves a sonic footprint. No way around it.

IME output caps are more transparent than any of the transformers. Try it, it’s not hard to do. Then try it with neither cap nor transformer :slight_smile: If you use a preamp that has input blocking caps you can easily just try it as a test. You can just jumper over the transformers for a test.

In sum, the Vocm mod + trafos upgrade is a real nice start. The DS DAC really starts singing when you change from the MFB LPF topology; I believe it’s likely because the step response is also corrected then.
Then, another, better sounding diff amp vs the 8139 makes a nice further upgrade. That setup + caps is becoming a truly great DAC; no distortion, albeit a bit of “softening” of the sound is about the only remaining flaw really. Move to a bipolar PSU and remove the caps and the transparency increases slightly further. It’s a lot of work for that though, but worth it to some of us.
Then finally, ditch the industrial diff amps and put in a new super good discrete analog output stage with no output caps. Joila, no more flaws; all DSD sonic goodness. Nothing left to complain about at all :slight_smile:

After all this, in hindsight it makes sense to me. The music (analog) is already there in the DSD stream. There is no digital to analog conversion per se, eg R2R or what have you not like other DAC solutions. All we have to do is low pass filter and the music is uncovered basically. It was there all along, just covered up by the high freq DSD aspect.
So now we have the DSD being initially passive low pass filtered, very little sonic imprint there because we use the best passives we can. That then passes to a super analog out stage for some gain and further low pass filtering, but this stage is also discrete high performance, extremely transparent and low distortion.
And that’s it, out comes the precious music signal with very little imprint on it (but with no more high freq DSD), passed out the back door connectors to our system and into our ears. Dig it!


This is perhaps the elusive TSS DAC everyone is waiting for. :slight_smile:

Well I think I understand the TSS approach, and was initially going that way myself as well (independently). I repurposed the 8132/39 diff amps as DSD bit amplifiers (they have the BW to do it), then fed that amplified DSD stream into a precision, passive RLCLC … low pass filter. It worked. But we didn’t remain on that path for long, esp once we realized the music is in there, it also still passes through the diff amps basically. There was no escaping it.


1 Like

What’s the chance of seeing this improvement in the MK1s owned by mere mortals? :slight_smile: I am pretty happy with the VOCM/APS Nickel Trans mod, and have no plan of upgrading to MK2 at this point. Would love to squeeze out more from this great DAC if possible.

There is a lot more possible :slight_smile: We’ll see how the big “studio version” debut goes in Dallas next Thurs in front of some of the hard core guys and take it from there accordingly.


This effect is only observed with AirLens when connected with an HDMI cable. If PWT transport is connected to the DS, then this effect is not observed. Then I decided to find out the reason for this phenomenon. I measured all operating voltages at the HDMI output of the AirLens and compared them with the PWT. It turned out that there are two voltages at the AirLens output that are not present on the PWT, these are: 3.28V on pin 16 and 4.35V on pin 18.
What happens next? This voltage from the digital board travels through the power line through the main power supply to the display. The magnitude of the voltage is not enough to start it, but this voltage also reaches the power capacitors, which are charged by it and at the moment of discharge they cause the display to flash, and since this is an oscillatory process, the display flashes with a certain frequency.
It is obvious that PSA did not take into account the topology features of the DS MKI digital board. How critical this is, I cannot say. I contacted technical support, albeit with a different question, but did not receive an answer, so I did not continue knocking on the closed door.
P.S. In the end, I solved this problem on my own.

Wow thanks, you’re a genius! I indeed have it since the AirLens.
Yes it would be interesting if there’s a risk of damaging something with rebooting.
The support at PSA in my experience is very friendly, but extremely slow and there seems to be hardly any efficient contact option for them to get answers to more than extremely simple questions from the engineering departments before months are over. Since Ted is quiet, there’s also no such discussion anymore in the forum, so I guess, to be save, one would have to shortly disconnect the HDMI cable before rebooting the DAC and then connect it again…

That sounds dangerous!
It means that PS Audio in AirLens are NOT compliant to their own i2s standard, doesn’t it?

Not really.
Most likely, PSA expanded the possibilities of using the HDMI bus, which they proposed 10 years ago, and at the same time they were also developing the DS MKI digital board. Since then this board has not been improved.
According to the datasheet, pins 15/16 are DAС control via I2C (not to be confused with the i2S data bus). It carries information, for example, about turning on a device that is in standby mode (for this you need +5 V transmitter power, contact 18…) as well as service signals, signal format, etc.
Антон_сс, эта проблема уже решена, так как мы разделили питание цифровой платы с питанием процессора.
P.S. But these are my assumptions; it would be more correct to hear a qualified explanation from PSA engineers!

1 Like

Thanks for the info. I used the silver transformers also in the MK1 and liked them.

Have you ever listed to the Bricasti M21 DAC? If so, do you have any opinions on it?

1 Like

I have not tried the M21. But I am a former Bricasti M1 user, only giving it up once PSA DS MK2 Massive came out and I did not hear a need to keep the M1 around for comparison anymore. Conceptually for me, the M21 having three different DACs in the box… one PCM using Analog Devices AD1955, one PCM ladder DAC and a native DSD converter to boot… is too much for me. Since I run HQPLayer, I could just output DSD to the much less expensive Bricasti M3, which has the native DSD converter, and be happy, without needing to scratch my head over how I wanted to convert PCM. I suppose one could use Roon to upsample to DSD too.

1 Like