According to Nelson’s SuSy article, the big advantage of SuSy is the cancellation of distortion and noise.
I still like to fiddle with the circuit and this afternoon I connected my scope to both + and – out of my AlephX.
I put the scope in substraction mode so I could see the cancellation effect. I made sure the scope was well calibrated because the slightest mismatch gives a non-ideal figure.
I assume that this approach is the correct one to see the cancellation effect. If not, don't read any further.
Ideally there would be a flat trace, meaning the effect of SuSy is optimal. However it wasn’t the case.
Finetuning one channel of the scope’s vertical amplifier I could further improve the trace till it was really flat, no signal showed up anymore, meaning that the X-effect of the Amp can be optimised.
Now, what could be done, apart from matching all the parts as good as possible to improve SuSy?
I know we shouldn’t use servos or other non- esthetical stuff in nice X-amps, but are there simple circuits possible that measure the SuSy effect and then compensate the slight mismatches of the circuit? Or is SuSy by nature self-compensating?
Hope I’m a bit clear. Ask if not!
/Hugo
I still like to fiddle with the circuit and this afternoon I connected my scope to both + and – out of my AlephX.
I put the scope in substraction mode so I could see the cancellation effect. I made sure the scope was well calibrated because the slightest mismatch gives a non-ideal figure.
I assume that this approach is the correct one to see the cancellation effect. If not, don't read any further.
Ideally there would be a flat trace, meaning the effect of SuSy is optimal. However it wasn’t the case.
Finetuning one channel of the scope’s vertical amplifier I could further improve the trace till it was really flat, no signal showed up anymore, meaning that the X-effect of the Amp can be optimised.
Now, what could be done, apart from matching all the parts as good as possible to improve SuSy?
I know we shouldn’t use servos or other non- esthetical stuff in nice X-amps, but are there simple circuits possible that measure the SuSy effect and then compensate the slight mismatches of the circuit? Or is SuSy by nature self-compensating?
Hope I’m a bit clear. Ask if not!
/Hugo
I'm certainly not claiming that my AlephX is build like a factory one. 
However, the P-P soldered board is fairly symmetrical,
mosfets well matched, wire about all the same length, bias at both sides equal and so on.
The main concern was if it is possible to measure the asymmetry and do something about it electronically with some simple components. A SuperSymmetry comparator or something like that.
Not being a designer, I wouldn't know where to start. I thought the idea wasn't a bad one, hence this thread.
I would like to find out if some of you have “been there and done that”.
/Hugo
However, the P-P soldered board is fairly symmetrical,
mosfets well matched, wire about all the same length, bias at both sides equal and so on.
The main concern was if it is possible to measure the asymmetry and do something about it electronically with some simple components. A SuperSymmetry comparator or something like that.
Not being a designer, I wouldn't know where to start. I thought the idea wasn't a bad one, hence this thread.
I would like to find out if some of you have “been there and done that”.
/Hugo
Petter
I bought 50 IRFP240’s and measured them all with the method proposed at Passdiy.
Then I took two pairs of 16 devices with the closest match to each other.
The middle 8 ones were used for the gain devices, the 2 X 4 outer ones for the current sources.
Your approach will probably give a better result.
Don’t get me wrong, I’m not complaining about my amps, the sole purpose of this thread is to know if it is possible to measure the SuSy effect electronically and within the amp and compensate for slight mismatches.
Some kind of “Constant Active SuSy Source”
/Hugo
I bought 50 IRFP240’s and measured them all with the method proposed at Passdiy.
Then I took two pairs of 16 devices with the closest match to each other.
The middle 8 ones were used for the gain devices, the 2 X 4 outer ones for the current sources.
Your approach will probably give a better result.
Don’t get me wrong, I’m not complaining about my amps, the sole purpose of this thread is to know if it is possible to measure the SuSy effect electronically and within the amp and compensate for slight mismatches.
Some kind of “Constant Active SuSy Source”
/Hugo
Hi Hugo,
where is this article about susy ?
As I focus on how to avoid excessive power dissipation, I had no time to try to really understand susy.
Could you please explain to me, where and why exactly cancel the distortions ?
What exactly did you measure ?
Do you have a scope with differential input ?
Another thing that I do not understand:
If a sine wave is distorted without susy, lets say it looks like a triangle, then both sides will be a triangle which means
(if you measure +signal + -signal you have 0 signal & 0 distortion wether distorted or not.)
If you make +signal - -signal you have 2xamplitudesignal & 2xamplitudedistortion.
My thinking error ?
If the sine is distorted with susy then if you make +signal - - signal you get 2xamplitude signal & 0 distortion.
(If you make +signal + -signal you get 0 signal & 2xamplitude distortion.)
Again thinking error ???
I guess your ascope makes -.
So you will have a doubleamplitude sognal with or without distortion.
How will you see if the sine is pure ?
I will get a spectrum analyzer in a few days, hope that helps me to understand....
Greetings, Bernhard
where is this article about susy ?
As I focus on how to avoid excessive power dissipation, I had no time to try to really understand susy.
Could you please explain to me, where and why exactly cancel the distortions ?
What exactly did you measure ?
Do you have a scope with differential input ?
Another thing that I do not understand:
If a sine wave is distorted without susy, lets say it looks like a triangle, then both sides will be a triangle which means
(if you measure +signal + -signal you have 0 signal & 0 distortion wether distorted or not.)
If you make +signal - -signal you have 2xamplitudesignal & 2xamplitudedistortion.
My thinking error ?
If the sine is distorted with susy then if you make +signal - - signal you get 2xamplitude signal & 0 distortion.
(If you make +signal + -signal you get 0 signal & 2xamplitude distortion.)
Again thinking error ???
I guess your ascope makes -.
So you will have a doubleamplitude sognal with or without distortion.
How will you see if the sine is pure ?
I will get a spectrum analyzer in a few days, hope that helps me to understand....
Greetings, Bernhard
Bernard
I’m still not sure that what I did with the scope and the result of my measurements where what I thought they were.
Let me explain. Every decent scope has two channels. If you’re lucky the signal from the second channel can be inverted. That means if you put two signals into the scope, in channel A the non-inverted and in Channel B the inverted signal, (+Out and –Out of the amp) and you switch the scope to the ‘add’ function, both signals will be added to each other.
If you push the invert button from channel B, both signals will be substracted from each other. Ideally when both parts of your amps (and AlephX’s are bridged amps and thus have two parts) are behaving identical, you should see nothing but a straight line on the scope.
The SuSy topology pushes the whole process a magnitude of 100:1 further according to Nelson’s article .
The whole point of my thinking is that when I substracted the two channels the line was not straight but I could see remains of asymmetry from both halves of the amp. We’re talking µV, perhaps mV here. Not a big deal.
Still, I wonder what I saw. Was it the ‘SuSy filtered’ remains of the good signal and thus the asymmetry of both halves or did I see the distortion of the amp.
If it was the SuSy-filtered signal then both halves should be better matched, if it was the distortion, could there be a way to make the SuSy effect smarter?
Surely an X-circuit can’t make an undistorted signal out of a distorted one. The X only cancels distortion generated by both halves of the amps.
/Hugo
I’m still not sure that what I did with the scope and the result of my measurements where what I thought they were.
Let me explain. Every decent scope has two channels. If you’re lucky the signal from the second channel can be inverted. That means if you put two signals into the scope, in channel A the non-inverted and in Channel B the inverted signal, (+Out and –Out of the amp) and you switch the scope to the ‘add’ function, both signals will be added to each other.
If you push the invert button from channel B, both signals will be substracted from each other. Ideally when both parts of your amps (and AlephX’s are bridged amps and thus have two parts) are behaving identical, you should see nothing but a straight line on the scope.
The SuSy topology pushes the whole process a magnitude of 100:1 further according to Nelson’s article .
The whole point of my thinking is that when I substracted the two channels the line was not straight but I could see remains of asymmetry from both halves of the amp. We’re talking µV, perhaps mV here. Not a big deal.
Still, I wonder what I saw. Was it the ‘SuSy filtered’ remains of the good signal and thus the asymmetry of both halves or did I see the distortion of the amp.
If it was the SuSy-filtered signal then both halves should be better matched, if it was the distortion, could there be a way to make the SuSy effect smarter?
Surely an X-circuit can’t make an undistorted signal out of a distorted one. The X only cancels distortion generated by both halves of the amps.
/Hugo
Hugo,
could you explain how the distortion cancels ?
Should be possible in just a few sentences...
If you have a positive sine half on chA and a negative on chB and you invert chB, you have two positive halfs. If you add theese, what will you get ???
Maybe I am tired, I should switch on my scope and see what happens
Bernhard
could you explain how the distortion cancels ?
Should be possible in just a few sentences...
If you have a positive sine half on chA and a negative on chB and you invert chB, you have two positive halfs. If you add theese, what will you get ???
Maybe I am tired, I should switch on my scope and see what happens
Bernhard
There's no way I could explain X better then Nelson's article.Bernhard said:
Maybe I am tired, I should switch on my scope and see what happens
You must be tired if you didn't see the link I provided.
And yes, get your scope tomorrow and see what happens.
Thanks stefanobilliani, looks like a very interesting thread. I must have missed it among the many others.
/Hugo
?
SuSy is just a buzzword
Common mode cancellation comes from the fact that the amp is differential, and has nothing to do with Susy.
You can either stop there and have nested differentials (BLS+ SOZ) or add feedback to connect the two stages. There are not many ways to achieve negative feedback with nested differentials.
When applicable, the idea is clever and it can be argued that the whole is more than the sum of the parts.
Without pulling Nelson's papers out, I don't recall seeing improved measurements for the SySy circuit. I can remember a lot of peanuts and chocolate that usually irritate my intestine.
SuSy is just a buzzword
Common mode cancellation comes from the fact that the amp is differential, and has nothing to do with Susy.
You can either stop there and have nested differentials (BLS+ SOZ) or add feedback to connect the two stages. There are not many ways to achieve negative feedback with nested differentials.
When applicable, the idea is clever and it can be argued that the whole is more than the sum of the parts.
Without pulling Nelson's papers out, I don't recall seeing improved measurements for the SySy circuit. I can remember a lot of peanuts and chocolate that usually irritate my intestine.
Hi Hugo,
I got it
Picture shows a 10kHz sine from a not so good Siemens function generator.
Marker is on 10kHz and cursor on 20kHz. 2nd harmonic is -55,7dB, there is a lot of other dirt
like strong third harmonic and very strong 7th harmonic and even disharmonic crap.
Another better sine generator shows only 2nd and very little higher harmonics.
I can see down to -75dB very clear frozen curves without noisfloor by averaging up to 500 or more measurements and storing them.
Averaged & stored curves may be subracted from or added to each other
That means I can test all my madness circuits and see what's up
Greetings, Bernhard
I got it
Picture shows a 10kHz sine from a not so good Siemens function generator.
Marker is on 10kHz and cursor on 20kHz. 2nd harmonic is -55,7dB, there is a lot of other dirt
like strong third harmonic and very strong 7th harmonic and even disharmonic crap.Another better sine generator shows only 2nd and very little higher harmonics.
I can see down to -75dB very clear frozen curves without noisfloor by averaging up to 500 or more measurements and storing them.
Averaged & stored curves may be subracted from or added to each other
That means I can test all my madness circuits and see what's up
Greetings, Bernhard
Attachments
grataku said:It's more like: "I got a big spectrum analyzer but I don't know how to use it"...
I'm sure your's is bigger than mine
Don't worry...
Also the sine generator curve can be subtracted from the amp (+generator) curve, so the pure amp distortion is left.
Before doing serious things I have to play a little.
Super Symmetry as I understand it
The thing about SuperSymmetry is that it uses feedback around each half to make each half look more alike.
As far as I can see, it really works the same way as a regular differential circuit... it's just that we now have two halves matched better than normal, thanks to the feedback applied to each side.
edit: I believe it is supposed to 10x better than just regular differential. (factor of 100 cancellation vs. factor of 10.) I, too, am going by memory... ('watakashi-no kyoku-ga tashika naraba...')
eL
The thing about SuperSymmetry is that it uses feedback around each half to make each half look more alike.
As far as I can see, it really works the same way as a regular differential circuit... it's just that we now have two halves matched better than normal, thanks to the feedback applied to each side.
edit: I believe it is supposed to 10x better than just regular differential. (factor of 100 cancellation vs. factor of 10.) I, too, am going by memory... ('watakashi-no kyoku-ga tashika naraba...')
eL
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.