2 or maybe 3 tone IMD - basically what the SMTE low+hi 2 tone IMD was developed for - but you could use 20 Hz, or even lower to excite "thermal modulations"
also there are "load soak" protocols - over the years real engineers haven't been near as stupid as audiophile gurus want to paint them
also there are "load soak" protocols - over the years real engineers haven't been near as stupid as audiophile gurus want to paint them
Member
Joined 2009
Paid Member
The problem is, the solutions touted for 'memory distortion' significantly change the performance in other ways - usually for the better, so that it's impossible to know if you fixed a problem with memory distortion or just made it a much more technically impressive circuit.
You are right. The Best Way to measure it, is to do an voltage step and analyze the response compared to the input signal. -> square wave signal.
Just some thoughts... Could be wrong in thinking this.
Is it just tempco that is to blame? As the amplifier output voltage changes a whole raft of device changes happen. I know NFB is there to cope with this but the PM/GM of the amp is changing too. This then changes the step response of the amp. For example, to get the fastest response I believe you need a PM of 75 degrees but the PM changes depending on the amp output voltage.
Is it just tempco that is to blame? As the amplifier output voltage changes a whole raft of device changes happen. I know NFB is there to cope with this but the PM/GM of the amp is changing too. This then changes the step response of the amp. For example, to get the fastest response I believe you need a PM of 75 degrees but the PM changes depending on the amp output voltage.
Once again...
On my side, i don't like those you call 'audiophiles'* more than you and use to call them 'idiophiles'.
* snake oil vendors and believers.
But i don't like more pure 'stupid' and pretentious objectivists with their intellectual way to simplify everything and believe they are able to understand everything, with an absolute miss of 'feeling'. Building audio equipment with their eyes and brain, forgetting... their ears !
Your IMD don't reproduce the various temps changes (short and long) that various pieces of music produce in an amp, while we ask it to reproduce so many different combinations of frequencies and levels, continuously changing.
There is some reasons why formula ones have failures during real races, while they are tested intensively in labs.
Last edited:
Problem with engineering is to gain the knowledge to apply the effort in the right direction.
Did you publish any schematics or have them to show with results. I would like to build a PA using this B-B topology.
Thx-RNMarsh
The second attachment is the topology that I use for my amps since about 5 years. Please note that its not a CFA any longer as correctly described by Analog Devices. Its capable of very high slewrates and bandwith like CFAs with advantages in offsets although carries a noise penalty. The inputs can be configured to operate in true Class AB which is the way I use it with folded cascode which in turn is augmented with hawksford cascode.
This topology is in use by Marantz in their top of the range amps and preamps although using current mirror TIS.
If one takes a look at scott wurcer discrete opamp, you find the Jfet version using the topology with the same type vas I like to use. Although I prefer the diamonds I think scotts amp is brilliant and simplifies things. Someone should seriously consider developing that design into a DIY power amp for the forum.
There is a short thread with sims using this topology, search for threads started by member "Moschfet".
You will have high slewrate, identical input impedance on In+ and In-, possible to use vfb compensation with This circuit.
A bit more complex than an ordinary vfb AMP . Only difference compared to a differential pair, is that there is no limit In max current draw set by the current source.
Good example to study for those wrongly claiming that "current on demand" is somehow a property of CFAs only. And for those assuming that a VFA has always some sort of long tail pair input stage, limiting the slew rate.
Waly, is this or Stochino what you were referring to when you mentioned VFAs can have current on demand?
Or did you have a more elegant circuit in mind?
And can you tell us more about your
Last edited:
Correct observations, it is in fact a VFB. Analog devices and other literature describe it as VFB design using CFB principles. First market examples was the LM6172 range of opamps.
You would love to have high slewrate but where you need the high slewrate is where you have the transition In the output stage where the negativ transistor turns off and the posetive switches on and the opposite Way around. This needs to be really fast, but with 21x we talk a step on the diamond buffer of less than 100mV. With 100r it is 1mA.
When we do slewrate test it is 1 volt step or more.
In This case jfet will perform equally good as bjt. But not at 100mV. Also noted by AD In the ad8610 datasheet.
When we do slewrate test it is 1 volt step or more.
In This case jfet will perform equally good as bjt. But not at 100mV. Also noted by AD In the ad8610 datasheet.
Last edited:
I put that up as a transistion point in the CFA discussion here as it is more like what I do and others and mfr as well. It is more practicle circuit and more universal to apply.
But thses things cause confusion when describing operation and a 'pure' CFA is what I strted thinking about at first. Now we can blend VFA and CFA and that is what has been done also by others.
We have some idea of CFA now and idea of VFA... pro-con... strengths and weakness. the B-B sought to take the strengths from each to best advantage. have they succeeded or is there better implementation still?
Thx-RNMarsh
But thses things cause confusion when describing operation and a 'pure' CFA is what I strted thinking about at first. Now we can blend VFA and CFA and that is what has been done also by others.
We have some idea of CFA now and idea of VFA... pro-con... strengths and weakness. the B-B sought to take the strengths from each to best advantage. have they succeeded or is there better implementation still?
Thx-RNMarsh
I would question dc stability over diff input pair In a discrete build. To gain advantage of it, i would Apply re of 10 - 20r. When we have drifting diamond buffers with a drift of delta 5 - 10mv i would start to feel that it is critical with a system gain of > 20gg
Can't you take look at the quoted schematic that manso posted two message above? Does that look like the Stochino amp?
There's nothing to say. The quoted property is obvious for everybody that prefers to live in the Quaternary Cenozoic, rather than in the Mesozoic Jurrasic.
Of course, it applies to true CFAs, not to those bastardized (by using high closed loop gains) CFAs used in audio, called "CFA" only because they may have high slew rate.
A 'failing' of diyAudio is that it doesn't have a patch reserved for that specific theme, like audiokarma.org does:
An externally hosted image should be here but it was not working when we last tested it.
The Cutting Edge (4 Viewing) 
Exceptional Audio: System Building and Fine Tuning in Search of That Last 10%.
I was hoping for something from Guru Waly that was simpler and more elegant than either manso/Marsh's circuit OR Stochino .. but with the same high performance.
Wait a minute .. there IS! It's commonly called a CFA! [deleted 41 pgs of rant over semantic / pedantic issues]
__________________
What about your
?
Last edited:
I thought I posted the LTspice up here way back. I'll have to root through my hard drive again to find the original.