Hi Allan,
Yes, I know it sounds like overkill, but over the years and in both solid state and valve/tube systems I have found that even if the output stage calls for say 30VA, a 180-300VA has ALWAYS made a substantial positive difference.
These benefits have normally come in the form of tighter/deeper bass, better soundstaging, better seperation and a cleaner overall presentation. And going back to the independant 30VA for the analog stage resulted in the benefits still being present vs using just one transformer for both analog and digital, but in considerably reduced form.
Notice that the likes of NAIM and others have made lots of money selling HICAP and SUPERCAP power supplies which do little else but power the analog stage from a bog-standard 317/337 based supply which has a 300VA-600VA (!!!) transformer attached.
Try it and see! And reports your results here!
Yes, I know it sounds like overkill, but over the years and in both solid state and valve/tube systems I have found that even if the output stage calls for say 30VA, a 180-300VA has ALWAYS made a substantial positive difference.
These benefits have normally come in the form of tighter/deeper bass, better soundstaging, better seperation and a cleaner overall presentation. And going back to the independant 30VA for the analog stage resulted in the benefits still being present vs using just one transformer for both analog and digital, but in considerably reduced form.
Notice that the likes of NAIM and others have made lots of money selling HICAP and SUPERCAP power supplies which do little else but power the analog stage from a bog-standard 317/337 based supply which has a 300VA-600VA (!!!) transformer attached.
Try it and see! And reports your results here!
deuginthesky said:
I have compared D-2.0A v.s. D-2.5A and found audible differences. D-2.0A's is more refined, hence more musical. I think it is due to a better power supply circuit of the D-2.0 unit. However, D-2.5 owners may not notice this out of good sonics they hear.
I doubt the D-3.0 would sound better. It is more towards PC audio gear with the new TG-Link. Adding more components increases more noise and interfers the power supply, IMHO.
Interesting findings....
Did some surgery on the Zhaolu 2.5 DAC (CS DAC using LM4xxx opamps):
The power supply for the opamps is feeding them +-9V! I found this surprisingly low, since these devices normally sound better when run close to their maximum (which in this case is +-17V). In fact that data sheet for the device shows distinct improvements in THD when the device is run at 17V rather than 15V.
So I whipped up a 300VA fed dual power supply using LM317/LM337 set for 15V and simply cut the power supply feed from the Zhaolu's PS board to the opamps/analog stage and plugged my own PS straight in.
The GOOD: Improvements in seperation, clarity/cleanness, soundstage, presence, bass appears tighter and more detailed, but STILL loose compare with the Gen Va (which is UNmodified). I must emphasis that this unit sounds amazingly good. To the extent that I find myself re-considering some digital projects I had planned simply because they'd take more time/money and I'd probably not end up with something sounding half as good for the money/time spent.
The BAD: I initially wired the mains input wires for the 300VA onto the Zhaolu's front panel switch. However, it seems not to be rated for the load and started behaving strangely. So I hardwired the AC inputs for the 300VA to the mains, which means that the analog stage is always on.
A worthwhile, if risky tweak. Not to be attempted if you're not familiar with dealing with AC mains voltages.
I'll shift the opamp voltages to 17V (currently at 15V) and report back.
Has anyone bypassed the relays and does that have much impact on the sound?
What about feeding the output from just the first stage of the dual opamp to the output?
Did some surgery on the Zhaolu 2.5 DAC (CS DAC using LM4xxx opamps):
The power supply for the opamps is feeding them +-9V! I found this surprisingly low, since these devices normally sound better when run close to their maximum (which in this case is +-17V). In fact that data sheet for the device shows distinct improvements in THD when the device is run at 17V rather than 15V.
So I whipped up a 300VA fed dual power supply using LM317/LM337 set for 15V and simply cut the power supply feed from the Zhaolu's PS board to the opamps/analog stage and plugged my own PS straight in.
The GOOD: Improvements in seperation, clarity/cleanness, soundstage, presence, bass appears tighter and more detailed, but STILL loose compare with the Gen Va (which is UNmodified). I must emphasis that this unit sounds amazingly good. To the extent that I find myself re-considering some digital projects I had planned simply because they'd take more time/money and I'd probably not end up with something sounding half as good for the money/time spent.
The BAD: I initially wired the mains input wires for the 300VA onto the Zhaolu's front panel switch. However, it seems not to be rated for the load and started behaving strangely. So I hardwired the AC inputs for the 300VA to the mains, which means that the analog stage is always on.
A worthwhile, if risky tweak. Not to be attempted if you're not familiar with dealing with AC mains voltages.
I'll shift the opamp voltages to 17V (currently at 15V) and report back.
Has anyone bypassed the relays and does that have much impact on the sound?
What about feeding the output from just the first stage of the dual opamp to the output?
Hi Ryan
Here is what i did to the 2.5.
- shorted the Muse coupling caps
- bypassed the second follower opamp. This involved changing a resistor value (i think from 2k to 200R and the corresponding filtering cap) so now the output impedance is 200R
- bypassed output relays - no untoward noises and clearly better sound
- used 4.7uF BG N at output
- got rid of the dreadful RCA block and put some decent jacks in place - as much improvement as any of the above.
A simple idea: if you are very allergic to caps but less so to opamps you can use the second opamp to compensate for the dac offset. No idea how it is with the CS chip but the AD has a very stable offset.
Btw your PS is likely defective, mine outputs exactly 15v.
Here is what i did to the 2.5.
- shorted the Muse coupling caps
- bypassed the second follower opamp. This involved changing a resistor value (i think from 2k to 200R and the corresponding filtering cap) so now the output impedance is 200R
- bypassed output relays - no untoward noises and clearly better sound
- used 4.7uF BG N at output
- got rid of the dreadful RCA block and put some decent jacks in place - as much improvement as any of the above.
A simple idea: if you are very allergic to caps but less so to opamps
you can use the second opamp to compensate for the dac offset. No idea how it is with the CS chip but the AD has a very stable offset.Btw your PS is likely defective, mine outputs exactly 15v.
Other fertile areas for discovery
The filter pin on the CS8416 will probably benefit as much from the wildmonkeysects RC setup as the CS8412 did.
What about the Vq, Va and Filt pins on the CS4398 DAC? Has anyone tried feeding the DAC with a separate analog power supply?
Biasing opamps into class A using JFETS?
BNC connector at the SPDIF input?
Clean PS to the 11.xxx clock?
I've got a few ALW (Andy Weekes) ultra low noise power supplies to play with, so I'll post results on some of these ideas over the next few weeks...
The filter pin on the CS8416 will probably benefit as much from the wildmonkeysects RC setup as the CS8412 did.
What about the Vq, Va and Filt pins on the CS4398 DAC? Has anyone tried feeding the DAC with a separate analog power supply?
Biasing opamps into class A using JFETS?
BNC connector at the SPDIF input?
Clean PS to the 11.xxx clock?
I've got a few ALW (Andy Weekes) ultra low noise power supplies to play with, so I'll post results on some of these ideas over the next few weeks...
analog_sa said:
Seems likely, mine also measures almost exactly +/-15V.
I'm going to try some of the other mods you guys have mentioned.
I have BG bipolar in place of the original caps from the dac chip, imo these caps might not be any better than the original ones in there.
Huge difference with LM4562 opamps over anything else I have tried.
Mine is the 2.5A with the AD1852 dac and I was initially going to buy the 3, but will now wait as it uses the CS chip..
I did do some tinkering on my 2.5A tonight with some interesting conclusions.
First I bypassed the input capacitors on the analog board, and have to admit I didn't like the result at all, and perhaps I am the only one here who found this less than scintillating.. I heard considerably increased sibilance on female vocals, edginess on the highs and a collapse in the depth of the sound stage. I am not sure of the effect several volts of common mode dc has on the inputs of the LM4562, theoretically none, but then who knows - it certainly did not sound better to me. It didn't affect my line stage which uses stepped attenuators and fixed bias (so there is a blocking cap before the 26 grid in the line stage.) Not sure what root cause was..
I quickly removed the jumpers and then re listened and the effect was gone.
My speaker system uses JBL 2702 annular ring radiators (horns) and 4333 mids (also horns) so it is very sensitive to small changes in mid and treble presentation.
Previous mods to this board included changing to LM4562 op amps (huge improvement) and replacing the nichicon muse with black gate n type caps - which apparently take some time to break in as I compared notes from when I first installed them and my impression now is much more positive.
I did disconnect the power to the headphone amplifier and disconnected all of the connectors from the analog board except the power and the digital board connection. This did seem to make a difference, perhaps a little more clarity, subtle improvements to imaging as well.
First I bypassed the input capacitors on the analog board, and have to admit I didn't like the result at all, and perhaps I am the only one here who found this less than scintillating..
I heard considerably increased sibilance on female vocals, edginess on the highs and a collapse in the depth of the sound stage. I am not sure of the effect several volts of common mode dc has on the inputs of the LM4562, theoretically none, but then who knows - it certainly did not sound better to me. It didn't affect my line stage which uses stepped attenuators and fixed bias (so there is a blocking cap before the 26 grid in the line stage.) Not sure what root cause was..I quickly removed the jumpers and then re listened and the effect was gone.
My speaker system uses JBL 2702 annular ring radiators (horns) and 4333 mids (also horns) so it is very sensitive to small changes in mid and treble presentation.
Previous mods to this board included changing to LM4562 op amps (huge improvement) and replacing the nichicon muse with black gate n type caps - which apparently take some time to break in as I compared notes from when I first installed them and my impression now is much more positive.
I did disconnect the power to the headphone amplifier and disconnected all of the connectors from the analog board except the power and the digital board connection. This did seem to make a difference, perhaps a little more clarity, subtle improvements to imaging as well.
YMMV
I second your findings... reading your post reminded me that something was " not quite right" since i did the dc block caps mod; and i undo it. It's "better"( i.e. impressive again) now, in my sytem.
( sony xb 930+2.5+dact ct 101+genelec 1031a). Don't know why,maybe because the gen's seem to have a dc block only on the tweeters ( 15u). Anyhow, bass is much better, but mids a bit recessed on vocals ( way deep in the sounstage)
I second your findings... reading your post reminded me that something was " not quite right" since i did the dc block caps mod; and i undo it. It's "better"( i.e. impressive again) now, in my sytem.
( sony xb 930+2.5+dact ct 101+genelec 1031a). Don't know why,maybe because the gen's seem to have a dc block only on the tweeters ( 15u). Anyhow, bass is much better, but mids a bit recessed on vocals ( way deep in the sounstage)
A different mod...
I have the Zhaolu 2.5C with the CS DAC, and previously have done a few mods, including the basic bypass of the DAC coupling caps, using LM4562 op-amps, and using Schottky diodes in the power supply. All have improved the sound over the stock unit, and was very satisfying, musically and sonically.
However, today, I put together a replacement output stage consisting of a of CineMag CMLI-15B 1:1 transformer for each channel coupling the differential DAC outputs to single ended feeding to an Intersil HA5002 buffer, which is direct coupled to the output. No capacitors, the transformers have a similar bandpass as the original filter network, so no anti-aliasing filter required.
The sound is incredibly open and clear. The microdynamics are another level over the previous mods, and I am hearing low level detail I have never heard before in digital. I do not have a circuit diagram to post, but it is conceptually simple.
This is a simple but worthwhile mod, the only drawback being the cost of the transformers (~$100, direct from CineMag). However I feel this takes the unit to a pretty high level of performance.
I have the Zhaolu 2.5C with the CS DAC, and previously have done a few mods, including the basic bypass of the DAC coupling caps, using LM4562 op-amps, and using Schottky diodes in the power supply. All have improved the sound over the stock unit, and was very satisfying, musically and sonically.
However, today, I put together a replacement output stage consisting of a of CineMag CMLI-15B 1:1 transformer for each channel coupling the differential DAC outputs to single ended feeding to an Intersil HA5002 buffer, which is direct coupled to the output. No capacitors, the transformers have a similar bandpass as the original filter network, so no anti-aliasing filter required.
The sound is incredibly open and clear. The microdynamics are another level over the previous mods, and I am hearing low level detail I have never heard before in digital. I do not have a circuit diagram to post, but it is conceptually simple.
This is a simple but worthwhile mod, the only drawback being the cost of the transformers (~$100, direct from CineMag). However I feel this takes the unit to a pretty high level of performance.
analog_sa said:
The CineMags are about 3 cm in diameter, and about 3 cm tall. I got the PC terminal version, and mounted them on a piece of Vector board cut the same size as the analog board, along with the buffer. DC shouldn't be a problem with the Crystal DAC since it has a low DC offset between the differential outputs, I understand the AD version has a higher offset that could pose a concern.
I will have to try bypassing the buffer, however I intentionally went with a buffer in the design because I am driving into a transformer volume control passive (TVC) instead of a linestage, and I didn't think the opamp in DAC was capable driving through two transformers and 3 meters of cable to my amp inputs.
David
Went looking in HK
Walked around Shim Sha Pui, saw a lot of audio stores, but amazingly didn't see any Zhaolu DACs.
I was walking in that area near dragon center, where the street stalls were set up. Saw a lot of stores including wing shin i think. This was after xmas.
All sold out? or did I miss something? Only my second visit to hong kong. was still disoriented.
Walked around Shim Sha Pui, saw a lot of audio stores, but amazingly didn't see any Zhaolu DACs.
I was walking in that area near dragon center, where the street stalls were set up. Saw a lot of stores including wing shin i think. This was after xmas.
All sold out? or did I miss something? Only my second visit to hong kong. was still disoriented.
Diykits still had 2.5C in stock the last time I checked, the current model is the 3.0..
http://www.diykits.com.hk/
http://www.diykits.com.hk/
kevinkr said:
I got my Zhaolu D2.0 from Eddie two years ago. I would suggest you talk with him when he will be at his store. It is easy to go there, but you may not meet him or you may need to wait for some mod that Eddie can do for you.
BTW, I am sure you will enjoy listening music at his store. It is one of the best acoustic listening rooms I have visited, although it looks like a work shop.
I just picked up a Zhaolu D3. It's a wonderful DAC especially at the price. My CDP is a California Audiolab Icon II. It has a pretty decent DAC despite its age. I have also been using the DAC in my Outlaw Audio preamp. It uses the CS4396 chip and is an improvement over the Icon II. The Zhaolu D3 is another step up in performance with better detail and clarity. I have run into a problem though which I can't seem to resolve. When I plug my BuffaloTech DVD network player in the the Zhaolu, I don't get any sound. I have some 24/96 disc that I like to play. The BuffaloTech plays fine through the Outlaw which identifies the input stream as 96kHz PCM. I have also tried a Toshiba DVD player and a Hitachi DVD player with the Zhaolu and they play fine through either DAC. I also tried a cheap Koss DVD player which also wouldn't play through the Zhaolu but would through the Outlaw. I've always though that PCM outputs were pretty standardized. Any ideas?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.