Modding the DirectStream DAC MKI

Ok didn’t know he was working on this

Once upon a time known as Ted Smith Signature DAC.

Final design and product TBD, AFAIK.


It does not appear on the screen when the resolution is selected as 705.6khz

Thank you Ted, Thank you turbodk for the photos. Today I was at my friend’s place and we made the 3D prints/cupholders for the additional ultracapcitors.
Thats the good news.

Now the bad ones: We couldnt find the voltage source for the motorola chip (is it 3.3V or 5V?). @tedsmith Is it OK to ask you for a small piece of the schematics of that section?

Regarding the power supply for the clock: it is supposed to be completley passive w/o active components.

The reclocker is 5V. Test Point (TP805) or the big top of Q800 is the easiest way to access the power for it. The chip itself has three 5V connections and one ground but some are buried underneath the chip.

I’m kind of literal minded. This is a statement. I’m not sure if you are asking a question or what said question might be. I would disagree with the statement as well… Altho the eight-legged low ESL capacitors around the clock and the reclocker do most of the work at around 22MHz IMO you want a low impedance voltage source for the clock and the reclocker which requires active components.


Wait, are you considering putting super caps on any of the power supplies on the analog board? That will take the voltage regulators way outside their safe operating area on each power up until the caps are charged. The clock regulator has a junction to ambient of 265 degrees Celsius per watt and charging a super cap will cause it to dissipate 0.72W at the beginning of the charging cycle (and slowly lessen until charged.) Each farad of the super caps’s value will take a minute or so at that power dissipation… This isn’t a problem with the capacitance that it was designed for, but super caps are basically a dead short when initially charging… The reclocker supply expects a bigger load but still a super cap will heat it way past its safe operating area too.

I’ve warned about adding capacitance to the analog board power supplies in the past but super caps are a whole different kettle of fish.


Hi Ted, thank you and please excuse the confusion. Attached to the crystek clock is an external power supply which is based on 2 ultra capacitors whichs has no active parts to generate the 3.3V.

The trace to the orginial 3.3V supply is cut. The crystek is like you wrote still one of the best clocks - no need to change that one. Since I really liked what that did to the sound I wish to do the same with a 2nd power supply for that motorola chip.

Link to description and photo of the pcb

1 Like

Ah, if you have your own supply that’s fine. I somehow didn’t catch that. To disconnect the reclocker from its supply you’ll need to simply remove Q800 and provide 5V to the big pad that Q800 was on. You could clip the two legs off of Q800 if you don’t have the equipment to remove it.


Like to ask if the Edcor XS4400 transformers removed from the new MKII will work on the old MKI?
Thank you.

1 Like

Yes, they will work fine (but, just in case: most other transformers aren’t interchangeable between the two units.)


Terry, I have a pair of Edcors on the JKR mounting boards if you’re interested. PM if so.

I noticed that the later production of MKI’s the toroidal power transformer 120 vac carries an ovalish shape and the older are totally round. Anyone knows why this change ?

Q800 transistor change with ultracapacitor power supply
Goal: To provide the best power to digital conversion chip
First listening test: Not as noticable as the power supply to the clock, but a bump up in soundstage and separation

The DAC with Q800 transistor removed

Close-up of Q800 removed

Wires attached

The two ultra capacitor boards from Ian Canada (makes great diy stuff)

The top plate is very heavy and the weights like a small amp

Next: the VA mod, perhaps 8139 replacement, opamp output buffer
and the q800 replacement sparked an idea: why not replace some other transistors like for the i2s and FPGA like that (not necessalry with a big power suply like for the the Q800 but somethin small like that LT3045 power supply)



Shameless plug:

Got any use for a JK Richards analogue power supply board or Edcor transformers?

Have fun.

1 Like

Thank you. No need for both. I too have some Edcors and a bridge2 to sell :smiley:

1 Like

Hi Jannik. Love your efforts here. Catch me up please, what is the VA mod?

And yes, you might consider moving from 8139 to the TI LMP8350 if I recall the number correctly. As I posted before, I preferred them to the AD choices. That wound up being the DAC DW was using, until the “Studio” version. But to be 100% clear, I also change the filter topology a bit and added some pre-filtering that isn’t present in the original circuit.

FAR better still though, as I shared above is a complete replacement of that output stage to an ultra analog “error correcting” topology, with a 2 pole supply. So no output caps or trafos required. It takes the DS MK1 about as far as it can go IMO. And that is pretty high up (ie, the “Studio” version). I tried most everything else along the way.

And we’re looking to try and get hands on a DS Mk2 this year, somehow. Lately I’ve been working on a super power amp development project, the Ultra_EX-B7 monos. The top to bottom re-design and proto construction/testing has been a lot of work, but they are fabulous (and if not, I’m wasting my time :wink: Unprecedented clarity and transparency IME, no distortion, no noise, etc. Should have a reviewer I respect a lot give them a listen after Axpona. And maybe another pretty well known guy as well. Meanwhile we got to shake down 2 pr in DW’s super system the last month or so. He has the full loom of Audioquest Dragon for example ($$$$$ lol). We can hear everything in there… And in DW’s system that also means the volume generally winds up nearer to max than it should be too :wink: Poor Audio Vector drivers, beaten like a circus monkey :wink: But wow! The audio bar is wayyy up there now… And there’s no going back.

Thanks, TK

Shaking down 2 pr of Ultra_EX-B7 in DW’s system- bi-amped with AQ Dragon cables. The “Studio” DAC still be holdin’ in there (and it’s amazing):

It’s a real blessing to be able to have access to test electronics in this system. Too bad DW moved away this week :frowning: Audiovector R6 Arreté review: exceptional musical performers | What Hi-Fi?

Some early Ultra_EX-B7 monos proto development work (whew):


So back to the DS Mk1 aspects… I’m really chomping at the bit to get the 2nd Studio version DAC built asap for my own system, Studio #2. I’d be on that now but my plate is full currently with still updating the mono power amplifiers. I have 2 pr to update, revamping the 1st 2 pr of protos. Basically am trying switching out from 4T caps in the PSU portion feeding the board rails, and will try some Litz pure Cu wire rather than the pure silver wire. I love Ag wire in general, it’s generally ultra transparent and quiet but am trying to get the midbass warmth up to what we want. Also have a Z foil in the very front of pr1 vs the AN silver Niobium I tried in pr2. We’ll see if things warm up to what we want… If they do, man, we’re really off to the races with this :slight_smile:

And we need to get this ~ Goldilocks t-ts up right as I have several significant parties (and potential customers) lined up now to hear them soon. And another new guy as well, has a fabulous system and a ton of listening and audio experience. He just bought a CDT3 and is loving it as well. Just called me raving about it :smiley:

And I find it amazing too that we’re at the level where the base circuit is ~ so out of the way the sound is essentially dominated now by the remaining components, the PSU and the wire. Never dreamed we’d be saying that.

Oh and while I’m remembering, I also ordered some 300W +/-36V SMPS to try in the amps as well. They do offer several potential upsides. But we’ll try it vs the excellent LPS and see; that’s all we can do, see where it actually winds up sonically.

So the #2 Studio DAC will differ from the #1, we’re going to use the same discrete circuit as the front end of the amp basically (minus the power output stage of course) as a “discrete op amp” for it. It will still fall into the category of what I’d call “ultra analog”, same as the Studio #1 and into “error correcting” territory with a lot of OL gain at 1KHz (over 105dB @ 1KHz). Cant wait to hear it… I expect that it will be even more amazing than Studio #1 :+1: And that is high praise indeed.

The output stage dominates the sound of the Mk1; get it right, get it out of the way sonically and it really amazes. I wish we could also get the latest, ultimate fw scheme in it as well.

Also pondering lately the possibility of maybe stitching in a hybrid PLL CLK cleaner to knock out CLK source jitter. It’s low already, but it can go (much) lower still. Also keeping in mind that TS had said he didn’t care for that when they listened to a CLK cleaner (not sure how or why not). From my own past experience, a CLK cleaner was a healthy step up in sound staging and such with an old DAC project (PCM, not DSD). Hmm… :thinking:

We just want to get it to be the best we can.


Here’s lookin’ at you kid :slight_smile: Soldering long into the night, here connecting some new un-plated Cardas Cu binding posts with 14ga UPOCC stranded wire. Those things take a ton of heat: