Steve Dunlap said:
Ok, try these:
- Distortion vs. frequency graph.
- Distortion vs. output power at constant load.
- Distortion vs. resistive load at constant output level.
- Spectral content at 400W into 4 ohm and 20KHz.
- Spectral content at 400W into 4ohm and 1KHz.
All experimental, of course. Based on your topology, explain the results.
G.Kleinschmidt said:
Hi,
I'm going to have to say that my circuit sim had the
output transistors biased so that the output current would cutoff
just like a normal double EF would. Which means that both
circuits are essentially the same.
It turns out that running my circuit sim at the point I did, yields
the lowest THD. Effectively the krill bias generator is disabled.
When I run the circuit so the output transistors are not
cutoff (enabling the bias generator), this does *not* yield the
lowest THD (by some 10db or more).
So I have to agree at this point, it looks like preserving the
output transistors from cutoff doesn't mean THD is also lowered.
Perhaps this is one of those mystery mechanisms.
Mike
Got Real 'Scope Pix??
Anyone who has built a version of Steve's amps got a good scope and a good square wave generator care to take and post some pix?
(on most modern scopes the "calibrator" output is a very clean fast square wave while usually only at 1khz is still very fast and clean... so that is fast and easy to use)
Like these: http://www.diyaudio.com/forums/showthread.php?s=&postid=1753905#post1753905 FYI...
Would love to see them! 😀
_-_-bear
Anyone who has built a version of Steve's amps got a good scope and a good square wave generator care to take and post some pix?
(on most modern scopes the "calibrator" output is a very clean fast square wave while usually only at 1khz is still very fast and clean... so that is fast and easy to use)
Like these: http://www.diyaudio.com/forums/showthread.php?s=&postid=1753905#post1753905 FYI...
Would love to see them! 😀
_-_-bear
Re: Got Real 'Scope Pix??
In sim for just the output stage, I see about 400v/usec.
20Khz squareware looks very sharp and not too shabby.
But keep in mind that a double EF output on its own, will probably
be even better.
Too bad we don't have a way to look at how good this
particular parameter needs to be...or what is really good.
bear said:
In sim for just the output stage, I see about 400v/usec.
20Khz squareware looks very sharp and not too shabby.
But keep in mind that a double EF output on its own, will probably
be even better.
Too bad we don't have a way to look at how good this
particular parameter needs to be...or what is really good.
I see this morning that Glen already did some sims of a plain old EF output.
I would think THD at full power is a rather insensitive test for crossover commutation effects. As syn08 points out comparison over a range of powers, frequencies, and loads would help. I was just interested in quantifying the improvement, if any. As Bob points out keeping the opposite output device on a little at signal peak is hardly doing anything at crossover except indirectly via dynamic effects. Our worst problem these days is the opposite, quasi-sat, where the base current changes sign when the most "on" device tries to turn off.
Maybe someone who actually built a Krill could quickly run a THD test with a huge capacitor in place of the 1uf that Nelson pointed out. This should turn the circuit into just a bias diode stack.
We build a forward non-linearity compensated video buffer a while back but the cancellation was perfect at only one load impedance. The diffgain/diffphase performance was nearly perfect, but no one though the idea had any commecial use.
I would think THD at full power is a rather insensitive test for crossover commutation effects. As syn08 points out comparison over a range of powers, frequencies, and loads would help. I was just interested in quantifying the improvement, if any. As Bob points out keeping the opposite output device on a little at signal peak is hardly doing anything at crossover except indirectly via dynamic effects. Our worst problem these days is the opposite, quasi-sat, where the base current changes sign when the most "on" device tries to turn off.
Maybe someone who actually built a Krill could quickly run a THD test with a huge capacitor in place of the 1uf that Nelson pointed out. This should turn the circuit into just a bias diode stack.
We build a forward non-linearity compensated video buffer a while back but the cancellation was perfect at only one load impedance. The diffgain/diffphase performance was nearly perfect, but no one though the idea had any commecial use.
andy_c said:
And I will continue to make that claim. I built and tested 4 of these. This was back when I could still do that sort of thing.
I thought I had summarised the topology. Do you have a specific question?
syn08 said:
But if just made up what I claimed, how could you believe anything I post. As long as you do not believe my results, there is little point in providing them.
heinz said:
Hi Hines,
I'm glad you like the amp. It is not unusual to need to adjust resistor values when substituting transistors, however a 50% change seems a bit much. What value are you using for R15? Is it going from +36V to -36V or to ground?
Any listening impressions yet?
Steve Dunlap said:
I am dissaponted Steve. Not willing to accept any challenges, not willing to substantiate any of your claims, not willing to accept that sometimes we make mistakes in our measurements or interpretations, not willing to listen to any reasoning or results. This kind of attitude really doesn't help your credibility, now and in the future. You may of course chose to ignore this - it's your call.
syn08 said:
Not to defend him but he has stated several times that the sims and actual measurements usually differ so to build it and measure / listen yourself for proof.
I intend on doing that myself but have a couple of other projects higher on the list.
My simulator data can be sent to anyone that want it.
So you can check and fix my errors..... distortion is small... i think i can listen above 0.2%.....but this seems nothing, as i have compared with 0.03% amplifier... listening... and perceived the Krill as much better in sonics.
Here is an image showing the DHT.
I would like to remember folks that we can change those numbers tweaking resistances, capacitors and transistors...so... this what i had without tweak.... it is showing 0.3% but i think i can reduce it to 0.1% without too much efforts.... but i do not care with those numbers to loose my time "tuning" it this way.... i love music.... so, it is listening the way i have to evaluate.... listening.
ALSO!... mine was modified and is using awfull transistors...so... the value i have may be one of the most bad pictures we can have from this amplifier....but ...... really.... listen this amplifier and you will know what to do with those numbers.
To my ears.... 0.2% and 0.000008% sounds the same...i cannot listen both distorted signals.... also i do not believe someone can perceive.... so.... it is a bull sheet (listed of animals written into a sheet of paper)
regards,
Carlos
So you can check and fix my errors..... distortion is small... i think i can listen above 0.2%.....but this seems nothing, as i have compared with 0.03% amplifier... listening... and perceived the Krill as much better in sonics.
Here is an image showing the DHT.
I would like to remember folks that we can change those numbers tweaking resistances, capacitors and transistors...so... this what i had without tweak.... it is showing 0.3% but i think i can reduce it to 0.1% without too much efforts.... but i do not care with those numbers to loose my time "tuning" it this way.... i love music.... so, it is listening the way i have to evaluate.... listening.
ALSO!... mine was modified and is using awfull transistors...so... the value i have may be one of the most bad pictures we can have from this amplifier....but ...... really.... listen this amplifier and you will know what to do with those numbers.
To my ears.... 0.2% and 0.000008% sounds the same...i cannot listen both distorted signals.... also i do not believe someone can perceive.... so.... it is a bull sheet (listed of animals written into a sheet of paper)
regards,
Carlos
Attachments
Just write to me an Email and you will have my Multisim 10 files
So, you can compare with your own, or to use it as a frame to produce a correct circuit reducing distortions and increasing performances...tuning to numbers.
But try the one.... build and listen...the only way we really can evaluate things.
Fourier analises will be attached...the THD (DHT in my language) was made using 1 Kilohertz defaut signal..... driven the power amplifier till 50 watts RMS into 8 ohms.... near the maximum unclipped power that is sligtly bigger than this one if continuous measurement is made.
panzertoo@yahoo.com
regards,
Carlos
So, you can compare with your own, or to use it as a frame to produce a correct circuit reducing distortions and increasing performances...tuning to numbers.
But try the one.... build and listen...the only way we really can evaluate things.
Fourier analises will be attached...the THD (DHT in my language) was made using 1 Kilohertz defaut signal..... driven the power amplifier till 50 watts RMS into 8 ohms.... near the maximum unclipped power that is sligtly bigger than this one if continuous measurement is made.
panzertoo@yahoo.com
regards,
Carlos
Attachments
Well... i would never have money to buy such kind of amplifiers
sophisticated and interesting design.... high performance (audible).... if you intended to buy, when factory was running, you would have to sell your Porshe to pay for one of those amplifiers.
And i have not a Porshe to sell.... so...he gave us a nice opportunity....not only to me... schematic is there..subjected to some tweaking and debugging but it is there...and it sounds nice...man!... the unit sounds nice.
Steve, for sure was not the company President, the accounting manager, the salesman, the enginner, the assembler, the laboratory man, the driver and all stuff...he made a part of the game but i think he never used his hands to build...or maybe one to twice...so.... a group, a staff did the work.... small resistances adjustments are needed into the production line....exactly the place were the theory and practice interact.
I am deeply gratefull to Mr. Steve Dunlap, as have shared this nice design to us... i cannot talk about others.... i hope others do not bother Steve.
He was friendly with all of us.... deserves respect and gratitude.
regards,
Carlos
sophisticated and interesting design.... high performance (audible).... if you intended to buy, when factory was running, you would have to sell your Porshe to pay for one of those amplifiers.
And i have not a Porshe to sell.... so...he gave us a nice opportunity....not only to me... schematic is there..subjected to some tweaking and debugging but it is there...and it sounds nice...man!... the unit sounds nice.
Steve, for sure was not the company President, the accounting manager, the salesman, the enginner, the assembler, the laboratory man, the driver and all stuff...he made a part of the game but i think he never used his hands to build...or maybe one to twice...so.... a group, a staff did the work.... small resistances adjustments are needed into the production line....exactly the place were the theory and practice interact.
I am deeply gratefull to Mr. Steve Dunlap, as have shared this nice design to us... i cannot talk about others.... i hope others do not bother Steve.
He was friendly with all of us.... deserves respect and gratitude.
regards,
Carlos
scott wurcer said:
It wouldn't be exactly the same due to the way the input signal is brought to the output stage - it is applied to the collectors of the biasing circuit. Usually the collector would be a current source so the signal wouldn't be able to get through but in this case the transistors are running almost saturated with about 0.2-0.3Vce. Some nonlinearity will be produced here but I don't know if it cancels some distortion from the output stage or if it increases it.
If output bias is increased, giving these transistors more Vce so they work in the linear (?) region there will be a region of decreased transconductance for the output stage, when saturation moves from the first to the other transistor. This will usually not be at 0A output current because of non-identical current gain of positive and negative half.
(see http://www.diyaudio.com/forums/attachment.php?s=&postid=1686992 for a plot of output impedance vs output current in this case. Middle hump in output is because of an underbiased output stage and the hump at 4A is when saturation moves from one bias transistor to the other)
Steve wrote earlier that there is a distortion null at some output stage bias current. If bias current is increased beyond this, distortion will rise again which can be explained this way (at low bias both transistors will always be saturated).
Steve Dunlap said:
Hi Steve,
Maybe I haven't looked hard enough, but I'm only familiar with the output stage you put up. Can you summarize a bit more about the complete amplifier topology, including the input stage, and the VAS.
I understand that the amplifier has no global negative feedback, but does it include local NFB loops? What I'm getting at here is that there are degrees of no negative feedback. For example, if one closed a loop from the output of the VAS (or a driver) back to the input, the amplifier would have no global negative feedback, but might have substantial inner NFB. In such a case, virtually all of the distortion might be expected to be from the output stage.
At the other extreme, if I understand him correctly, Charles Hansen of Ayre has no feedback anywhere at all, with the exception of emitter degeneration.
Where in this spectrum does your amplifier design fall?
Thanks,
Bob
Re: My simulator data can be sent to anyone that want it.
Carlos,
There is 0.2% and then there is 0.2%. It is statements like yours here and un-detailed interpretations of THD measurements that have given THD as a measurement an underserved bad rap.
Some of the finest vacuum tube amplifiers have 0.2% of soft THD, while some terrible solid state amplifiers in the throws of TIM and near-slew rate limiting or awful crossover distortion have 0.2% THD. There is a difference in the 0.2% THD numbers and it is not evident in the simple lumped number. This is what so many people unfortunately DON'T GET.
THD-20 of 0.2% in a modern solid state amplifier is pretty BAD (although even here there are exceptions; you have to look at the spectra).
Don't throw the baby out with the bathwater with such reckless statements.
Cheers,
Bob
destroyer X said:
Carlos,
There is 0.2% and then there is 0.2%. It is statements like yours here and un-detailed interpretations of THD measurements that have given THD as a measurement an underserved bad rap.
Some of the finest vacuum tube amplifiers have 0.2% of soft THD, while some terrible solid state amplifiers in the throws of TIM and near-slew rate limiting or awful crossover distortion have 0.2% THD. There is a difference in the 0.2% THD numbers and it is not evident in the simple lumped number. This is what so many people unfortunately DON'T GET.
THD-20 of 0.2% in a modern solid state amplifier is pretty BAD (although even here there are exceptions; you have to look at the spectra).
Don't throw the baby out with the bathwater with such reckless statements.
Cheers,
Bob
syn08 said:
What are you after?
That Steve will ignore the working circuit and the actual measurements in favor of your hypothesized ideas?
- Status
- Not open for further replies.