As promised here are the very subjective results of the casual listening tests comparing the Benchmark Dac 1 with the Zhaolu 2.5A.
Some notes:
16/44 and 24/96 material furnished by the media server via an m-audio 2496. 24/96 material was downloaded from Linn Records in the past.
16/44 also provided by the Sony and delivered by coax to the Zhaolu and the Benchmark Dac 1. (This appears to be an early version ~ 3yrs old)
Most 16/44 material was from Il Sogno, the 24/96 material was from the Linn Classical library. (I can provide the specific file name, don't have it to hand here.)
Spdif coax connections were used for all tests.
Audio signal paths in all devices compared use op-amps. (No fancy discrete I/V etc.)
Source - Sony utilized as transport
The Benchmark sounds fuller and tonally balanced, good level of detail, smooth, extended highs.. Good sound stage depth, and width. Did I say smooth?
The Zhaolu by comparison sounds wiry, thin and strident. Much more ascetic (yes vinegary) and a bit grating on the nerves. Paper thin imaging compared to the Sony, very lean and wiry compared to the Benchmark.
The Sony has the most sound stage depth and width, great ambiance recovery and detail, leaner tonally than the Benchmark. Neither dac imaged as well or managed anything like the same level of ambiance recovery in comparison to the Sony.
Source - media server:
24/96 The Zhaolu sounds very unpleasant, lean, and choked off, stridency audible on some passages - no hint that this file from Linn Records is high rez. Violin sound harsh, thin and reedy. Repellent...
Benchmark by comparison sounds tonally balanced, lots of detail, great ambience recovery and good sound stage height and depth. Makes you want to continue to listen, draws you into the music.
The Zhaolu just does not sound like it is operating in 24 bit mode..
16/44 The Zhaolu actually seems to sound better at lower bit rates - this seems evident when resampling 16/44 to 24/96 - there is no audible hint of improvement and some artifacts which in casual listening are hard to pin down. Again though the Benchmark provides a much more tightly integrated and pleasing sound - the best I have ever heard the media server sound in fact.
Overall I have to admit that I was disappointed, the Zhaolu 2.5A is nowhere close in performance to this early Benchmark, and on critical listening had a much leaner tonal balance than the Sony and far less detail, ambiance or sound stage.
The Benchmark (this version) does not offer the ultimate in detail or sound staging, but the tonal balance sounds very natural and is pleasing to listen to for extended periods of time.
In the end I'll probably end up just getting a Benchmark dac, in the meantime I will try to see whether or not I can close the gap a little more.
Some have said the Zhaolu trounced the Benchmark, perhaps this might be true of the 2.0, or the 2.5C, but sadly I heard absolutely no evidence of this with the 2.5A. I'm really not that surprised given the obvious differences in quality of execution of the respective designs. The Benchmark is beautifully built, and the Zhaolu isn't below the surface.
The Zhaolu is a good value, but it probably does not compare to a dac approaching the state of the art. I do hope to prove myself wrong with some more mods.
DIR9001 and PCM1794 anyone?
Some notes:
16/44 and 24/96 material furnished by the media server via an m-audio 2496. 24/96 material was downloaded from Linn Records in the past.
16/44 also provided by the Sony and delivered by coax to the Zhaolu and the Benchmark Dac 1. (This appears to be an early version ~ 3yrs old)
Most 16/44 material was from Il Sogno, the 24/96 material was from the Linn Classical library. (I can provide the specific file name, don't have it to hand here.)
Spdif coax connections were used for all tests.
Audio signal paths in all devices compared use op-amps. (No fancy discrete I/V etc.)
Source - Sony utilized as transport
The Benchmark sounds fuller and tonally balanced, good level of detail, smooth, extended highs.. Good sound stage depth, and width. Did I say smooth?
The Zhaolu by comparison sounds wiry, thin and strident. Much more ascetic (yes vinegary) and a bit grating on the nerves. Paper thin imaging compared to the Sony, very lean and wiry compared to the Benchmark.
The Sony has the most sound stage depth and width, great ambiance recovery and detail, leaner tonally than the Benchmark. Neither dac imaged as well or managed anything like the same level of ambiance recovery in comparison to the Sony.
Source - media server:
24/96 The Zhaolu sounds very unpleasant, lean, and choked off, stridency audible on some passages - no hint that this file from Linn Records is high rez. Violin sound harsh, thin and reedy. Repellent...
Benchmark by comparison sounds tonally balanced, lots of detail, great ambience recovery and good sound stage height and depth. Makes you want to continue to listen, draws you into the music.
The Zhaolu just does not sound like it is operating in 24 bit mode..
16/44 The Zhaolu actually seems to sound better at lower bit rates - this seems evident when resampling 16/44 to 24/96 - there is no audible hint of improvement and some artifacts which in casual listening are hard to pin down. Again though the Benchmark provides a much more tightly integrated and pleasing sound - the best I have ever heard the media server sound in fact.
Overall I have to admit that I was disappointed, the Zhaolu 2.5A is nowhere close in performance to this early Benchmark, and on critical listening had a much leaner tonal balance than the Sony and far less detail, ambiance or sound stage.
The Benchmark (this version) does not offer the ultimate in detail or sound staging, but the tonal balance sounds very natural and is pleasing to listen to for extended periods of time.
In the end I'll probably end up just getting a Benchmark dac, in the meantime I will try to see whether or not I can close the gap a little more.
Some have said the Zhaolu trounced the Benchmark, perhaps this might be true of the 2.0, or the 2.5C, but sadly I heard absolutely no evidence of this with the 2.5A. I'm really not that surprised given the obvious differences in quality of execution of the respective designs. The Benchmark is beautifully built, and the Zhaolu isn't below the surface.
The Zhaolu is a good value, but it probably does not compare to a dac approaching the state of the art. I do hope to prove myself wrong with some more mods.
DIR9001 and PCM1794 anyone?
Thanks Kevin. That was a very interesting comparison. I am a little surprised that you said the Benchmark was fuller. Most reports say it is lean and overly detailed. The Zhaolu must sound very lean indeed.
I think you said you were using the LM4562 op amps. I also found these to be very light sounding. Almost overly transparent with no body to them.
I think you may have a more pleasant result with the OPA2107 op amps. You might want to give them a try before you give up.
I also found the Zhaolu had better body and punch using as AD823 op amp in the power supply section.
Good luck,
Bob
I think you said you were using the LM4562 op amps. I also found these to be very light sounding. Almost overly transparent with no body to them.
I think you may have a more pleasant result with the OPA2107 op amps. You might want to give them a try before you give up.
I also found the Zhaolu had better body and punch using as AD823 op amp in the power supply section.
Good luck,
Bob
Hi Bob,
I'm going to give your suggestions a try. I think the Zhaolu is extremely lean sounding, and perhaps a slightly more euphonic op-amp would help here. I've generally found the LM4562 to be preferable in most applications and in the stock dac they were a major improvement over the 2604 and 1570 that it originally came with. I might just try the 2604 and see how that does. I will also try some 5532 which is what the Benchmark uses.
I think the overall sound depends heavily on the sources used and on a lot of other aspects of system integration and design. My system is tube based with onken style bass cabinets and JBL horns on mid and tweeter. It is rather revealing, and can be a bit unforgiving if the source is not totally up to snuff.
I had expected the modded Zhaolu to do much better than it did, and deliberately did not listen to the system Monday evening in order to put some water under the bridge between me and the last set of mods. I think sometimes in the heat of the moment after spending hours doing mods you can almost will yourself into hearing improvements that aren't really there. What I heard last night makes me believe that I have either reached the limitations of the chips used in this design or that the overall implementation is so poor that I have a tremendous amount of work ahead of me.
Realistically speaking I did not expect it to better the Benchmark, but it wasn't even close.. Seriously.. And the Sony clearly outperformed them both, although the Benchmark sounded really good, just lacking the edge in sound stage and detail recovery. The Sony just looked much deeper into the mix, and things were audible on its outputs that neither of the two dacs picked up..
I can't justify the acquisition of a zap filter for this dac as there are so many obvious design issues present and the price of purchase would get me well on the way to covering the cost of materials and pcb for my own design which at this point would be based on the DIR9001 and PCM1794 or 1798. The same kind of money would get about 1/3rd of the way to a good used Benchmark.
I'm going to give your suggestions a try. I think the Zhaolu is extremely lean sounding, and perhaps a slightly more euphonic op-amp would help here. I've generally found the LM4562 to be preferable in most applications and in the stock dac they were a major improvement over the 2604 and 1570 that it originally came with. I might just try the 2604 and see how that does. I will also try some 5532 which is what the Benchmark uses.
I think the overall sound depends heavily on the sources used and on a lot of other aspects of system integration and design. My system is tube based with onken style bass cabinets and JBL horns on mid and tweeter. It is rather revealing, and can be a bit unforgiving if the source is not totally up to snuff.
I had expected the modded Zhaolu to do much better than it did, and deliberately did not listen to the system Monday evening in order to put some water under the bridge between me and the last set of mods. I think sometimes in the heat of the moment after spending hours doing mods you can almost will yourself into hearing improvements that aren't really there. What I heard last night makes me believe that I have either reached the limitations of the chips used in this design or that the overall implementation is so poor that I have a tremendous amount of work ahead of me.
Realistically speaking I did not expect it to better the Benchmark, but it wasn't even close.. Seriously.. And the Sony clearly outperformed them both, although the Benchmark sounded really good, just lacking the edge in sound stage and detail recovery. The Sony just looked much deeper into the mix, and things were audible on its outputs that neither of the two dacs picked up..
I can't justify the acquisition of a zap filter for this dac as there are so many obvious design issues present and the price of purchase would get me well on the way to covering the cost of materials and pcb for my own design which at this point would be based on the DIR9001 and PCM1794 or 1798. The same kind of money would get about 1/3rd of the way to a good used Benchmark.
kevinkr said:
Hi Kevin,
I'm not disputing your findings or opinion but I can't agree that the design is flawed or that you can't achieve top notch sound quality with it. Now, I do have the C version which uses the Cirrus Logic CS4398 DAC that is their flagship so to speak. I did find the analog section to be limiting. The Zapfilter changes everything.
I have compared my Zapped Zhaolu to a friends Cary 302/200 (BB1704) that I think sounds wonderful. The Zhaolu is comparable, IMHO. I've also had a stock Sony ES777 in my system but it's been a while. The Sony was very nice but not of the Zapped Zhaolu's caliber based on my recollection.
Ori of Oritek offers the A version modded with his own discrete output board and many other mods. He speaks highly of the power supply etc. in the 2.5. His DAC's are extremely highly regarded. I'm curious why you two differ in your assessment.
Anyway, I think you're missing the key to great sound by not going discrete on the analog stage.
My $.02 worth...
BobM said:
Hi Bob,
The Zapfilter changed everything. With the discrete output stage the clarity became exceptional and all hints of stridency went away. The detail level increased dramatically and there was that "real" sound of presence that only the best DAC have. It went from very good for the price after I did the opamp changes etc to great with the discrete output stage. The differences are not subtle.
I've not heard the level of detail it now has in any other DAC. These are my opinions of course but they agree with many, many others. I think you post at Head-Fi. Read the mod your Zhaolu thread. It's huge but very enlightening.
Ori likes the AD1852 version that you guys are playing with. I'm really happy with the CS4398 version. BTW, Audiomagus has the DAC boards available individually so you can swap back and forth if you're so inclined. You can get them from Eddie the distributor on HK too.
I'd like to A/B it against a CI DAC some day if I could get my hands on one.
HTH
Jim
jholtz said:
While I very much appreciate the input have you really looked closely at the board layout in the Zhaolu? And yes I agree that the analog board is a particular weak spot, but I am hesitant to spend 2 times what the dac cost for a zapfilter only to find that the issues I have with it are not addressed. Admittedly the zapfilter has its own supplies which probably addresses a fair number of the issues I identified on the supply pcb. Note that the AD1852 has op-amps in it, which presumably are not in the same league as the 4562 or the other op-amps I have evaluated in this dac.
Bear in mind also that the other dac had all op-amps in its signal path so it is possible to get reasonable if not stellar sound using this approach.
For the same money from Twisted Pear I can get a pair of cods, the spdif receiver and asrc module and do my own discrete I/V converters. I am coming to believe this might be a better place to spend my money. I think the Zhaolu is a tremendous value for $130 plus shipping, but I don't as I indicated want to make it into a $400 dac..
The sony referred to is an SCD-777ES SACD player, not sure if the one you cited is the same. (Not to be confused with the XA777etc.) This one is moderately modified and performs quite well.
The Zapfilter is $214 from Parts Connection plus. The Zhaolu 2.5C is $200 rather than the $130 the 2.5A version costs. So, yes, the Zapfilter doubles the cost.
As far as looking at the board for design deficiencies, I'm not an engineer so I could look all day and I wouldn't know the difference. However, I have an exceptionally revealing system and it is very, very easy to hear the differences when changes were made.
No quibbles if you'd prefer to go the Pear route. That would be a project that I wouldn't attempt. Too complex for me. I could sell the Zapped Zhaolu and get the majority of my money back which is why I gave it a try. I'm now extremely glad I did.
BTW, the Sony was an ES model and was very nice as I said. Amazingly so from a bone stock component from a consumer electronics company. It also belonged to a friend that ultimately had it modded. I haven't seen him in a long time so I'm not sure what the results of the mods were.
Good luck with your new project.
Best regards.
Jim
As far as looking at the board for design deficiencies, I'm not an engineer so I could look all day and I wouldn't know the difference. However, I have an exceptionally revealing system and it is very, very easy to hear the differences when changes were made.
No quibbles if you'd prefer to go the Pear route. That would be a project that I wouldn't attempt. Too complex for me. I could sell the Zapped Zhaolu and get the majority of my money back which is why I gave it a try. I'm now extremely glad I did.
BTW, the Sony was an ES model and was very nice as I said. Amazingly so from a bone stock component from a consumer electronics company. It also belonged to a friend that ultimately had it modded. I haven't seen him in a long time so I'm not sure what the results of the mods were.
Good luck with your new project.
Best regards.
Jim
I really wish I knew someone in the NY area with a Zapped Zhaolu to see what the real differences are, for myself. Right now I think my 2.5C sounds pretty damn good. That pulse transformer and recent mods from Kevin's posts has brought it up another level of transparency and I'm enjoying it a lot.
I think I'm going to enjoy it for a while. My Birthday is in April - perhaps I'll get myself a present and buy a Zap board.
Bob
I think I'm going to enjoy it for a while. My Birthday is in April - perhaps I'll get myself a present and buy a Zap board.
Bob
BobM said:
Hi Bob,
The Zapfilter made me a believer in discrete output stages. I know you can get good sound with opamps too but I'm of the opinion that discrete takes it up a notch. In the case of the Zhaolu, the differences were anything but subtle and that was after I changed opamps and shorted the blocking caps, both of which helped.
I think that's why Ori has developed his own discrete analog stage for his Zhaolu modded units. He does do quite a few other things too that I can't comment on but everyone seems to love them.
Fun stuff!
Jim
Hi Bob,
Keep us posted when you get the Zapfilter. I had at one point thought about doing an all discrete audio board for the Zhaolu so it will be interesting to see what you think with the Zapfilter fitted.
I going to keep looking for obvious things I have missed and I will continue to post here. I haven't hit my budgetary limit yet so if I identify anything else I can "fix" or perhaps that should be "break", I'll post here
I am going to get the Twisted Pear modules when the next batch are available, but in the meantime I will keep tinkering with the 2.5A.
Keep us posted when you get the Zapfilter. I had at one point thought about doing an all discrete audio board for the Zhaolu so it will be interesting to see what you think with the Zapfilter fitted.
I going to keep looking for obvious things I have missed and I will continue to post here. I haven't hit my budgetary limit yet so if I identify anything else I can "fix" or perhaps that should be "break"
, I'll post hereI am going to get the Twisted Pear modules when the next batch are available, but in the meantime I will keep tinkering with the 2.5A.
Member
Joined 2003
Circuit simulations
Let's hope I reverse engineered the analog filter pcb correctly. Here's what I've come up with for a schematic. The output of the op-amp is connected in series with a 2K resistor, then to a buffer. For the most part, this can be omitted. Just cut the track on the board, and wire the output of the first op-amp to the output of the DAC.
Image: Zhaolu output fitler schematic
I'm not sure what the 100K resistors are for, but they're completely useless from what I can tell. For comparison, here's some graphs I made comparing the recommended filter in the AD1852 datasheet to the one on the Zhaolu DAC:
Image: Datasheet output filter response
Image: Datasheet output filter response omitting the 200 Ohm resistor and 10nF cap
Image: Zhaolu output filter response
Image: Zhaolu Output filter response omitting 100uF caps.
[Edit: missed some parts, Zhaolu filter looks more reasonable now]
Let's hope I reverse engineered the analog filter pcb correctly. Here's what I've come up with for a schematic. The output of the op-amp is connected in series with a 2K resistor, then to a buffer. For the most part, this can be omitted. Just cut the track on the board, and wire the output of the first op-amp to the output of the DAC.
Image: Zhaolu output fitler schematic
I'm not sure what the 100K resistors are for, but they're completely useless from what I can tell. For comparison, here's some graphs I made comparing the recommended filter in the AD1852 datasheet to the one on the Zhaolu DAC:
Image: Datasheet output filter response
Image: Datasheet output filter response omitting the 200 Ohm resistor and 10nF cap
Image: Zhaolu output filter response
Image: Zhaolu Output filter response omitting 100uF caps.
[Edit: missed some parts, Zhaolu filter looks more reasonable now]
Re: Circuit simulations
DcibeL said:Let's hope I reverse engineered the analog filter pcb correctly. Here's what I've come up with for a schematic. The output of the op-amp is connected in series with a 2K resistor, then to a buffer. For the most part, this can be omitted. Just cut the track on the board, and wire the output of the first op-amp to the output of the DAC.
<snip>
Great work and it looks correct to me. Incidentally you are right about the 100K resistors - they are totally useless..
To get best cmrr you should replace the 100K resistor on the non inverting input with a 2K resistor to make the input impedance the same on both the inverting and non-inverting inputs. Delete the 100K on the other input.
One other thought I had particularly if dc coupling to the diff amp might be to scale of all those resistors and caps by 10X so resistors go to 10K and those 2.2NF caps become 220pF - this should reduce the load on the dac op-amps and might help linearity. Why I ordered 1K resistors instead I don't know..
I removed the coupling caps, but I am not sure that this is a great idea, and I am going to put something back. I don't think the BG N types I had in there sounded too great, but perhaps they were interacting with the 1uF caps that I have removed.
I am planning to spend a little time on the dac this week-end and see what further improvement I can make without resorting to spending any significant money. I have plenty of parts on hand so I can do plenty. Big snow storm Saturday so I am going to stay indoors.
I think there is a very simple way to do a dc servo based on taking the output from the first op-amp and using the second as an integrator. I'm going to think about that and if I manage to get it to work I will post the details here.
Obviously I will tap the signal at the output of the first op-amp based on your recommendation, had been thinking of doing this anyway - simple with the mods I have done to this board already.
I am going to get the Twisted Pear kits, but I'll noodle around a bit more with this one, and then use it as a reference point for the next big thing..
I just realized I see a really big no no in the design of this diff-amp - that 2.2nF cap should not be directly on the inverting input.. I'll have to think about how to fix that. Damn should have caught that..
Edit - I've got it, I'll post the solution when I have tested it. Happily it will result in 2K input impedance at both inputs reducing the load on the dac op-amps and allowing the deletion of the 2K resistor I mentioned in the previous post about the non-inverting input. This will also allow smaller electrolytic caps if ac coupling is desired. 22uF ought to be more than enough.
Edit - I've got it, I'll post the solution when I have tested it. Happily it will result in 2K input impedance at both inputs reducing the load on the dac op-amps and allowing the deletion of the 2K resistor I mentioned in the previous post about the non-inverting input. This will also allow smaller electrolytic caps if ac coupling is desired. 22uF ought to be more than enough.
Member
Joined 2003
There's a 1K resistor in between. I don't have the board in front of my to be sure of my schematic, but the topology of this circuit is the same as what is in the AD1852 datasheet, just with different part values. Are you suggesting that the suggested filter in the datasheet also has a big no-no?
I'll have a closer look at this board over the weekend, and come up with what I think is a good improvement and order some parts. I'd very much like to see what you have come up with for a solution as well, provided it's fairly simple to implement.
I'll have a closer look at this board over the weekend, and come up with what I think is a good improvement and order some parts. I'd very much like to see what you have come up with for a solution as well, provided it's fairly simple to implement.
I don't think this is a good idea, as it will decrease the 47pF cap to 4.7pF. At these capacitance levels the pcb traces will become problematic.
This is a good idea, I think.
Hi,
Take a look at what I have come up with - I will be implementing this tonight.
Flat with .1dB or so at 20kHz, down about a dB at 100kHz.
Gain is a little over unity which is what I need in order to have close to standard redbook output levels - currently the dac outputs 1.8VrmsFS.
CMRR with 1% resistors is less than 40dB, but can be improved to >60dB by simply matching them to better than 0.1% and selecting mirrored pairs.
You can probably ignore R9 in most applications - it makes sense if using ac coupling with reasonably well matched electrolytics to maximize cmrr.
Take a look at what I have come up with - I will be implementing this tonight.
Flat with .1dB or so at 20kHz, down about a dB at 100kHz.
Gain is a little over unity which is what I need in order to have close to standard redbook output levels - currently the dac outputs 1.8VrmsFS.
CMRR with 1% resistors is less than 40dB, but can be improved to >60dB by simply matching them to better than 0.1% and selecting mirrored pairs.
You can probably ignore R9 in most applications - it makes sense if using ac coupling with reasonably well matched electrolytics to maximize cmrr.
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.