Hi Folk!
Is the LM4562 really good at DAC output use? Or its spec need a special design for best use?
For testing purpose, I installed this op-amp in different equipments. In pre-amplifier circuit, in buffer and filter with nice result. But I tried to use it in DAC output (LPF circuit) in 2 different equipments, and each time, I had a very bad sound.
The sound is very detailed but metallic and flat. If I replace it with an AD8620 or an OP627 or another op-amp as OPA2134, the sound is ok.
The last tried was inside a Tag AV32R. The circuit is designed for an OPA2134. It's at DAC output (AK4393), the next circuit in the chain is an CS3310. The op-amp output is directly connected to the input of the CS3310. PSU is 12,5v. I've enclosed the approximate schematic to this post. I've not traced all the circuit yet. Some values are missing and may be some component too. The equipments is curently in use and only worked with some picture I made.
Any idea ?
Is the LM4562 really good at DAC output use? Or its spec need a special design for best use?
For testing purpose, I installed this op-amp in different equipments. In pre-amplifier circuit, in buffer and filter with nice result. But I tried to use it in DAC output (LPF circuit) in 2 different equipments, and each time, I had a very bad sound.
The sound is very detailed but metallic and flat. If I replace it with an AD8620 or an OP627 or another op-amp as OPA2134, the sound is ok.
The last tried was inside a Tag AV32R. The circuit is designed for an OPA2134. It's at DAC output (AK4393), the next circuit in the chain is an CS3310. The op-amp output is directly connected to the input of the CS3310. PSU is 12,5v. I've enclosed the approximate schematic to this post. I've not traced all the circuit yet. Some values are missing and may be some component too. The equipments is curently in use and only worked with some picture I made.
Any idea ?
Attachments
Hi,
Let your ears be the final judge.Could it be the filter network
is not suitable for LM4562?Is the CS3310 "perfect" soundwise?
Cheers.
Let your ears be the final judge.Could it be the filter network
is not suitable for LM4562?Is the CS3310 "perfect" soundwise?
Cheers.
I've made a list of the possibilities:
- filter network and sum not adapted (too high impedance)
- output from LM4562 not adapted with CS3310 input. In fact, the CS3310 have been replaced by a PGA2311A.
- PSU: on the LM4562 datasheet, they specify 47uF on the rail close to the chip. We "only" have 10uF in this circuit.
- PSU: 12v too low for optimum performance
- the op-amp itself -> sonic signature too chirurgical. As I tested the mod with a pair of Dynaudio Contour, each change is really easy to heard.
I don't think that the PCB can cause this. As you can see in the picture, its well done.
I replaced it with an AD8620. In this circuit and with the rest of my equipments, it's the best for me.
- filter network and sum not adapted (too high impedance)
- output from LM4562 not adapted with CS3310 input. In fact, the CS3310 have been replaced by a PGA2311A.
- PSU: on the LM4562 datasheet, they specify 47uF on the rail close to the chip. We "only" have 10uF in this circuit.
- PSU: 12v too low for optimum performance
- the op-amp itself -> sonic signature too chirurgical. As I tested the mod with a pair of Dynaudio Contour, each change is really easy to heard.
I don't think that the PCB can cause this. As you can see in the picture, its well done.
I replaced it with an AD8620. In this circuit and with the rest of my equipments, it's the best for me.
Attachments
Don't worry Stef, trust your ears. A few of us have had similar experience with the 4562 and its single derivative (LME47910) - all the notes but little of the expression.
Your list of possibilities looks good to me. I've experimented quite a lot with PSu/regulators, operating voltages and decoupling strategies and just couldn't get a compelling result from the 47910 in a CD player output stage. I'll try it elsewhere another time.
Your list of possibilities looks good to me. I've experimented quite a lot with PSu/regulators, operating voltages and decoupling strategies and just couldn't get a compelling result from the 47910 in a CD player output stage. I'll try it elsewhere another time.
Great sentence. I've the same thinking. You can count and place all instruments but each one is playing alone and not with the same tempo. C'est la cacophonie !
I installed it at the input stage of a modled Quad 405 and at input stage of an UcD180 with great result.
Just give this a shot 🙂
http://midiserver.student.utwente.nl/sitestuff/apparatuur/sony_md_mds_je530/minidisk.htm
http://midiserver.student.utwente.nl/sitestuff/apparatuur/sony_md_mds_je530/minidisk.htm
hi steff i am a little confused with you
you have mentioned in another thread that lm4562 is the winner compared to other opamps when using the TAG with other people listening also is it the sapme you used since i think you havent ordered it before?
i also want to try LM4562 to my nad cdp now i'm using lm6172🙁
you have mentioned in another thread that lm4562 is the winner compared to other opamps when using the TAG with other people listening also is it the sapme you used since i think you havent ordered it before?
i also want to try LM4562 to my nad cdp now i'm using lm6172🙁
RICMAN said:
Yes, you're true. It was a big mistake or "wrote too fast".
In fact, I first tested the tag mods on a pair of small speaker with an old sono amplifier in stereo. The sound was apparently fine. The details added by the LM were impressive. But after, I tested it on my main install (with a big Dynaudio Contour). This system is very transparent and reveal all problem in sound as a magnifying glass. It was impossible to listen something more than 10 minutes.
I opened again the Tag and done again an op-amp swap session but this time on my main system.
The LM4562 stay the better op-amp for details, quality of bass and tone but the sound is too transparent. Too many details hide the music. Music became like a "cacophony" of different instrument. Each sound competing to be the first.
Used in surround mode, it was worst than in stereo. All this transparency made movie listening close to impossible. Dialog was completely hidden in background music and effect.
I installed the LM inside an UcD180 and in a QUAD 405 (input buffer) with success compared to others. The LM fit also fine in the output buffer of the Tag. I also installed a pair of LM4562 at DAC output of a DEQ2496 without problem.
But definitively not in the LPF at DAC output of the Tag.
I also tried the lm6172 in the Tag but preferred the AD8620. I was surprised because, I'm not a 8620 lover.
To end, I'll say that the LM4562 is a great op-amp but less "universal" that the OPA2134 or NE5532. Depending of the circuit or the others parts of the system, you can have the best or the worse. You should try.
.
Attachments
MidiMaze said:
I love the gameboy used as a spectrum analyser. ;-)
I'll do some FFT spectrum test by the end of the week.
.
as i am searching the web about lm4562 reading comments by others made me think that 4562 is a good opamp in other circuits like in tone controls, preamp, buffer, phono preamp but not in some circuits like I/V, cd output stage, DAC output...IMHO?
steff you may try THS4032 maybe it suits your tag and let us know the results i used this in my philips cd output stage with good results just dont use it in unity gain circuit it is unstable
steff you may try THS4032 maybe it suits your tag and let us know the results i used this in my philips cd output stage with good results just dont use it in unity gain circuit it is unstable
Hi Stef1777,
A few thoughts from this humble, empyrical mate...
Have you ever heard Ak4393 direct output? It is very detailed but lacks "flesh". Maybe the LM4562 is just letting hear the DAC chips "as is". Not a defect in itself, IMHO. Maybe the other BB opamps are a better match with that specific DAC because they make it sound darker, conceiling details...
I think is a matter of "tuning" the sound to your liking: play a little with PS caps and surtout allow a long burn-in period.
I recommend Ruby ZL 47-100uF//BG NX 0u1 for the opamps (never heard ZA caps).
Maybe some Elna electros if more warmth is desired...
Portlandmike recommended, and Ecdesigns uses, much lower R. Maybe worth trying.
Good luck.
M
A few thoughts from this humble, empyrical mate...
Have you ever heard Ak4393 direct output? It is very detailed but lacks "flesh". Maybe the LM4562 is just letting hear the DAC chips "as is". Not a defect in itself, IMHO. Maybe the other BB opamps are a better match with that specific DAC because they make it sound darker, conceiling details...
I think is a matter of "tuning" the sound to your liking: play a little with PS caps and surtout allow a long burn-in period.
I recommend Ruby ZL 47-100uF//BG NX 0u1 for the opamps (never heard ZA caps).
Maybe some Elna electros if more warmth is desired...
Portlandmike recommended, and Ecdesigns uses, much lower R. Maybe worth trying.
Good luck.
M
Hi,
Before you toss the LM4562 under the bus, I recommend you match the diff amp of your circuit.
The ??? cap is of no matter in that it is a balanced affect, but the fact that the feedback cap is 330pF and the grounded cap is 1nF is a gross issue and will likely cause more negative affect if your op amp has better bandwidth, the LM4562 is higher bandwidth.
I'm not sure why they would have mismatched the circuit otherwise, but I suspect they are attempting to compensate for the limited bandwidth of the op amp and trying to correct the diff amp function accordingly. (actually, that's only a guess as the simulation says you will get horse dung common mode rejection with the circuit as is (like 6dB vs >60dB at 20kHz with a LM4562 with two 330pF caps, depending on the match of coarse, 1% being 40dB)
If so inclined, I recommend you try changing the 1nF grounded cap to be 330pF, or actually, match them with two identicle 330pF caps. (simulation says 1nF will roll off your highs before 20k)
I suspect your ??? cap is 1nF because it appears to give a nice 20kHz 12dB per octave roll off.
Take what I say with a grain of salt, and by all means, let your ears be the final judge. Likely it will improve the 8620 too.
As for lower values of resistance (maricio), in this case, make a simple change first, then maybe scale things. Clearly this circuit insn't optimized for the best possible noise performance, but likley the original op amps drive capability. That or bigger value caps were to costly. Or, the DAC can't drive lower values well. Either way, leave them for now.
Also, to add to the list of possibilities of reasons, an PCB designed for a 10Mhz op amp may be not good enough for a higher bandwidth op amp.
I've read a post were a guy said a AD712 sounded better than a LM4562. To me, I'd guess that is because the layout and bypassing method sufficient for a 1Mhz op amp just doesn't cut it at 56Mhz.
If your really lazy, remove both the 1nF and the 330pF and see how it sounds. You will get excellent common mode rejection I limited only by the resistor tolerence (the ??? cap has not affect) and a nice 6dB per octave roll off of the noise.
Compare your differnet op amps with that circuit, then decide if its worth getting some good 2% 330pF caps to try.
( This also runs the op amp at greater than unity gain at HF thus greatly improves the phase margin and transient response. This is what I'd recommend for the LM4562, that's good at untity gain, but clearly better in a D/A application with more phase margine.)
portlandmike
Before you toss the LM4562 under the bus, I recommend you match the diff amp of your circuit.
The ??? cap is of no matter in that it is a balanced affect, but the fact that the feedback cap is 330pF and the grounded cap is 1nF is a gross issue and will likely cause more negative affect if your op amp has better bandwidth, the LM4562 is higher bandwidth.
I'm not sure why they would have mismatched the circuit otherwise, but I suspect they are attempting to compensate for the limited bandwidth of the op amp and trying to correct the diff amp function accordingly. (actually, that's only a guess as the simulation says you will get horse dung common mode rejection with the circuit as is (like 6dB vs >60dB at 20kHz with a LM4562 with two 330pF caps, depending on the match of coarse, 1% being 40dB)
If so inclined, I recommend you try changing the 1nF grounded cap to be 330pF, or actually, match them with two identicle 330pF caps. (simulation says 1nF will roll off your highs before 20k)
I suspect your ??? cap is 1nF because it appears to give a nice 20kHz 12dB per octave roll off.
Take what I say with a grain of salt, and by all means, let your ears be the final judge. Likely it will improve the 8620 too.
As for lower values of resistance (maricio), in this case, make a simple change first, then maybe scale things. Clearly this circuit insn't optimized for the best possible noise performance, but likley the original op amps drive capability. That or bigger value caps were to costly. Or, the DAC can't drive lower values well. Either way, leave them for now.
Also, to add to the list of possibilities of reasons, an PCB designed for a 10Mhz op amp may be not good enough for a higher bandwidth op amp.
I've read a post were a guy said a AD712 sounded better than a LM4562. To me, I'd guess that is because the layout and bypassing method sufficient for a 1Mhz op amp just doesn't cut it at 56Mhz.
If your really lazy, remove both the 1nF and the 330pF and see how it sounds. You will get excellent common mode rejection I limited only by the resistor tolerence (the ??? cap has not affect) and a nice 6dB per octave roll off of the noise.
Compare your differnet op amps with that circuit, then decide if its worth getting some good 2% 330pF caps to try.
( This also runs the op amp at greater than unity gain at HF thus greatly improves the phase margin and transient response. This is what I'd recommend for the LM4562, that's good at untity gain, but clearly better in a D/A application with more phase margine.)
portlandmike
I tried this opamp in my CD player and I get distortion/noise now and then. I put back the opa627BP (mounted in a brown-dog adapter).
jarthel said:
The devil is in the details!
Distortion is a sign of a gross miss implementation in most cases.
Portlandmike
Thanks Mike and others for all these advices.
Concerning the Wima FKP2 caps. Farnell doesn't have the 1% version. Can I use 5% version or try to use another 1% suppliers.
Farnell have this in 1%: 1264887 or 9520210 (this one will not fit I think).
I've the possibility to do some FFT spectrum test. Do you will like to have some specific curve to be useful?
I also have a Tag AV30R. The circuit is close and inserting LM4562 done the same problem. I plan to try all this with the AV3OR less complicated to open and unmount.
What about the PSU and LM4562? The current analog PSU is per rail 2200uF (for 9 opamp + 3 AKM + 3 CS3310) + LM3xx + 100uF for 3 op-amp + 10uF tantalum close to each op-amp. Chimical are Silmic II.
Do I need to change the 10uF Tantalum caps close to the opamp by 47uF Tantalum? As indicated in the LM4562 datasheet.
I added 1 curve done on the AV30 with OPA2134.
.
Concerning the Wima FKP2 caps. Farnell doesn't have the 1% version. Can I use 5% version or try to use another 1% suppliers.
Farnell have this in 1%: 1264887 or 9520210 (this one will not fit I think).
I've the possibility to do some FFT spectrum test. Do you will like to have some specific curve to be useful?
I also have a Tag AV30R. The circuit is close and inserting LM4562 done the same problem. I plan to try all this with the AV3OR less complicated to open and unmount.
What about the PSU and LM4562? The current analog PSU is per rail 2200uF (for 9 opamp + 3 AKM + 3 CS3310) + LM3xx + 100uF for 3 op-amp + 10uF tantalum close to each op-amp. Chimical are Silmic II.
Do I need to change the 10uF Tantalum caps close to the opamp by 47uF Tantalum? As indicated in the LM4562 datasheet.
I added 1 curve done on the AV30 with OPA2134.
.
Attachments
stef1777 said:
Nice curves.
What you can't measure is the common mode stuff.
Just a note on the frequency response, how was it measured. If that white noise FFT'd?
Diff output DACs usually have lots of nasty common mode noise, and you don't want that. The more CMMR the better.
Also, you 1kHz has 1% 3rd harmonic. Perhaps you like that but I find that to be to high IMO. 1% 3rd doesn't sound bad. I'd be more concerned about the .1% 9th harmonic.
Getting rid of that will make things much better, but its hard to get good measurements. That may be just measurement error.
As for the diff amp caps, I stand by my comment to just remove them first. I understand op amp swaps can be tough, but it will be insightful, and in your application DAC amp, I highly recommend running the LM4562 at a gain greater than 1 at HF. The 330pF cap makes it unity gain and you can see from the DS that it rings a bit. One thing that helps reduce potential ringing with this op amp is to load the output to a current source to the positive rail. Say a 5mA CRD.
If you want the cap in there, just use two that are the same. 5% is a good start, just not ideal.
As for bypassing. I highly recommend you isolate the op amp power supply from the DAC power supply. Even a 22ohm resistor to a seperate supply cap will be fine. Far more impact than changing it from 10uF to 47uF. All this assumes its easy to do, and sometimes it isn't.
I'll open the AV32R this week-end and try to finish to trace the circuit to be sure. And will place the 2 caps on socket for testing purpose.
I'll try to do some better mesurement with a CD directly with the digital input. I'm not sure of the first test done as the software is new for me. Look at the new curve inclosed. It's looking different.
I plan white, 100 Hz, 1K and 10K, and frequency reponse (I know how to do it with the tool now). I've not the equipement to do THD test.
In the AV32R, digital and analog psu are completly separated. In fact, 7 differents PSU drive all the stuff.
.
Attachments
New and complete design of the DAC output circuit. I've not done hardware modifications on it yet. I'm waiting for some parts first (and more advice).
.
.
- Status
- Not open for further replies.