DcibeL said:
FWIW the 47pF/1K pole is probably high enough even with the changes I proposed to put it far enough out of the audio band as not to be an issue. I omitted to mention that I didn't intend to scale this network.
I don't like the implementation and will change it to the classical diff amp I outlined in the previous post. No special considerations are required for HF stability and the input poles while still not symmetrical are much closer.
I may also add the dc servo mentioned later..
kevinkr said:
I believe this circuit is often used in situations where the diff amp is expected to do all of the filtering in lieu of a separate LPF on the output. The values Zhaolu chose don't seem to be particularly effective based on the responses post. I would need to do further analysis to figure out what I would do were I to stick to this design which given my intentions to change to the circuit posted above probably ain't going to happen.. LOL
Member
Joined 2003
Hi Kevin,
Your circuit looks good. I'll simulate it when I get home from work to see how the response looks, but I think it will be okay.
I'll have to order some parts, so the modifications will be a while before I can have something to listen to, but here's what I have in mind:
-various modifications to the power supply, mainly fixing the grounding, replacing high voltage diodes with schottky diodes, replacing a cap or two...
-replacing tantalum caps on the analog supply to the AD1852 IC with good low ESR electrolytics
-remove ferrite beads from analog supply, and replace filter caps with good low ESR caps of proper voltage rating.
-fix the output filter
-insert my Kookaburra pre-amp into the chassis as a pre section, and power it from the Zhaolu supply instead of the simple onboard 78XX/79XX regulators. The only bad part of this idea is that my pre-amp is currently in a case that matches my power amp, and unfortunately the Zhaolu will not fit in this chassis, so I will have to put the pre into the Zhaolu chassis (in place of the mediocre headphone amp) and find a new use for the matching box.
Your circuit looks good. I'll simulate it when I get home from work to see how the response looks, but I think it will be okay.
I'll have to order some parts, so the modifications will be a while before I can have something to listen to, but here's what I have in mind:
-various modifications to the power supply, mainly fixing the grounding, replacing high voltage diodes with schottky diodes, replacing a cap or two...
-replacing tantalum caps on the analog supply to the AD1852 IC with good low ESR electrolytics
-remove ferrite beads from analog supply, and replace filter caps with good low ESR caps of proper voltage rating.
-fix the output filter
-insert my Kookaburra pre-amp into the chassis as a pre section, and power it from the Zhaolu supply instead of the simple onboard 78XX/79XX regulators. The only bad part of this idea is that my pre-amp is currently in a case that matches my power amp, and unfortunately the Zhaolu will not fit in this chassis, so I will have to put the pre into the Zhaolu chassis (in place of the mediocre headphone amp) and find a new use for the matching box.
Hi Dcibel,
I did run simulations on the new version of the circuit and it looked very good.
I have also come up with a version using the other half of the op-amp as a dc servo, and this looks very promising. I plan to use some of the salvaged 1uF Caps in the servo circuit hence the values chosen.
Incidentally the attenuation at 1MHz is roughly -17dB relative to 1kHz for both versions.
I'm going to implement the new audio stage design tonight if I have enough time before my wife and I go out for dinner. (About 3 hrs after I get home) Hope to have something positive to report.
Regarding your other changes they are mostly the same things I did to mine with the exception of moving the pre-amp into the Zhaolu. (My 26 dht based one won't fit..lol) Possibly you might want to hold off on that - I'm not convinced of the ultimate quality of the Zhaolu analog power supply. It's rather odd. I think you might get more bang for bucks just changing the LM78XX/79XX in your existing pre to LM317/337 plus that way you are not wed to the Zhaolu when possibly in a few more years you might want something else. Plus the matching issue you mentioned.
I did run simulations on the new version of the circuit and it looked very good.
I have also come up with a version using the other half of the op-amp as a dc servo, and this looks very promising. I plan to use some of the salvaged 1uF Caps in the servo circuit hence the values chosen.
Incidentally the attenuation at 1MHz is roughly -17dB relative to 1kHz for both versions.
I'm going to implement the new audio stage design tonight if I have enough time before my wife and I go out for dinner. (About 3 hrs after I get home) Hope to have something positive to report.
Regarding your other changes they are mostly the same things I did to mine with the exception of moving the pre-amp into the Zhaolu. (My 26 dht based one won't fit..lol) Possibly you might want to hold off on that - I'm not convinced of the ultimate quality of the Zhaolu analog power supply. It's rather odd. I think you might get more bang for bucks just changing the LM78XX/79XX in your existing pre to LM317/337 plus that way you are not wed to the Zhaolu when possibly in a few more years you might want something else. Plus the matching issue you mentioned.
Attachments
Member
Joined 2003
The DC servo I think I am going to skip, because I don't think it's the easiest thing to add on to this circuit, and I don't think it's really necessary in my application.
I'm not going to be wed to the Zhaolu in any way by putting the preamp in it. What goes in can come out. It should be real simple task of removing the regulators on the preamp, and wiring in the zhaolu supply on the pcb where the regulator output pin connects. I forgot my Kookaburra already has LM317/337 regulators on it that I built on a small perfboard. Should be interesting to compare.
I'm not going to be wed to the Zhaolu in any way by putting the preamp in it. What goes in can come out. It should be real simple task of removing the regulators on the preamp, and wiring in the zhaolu supply on the pcb where the regulator output pin connects. I forgot my Kookaburra already has LM317/337 regulators on it that I built on a small perfboard. Should be interesting to compare.
DcibeL said:
I took a look at the diodes in my 2.5C and decided to not upgrade them. They are actually fairly high quality, fast Vishay diodes (2A if I remember correctly). I have some Shottky's on hand but I didn't think I could really improve much on these Vishay's. Do you think it is worth the swap-out?
Most of the caps in my version, which is fairly recent, look to be decent Nichicon's. Not sure changing them out is going to prove worthwhile.
Bob
Hi Bob,
I couldn't identify what kind of diodes are fitted to my unit. There seems to be quite a lot of variation in components used. Mine had just 6 nichicons (4 audio, 2 ac fdbk in the +/-15V supplies) and all the rest were low end elna or worse. I did notice some significant improvement when I replaced them all with standard grade Black Gates.
I'm going to do the analog board surgery tonight hopefully (servo version) and see where that gets me.
I am definitely going the Twisted Pear route down the road, but as long as I don't spend more than another $20 or so on this dac I will keep tweaking to see whether there is more fruit, low hanging or otherwise to be had..
I couldn't identify what kind of diodes are fitted to my unit. There seems to be quite a lot of variation in components used. Mine had just 6 nichicons (4 audio, 2 ac fdbk in the +/-15V supplies) and all the rest were low end elna or worse. I did notice some significant improvement when I replaced them all with standard grade Black Gates.
I'm going to do the analog board surgery tonight hopefully (servo version) and see where that gets me.
I am definitely going the Twisted Pear route down the road, but as long as I don't spend more than another $20 or so on this dac I will keep tweaking to see whether there is more fruit, low hanging or otherwise to be had..
kevinkr and DcibeL
Does the AD need the passive filter?
You can always try the "marantz 63 thread" passive filter/buffer stage
http://home.quicknet.nl/qn/prive/ra.vdsteen/pdfs/CD_outputstage.pdf
allan
Does the AD need the passive filter?
You can always try the "marantz 63 thread" passive filter/buffer stage
http://home.quicknet.nl/qn/prive/ra.vdsteen/pdfs/CD_outputstage.pdf
allan
Well here is the latest update, and there seems to be some good news.
I modified the audio board using the non-servo design I posted. (I, uh didn't have any 1uF film caps for the servo..)
While I'm not going wild with enthusiasm, and no it isn't better than the Sony now - it does sound significantly more relaxed, there is still a hint of the leanness previously mentioned but there is a definite audible improvement right where I wanted it - tonality.
I might want to try something like a BB2132 or BB2604 both of which I have or the BB2107 which I will have to order.
The AD1852 does have a significant amount of HF stuff on its outputs - I can see it on my scope, although at this point I don't recollect the exact levels. I do recommend the LPF as implemented in either of the circuits posted.
Incidentally R9 should be about 3.92K
Edit:
Just put a vintage pair of PMI OP275 (BB has these now) and it seems like these might be a good choice. A little fuller and good detail.
I modified the audio board using the non-servo design I posted. (I, uh didn't have any 1uF film caps for the servo..)
While I'm not going wild with enthusiasm, and no it isn't better than the Sony now - it does sound significantly more relaxed, there is still a hint of the leanness previously mentioned but there is a definite audible improvement right where I wanted it - tonality.
I might want to try something like a BB2132 or BB2604 both of which I have or the BB2107 which I will have to order.
The AD1852 does have a significant amount of HF stuff on its outputs - I can see it on my scope, although at this point I don't recollect the exact levels. I do recommend the LPF as implemented in either of the circuits posted.
Incidentally R9 should be about 3.92K
Edit:
Just put a vintage pair of PMI OP275 (BB has these now) and it seems like these might be a good choice. A little fuller and good detail.
Well this is what I have settled on for the time being:
The LM4562NA are back in the audio board, there is now an OP275 in the voltage regulator circuit.
I ripped out the Wima 2.2nF and replaced them with Xicon polystyrene and foil 3.3nF which has an insignificant effect on the frequency response out to 20kHz and a couple of dB at 1MHz or so.
I never have been a fan of Wima caps, and I think there was a very small but audible improvement due to their replacement.
I think overall the configuration change was worthwhile.
The BB2604 sounded pretty decent, but overall I would give the edge to the LM4562. The OP275 was also quite good, I'd say just a bit fuller and more balanced sounding than the 2604, but a little less detailed than the 4562. Probably a pretty good choice actually.
Edit:
At the risk of being boring, this is definitely an improvement. As it warms up it seems like it might be major.
The LM4562NA are back in the audio board, there is now an OP275 in the voltage regulator circuit.
I ripped out the Wima 2.2nF and replaced them with Xicon polystyrene and foil 3.3nF which has an insignificant effect on the frequency response out to 20kHz and a couple of dB at 1MHz or so.
I never have been a fan of Wima caps, and I think there was a very small but audible improvement due to their replacement.
I think overall the configuration change was worthwhile.
The BB2604 sounded pretty decent, but overall I would give the edge to the LM4562. The OP275 was also quite good, I'd say just a bit fuller and more balanced sounding than the 2604, but a little less detailed than the 4562. Probably a pretty good choice actually.
Edit:
At the risk of being boring, this is definitely an improvement. As it warms up it seems like it might be major.
Kevin, it sounds like another breakthrough. Care to post a picture of the mod? I know you've done a lot to this board, but looking through the recent threads(for ma at least) makes it a bit difficult to pick out exactly which components were removed, jumpered or changed in value.
Could you summarize in a concise list for me, please? I'm not an engineer, and I can follow a schematic, but I wouldn't know off hand if I was doing something wrong per your diagrams.
Thanks for all your efforts on improving this DAC,
Bob
Could you summarize in a concise list for me, please? I'm not an engineer, and I can follow a schematic, but I wouldn't know off hand if I was doing something wrong per your diagrams.
Thanks for all your efforts on improving this DAC,
Bob
Hi Bob,
Yes I'll try to post a picture of the mod in the next day or so.. And FWIW it definitely is an improvement, enough so that I now can believe that a zapfilter would be a major improvement over the existing audio board. It seems that despite all of the previous mods there was a real problem with the basic design of the audio board, in addition to the ones I identified on the psu pcb.
I suspect that the AD1852 is not all that happy driving a 1K load, it is much happier with the 2K load it now drives. The topology change and the elimination of an op-amp from the path also help a lot.
The dc servo can be probably be considered optional, the dc offsets present at the output of my audio board have not caused undue problems with my stepped attenuators. I'll probably add it at some point in the future just to get rid of the offset.
The sound is much fuller and more pleasing, the irksome thinness is almost completely gone, and as it warms up it fleshes out further. It is not quite in the same league as the Sony, but the differences are notnearly as glaring as before.
Yes I'll try to post a picture of the mod in the next day or so.. And FWIW it definitely is an improvement, enough so that I now can believe that a zapfilter would be a major improvement over the existing audio board. It seems that despite all of the previous mods there was a real problem with the basic design of the audio board, in addition to the ones I identified on the psu pcb.
I suspect that the AD1852 is not all that happy driving a 1K load, it is much happier with the 2K load it now drives. The topology change and the elimination of an op-amp from the path also help a lot.
The dc servo can be probably be considered optional, the dc offsets present at the output of my audio board have not caused undue problems with my stepped attenuators. I'll probably add it at some point in the future just to get rid of the offset.
The sound is much fuller and more pleasing, the irksome thinness is almost completely gone, and as it warms up it fleshes out further. It is not quite in the same league as the Sony, but the differences are notnearly as glaring as before.
I think that's its major flaw right now. There's an overabundance of detail on the top, causing the mids and bottom end to sound thin and undynamic. I think the OPA2107 helps in this regard, as I find the LM4562 to be even thinner sounding.
I compared my Zhaolu this morning to a friend's Theta Gen III and that was definitely the missing piece. The Zhaolu was thin soounding and overly detailed. Not balanced top to bottom. The Theta was just natural everywhere, and powerful too. Imaging and soundstaging were about the same, so I think if we can find a way to warm up the Zhaolu a little it may be the ticket.
I have an AD823 in the power supply, but I may try going back to the AD08 (I think that was what was in there before) and see what the true difference is there now. Sometimes going back brings out more differences than always going forward.
Perhaps a Zapfilter is in my future but first I want to see if your newest changes make up for the difference.
Bob
I compared my Zhaolu this morning to a friend's Theta Gen III and that was definitely the missing piece. The Zhaolu was thin soounding and overly detailed. Not balanced top to bottom. The Theta was just natural everywhere, and powerful too. Imaging and soundstaging were about the same, so I think if we can find a way to warm up the Zhaolu a little it may be the ticket.
I have an AD823 in the power supply, but I may try going back to the AD08 (I think that was what was in there before) and see what the true difference is there now. Sometimes going back brings out more differences than always going forward.
Perhaps a Zapfilter is in my future but first I want to see if your newest changes make up for the difference.
Bob
Now that I think on it I have those 4 analogue input coupling caps jumpered. Perhaps if I replaced them with something else, or removed the jumpers I might cut back on the over extension on the top.
Anyone have a suggestion on what to replace them with? I'd like to put some poly caps in there or something decent. Then again, perhaps there should be a decent quality coupling cap on the output of each channel just at the RCA's instead of coming into the analogue stage.
Hmmmm,
Bob
Anyone have a suggestion on what to replace them with? I'd like to put some poly caps in there or something decent. Then again, perhaps there should be a decent quality coupling cap on the output of each channel just at the RCA's instead of coming into the analogue stage.
Hmmmm,
Bob
Hi Nick,
You'll recognize the dac because it doesn't look all that different.. There are fewer parts on the top of the audio board than before, and a few more on the bottom.. LOL Plus I cleaned it carefully.
It doesn't sound much like it did before, much more top to bottom balance. The overall sound is much more cohesive, it kept what I liked about it before, but it is so much better it is hard to recognize it as being the same dac of a few days ago. Unlike the last set of mods where the next time I heard it I realized it still sounded bad (when Ryan was here with the Benchmark) this seems to be the real mccoy - no it isn't perfect, but it really did improve, and this is an ongoing impression I keep getting when I listen to it.
Hi Bob,
I have no coupling caps anywhere in the signal path and I changed out the Wimas for some Xicon polystyrene caps. I' m using the LM4562 for audio and an OP275 in the power supply and I'm pleased. I'll try to post a picture tomorrow. You'll find the changes worthwhile. I'm not planning on a Zapfilter incidentally, it is just that I now I can tell that the digital section is good enough to justify it - I think this particular mod provides a good level of improvement for minimal investment.
You'll recognize the dac because it doesn't look all that different.. There are fewer parts on the top of the audio board than before, and a few more on the bottom.. LOL Plus I cleaned it carefully.
It doesn't sound much like it did before, much more top to bottom balance. The overall sound is much more cohesive, it kept what I liked about it before, but it is so much better it is hard to recognize it as being the same dac of a few days ago. Unlike the last set of mods where the next time I heard it I realized it still sounded bad (when Ryan was here with the Benchmark) this seems to be the real mccoy - no it isn't perfect, but it really did improve, and this is an ongoing impression I keep getting when I listen to it.
Hi Bob,
I have no coupling caps anywhere in the signal path and I changed out the Wimas for some Xicon polystyrene caps. I' m using the LM4562 for audio and an OP275 in the power supply and I'm pleased. I'll try to post a picture tomorrow. You'll find the changes worthwhile. I'm not planning on a Zapfilter incidentally, it is just that I now I can tell that the digital section is good enough to justify it - I think this particular mod provides a good level of improvement for minimal investment.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.