That's definitely supposed to make me feel better about my build... Sigh, it does make sense and I believe you...
The other major heat sources in my build are the two transformers (12V 25W toroidal ~45C, SK Lite PCB transformer, possibly split-bobbin type, ~50C) and UltraBiB. Do you know if anyone ever found out whether the power draw is really constant? I put in 50% for peak draw as Soren suggested earlier in the thread but if I can cut on that maybe it'll cool the ladders by a couple of degrees due to a slightly lower internal temp.
Also, do you know if anyone tried to remove the power filter caps (and bridge rectifier)? I feel that more power output capacitance should probably be on UltraBiB not dam1021. Seems somewhat important with so much speculation on the sound of caps.
You're so polite. Are you sure you don't want to try adding polymer caps to 1121?... I'm not 100% sure why it helps so much and for a while I was very worried that the greatly improved naturalness was due to noise somehow introduced by the caps (not that it sounds very scientific...). But when I stopped holding my breath and listened more carefully, I realized that all the micro-details are there, only perhaps I was accustomed to the harshness in the sound that's so easily taken for details.
I think the word you are looking for is "frank". I have done the cap mod on my Rev 1 1021. From measurements presented in this thread, it was evident that it could have some impact. I also isolate/damp the clocks physically. And use proper power - tried SLA battery but really didn't care for it in the end. Most, if not all of your ideas has already been discussed and tried - read the thread if you have the guts - at least up to say - rev2. Is your 1021 the latest rev?
I'm contemplating caps on the 1121 but not for now. Out of the box, I felt 1121 superior to my moded 1021. I ran it in balanced mode for a while but is back on SE - want to revisit balanced at a later time.
In tests, again and again, distorsion is preferred before clean. I believe it's a relief, not a cure. This is not the route I wish to pursue. The best DAC to come is one that never needed any auditioning - this is why a did put my bet on the DAM - I like and believe in the technical approach. It felt like the DAM was made in this way - but alas, it was not as the measurements revealed (1021 R1).
Still I enjoy it a lot and it's the best DAC I have owned. And when not I think about my speakers as a source for a lot more errors than my DAC and think about how to correct them
//
Last edited:
I think the word you are looking for is "frank". I have done the cap mod on my Rev 1 1021. From measurements presented in this thread, it was evident that it could have some impact. I also isolate/damp the clocks physically. And use proper power - tried SLA battery but really didn't care for it in the end. Most, if not all of your ideas has already been discussed and tried - read the thread if you have the guts - at least up to say - rev2. Is your 1021 the latest rev?
I'm contemplating caps on the 1121 but not for now. Out of the box, I felt 1121 superior to my moded 1021. I ran it in balanced mode for a while but is back on SE - want to revisit balanced at a later time.
In tests, again and again, distorsion is preferred before clean. I believe it's a relief, not a cure. This is not the route I wish to pursue. The best DAC to come is one that never needed any auditioning - this is why a did put my bet on the DAM - I like and believe in the technical approach. It felt like the DAM was made in this way - but alas, it was not as the measurements revealed (1021 R1).
Still I enjoy it a lot and it's the best DAC I have owned. And when not I think about my speakers as a source for a lot more errors than my DAC and think about how to correct them
//
Nope I wasn’t proposing anything new. But it appears that we all believed 1000uf per rail was enough in rev1/2. And rev4/5 need absolutely none extra, like all commercial line products and 1121, which has 10*100uf per rail at least. Living sounds changed my mind and I’m hearing large impacts of the mod. I’m not sure how it’d show up on measurements though - the posted white noise playback was confusing at best... Was stock+2200uf better than factory mod or worse?... living sounds also said measurements didn’t change much even though recorded sound did, I’m assuming the measurements.
As for thermals... definitely no fans, but maybe I’ll test how much help some cutouts can be... but like Soren said there’s probably only limited improvements to be had.
Last edited:
You will also tend to love your own efforts
//
Don’t worry my attachment to the mods is nothing compared to the other distortions in my thinking... at one point I was convinced that my laptop had tighter bass and brighter highs than the modded dam1021, but when I finally relaxed and just listened, I realized dam1021 sounded way more natural, dynamic, and had more low-level details; the perceived advantages of my laptop are likely to be artifacts, though somewhat impressive when first listened to. I worried about other things before the mod, mostly related to “harshness”. I remember last year when I shorted the MOSFET on my sigma22 I was really upset but trying my laptop audio again it almost sounded better, more natural, and less fatiguing. I can’t see a reason to doubt the perceived differences. Now, it’s almost the other way around, with dam1021 clearly more natural, and now also much more dynamic and involving. I’m no audiophile and I don’t intend on becoming one. So it would be really funny if those observations turn out to be products of my imagination, and I’ll probably never believe my ears ever again, maybe stop worrying about equipment altogether...
How would you test for distortion vs improvement without an analyzer? How could the caps, properly installed, introduce distortions?....
P.S. your signature line finally makes sense now... I agree, except sometimes not even the distortion is audible... And thanks for reminding me that cures should be in most cases preferable to reliefs. Things get confusing when cures are not an option and relief is actually desirable, easy to forget when to start looking for cures again
Last edited:
Nope it's rev4 but close enough. Added 3*1500uF polymer caps (to the original 3*100uF + 47uF). Unbelievable change.
Same. A bit of a let down but it was mostly "misjudgement" on my part. If it was truly fully justified many professional engineers and researchers developing sigma-delta chips and algorithms and whom care about sound would be very very disappointed...
But still, dam1021 (at least later revisions) sounds very lovely
Sorry to burst your bubble but you show all the telltale symptoms of the audiophile
thanks for the diagnosis...
I'm very very happy with the sound of the dac1101, several small improvements, although the biggest improvement compared to a dam1021 is the new output buffer / headphone amplifier, in my humble opinion it's kinda fantastic, can drive 4V RMS into 32 ohm from just +-6.5V, fully discrete, no nfb, high bias class AB, my Sennheiser HD650 have never sounded better.... I now need to upgrade my headphones, maybe the new HD800 S, good to have a company to pay
I'm considering making an add-on balanced output buffer / headphone amplifier for the dam1021 based on that amplifier design, any interest ?
Soren, is this still going to happen...?
The slightly lower drive voltage alone would be great news for the toasty dam1021 3.3V LDO...This could also be very valuable to 1941 users needing proper headphone amplification and easy integration
Last edited:
So I ordered the polymer caps, really curious what they will do. There are so many variables in these dac projects... Somehow the other day the USB receiver board died, I seem to have bad karma. Fortunately I had another one, the Waveio, and installed it.
Big difference in sound, was surprised by that. Don't know if better or worse, just different. But although I use batteries and Sakas shunts, there is still something in the sound which makes me constantly thinking about more possible tweaks to remove that artificial taste in the sound which gives me a headache.
During the time the USB was not working I switched on my cheap Chinese SD card player, which is running a AK4495 chip with output transformers, also tweaked in various ways.
Unfortunately it is doing much better in most musical departments than the Soekris, it just sounds so much more natural and easy and with a sense of stillness between the notes. I don't think about the next tweak, I think about which music next.
Is it possible that it is the crappy USB connection which makes all the difference? The big difference between the USB boards made me think.
I am really curious how the Soekris would sound with an SD player, maybe that's the holy grail, who knows.
But first the Vref caps...
Big difference in sound, was surprised by that. Don't know if better or worse, just different. But although I use batteries and Sakas shunts, there is still something in the sound which makes me constantly thinking about more possible tweaks to remove that artificial taste in the sound which gives me a headache.
During the time the USB was not working I switched on my cheap Chinese SD card player, which is running a AK4495 chip with output transformers, also tweaked in various ways.
Unfortunately it is doing much better in most musical departments than the Soekris, it just sounds so much more natural and easy and with a sense of stillness between the notes. I don't think about the next tweak, I think about which music next.
Is it possible that it is the crappy USB connection which makes all the difference? The big difference between the USB boards made me think.
I am really curious how the Soekris would sound with an SD player, maybe that's the holy grail, who knows.
But first the Vref caps...
Soren, is this still going to happen...?
This could also be very valuable to 1941 users needing proper headphone amplification and easy integration
Not a priority right now, so nothing short term.
I'm sorry to hear that your USB card died... But are you sure all wiring is identical?... I recall spikestabber reporting improvements going from Amanero to Singxer XMOS but Soren said it's impossible which makes sense theoretically...
Good luck on the vref mod!
P.S. Your description of how we think about "next tweaks" is very interesting
Last edited:
Yes, I know a dac or Amp or whatever sounds right when I don't think about tweaking.And yes, theoretically USB modules should sound same, also because of reclocking and all that. But reality doesn't care about theories. To me a theory needs to describe reality, and if it doesn't, we have to question that theory and adjust it. Means, opening the door in the mind to accept, that we probably don't know what's going on.
For example... I use the isolated output of the Waveio. But still it makes a substantial improvement when I cut the ground and v+ wire of the USB cable from laptop to Waveio after handshake. BTW a worthwhile tweak
, when your USB board has external power.I'm sorry to hear that your USB card died... But are you sure all wiring is identical?... I recall spikestabber reporting improvements going from Amanero to Singxer XMOS but Soren said it's impossible which makes sense theoretically...
Good luck on the vref mod!
P.S. Your description of how we think about "next tweaks" is very interesting
That is another can of worms, but the general consensus is that each USB interface actually does sound different, everyone seems to have their own theory as to why. I know the Singxer has significantly improved power circuitry so that could be one aspect.
Yes, dramatic differences.
Surprisingly few of the usb boards are fit for musical consumption. Amanero and diyinhk were perhaps the worst i've heard, while EXA and JLSounds being the best. A fifo does nothing to decrease these subjective differences.
The changes due to the VREF mod are not subtle.
I do think total capacitance matters more than all of it being ultra-low-ESR: The extra added caps I used are 8200 uf Panasonic FC. Not the lowest ESR and their size puts them at a disadvantage there. Plus I used the original several cm long leads. Still, adding them to the already VREF-modded rails made another very audible improvement.
With the somewhat extensive power supply filtering and using the 4k soft filter I can listen to this DAC in the studio all day without fatigue. It paints an accurate picture and is dynamically alive in a way most DACs aren't.
The linear filter has more high end, but sounds more contrieved, more stressfull and dynamically constrained than the soft filter to my ears.
A good indicator is that I make much better mix decisions with the way the DAC is now. And signal originating from it takes processing better.
I do think total capacitance matters more than all of it being ultra-low-ESR: The extra added caps I used are 8200 uf Panasonic FC. Not the lowest ESR and their size puts them at a disadvantage there. Plus I used the original several cm long leads. Still, adding them to the already VREF-modded rails made another very audible improvement.
With the somewhat extensive power supply filtering and using the 4k soft filter I can listen to this DAC in the studio all day without fatigue. It paints an accurate picture and is dynamically alive in a way most DACs aren't.
The linear filter has more high end, but sounds more contrieved, more stressfull and dynamically constrained than the soft filter to my ears.
A good indicator is that I make much better mix decisions with the way the DAC is now. And signal originating from it takes processing better.
For example... I use the isolated output of the Waveio. But still it makes a substantial improvement when I cut the ground and v+ wire of the USB cable from laptop to Waveio after handshake. BTW a worthwhile tweak
Then I should also mention that I did the vref mod at the same time as I finally got rid of the triple core I2S cable for proper cat7 cable. I didn't feel comfortable using the shield as GND (not that any good DIYer should do that...) even though the GND connection helped me get rid of all pop/clicks I believe. I also swapped the Amanero copy for a recent Amanero, and replaced USB 5V supply with ultra-low noise regulated external supply.
However, I don't believe that the USB card can change the sound. I only hoped for more stability in my USB mods, not SQ gain. Also nice to do things proper so as to eliminate weak links when problem does arise. Soren was very confident that properly connected USB cannot impact sound, and he knows the FIFO implementation. Theoretical predictions here seem complete, which is different with the vref mod. Perhaps the next step is to get some measurements from each USB interface and see if anything shows up?... I'm inclined to trust our best theoretical conclusions more than subjective reports on this issue. If sound does change with I2S input through isolated FIFO reclocker, it's probably either Soren drew a wire wrong, or we found something publication-worthy...
Not subtle at all, worth reiterating to those who are just starting to consider this mod...
You do have 4 electrolytic caps in parallel each rail so I'm not sure if we can rule out low ESR as beneficial, but good to know that we don't need absolutely minimal ESR to get really good results (or maybe we should lament the high impedance of the shift registers limiting performance? I don't even want to ask Soren if the shift registers on the newer models are better... they look cuter...).
I've been using 4k linear with 1.20 FPGA and enjoyed it a lot, but I just finally A/B'ed between linear and soft and I hear the differences you describe. Softer, more dynamic and smoother sound. On HD650 it can sound scarily smooth, but the details and high end are still there. I would not believe it if someone called this fatiguing or overly analytical. I just wish I have a pair of SDR-modded HD800 in front of me to try out...
Last edited:
Also keep in mind that with a similar configuration last year, except with adapted sigma22 PSU which might be way worse than UltraBiB and without the vref caps on rev.4, the sound was unmistakably harsh/analytical (without reference to the "right" sound) to my ears, and I remember agreeing in my mind with Soren that Soft filters are not soft enough....
Btw, 0.01% of 4V is 0.4mV. Perhaps Soren wasn't ambitious enough with the 1mV ripple target in the factory mod?... When you have inevitable 0.01% error in one place, you would want to eliminate problems everywhere else to keep error at 0.01%, not set an error tolerance of 0.01% so 10 things add (if not multiple) into a bigger error. Of course, not that many things have such an important place in the system, but maybe Vref does. It's also crazy that this simple question has been discussed for over 2 years now...
Btw, 0.01% of 4V is 0.4mV. Perhaps Soren wasn't ambitious enough with the 1mV ripple target in the factory mod?... When you have inevitable 0.01% error in one place, you would want to eliminate problems everywhere else to keep error at 0.01%, not set an error tolerance of 0.01% so 10 things add (if not multiple) into a bigger error. Of course, not that many things have such an important place in the system, but maybe Vref does. It's also crazy that this simple question has been discussed for over 2 years now...
Last edited: