Hi SteveG
My statement about slew-limitation crossed my mind because you mentioned it happening mainly when driving it with high frequencies. Having read the other posts I am not convinced about this anymore because the other ideas make much more sense.
Slew-rate limiting usually looks more like "straightening a sinusoidal in the area of it's zero crossing".
While SINGLE ENDED class-A amplifiers like the Aleph definitely never generate any crossover distorion the statement is not generally true for all class-A amplifiers. Push-pull class A amplifiers can generate some crossover distortion, though definitely much less than class B and class AB.
To test if your circuit is really working as intended, it might be worth having a look at how the currents through the current source and the lower MOSFETs behave over one cycle.
grataku
While I believe that the Pass amps belong to the most generously dimensioned , carefully -designed and -built amplifiers on the market, I have problems to believe that they are 100% "cost no object". In the end Nelson (and his employees as well !) has to earn money by selling them.
I also belive that he is a very skilled and experienced audio-engineer but that doesn't mean that it is impossible that one day somebody will come up with an idea that beats everything we know so far, although it didn't work out with SteveG's try (at least this time !).
Regards
Charles
My statement about slew-limitation crossed my mind because you mentioned it happening mainly when driving it with high frequencies. Having read the other posts I am not convinced about this anymore because the other ideas make much more sense.
Slew-rate limiting usually looks more like "straightening a sinusoidal in the area of it's zero crossing".
While SINGLE ENDED class-A amplifiers like the Aleph definitely never generate any crossover distorion the statement is not generally true for all class-A amplifiers. Push-pull class A amplifiers can generate some crossover distortion, though definitely much less than class B and class AB.
To test if your circuit is really working as intended, it might be worth having a look at how the currents through the current source and the lower MOSFETs behave over one cycle.
grataku
While I believe that the Pass amps belong to the most generously dimensioned , carefully -designed and -built amplifiers on the market, I have problems to believe that they are 100% "cost no object". In the end Nelson (and his employees as well !) has to earn money by selling them.
I also belive that he is a very skilled and experienced audio-engineer but that doesn't mean that it is impossible that one day somebody will come up with an idea that beats everything we know so far, although it didn't work out with SteveG's try (at least this time !).
Regards
Charles
grataku,
I really don't have any idea why you're so ticked off about me modifying the circuit...
As I stated before, my main objective was to gain an understanding of the workings of the variable current source. Nelson himself suggested playing around with the percentage of current in that circuit- suggesting maybe something sounded better to someone else than the typical 50% ratio (see zen revisited part 2). Don't get me wrong here- I definitely do not claim to be an expert designer! I am a total beginner in this, and that is why I wanted to know what I was seeing, and what was causing it. I would like to understand what is going on in this circuit so I can use that info in the future. My simulation is for learning.
I have adjusted many different values in an attempt to figure out what the cause is, so far with no success. If I were to plug in the standard circuit unmodified, what did I learn? I still would not know where my design was wrong.
Yes, it looks quite a bit like that, but with more exaggerated "humps" in the falling side. I guess it could be that I'm just driving it hard(near the point of clipping). Is it typical to put in +/- 1 volt and get the amp near clipping? If so, why does the balanced line stage swing a horrendous amount of voltage? Will the bridged design of the AlephX cancel out any of the even order harmonics? (dosen't seem like it works that way like push pull does, correct me if I'm wrong on this).
Anthony,
Thanks for the suggestions. I will try adjusting the source resistors. I was wondering if maybe a lack of open loop gain could be responsible, and I was thinking that maybe my feedback was not working at high frequencies. I really need a good book on feedback- I do not understand very well how to go about using it. So much to learn!
Thanks everyone. I will go home tonight and try a few things and see what I come up with.
Steve
I really don't have any idea why you're so ticked off about me modifying the circuit...
As I stated before, my main objective was to gain an understanding of the workings of the variable current source. Nelson himself suggested playing around with the percentage of current in that circuit- suggesting maybe something sounded better to someone else than the typical 50% ratio (see zen revisited part 2). Don't get me wrong here- I definitely do not claim to be an expert designer! I am a total beginner in this, and that is why I wanted to know what I was seeing, and what was causing it. I would like to understand what is going on in this circuit so I can use that info in the future. My simulation is for learning.
I have adjusted many different values in an attempt to figure out what the cause is, so far with no success. If I were to plug in the standard circuit unmodified, what did I learn? I still would not know where my design was wrong.
Yes, it looks quite a bit like that, but with more exaggerated "humps" in the falling side. I guess it could be that I'm just driving it hard(near the point of clipping). Is it typical to put in +/- 1 volt and get the amp near clipping? If so, why does the balanced line stage swing a horrendous amount of voltage? Will the bridged design of the AlephX cancel out any of the even order harmonics? (dosen't seem like it works that way like push pull does, correct me if I'm wrong on this).
Anthony,
Thanks for the suggestions. I will try adjusting the source resistors. I was wondering if maybe a lack of open loop gain could be responsible, and I was thinking that maybe my feedback was not working at high frequencies. I really need a good book on feedback- I do not understand very well how to go about using it. So much to learn!
Thanks everyone. I will go home tonight and try a few things and see what I come up with.
Steve
I have no problem with people playing with the circuit
or running simulations.
Looking at it, the topology is fine, but I am not familiar
with the output devices, and of course the gain of the
current source may or may not be 1/2 the output, etc.
Keep in mind that the power MOSFET models are not
particularly good at predicting distortion, and this might
be part of the problem.
Of course you are clipping this amp at these levels
with 16 volt rails, and I think you are also seeing
loop saturation recovery.
Properly set up with IRF240's or something similar, I
don't get this waveform with 25 volt rails, but I might
if I clipped it at +/- 24 volt output.
or running simulations.
Looking at it, the topology is fine, but I am not familiar
with the output devices, and of course the gain of the
current source may or may not be 1/2 the output, etc.
Keep in mind that the power MOSFET models are not
particularly good at predicting distortion, and this might
be part of the problem.
Of course you are clipping this amp at these levels
with 16 volt rails, and I think you are also seeing
loop saturation recovery.
Properly set up with IRF240's or something similar, I
don't get this waveform with 25 volt rails, but I might
if I clipped it at +/- 24 volt output.
SteveG said:
The stock Aleph 30 had an input sensitivity of about 775mV rms, that is 2.19V peak to peak. So yes +/-1V will run the amp almost upto full power and if you've played with resistor values it is entirely possible it is now running into clipping.
The balanced line stage was primarily designed to be coupled with the Son of Zen which needs the swing.
Yes bridged amps cancel a fair amount of even harmonics.
Gabevee, it is harmonics.
Nelson,
Thanks for allowing us to play with your designs. I used the models available in the simulator- probably not optimal. I do have the output sitting very close to 0V (millivolts difference). In playing with the circuit, I find that the lower rail voltages are what is causing the problem (or something associated with the lower voltage). I upped the voltage to the 25 volts of the Aleph 30, and the wave is clean up to clipping. So, I am a little bummed, as far as my project is concerned. I planned on using the 16 volt rails for my alephX, so I bought transformers for that purpose.
When you are using multiple output MOSFETS, is the ratio of AC current through the output resistors compared to the current source sense resistors 2:1 for one transistor of the set, or do you multiply it by the number of output transistors? Also, what is loop saturation recovery?
Steve
Thanks for allowing us to play with your designs. I used the models available in the simulator- probably not optimal. I do have the output sitting very close to 0V (millivolts difference). In playing with the circuit, I find that the lower rail voltages are what is causing the problem (or something associated with the lower voltage). I upped the voltage to the 25 volts of the Aleph 30, and the wave is clean up to clipping. So, I am a little bummed, as far as my project is concerned. I planned on using the 16 volt rails for my alephX, so I bought transformers for that purpose.
When you are using multiple output MOSFETS, is the ratio of AC current through the output resistors compared to the current source sense resistors 2:1 for one transistor of the set, or do you multiply it by the number of output transistors? Also, what is loop saturation recovery?
Steve
Re: "high end" cost no object
-Anyway,
-a LED is a component that is easily damaged.
-BY
-1) Connecting in wrong direction, with anything
-but very low current.
-2) When soldering, be careful.
-Apply only as much heat as needed. It melts easy.
-3) When connecting with to much voltage over it.
-The same as applying to much current.
-I am not totally sure about 1)
-I guess your LED shows light by now.
-LED:s are also very useful as Voltage references,
-often used to set working point for current-sources
-groman
-lights up of kind people
------------------------------------------------------------------
The Above is from one of my earlier posts
in Pass Community.
To spare the LAZY guys the trouble to search,
which only a few do, I Quote myself
Thanks for your WISE thoughts HarryHaller
KEEP on Working HARD! Never be LAZY here at DIYaudio.com
I do my work! 😀
------------------------------------------------------------------HarryHaller said:
-Anyway,
-a LED is a component that is easily damaged.
-BY
-1) Connecting in wrong direction, with anything
-but very low current.
-2) When soldering, be careful.
-Apply only as much heat as needed. It melts easy.
-3) When connecting with to much voltage over it.
-The same as applying to much current.
-I am not totally sure about 1)
-I guess your LED shows light by now.
-LED:s are also very useful as Voltage references,
-often used to set working point for current-sources
-groman
-lights up of kind people
------------------------------------------------------------------
The Above is from one of my earlier posts
in Pass Community.
To spare the LAZY guys the trouble to search,
which only a few do, I Quote myself
Thanks for your WISE thoughts HarryHaller
KEEP on Working HARD! Never be LAZY here at DIYaudio.com
I do my work! 😀
So! The constant current source is not constant. It is variable.
Question:
Would that not make it a... quasi push-pull circuit?????😕
Of course, even a P-P circuit biased class-A would not have cross over distortion. But I am curious.
Gabe
Question:
Would that not make it a... quasi push-pull circuit?????😕
Of course, even a P-P circuit biased class-A would not have cross over distortion. But I am curious.
Gabe
Hi Gabevee
Of course would an ideal push-pull class-A amp generate no crossover distortion at all but as there are no ideal amplifiers around ..................
Because the Aleph's current source isn't constant current doesn't mean it is a push-pull amp. It is in my opinion an improved SE amplifier. But that doesn't mean that it couldn't show effects looking like crossover distortion if not done right.
Regards
Charles
Of course would an ideal push-pull class-A amp generate no crossover distortion at all but as there are no ideal amplifiers around ..................
Because the Aleph's current source isn't constant current doesn't mean it is a push-pull amp. It is in my opinion an improved SE amplifier. But that doesn't mean that it couldn't show effects looking like crossover distortion if not done right.
Regards
Charles
Gabevee,
Have you read the zen variations, part 2? It shows how the variable source works.
Steve
Have you read the zen variations, part 2? It shows how the variable source works.
Steve
I have nothing against simulations, mods, and redesign per se. My point, if I even remember what that was, is that if one changes several variables at one time it becomes impossible to figure out which variable did what.
Gabevee,
I think you should join Gromanswe for his next electroshock therapy session.
HH
While adding an led on the CS, regulator on the DC maybe worthwhile improvements, they don't really qualify as redisign.
Gabevee,
I think you should join Gromanswe for his next electroshock therapy session.
HH
While adding an led on the CS, regulator on the DC maybe worthwhile improvements, they don't really qualify as redisign.
WHy? Because there is soemthing electronic I just don't know?
I already know I ain't normal!
Gabe
Steve, consider using two transformers to get +-32
volts instead. When paralleling devices in the current
source, the best approach is to just do it, and then set
the current gain with the resistor that goes to the base
of the NPN from the output. Set it so that the current
source provides 1/2 the output current.
Loop saturation is where the output can't be driven any harder,
as in clipping, and the feedback loop, such as it is, saturates
trying to get more output. It takes time to recover from this.
In your case the biggest culprit is that the MOSFET input
capacitance is very much a function of DS voltage at low
voltages, and when the DS voltage gets down to a volt
or so, the capacitive nonlinearity goes to hell in a bucket.
This shows up at high frequency clipping with low rails.
On another note:
You can get a real nice crossover notch on the Aleph
circuit if you set the gain of the current source greater
than 50% and then overdrive it.
volts instead. When paralleling devices in the current
source, the best approach is to just do it, and then set
the current gain with the resistor that goes to the base
of the NPN from the output. Set it so that the current
source provides 1/2 the output current.
Loop saturation is where the output can't be driven any harder,
as in clipping, and the feedback loop, such as it is, saturates
trying to get more output. It takes time to recover from this.
In your case the biggest culprit is that the MOSFET input
capacitance is very much a function of DS voltage at low
voltages, and when the DS voltage gets down to a volt
or so, the capacitive nonlinearity goes to hell in a bucket.
This shows up at high frequency clipping with low rails.
On another note:
You can get a real nice crossover notch on the Aleph
circuit if you set the gain of the current source greater
than 50% and then overdrive it.
Nelson,
Thanks for the suggestion, and the hints on the current sources.
My transformers have only 1 primary and 1 secondary 120V:13.5V@20amps. What is the best way to go about putting them in series? I would assume at the capacitors? Is this alright to do? I don't have any problem using 4 transformers per amp, if it is necessary.
Thanks for the suggestion, and the hints on the current sources.
My transformers have only 1 primary and 1 secondary 120V:13.5V@20amps. What is the best way to go about putting them in series? I would assume at the capacitors? Is this alright to do? I don't have any problem using 4 transformers per amp, if it is necessary.
Ah! So there IS crossover distortion there!!!!!! From the master's lips to our ears (OK, fingertips to eyes).
Take it back Grataku and others!!!😛
Gabe
I noticed the idea of influence of current source gain.
Is it of worth to set up real gain (changing the value of R2) of the source by measuring the distorsion on the output? I am afraid original values of A3 ( or A30) are not the best (due to sligtly different parameters of used components). I found lowest distorsion at lower values ( about 500ohm instead of 750 or 850). And another problem - distorsion depends on the frequency - rising gain you lower THD at low freq. and rise at high.Solution? Choosing value in between?
Is it of worth to set up real gain (changing the value of R2) of the source by measuring the distorsion on the output? I am afraid original values of A3 ( or A30) are not the best (due to sligtly different parameters of used components). I found lowest distorsion at lower values ( about 500ohm instead of 750 or 850). And another problem - distorsion depends on the frequency - rising gain you lower THD at low freq. and rise at high.Solution? Choosing value in between?
a)
4 of those transformers, each with its own bridge and cap
will give you a nice +/- 36 to 40 volts.
b)
You can indeed set the gain of the current source by
looking at distortion curves. Just adjust to taste.
50% is still the optimal number. Higher than that gives
lower distortion at low wattage levels, but the current
source starts shutting off at the highest levels.
Lower than that gives higher distortion at low levels, and
early clipping on the positive waveform.
4 of those transformers, each with its own bridge and cap
will give you a nice +/- 36 to 40 volts.
b)
You can indeed set the gain of the current source by
looking at distortion curves. Just adjust to taste.
50% is still the optimal number. Higher than that gives
lower distortion at low wattage levels, but the current
source starts shutting off at the highest levels.
Lower than that gives higher distortion at low levels, and
early clipping on the positive waveform.
Thanks yet again, Nelson. Looks like I may build an AlephX-200 instead of the 40 watt version I had planned!
Steve
Steve
- Status
- Not open for further replies.