Bear,
I have the Shiit Gungnir non-multibit ($1100) and the Soekris DAM1023 ($350) DACs.
Using the 1023 right off the resistor array, so need to boost the preamp a couple steps to get the same output as the Gungnir.
My 1023 uses two $11 regulated power supplies and 1 transformer. Also using the Amanero USB to I2S for use with music server.
The sound is very close, but the Soekris DAC is more enjoyable ($500). It doesn't feel forced like the Gungnir. Now, I''ve listened with others and they prefer the Gungnir for one reason or another, but there is something more organic about the DAM1023 to me.
I've heard the Yiggy and Gungnir with multi-bit upgrade. I'm not sure what the big deal is. I mean I like it, but I can't hear that it's that much better. I know several people that have both, so I might have to have another listen, but for all the talk, it better knock me over.
I plan to put a Didden SuperReg on the +/- 12v and a LT1085 on the +3.3v side. Will also try independent transformers.
The Soekris filters are another interesting feature you get to play with.
Just my 2 Cents.
Vince
I have the Shiit Gungnir non-multibit ($1100) and the Soekris DAM1023 ($350) DACs.
Using the 1023 right off the resistor array, so need to boost the preamp a couple steps to get the same output as the Gungnir.
My 1023 uses two $11 regulated power supplies and 1 transformer. Also using the Amanero USB to I2S for use with music server.
The sound is very close, but the Soekris DAC is more enjoyable ($500). It doesn't feel forced like the Gungnir. Now, I''ve listened with others and they prefer the Gungnir for one reason or another, but there is something more organic about the DAM1023 to me.
I've heard the Yiggy and Gungnir with multi-bit upgrade. I'm not sure what the big deal is. I mean I like it, but I can't hear that it's that much better. I know several people that have both, so I might have to have another listen, but for all the talk, it better knock me over.
I plan to put a Didden SuperReg on the +/- 12v and a LT1085 on the +3.3v side. Will also try independent transformers.
The Soekris filters are another interesting feature you get to play with.
Just my 2 Cents.
Vince
I fully agree. I've even bought an SMSL M8 for our big hall control room laptop.
I haven't followed the whole initial DAC thread so apologies if my questions are answered in there.
WRT R2R monolythic sign magnitude ladder DAC's, they generally need to
be terminated into virtual ground otherwise additional distortion is created.
Is this the case with this discrete DAC - have you measured any difference
with OP voltage swing and running into virtual Gnd I-V?
WRT 0.01% resistors, have you taken into account the finite resistance of Cmos logic 'bit drivers'? Also they have different OP Z between high and low
states.
Thanks - BTW very nice project.
Not relevant, the dam1021 is native voltage out....
Of course, beside studying the datasheets I even measured some lvc595 samples at different voltages. The recents revision use resistors manufactured to order with custom values.
I have a dam1021 v3 board, in its original form without tweak other than using another board for DC input. It is a very pleasant sounding dac but I wouldn't call it truly outstanding. Compare to the cheap Chinese made es9018 dac (with some mod), the Soekris is a bit more pleasant sounding but did not completely out class the ES9018 dac.
I later modify the soekris 3.3v and +/- 5V (personally, I think the power supply for this dac is fairly poor using the 1117 and lm7x05 regulator, and very thin power trace) to low noise tps7a7400 regulator, it improve the sound up a notch, both detail and dynamic improve a bit and becomes a really nice sounding dac.
That said, I am not using the Soekris right now as I found the lifepo4 battery power ak4495 dac with transformer output to be even better. More organic sound, better detail and impact. I found Eva Cassidy sound better with ak4495 in my system than the Soekris.
Beeng an owner of dam1021 v2 (heavily modified) I am also very interested in comparisons with other dacs. Especially Totaldac d1 that is also built as an r2r dac but with much higher quality (and more expensive) resistors. And without oversampling and with very short digital filters.
As in my system, when I make upgrades/modifcations to dam1021, I hear very distinctive and clear SQ changes (for example, after removing filter capacitor the SQ has improved very much; and swapping Li-ion and alkaline batteries has a profound effect on the SQ) I am wondering how would the 'final' version of the dam1021 sound (it would not be dam1021 anymore, BTW).
So, depending on your objectives, I guess the question is if you are interested in comparing unmodified versions of different dacs or get a grasp of how modifications of dam1021 affect the SQ.
Last edited:
You might hear what the SOEKRIS can really sound like if you did the nige2000 26650 (I used the A123 variety) mod. Souinds like you are on the right track. Give the SOEKRIS another chance with good batteries.
Even when both AK4495 and Soekris is using AC power, I can't say I prefer the Soekris more. The AK4495 is a different beast when I use output transformer. My Soekris V3 already has 3 ultra low noise tps7a7400 on the 3.3v, +/- 5V and DC input to the +/- input with most of the main cap changed to oscon. Panasonic PPS smd caps on the +/-5V. I could add more cap decoupling on the resistor rail, but nevertheless I think you would agree that there is already quite an extensive mod on the Soekris power supply.
It is so much easier to use battery on the ak4495, only one 3.3v and 7v (i use 2 batteries) and done. It would be more difficult to put battery on the soekris and I don't have a motivation to do it unless battery has a huge improvement over the existing low noise tps7a4700.
chuckt,
I think you can expect a significant improvement.
Use those regs to "float charge" the batteries - they are being used as giant fast capacitors - that is all there is to it.
Yes, it is complicated since you have to incorporate some kind of switch for each battery.
I am using those same regs to "charge" the batteries on the resistor stack. 3.3 volts is plenty there. SOEKRIS has never explained why he chose 4 volts. I am sure he had his reasons but there is not a sonic penalty that I can hear (which is next to meaningless I readily admit!). I am using a DDDAC shunt reg for the input battery since it was not being used anywhere else.
Another aspect of the mod is removing many of those ceramic caps and the existing vRef supply components entirely. The clock is powered by a 26650 which is also used to power a 1.2 volts regulator for the FPGA instead of the switcher as supplied. While you are at it remove all of the components for the 5 vols regulators and the caps for the AC input. My board no longer has any elelctrolytics and the only ceramics remaining are those surrounding the FPGA and those between the FPGA and shift registers.
I have found there is nothing to fear from removing the low pass capacitor on the output of each channel. Makes a big difference in quality. If you want to retain it put a better cap somewhere else in the chain. You cannot fit a good capacitor in its place on the board.
One battery for the "input" section and four (two for each polarity and a pair for each channel) for the resistor stack. Using the vias for power input on the output and the removed regulator pads for the "input" power input. I used a cloth covered silver wire between battery/switch and the vias.
Depending on where your signal is coming from you should remove those isolators or at least bypass them. In my case I take the wires from the output of the SDTrans and solder direct to the isolator pads.
All of these things add up.
Something to do when you get bored ...
I think you can expect a significant improvement.
Use those regs to "float charge" the batteries - they are being used as giant fast capacitors - that is all there is to it.
Yes, it is complicated since you have to incorporate some kind of switch for each battery.
I am using those same regs to "charge" the batteries on the resistor stack. 3.3 volts is plenty there. SOEKRIS has never explained why he chose 4 volts. I am sure he had his reasons but there is not a sonic penalty that I can hear (which is next to meaningless I readily admit!). I am using a DDDAC shunt reg for the input battery since it was not being used anywhere else.
Another aspect of the mod is removing many of those ceramic caps and the existing vRef supply components entirely. The clock is powered by a 26650 which is also used to power a 1.2 volts regulator for the FPGA instead of the switcher as supplied. While you are at it remove all of the components for the 5 vols regulators and the caps for the AC input. My board no longer has any elelctrolytics and the only ceramics remaining are those surrounding the FPGA and those between the FPGA and shift registers.
I have found there is nothing to fear from removing the low pass capacitor on the output of each channel. Makes a big difference in quality. If you want to retain it put a better cap somewhere else in the chain. You cannot fit a good capacitor in its place on the board.
One battery for the "input" section and four (two for each polarity and a pair for each channel) for the resistor stack. Using the vias for power input on the output and the removed regulator pads for the "input" power input. I used a cloth covered silver wire between battery/switch and the vias.
Depending on where your signal is coming from you should remove those isolators or at least bypass them. In my case I take the wires from the output of the SDTrans and solder direct to the isolator pads.
All of these things add up.
Something to do when you get bored ...
Say what? Who?
Link?
But I still have no points of reference in terms of what sort of systems are being used here to listen through (from the DAC out).
What sort of "SQ improvements" are happening vs. stock.
Also while I have not seen any recent tests on LiPo battery noise, previous batteries were inferior in terms of noise to the best regulators.
Also, I thought the direct out from the resistors was the way to go? No?
The buffer opamps are cap coupled outputs?
And the OEM DAM 1121, I guess i need to find the manual for it, and read, but how are the two filter "banks" usable, and what of the "biquad" section??
Trying to get a handle on that part too...
And, has anyone RUN FFT tests on the stock vs. modded versions??
Any change in the harmonics from the published?
_-_-bear
PS. PM me if you wish to keep some thoughts/comments private...
Wow, that is a lot of work, I guess some time far down in the road. So I can bypass the 5v and its opamp and directly input the +/-3.3v for the resistor rail? what if that is a large voltage diff between +/- 3.3v?
My digital input is xmos xu208 with i2s into the isolators powered by on board tps7a74 3.3v with 330mh choke and oscon in between. Any reason why one should bypass the isolator?
My existing battery for ak4495 set up is very simple and I am using a 10AH one. So those last a long time and I only have to worry recharging 3 batteries probably every other month. The sound from ak4495 is very pure and I doubt even the fully battery soekris can really over lapse it in any significant order, maybe just a different sound signature. But maybe far down the road, I will find out.
If the Soekris DAC still requires a serious battery supply installation and numerous other mods to show its full potential - I mean "significant improvements" is a word - I'd rather stay away from it for another while.
Thx for letting us know.
Though Im not using this dac but I believe that it can turn into something really really
good. What's not nice is that it's all smd which limits the amount of tweaks. Example
caps & regs. You guys may shoot me but different caps at different locations as well as
different regs have mega effect on Sq . For regs low noise , low impedance does not
assure great SQ. Pls check this thread : another look at Lm317 . This gentleman is on
to something. I've build this reg & compared to shunt regs on a buffer circuit. Guess
what Sq of the LM 317/LT1033 combi sounded much much better more organic if I may
say so with no hint of grain etc. yes its another can of worms that's open for arguement
but then again it's the spirit of diy & one really has to muck around to gain experience
& knowledge.
Cheers
good. What's not nice is that it's all smd which limits the amount of tweaks. Example
caps & regs. You guys may shoot me but different caps at different locations as well as
different regs have mega effect on Sq . For regs low noise , low impedance does not
assure great SQ. Pls check this thread : another look at Lm317 . This gentleman is on
to something. I've build this reg & compared to shunt regs on a buffer circuit. Guess
what Sq of the LM 317/LT1033 combi sounded much much better more organic if I may
say so with no hint of grain etc. yes its another can of worms that's open for arguement
but then again it's the spirit of diy & one really has to muck around to gain experience
& knowledge.
Cheers
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.