Modding the DirectStream DAC MKI

Audio Components DACs
3680

Well I’ve become super interested in DAC jitter again lately. While we wait for the super duper inductor treatment to get done for the DACS. Thanks to Ted for all his help on this here.

I don’t have the gear to do it here really but wanted to see if maybe I could come up with a way to at least get a baseline feel for what’s going on. Alas, I have only a 100MHz DSO here. At the old j.o.b. we had 8GHz sampling machines, and that was 10+ years ago already. Anyway I soldered a coax cable directly to one of the OA bit buffer output with a BNC on the other end for the scope. Keep it a tiny loop area at the probe end and it is indeed a high fidelity connection, no question :wink: However it is not transmission line terminated but we can ignore that for this purpose as the reflection artifacts are away from the edges. We’re interested in the edges here.

Years ago at the j.o.b. I used same cable once with a differential amp to take some stunning clear scope images of an SMPS phase node, right down in the noisy heart of it all.

Anyway, was wondering if maybe we could hedge our bets here and see something useful if we moved the scope trigger way upstream and look with the scope at the bit edges like 50us downstream. And allow the scope to catch an infinite persistence history there basically and examine how wide the resulting trace capture is, which should indicate some amount of jitter. Maybe (probably) I’m not thinking clearly, who knows lol…

But this is the best I could get with the baseline characterization, below. Probably not much use but there it is. Those bits are occurring pretty darn steady (no surprise) :wink: Wish I had some better gear… T

Scope turned up as high as she’ll go (5ns/Div)… This is 50us downstream of the trigger, and infinite persistence with many captures. Sorry for the washed out cam tonight:

image
image1920×1440 186 KB

image
image1920×1440 238 KB

3681

Thanks Ted!

-Alan

3682

Ok, new baseline… I put a shielded ribbon cable I had made and it does appear cleaner. On the DS the cable does connect right over the VCXO and control circuitry. So we look at it with raised eyebrow at least…

Also have settled on using a 44K1 sine wave to test.

And very interesting too, can see the VCXO being adjusted in real time… It complicates this when it moves but it will stay steady for long periods as well. I’m glad Ted explained that to me, otherwise I would have been calling an exorcist or something while watching the traces slide over in real time lol. T

New baseline- I’d call it less than 800ps :slight_smile:

image
image1920×1440 157 KB

Sometimes the traces do the electric slide with VCXO in action:

image
image1920×1440 152 KB

3683

Remember that your scope has to have a better clock than the device being measured. Also, the scope’s intrinsic jitter has to be accounted for and “subtracted out.” (It’s more complicated than that, but you get the idea.) In general, to get reliable results you need to sample at at least 5 times the clock signal bandwidth. In the case of the DS the clock bandwidth is approx 5GHz (this comes from the spec’ed rise time of the reclocker.) Also the jitter that matters for audio is the phase noise lower than, say, 20kHz That needs some long long measurements to find the 20Hz jitter.

3684

Yeah… I’m trying to do something I just don’t have the equipment for. It’s below what the scope can go. Even through the op-amp bit buffer. That’s a good thing IMO. Now I know.

I have to laugh, thinking about the allowable jitter spec for AES… something like 1ns is “ok” to them… Pro audio guys sigh… I’m sure the audio science review herd would agree as well- Science! :wink:

Speaking of 20Hz… I also checked the DAC1 transformer output using the Keithley 2015 THD+N measurement. At 20Hz the reading gets worse by some ~ +30dB vs same at 1KHz. Even with an amorphous core trafo. Wow. Ouch.

Will have to compare that vs the DAC2 with no trafos. Actually I already did I guess; and I recall it was ~ same at 20Hz (or better) as 1KHz. Using 176KHz FS full scale sine waves. Will check it again soon.

BTW The DAC1 is playing again here this evening, after 2-3 days of another highly regarded expensive DAC. My goodness- the clarity, the realism, the low noise floor- simply unprecedented IME. I cant wait to get the DAC2 out for demo. The case arrived Friday as well. Its going to work. T

3685

Ted,

Is there a C704 on both output transformers or just one? If just one which one?

3686

One for each channel. I just showed one in the picture above, but the other one is trivial to find, especially since it has to be tween the transformers and the relays.

3687

We are one (1) short of placing another batch order of adapters.
PM me if you would like to be placed on the list.

I will have a test transformer here today (Thanks Turbo) to do the final fit and test of the adapters and will begin shipping them out.

3688

Many repeated questions are being asked again so here are the post #.

Here are some indexes of MODS:
This is a temporary solution to help find important post.

Video instructions to help open the top cover 1524

XS4400 Transformer mod 160
Link to manufacture 630
Circuit board explanation for transformers by Ted 1133
Document transformer mod install 1141
XS4400 Transformer layout 1749
Transformer comparison to pro gear 1662
Recommended de-solder for DYI transformer mod 1664
Not to use both XLR and RCA by Ted 1180
Transformer burn in recommendation by Ted 1592

Internal Analog power connector part number 195
Power wire twist 1503

First external power mod 312
Power supplies used 1722 & 1870

First internal power mod 434
Ted Smith linear power supply notes 497
Ted Smith on power up sequence 749
Internal power supply noise 1827

Block EMI / RF with stilpoints 830 & 1481
3M Sheet that seems to cover the emissions range 1647
Shielding photos 1713
DYI screen shield 1980
From Ted a few years ago regarding EMI absorption inside the DS 937

Teds original prototype 1106

Vented cover mod 1233 & 1377

First talk about The Power Supply Controller Board 1683 & 1760
Introduction to The External Power Supply Controller Board 1852
External Power Supply Controller Board Description of operation 1974
2-pin power connector used in the Power Supply Controller Board 1886
Install instructions for the External Power Supply Controller Board 1898
First photos of installed boards 1969
Photo of finished power supply board 2055, 2071 & 2090

From Turbotk:
Here are some links that at least start I think:

VOCM mod. 2762
No transformers early test: Modding the DirectStream DAC - #2111 by Turbotk
Modding the DirectStream DAC - #2112 by Turbotk
Vocm mod: Modding the DirectStream DAC - #2674 by Turbotk
Modding the DirectStream DAC - #2685 by Turbotk
Change OA circuit to perfect step response: Modding the DirectStream DAC - #2835 by Turbotk
Modding the DirectStream DAC - #2836 by Turbotk
Swap OA to higher perf versions: Modding the DirectStream DAC - #2726 by Turbotk
Modding the DirectStream DAC - #2858 by Turbotk
Final approach- convert the OA to bit buffers: Modding the DirectStream DAC - #2889 by Turbotk
Final approach- passive LPF D/A: Modding the DirectStream DAC - #2976 by Turbotk
Modding the DirectStream DAC - #2996 by Turbotk
Bybee Purifiers: Modding the DirectStream DAC - #2998 by Turbotk
Recs: Modding the DirectStream DAC - #3002 by Turbotk
Some results: Modding the DirectStream DAC - #3015 by Turbotk
Modding the DirectStream DAC - #3027 by Turbotk

LL1948/ LL1949 Transformer adapters.
https://forum.psaudio.com/t/signup-and-order-for-the-lundahl-ll1948-transformer-adapter-board-kit-for-the-directstream-dac/29876

Location of the output CAP C704 that was not installed from the factory 1379.

3689

Nice summary of mods. FWIW, the first internal power supply mods were done back in 2017, just mentioned within a different thread: “Ultimate Directstream” :slightly_smiling_face:

3691

Well today spent some time tweaking the OA bit buffer resistors a bit. Simulation was promising some nice things so tried it. Got the bits looking really great on the scope IMO (per below), but the meter didn’t measure any lower THD+N. Think I’ll go with these new values anyway though since the bits look so good.

Next, think we’ll try a new PSU experiment for the OA section. If I can hear an improvement we’ll pursue it further. T

image
image1600×1200 126 KB

3693

Sorry about the delay. I’ll get the LL1948 out of DAC1 in the next few days and off to you. T

3694

No problem Turbo and no hurry. I’m just trying to get these adapters kits out to members so they can give the LL1948’s a listen but do want to make sure they fit properly. I don’t think there will be any issue as they test perfect and the LL1949’s pin spacing fits just the board is too short for them.

3695

I ordered transformers for me and @yuriy in Europe but haven’t get production date yet. Normally this kind of transformers supplied either in 1 week or of it takes some days to produce - in 2-3-weeks. But we wait for confirmation from Lundahl.

3696

@jkrichards, I received the transformers (1948’s) last week but won’t be able to do anything until the first week in Sep. Probably won’t work into your schedule but let me know if I can help at that time. FWIW, for timeline on receiving my order/boards, I’m in no rush as I have family visiting and am also neck deep in house renovations…but, would gladly help, then, to move things along for everyone else if still needed.

3697

I haven’t made a shielded ribbon cable. How did you make it?

Thanks!
Vince

3698

For the ribbon between the analog and digital boards, it probably has a characteristic impedance that could be affected by shielding. So I covered the cable with cotton ball fluff - if that makes sense - then loosely wrapped copper shielding around that and soldered on a lead and grounded it to the chassis. Wrapped the shielding with electrical tape in case is touched something live - one could use heat shrink but I think that would distort the shape of the ribbon. It looks awful but sounds good. All the other ribbons I just wrapped copper tape directly around the ribbons, sans cotton, and grounded.

3699

Yes, if you wrap the ribbon with a grounded Cu or alum foil for example (rather than surrounded by air) that changes the ribbon characteristic impedance.

The wrap provides voltage shielding (the E field in EMI) but not magnetic shielding (the M field in EMI).

However we could debate as to whether that makes any significant difference to the actual operation here. Maybe, maybe not. In my own experience I haven’t heard or experienced anything detrimental from doing similar with 3M EMI stuff. However YMMV and we do not recommend simply wrapping the cable with foil shield.

A possible recipe here. You might consider wrapping the cable first in a very thin expanded cell foam, like used for packing. The air bubbles make it probably close to a teflon dielectric, around 2.x’ish vs the 1 of air.

By wrapping, instead of a spiral wrap process you could cut a ~ 2x width of it of the proper cable length and simply fold it over the cable like a sandwich. Secure the edge with some tape. We’d like this dielectric spacer layer to be evenly distributed along the cable length basically.

Then wrap (fold) that in a layer of the E field foil- alum or Cu. Also run a small ga flexible stranded drain wire along the length of that foil so you can connect the shield to a ground. I suggest perhaps the local large caps (-) pad near the cable connection head at the digital board for that connection. That drain connection length needs to be minimized/short for any hope of actual ~ high freq service.

Then finally encase/wrap (fold) that well prepared Dagwood sandwich with some 3M adhesive EMI stuff to contain it all. Looks pretty slick and neat with the color as well :slight_smile:

As usual YMMV, none of the above is officially recommended and you may have better ideas. T

image
image382×713 64.4 KB

3700

Very nice. Thanks!

3701

No problem.
Turbo came to the rescue and is sending me 2 LL1948 Transformers :smiley: Should have them later this week or early next week.
Thanks!