ROFL! I was trying to translate that into techno-speak.
797 at 20 kHz, 2 Volts peak: .000001 percent (-160 dB)
5532 at 20 kHz, 2 Volts peak: .001 percent (-100 dB)
This assumes that only the input stage distorts and the rest of the amp is distortionless per Gilbert's paper.
I'm going to listen to some tunes before I crash.
797 at 20 kHz, 2 Volts peak: .000001 percent (-160 dB)
5532 at 20 kHz, 2 Volts peak: .001 percent (-100 dB)
This assumes that only the input stage distorts and the rest of the amp is distortionless per Gilbert's paper.
I'm going to listen to some tunes before I crash.
Re: (->s.w.r<-) Üü
Michael,
I have tried it. The improvement was lower compared to that of buffered stage. Also LM6171/2 was tried with results inferior to the buffer circuits. I have succesfuly tested BUF634T and LT1010CT buffers inside feedback loop. I do not suppose this is a question of impedance matching, but reducing of HF/RF components and/or higher power of link signal transfer (with terminated cable). The second circuit that works is the buffer with very low output series resistor (5 Ohm) and no cable termination. The output impedance must remain very low even for high frequencies. I recommend careful mechanical design and shielded boxes. Reducing RFI/EMI greatly improves the sound. Careful design of input filters is highly recommended, especially for digital sources, as signals up to 16 MHz can be easily measured at their outputs. Turntable is not that sensitive for HF supression.
Regards,
Pavel
P.S. It was mentioned (not by you) that less components is better. From my experience, sonically the worst audio preamp solution is the passive high impedance (20 kOhm) "preamp" connected by 2 cables into the audio chain.
Ultima Thule said:
Michael,
I have tried it. The improvement was lower compared to that of buffered stage. Also LM6171/2 was tried with results inferior to the buffer circuits. I have succesfuly tested BUF634T and LT1010CT buffers inside feedback loop. I do not suppose this is a question of impedance matching, but reducing of HF/RF components and/or higher power of link signal transfer (with terminated cable). The second circuit that works is the buffer with very low output series resistor (5 Ohm) and no cable termination. The output impedance must remain very low even for high frequencies. I recommend careful mechanical design and shielded boxes. Reducing RFI/EMI greatly improves the sound. Careful design of input filters is highly recommended, especially for digital sources, as signals up to 16 MHz can be easily measured at their outputs. Turntable is not that sensitive for HF supression.
Regards,
Pavel
P.S. It was mentioned (not by you) that less components is better. From my experience, sonically the worst audio preamp solution is the passive high impedance (20 kOhm) "preamp" connected by 2 cables into the audio chain.
WaltJ said:
Walt,
I completely agree that it is confusing enough as it is. I was supposing that originally the thread was not limited to input stage distortion. As I was getting impression that opamp based circuits sound problem was being explained by PIM of the input stage, I tried to show that there were another kind of problems as well. According to my experience these problems (output stage + cable interaction) affect the sound more, provided the suitable audio opamps are used (not NE5532, NJM2114 etc.).
Pavel
Andy C, what about the 4558? That is a more realistic IC that used in typical audio equipment. The AD797 is one of the best examples of IC design, today. Still, a discrete design can still have certain advantages over it, such as pure class A, full complementary differential folded cascode, all fet input. That should be more linear than any differential transistor input stage, and that is what I use, when I am able to.
Hi, PMA,
If the main target is "buffering" for opamp output, what do you think about buffering opamp output with single emitor follower transistor (with R in emitors, Colector tied to +voltage)? It will drop 0.65V from the output of opamp, but if feedback is taken from this emitor follower, it will be 0volt again. But what about this configuration's performance? (compared to make a complete buffer like you do)?
If the main target is "buffering" for opamp output, what do you think about buffering opamp output with single emitor follower transistor (with R in emitors, Colector tied to +voltage)? It will drop 0.65V from the output of opamp, but if feedback is taken from this emitor follower, it will be 0volt again. But what about this configuration's performance? (compared to make a complete buffer like you do)?
Lumanauw,
I agree with Upupa Epops. From the monolithic circuits, I prefer the LT1010CT. The BUF634T is also very good. The diamond buffer according to Walt Jung can be modified to class A and CFP output stage and then gives excellent results.
The case is that the buffer should be the lightest possible load for the opamp and should have very low distortion. This is not the case of the SE follower transistor and feedback has to do more work with it.
I agree with Upupa Epops. From the monolithic circuits, I prefer the LT1010CT. The BUF634T is also very good. The diamond buffer according to Walt Jung can be modified to class A and CFP output stage and then gives excellent results.
The case is that the buffer should be the lightest possible load for the opamp and should have very low distortion. This is not the case of the SE follower transistor and feedback has to do more work with it.
andy_c, and/or anyone else
Based on your math I have the following questions/assumptions/conclusions;
1 For the 797 in the configuration you ran the numbers for, 3rd H is insignificant.
2 This should also be true for other VFB opamps used by DIYers, i.e. OPA627, AD8610, etc.
3 IYHO at what point is the 3rd H hearable and start to make a difference? Assuming only the mechanism you listed and others are not contributing.
4 The sonic character of op amps used by DIYers would have to come from a different mechanism than PIM based on your math. PSRR, CMRR, decreasing OLG/freq., etc.
5 Or, IYHO, does even the smallest amount of 3rd add to the character of the amp due to higher order harmonics being added.
6 Personally, I like the sound of a zero GFB design. This thread has examined some distortion mechanisms that could potentially affect the sound of a GFB design. The ultimate question then becomes(I'm adding fuel to the fire here) IMHO why does to much GFB seem the kill the harmonics and give a sound that to me becomes tiring compared to 0 GFB? (I think I'm gonna get flamed here)
7 What else does anybody want to add?
Mike
Based on your math I have the following questions/assumptions/conclusions;
1 For the 797 in the configuration you ran the numbers for, 3rd H is insignificant.
2 This should also be true for other VFB opamps used by DIYers, i.e. OPA627, AD8610, etc.
3 IYHO at what point is the 3rd H hearable and start to make a difference? Assuming only the mechanism you listed and others are not contributing.
4 The sonic character of op amps used by DIYers would have to come from a different mechanism than PIM based on your math. PSRR, CMRR, decreasing OLG/freq., etc.
5 Or, IYHO, does even the smallest amount of 3rd add to the character of the amp due to higher order harmonics being added.
6 Personally, I like the sound of a zero GFB design. This thread has examined some distortion mechanisms that could potentially affect the sound of a GFB design. The ultimate question then becomes(I'm adding fuel to the fire here) IMHO why does to much GFB seem the kill the harmonics and give a sound that to me becomes tiring compared to 0 GFB? (I think I'm gonna get flamed here)
7 What else does anybody want to add?
Mike
john curl said:
I'll run the numbers on that tonight when I get home. Looks like it has a 3 MHz GBW, so results will be somewhat worse than the 5532.
I agree. Kudos to Scott for an outstanding job!
I remember seeing somewhere, but I can't remember where, that there's some kind of process-related common-mode distortion problem with IC op-amps having FET inputs? I'm not an IC guy, so I don't have a clue about this. If so, and you wanted optimum low noise with good input stage linearity, it seems a discrete design with FET input would really fit the bill.
You folks have me very confused. The LP and almost any speaker has orders of magnitude more PIM (IM and AM too for that matter) than a modern high speed opamp. In fact, years ago I wondered if the slight modulations in LP reproduction contributed to making the sound more pleasing to some.
Lowering the noise floor continues to do the most for me. My latest pre allows me to hear pre-echo on some half speed mastered LP's that I simply never noticed before.
Lowering the noise floor continues to do the most for me. My latest pre allows me to hear pre-echo on some half speed mastered LP's that I simply never noticed before.
JC correct me if I'm wrong, if the echo matches up closely with the 1.8 seconds it takes for the LP to go around, it would be an unlikley coincidence that it be the tape print through. BTW I also hear hall ambience like never before. A funny thing too, it is VERY hard to set the recording level at 16/44 on a good LP. Using total averaged noise to state the dynamic range on an LP seems pretty bogus to me. If you look at an FFT most of the bins are as empty as a shorted input.
My preamp is cacoded diffpairs of some now unavailable Toshiba low noise JFETs, open loop, with passive equalization. I still use the low noise power supply and AD823/AD815 line stage that Walt sent me years ago. All powered from two 18v wall warts.
Since I don't believe in posting anonymously, I hope people will not take anything I say as an endorsement or lack of endorsement of any AD products or those of our competitors or anyone else.
My preamp is cacoded diffpairs of some now unavailable Toshiba low noise JFETs, open loop, with passive equalization. I still use the low noise power supply and AD823/AD815 line stage that Walt sent me years ago. All powered from two 18v wall warts.
Since I don't believe in posting anonymously, I hope people will not take anything I say as an endorsement or lack of endorsement of any AD products or those of our competitors or anyone else.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.