Current Feedback Amplifiers, not only a semantic problem?

Jump to Latest
#261
Waly said:
A good example of how toxic simulations can be in a discussion. You are trying to make a point, then you got slapped with a simulation. It could be correct, it could be wrong, no way to find out up front. What are you supposed to do? Spend hours in extracting information from the authors, capture the schematic in your tool, verify the models, compare results, etc...? And for what purpose? Meantime, the original topic got diluted to the point of no return.
Click to expand...

So true!
 
#262
Bonsai said:
Not that I am trying to take anything away from your excellent amps Dadod, but that is a sim and you will get the same result with a well designed VFA.

The reason I ask Wavebourne for clarification is this:-

If you measure any amp - or OPA - the distortion will RISE as the input level decreases because the test instrument noise floor is increasing.

The AP SYS272 has a resolution of 2 ppm or circa -114 dBV. It should not come as a surprise therefore that when you measure a good opamp, the best performance is usually around 1~3V output - the AP sweet spot and somewhere before any LSN from the DUT creeps in. Note the AP has gain ranging/attenuation hence the sweet spot spread. If you measure the OPA with a 1mV input, the distortion is correspondingly higher. This has little to do with cross over or any other DUT artefacts - most OPA's are running in class A when delivering only a few mV into a reasonable load - 2k and up though some will do this trick into lower load resistances.

The reason why you don't see rising distortion in tube amplifiers is that the amplifier distortion noise floor is orders of magnitude higher - the test instrument noise floor is buried way below the DUT noise and distortion floor. The result is you get a flat plot until the DUT distortion starts to increases at high output levels.

Sims of course will not show this effect at all - the resolution is for all intent and purposes orders of magnitude below any real world test gear and your plot is an example of this.

Of course, if the DUT suffers gross distortion, all bets are off. However, nowadays, there are precious few opamps designed for audio (or even general purpose types) that would have this problem - even the humble TLO72 is not bad for a 35c device.
Click to expand...

Well summarised - thanks.

Jan
 
#263
Waly said:
A good example of how toxic simulations can be in a discussion. You are trying to make a point, then you got slapped with a simulation. It could be correct, it could be wrong, no way to find out up front. What are you supposed to do? Spend hours in extracting information from the authors, capture the schematic in your tool, verify the models, compare results, etc...? And for what purpose? Meantime, the original topic got diluted to the point of no return.
Click to expand...

Toxic simulations,... or toxic tongue.
 
#264
Bonsai said:
Not that I am trying to take anything away from your excellent amps Dadod, but that is a sim and you will get the same result with a well designed VFA.

The reason I ask Wavebourne for clarification is this:-

If you measure any amp - or OPA - the distortion will RISE as the input level decreases because the test instrument noise floor is increasing.

The AP SYS272 has a resolution of 2 ppm or circa -114 dBV. It should not come as a surprise therefore that when you measure a good opamp, the best performance is usually around 1~3V output - the AP sweet spot and somewhere before any LSN from the DUT creeps in. Note the AP has gain ranging/attenuation hence the sweet spot spread. If you measure the OPA with a 1mV input, the distortion is correspondingly higher. This has little to do with cross over or any other DUT artefacts - most OPA's are running in class A when delivering only a few mV into a reasonable load - 2k and up though some will do this trick into lower load resistances.

The reason why you don't see rising distortion in tube amplifiers is that the amplifier distortion noise floor is orders of magnitude higher - the test instrument noise floor is buried way below the DUT noise and distortion floor. The result is you get a flat plot until the DUT distortion starts to increases at high output levels.

Sims of course will not show this effect at all - the resolution is for all intent and purposes orders of magnitude below any real world test gear and your plot is an example of this.

Of course, if the DUT suffers gross distortion, all bets are off. However, nowadays, there are precious few opamps designed for audio (or even general purpose types) that would have this problem - even the humble TLO72 is not bad for a 35c device.
Click to expand...

Exactly right. 🙂

S
 
#266
Bonsai said:
Blanket statements like 4nV/rt Hz and 16 pA/ rt Hz specifications suck show no appreciation what the specific performsnce requirements of each of the signal chain elements. And 3ppm distortion is totally inaudible.

Again, wild opinions not backed up by experience or any technical consideration whatsoever.
Click to expand...

And shifting the topic from "who's better" to "who's good enough". Nice try.
 
#267
Bonsai said:
Not that I am trying to take anything away from your excellent amps Dadod, but that is a sim and you will get the same result with a well designed VFA.
Click to expand...


After Dadod finished his sim, I helped him to make a physical prototype pair that I could test.

I used the AP2722 and the ShibaSoku 725D and then its output went to a QA401/FFT; The combo allows accurate measurement of harmonics to -160dBv.

I used R loads and both electrostatic and dynamic speaker loads. And, I listened to it.

After several days of carefully run grounding etc...... the amplifier measures as good as it SIM'ed.



THx-RNMarsh
 
Last edited:
#268
RNMarsh said:
After Dadod finished his sim, I helped him to make a physical prototype pair that I could test.

I used the AP2722 and the ShibaSoku 725D and then its output went to a QA401/FFT; The combo allows accurate measurement of harmonics to -160dBv.

I used R loads and both electrostatic and dynamic speaker loads. And, I listened to it.

After several days of carefully run grounding etc...... the amplifier measures as good as it SIM'ed.



THx-RNMarsh
Click to expand...

Richard,

I have no doubt Dadod's amps are good and measure very well and for the most part as simm'd (I've had very good correlation in this regard on my newer amps as well).

The point I was making was that you have to be circumspect when correlating real world measurements with sim projections. You will get ppb performance at moderate to low powers on any high feedback amp - CFA or VFA, small signal or PA. However, you will not be able to measure it (special test configurations a la LM4562 notwithstanding).

(Jan, looks like we cross posted)
 
#270
RNMarsh said:
You missed the point entirely ---- It wasn't about a comparison to a particular VFB amp.

There was an implied general statement that all CMode amps are grossly inferior/bad when it comes to noise......I showed this isn't true.
Click to expand...

Yes, you did, based on your own, proudly announced, criteria "10nV/rtHz is considered low noise".

Flashnews, it is not in my book, 10nV/rtHz is a mediocre noise performance for an op amp, at best. You may also need to check with your otherwise useless wall of books about input current noise and how is that affecting the overall noise performance. And no, the Bentley would not know about, no need to ask.

So sorry, you didn't show ****, other than once again the basic lack of EE101 understanding.
 
#271
RNMarsh said:
is there a way to do a name search for patents somewhere? I want to do a patent search on my name and see which patents show up and how many I have.

Never mind, I found it....... about 11 patents and several new designs not patented but released to public via magazine articles

THx-RNMarsh
Click to expand...

Mr.
Is this the patents you are referring to?
Link

"and several new designs not patented but released to public via magazine articles"

Would that be the Headphone amp. in Jan's Mag?

S
 
#273
Waly, you're making another blanket, poorly thought through assertion again . . .

Most CFA OPA's get fed from lo Z sources (the bias current is a clue) - the higher noise current in these applications is not a limiting factor. If you are feeding from a volume control pot or something like that then no - a low noise current device is better.

But you knew that didn't you - so why the troll? You're thrashing around like a fish out of water simply to score points . . . that you aren't scoring.
 
#274
Bonsai said:
Waly, you're making another blanket, poorly thought through assertion again . . .

Most CFA OPA's get fed from lo Z sources (the bias current is a clue) - the higher noise current in these applications is not a limiting factor.
Click to expand...

Most CFA's by their nature are highly asymmetric in input current errors both DC and noise. You always need to clarify the application and all the real input impedance levels especially inverting vs. non-inverting.
 
#276
sf1412 said:
Mr.
Is this the patents you are referring to?
Link

"and several new designs not patented but released to public via magazine articles"

Would that be the Headphone amp. in Jan's Mag?

S
Click to expand...

Oh. Thank you.... that page shows more than what I found.... 14. but I know there are more.


One circuit topology that I published was for a MC pre=pre. The LLNL lawyers said it was patentable but had no application to what we did at LLNL so released it... so I could patent it at my expense. I published it instead.

Another was the first to take a 2 transistor (Linn?) and make it a push-pull complimentary current mode topology.... published as a line stage.

Another was the first consumer RTA --- Real Time (audio) Anaylzer... 1/2 octave. Published design. We all know this now and many products still use same concept for 'graphic' display.

The concept of a DC-Servo to eliminate the coupling caps.... reduce off-set and drift. First use was in WJ's phono preamp. Published.

Probably a few others I have forgotten about.


THx-RNMarsh
 
Last edited:
#277
Bonsai said:
But you knew that didn't you - so why the troll?
Click to expand...

Because I love watching you and your pseudo engineering fan club gasping for air, deflecting, sliding the target, obfuscating, sacrificing and ignoring every basic engineering fact, in general making a fool of themselves and sometimes not even realizing it, for the sake of defending their religious type of beliefs. But I have to admit, it's getting tiresome.

So point in case, talking about trolling: I made exactly three statements:

1. 10 nV/rtHz is not "low noise"
2. CFA opamps have mediocre noise performance compared to VFAs.
3. Noise current in CFAs is high, usually higher than in VFAs

You jump straight at 3. above and introduce in the discussion the new variable of sources having low impedance, and conclude the current noise would mostly not matter. Set aside it could be argued that op amps could be fed from a low resistive, but high reactive source impedance (and then the noise current could be very bad, contributing more than the voltage noise), the source impedance was not under discussion here. It was just a fair, apple to apple comparison between data sheets. Oh, and you missed addressing 1 and 2 above, but instead talk about trolling, how convenient.

Long story short, I don't see why one would choose the LME49713 CFA over the, say, OPA1611 VFA. Fortunately TI made the selection easy, by obsoleting that LME49713 "audio CFA op amp" absurdity. But I'm sure Bentley owners will quickly buy the remaining distributor stock, in TO99-8 metal cans, at 10 quid a pop.
 
Last edited:
#278
🙂 You are a funny guy, Waly.

Analog Devices calls <10nv low noise.

And then there is the high Z buffer on the CMA - port. Noise issues anyone?

This never ends because there are always work-arounds or exceptions etc.

BTW - I like my new CTS-V better than the Bentley.


THx-RNMarsh
 
#279
Come on folks, some here have invested a lot of time in realizing their ideas and I have seen enough to believe dadod, Bonsai, and ex participants like Ovidu and Glen K have all built exemplary PA's. Why get hung up on what they're called? It might be that it's not literally "current feedback" but I proposed an angle where the concept is not completely bogus especially as a heads up to the average designer. This acrimony serves no purpose except to drive people away. We all sometimes disagree we all sometimes are not totally correct and we all depend on a lot of previous work by many other people.
 
Last edited:
You must log in or register to reply here.
Top Bottom