Modding the DirectStream DAC MKI

Audio Components DACs
3637

Yes good news, and I’ve been able to listen to DAC1 with the LL1948 again vs DAC2 with no trafos. DAC1 still sounds very good, the LL1948 are quite good. In fact good enough that I’m forced now to try and drill down a little deeper the next few days to be fair and compare them more apples to apples (same wire, etc) and see where they stand. Get a true handle on how they compare vs no trafo.

I spent several hours last night re-wiring DAC2 with the winning wire. It’s truly amazing to me what a difference even a few inches of wire can make. Of course in DAC2 we’re pretty much only listening to the sound of a couple of passives basically, including the wire. About as pure as it will get. And yes, all those passives have big, big differences at this level. So it has to be optimized to get the best result.

Post a clip below of it playing a Mister Magic live (Bob James) with some interesting (and clear as h-ck) drum stick and drum solo about midway through. If/when Google will allow it that is :roll_eyes: The DAC2 boards are in the silver DAC1 chassis here. This is with 1 pr of Front Row ic in the system so far; others are cooking and getting ready to ship soon. T

3640

I have received the adapter kit boards. As soon as the transformers arrive I can do a final fit and test and ship the kits and kits/transformers out. The adapter boards look and electrically test good.

3641

Edcors are not even close to the Jensen’s as far as open top end and tight clear bass. I would imagine the 1948’s will be very close maybe better? Everyone who ordered them will know in a few weeks.

3642

For anyone that hasn’t purchased the LL1948’s…I was half expecting a wait time for them. Yesterday, I emailed my request to Erhard Audio. In stock and I had a purchase order/USPS tracking number within an hour of payment.

Not sure when I’ll have time to install them along w/the adapter boards but am looking forward to it. I have a pair of 4400’s that haven’t been installed, so if anyone’s interested I’d be willing to send at cost and I’ll cover shipping. Please PM and we can work out details.

Thanks, Jeff, for moving this along so fast. Nice to have such competent and trusted folks on this forum. Same goes for the Ted, T and other’s that have contributed to all the experimentation and mods.

3643

Yes, Erhard-audio is the US distributor for Lundahl. Glad you found them in the states.

Google finally released the video clip lol. I hope the AI bots enjoyed.
This is what I mean by “Low Noise Floor Clarity” (LNFC) :slight_smile:

3644

Well I think the DAC2 is about ready to go out here. In silver chassis- yes, they’re swapped because the silver had a proto shield already built. I put it back together this afternoon for good I think. Just lacks the lid.

-I need to find a good shipping box for it this week. And do some comparing with the updates in progress on DAC1. Then its ready to go.

DAC1 will soon go to the DFW area for similar. A request for it was also add a 2nd pair of XLR outs to allow bi-amping directly. It will be a bit challenging to fit but we’ll fire up the garage machine shop (lol) and do it.

-This is the full current DAC2 no trafos spec. OA converted to DSD bit buffers, which feed precision passive low pass filters for D/A conversion which drive the output directly. No transformers. It uses the best parts I know of basically. It sounds amazing.

-PLEASE pm me with your name and complete shipping address and phone and emall (for Fedex). I’ll keep the personal information confidential of course, other than first name or handle used here.

-PLEASE also indicate if you use RCA or XLR outputs. I’m thinking about maybe also including a pair of Front Row interconnect for trial use. Maybe. Please let me know your thought on this in the pm.

I’m also thinking about 1 week or so should be good to check it out. That’s not set in stone and we can discuss. I don’t expect it but reality is some may not like it and be done in 1 day. Otherwise, the list may grow dynamically. We’ll see. But please be ready and available and system ready to try out the DAC; otherwise please postpone your trial until ready. Lets try and keep it moving along crisply.

-PLEASE- Terms: -Handle with care- it is a hand made prototype. And it was not cheap. We’d also like to preserve the cosmetics as much as possible please. -Do not open the DAC. If you have questions, just ask. -Please share what you heard and think. -When finished with it please prepare it (and the cables possibly) very carefully to ship to the next destination. -Then please ship it promptly, with tracking updates provided for the next destination and myself. -If not using Fedex please insure for $3000 or declare the value $3000 (Fedex). I am willing to create the Fedex labels for this.

-I will have a Fedex bag on the box for the shipping label. The plan is that this bag will serve for all labels; please handle it with care. I can create the Fedex shipping label and email it if need be. Then just print it, fold in half and insert in the bag. Please then PP me the 1 way shipping cost for reimbursement (from you to next destination). I’ll cover the initial send out. I would prefer to not use the USPS please- either Fedex or UPS only.

Update- Think I’m going to buy a hard case (they’re not cheap!) with some combination locks for the DAC travels. And think I’ll do all the Fedex shipping labels from here. And cables, think they should go separate in a Fedex small box to where they’re needed, as not everyone will.
I have ordered 1 pr of 1.5m RCA Front Row and 1 pr 1.5m XLR Front Row for demo.

Thanks! T

Hard case and locks:

image
image771×520 74.9 KB

Front Row :+1:

image
image486×632 47.9 KB

3645

@tedsmith
I was wondering if we disconnected the 2nd set of center tapped secondary windings from the 4400 transformer wouldn’t that gain 6dB output versus paralleling them. The primary would still be paralleled as they are.

3646

In the stock DS (and with a DS with the XS4400 swapped in) both the primaries and the secondaries of the transformer are in parallel. Disconnecting one half of either the primaries or the secondaries won’t change the voltages. It will change the output impedance of the DAC, but since the output impedance is on the order of 100 ohms and the input to your amp or preamp is closer to 100,000 Ohms any difference caused by output impedance in a typical system is quite small.

When you change the connections on the audio transformer by converting the parallel connection on the input or the output to series you either half or double the output voltage (or leave it alone if you change both the input and output impedance.

3647

What configuration are we listening to now?

3648

The full DAC2 config is playing here now. Later this evening I’ll have the final DAC2 vs DAC1 showdown and see where the trafo output stands. T

3649

I’m getting an odd frequency response, and seemingly reduced dynamics, with the XS4400 transformers in my DS, and I’d like to know if others have had similar results.

My setup is somewhat unique, however, as the XS4400 secondary’s center-taps have been lifted from ground (floating), resistors R705/706 and R707/708 have been bypassed, and I’m using the XLR outputs single-ended, with the -ve being referenced to ground by the amplifier. The Signal Generator from Room EQ Wizard software was used to feed the DS via USB at 352.8kHz at 24 bits. My typical load is 27kohm, but I also tried 100k and it didn’t make much difference.

Basically, relative to a 2.82VRMS output at 1kHz (0dBFS in the software setup), it remains basically flat down to 160Hz, but then it climbs slightly to 3.12V at 63Hz and then declines slowly to 2.74V at 20Hz and 2.53V at 10Hz (roughly -1dB). Other than the slight rise in the bass, this is not that unusual and seems similar to the measured results by Stereophile way back when, although I would have expected more extended bass response from the XS4400s.

Going up from 1kHz, it slowly declines to about 2.71V at 20kHz, then falls off more rapidly: 2.45V @ 30kHz (just over 1dB down) and 1.80V at 50kHz (over 3dB down). This is falling off considerably faster than the test results from Stereophile with the stock transformers and older firmware.

I took the measurements at the ends of 2ft interconnects, so that may have something to do with it. And lowering the input signal for a 1VRMS output at 1kHz gave similar results. Anyone else get results like this?

3650

I’m not sure the top end is falling off faster than the Stereophile measurements. Here is a simulation of the originally designed frequency response:

image
image2179×1695 195 KB

We liked the sound better without the final filtering cap so this is what the frequency response looks like as shipped:

image
image2189×1677 187 KB

But frequency responces don’t explain reduced dynamics.

I have to wonder if your transformer groundlift, XLR connection, cable, … is working as expected. If you remove the center tap but intend to use the output single ended, you must also make sure that there is no ground connection with pin 1 of the XLR, the cable or it’s shield on the recieving end. Otherwise you will be messing things up and the bass would suffer the most.

3651

Thanks so much, Ted. Indeed, my DS is not too much worse than what you show (-18.25dB at 100kHz according to my oscilloscope), but I was expecting better than the original output transformer.

It’s super clear and detailed and, as you say, I think the reduced dynamics have something to do with the lifted ground on the XS4400s. The same strategy was fine with the stock transformers, and when I lifted the grounds on them I was shocked by the increased dynamics of doubling the output voltage.

I’m getting no connection to circuit ground on either the + or - pins, unless I hit the filter and then both pins drop to about 15 ohms (about 22R between pins), as one would expect.

I’ll have to re-test without the interconnects. Who knows, maybe the output from the software isn’t flat (highly unlikely).

3652

Perhaps C704 is also playing a different role in my setup, or capacitance to ground in the XS4400 is higher than stock.

3653

C704 isn’t there in the DS, you will get a technically better response with a good 15nF film cap there. I suspect that taking out the 15nF cap in the DS was a patch for the Vocm bug and as the software has gotten better and people have replaced the resistor divider and add caps to Vcom that putting the cap back would be a good thing.

I also meant to say that changing the transformer only would not change the FR noticeably. It would extend the low end slightly but the other filtering in the DS would hide most of the wider bandwidth on the top end. The FR should be flat within a fraction of a dB from the bottom to near 20k. It matches the simulation results above very well.

3654

Well last night I had the final trafo vs no trafo showdown. It was closer than I expected actually. On the first song I couldn’t really pick a clear winner actually, even though there was a slight difference to the sound with a very slight veiling perception with trafo relative to no trafo. But it wasn’t enough to make me stop with a winner yet.This music was simple, well recorded pure piano chords and some percussion. They both played this gorgeously with excellent clarity and resolution. We need a different song for this it seems.

On the next song one aspect became ~ readily apparent after a few passes through it; when the music, which wasn’t that well recorded really, got complicated with lots of instruments playing the no trafos maintained the individual instruments better, more distinctly. And those instruments sounded more real. And the image was more spread out (more like it should be).

But it was close otherwise and the LL1948 is a good transformer, no question. I think most people would be ~ thrilled with the DAC1 with trafos. Even me. And again, I’m not sure an amorphous core is required to get that level of performance. A pure wire mu metal version might ~ equal it at lower cost. Going from memory, I liked them both a lot.

But we’re after the absolute most pure, clear, natural sound we can get and that is DAC2 in this case. I declare it the winner. And will convert DAC1 → DAC2 spec. And the LL1948 will go to JK. T

3655

Hi T will you share what the defenitiv OTL schematic of the DAC looks like?

3656

When we get the 1948 transformers in I can A/B the Jensen’s to the LL1948’s The Jensen’s are MU nickel based where the LL1948’s are amorphous cobalt based.

3657

I cannot wait for the outcome.

3658

The Jensen’s are not cheaper and are harder to find. What other MU Transformers are available?