Re: Phase Intermodulation and FM distortion
This is confirmed in Cabot's paper as being manifest in THD+N analysis.
Bob Cordell said:
This is confirmed in Cabot's paper as being manifest in THD+N analysis.
bobolix said:
I have repeatedly posted my arguments on the matter, this post collects a few:
http://www.diyaudio.com/forums/showthread.php?postid=843470#post843470
PIM measurement with a low jitter soundcard doesn't seem too challenging, just keep the complex amplitudes instead of looking only at magnitude in the FFT
but the crux is the audibilty of PIM/FM distortion products vs AM distortion products, I haven't seen published evidence of large differences in their audibility
PIM
I have a hard time looking at PIM in the same way as clock jitter on a DAC, but I'll give it some thought.
I don't know the audibility of PIM compared to other distortions, but the numbers I measured seemed pretty small.
The main point of my original JAES paper in the early 80's on PIM was that the introduction of negative feedback does not, in general, cause an increase in PIM, but most often a reduction. The other point that I tried to make was that PIM exists in amplifiers with NO negative feedback.
Bob Cordell
I have a hard time looking at PIM in the same way as clock jitter on a DAC, but I'll give it some thought.
I don't know the audibility of PIM compared to other distortions, but the numbers I measured seemed pretty small.
The main point of my original JAES paper in the early 80's on PIM was that the introduction of negative feedback does not, in general, cause an increase in PIM, but most often a reduction. The other point that I tried to make was that PIM exists in amplifiers with NO negative feedback.
Bob Cordell
PIM stuff,
not only miller caps can give the same effect, see my answer to PB2: http://www.diyaudio.com/forums/showthread.php?postid=594792#post594792
Cheers Michael
not only miller caps can give the same effect, see my answer to PB2: http://www.diyaudio.com/forums/showthread.php?postid=594792#post594792
Cheers Michael
Lumanauw,lumanauw said:
I delayed responding until I could check this out. Using two different SMPSs (a .75amp phone supply loaded with a 27ohm 20watt resistor and a Dell laptop supply) Monitoring the amplifier output on my Picoscope, (peakhold, FFT) I found no noise pickup of any kind placing the supply at any position next to the speaker wires, powersupply or external to the chassis. The only part of both the amp or the preamp that was sensitive was the FET differentials. These only picked up the broadband radiation from the Dell supply when I got 3-4" above the fets. I narrowed it down to the fets themselves using the radiation from the resistor that I added to load the phone supply, which I used as a probe.
This was actually better performance than I expected, in fact I was a bit concerned what the results would be after reading your suggestion.
It does make sense though (at least to me), when I think back on of amount of thought that went into the layouts, not to mention the lessons learned from the 30 or so versions that preceded it over the years. (actually I never kept count, 30 sounded right though).
I would be curious as to anyone elses experience with this.
Sorry to highjack Bob's thread again, I just wanted to follow this thought up.
Regards, Mike.
lumanauw said:
Q-7 & Q-8 are run in Class A, and usually at a fairly high current, so that they quickly drain away the Base Emitter capacitance of the Output transistors, for quick Voltage swings, and low distortion.
Just for Vey quick and approx calculations, you can approximate the Base Emitter capacitance of Power Transistors at 11 pf per mA of Emmitter-collector current.
For smaller transistors 17 pf per milli amp ( ofcourse its a MUCH smaller current ) is a nice approx.
Hi, Mike,
If you look only the output trace, you won't find any difference. I cannot see the difference myself if I only looked at the output trace.
The difference appears only if you have filtered the fundamental with "distortion meter" first. Do you use distortion meter in your test or only look at the output signal? In post #156-#159, I use fundamental of 1khz then it is filtered with distortion meter.
The upper track is the fundamental output (you can see it is no difference whether there is no load, loaded with dummy load or loaded with speaker).
The lower track is the "residual". You can see, there is difference (spikes) in residuals (not in the output trace itself) when loaded with speaker cable, both with dummy load and with speaker.
If you look only the output trace, you won't find any difference. I cannot see the difference myself if I only looked at the output trace.
The difference appears only if you have filtered the fundamental with "distortion meter" first. Do you use distortion meter in your test or only look at the output signal? In post #156-#159, I use fundamental of 1khz then it is filtered with distortion meter.
The upper track is the fundamental output (you can see it is no difference whether there is no load, loaded with dummy load or loaded with speaker).
The lower track is the "residual". You can see, there is difference (spikes) in residuals (not in the output trace itself) when loaded with speaker cable, both with dummy load and with speaker.
Hi, MikeB,
In my experiment, in post #157 (no load) and post #158 (dummy load attached to speaker cable), the residual traces (bottom trace) are in the same shape, the difference is only there are spikes.
But why is that in post #159 (speaker load attached to speaker cable), besides the spikes, the residual trace (lower) have different shape than if there is no load (or if the load is only resistive dummy load)?
What makes different shape residual when the test is connected with real speaker?
In my experiment, in post #157 (no load) and post #158 (dummy load attached to speaker cable), the residual traces (bottom trace) are in the same shape, the difference is only there are spikes.
But why is that in post #159 (speaker load attached to speaker cable), besides the spikes, the residual trace (lower) have different shape than if there is no load (or if the load is only resistive dummy load)?
What makes different shape residual when the test is connected with real speaker?
Hi, Janneman,
No, I'm not smarter than anyone
. It's just strange to me that Q5-Q6 have different connection from Q7-Q8. If I Q7-Q8 is connected in classA fashion, then Q5-Q6 better use classA connection too. But why Q5-Q6 is not?
This connection also in Alesis power amp and Fostex power amp. I know both factory (Alesis and Fostex) have their reasons to do this, and I wanted to know why
No, I'm not smarter than anyone
. It's just strange to me that Q5-Q6 have different connection from Q7-Q8. If I Q7-Q8 is connected in classA fashion, then Q5-Q6 better use classA connection too. But why Q5-Q6 is not? This connection also in Alesis power amp and Fostex power amp. I know both factory (Alesis and Fostex) have their reasons to do this, and I wanted to know why
lumanauw said:Hi, Janneman,
No, I'm not smarter than anyone
This connection also in Alesis power amp and Fostex power amp. I know both factory (Alesis and Fostex) have their reasons to do this, and I wanted to know why
There appears to be no advantage to this connection, at all.