Hi guys 
Thank you very much for very pleasant and encouraging words I really appreciate them a lot.
I am glad of latest activity going on in this thread, hopefully will produce some new incarnation of SSA. Although Bigun also made TGM-5 all BJT SSA it would be refreshing to see new all BJT version by hahfran.
Regarding all requests for PCB group buy in this phase I don't want to hurry too much. SSA BIGBT HP PCB is functioning perfectly with no wrong tracks. As you may remember I add one NTC to input BJT pair and this, if in the phase of testing prove to be reliable solution, should be included in new modified PCB - minor changes. Now the power band-width tests will be performed to calibrate some parts to optimal values, etc. Anyway please give me a little more time and patience that I conclude this version to final product and than we can certainly discuss about group buy.
One very nice and prosperous day to all of you, regards Andrej
Thank you very much for very pleasant and encouraging words I really appreciate them a lot.
I am glad of latest activity going on in this thread, hopefully will produce some new incarnation of SSA. Although Bigun also made TGM-5 all BJT SSA it would be refreshing to see new all BJT version by hahfran.
Regarding all requests for PCB group buy in this phase I don't want to hurry too much. SSA BIGBT HP PCB is functioning perfectly with no wrong tracks. As you may remember I add one NTC to input BJT pair and this, if in the phase of testing prove to be reliable solution, should be included in new modified PCB - minor changes. Now the power band-width tests will be performed to calibrate some parts to optimal values, etc. Anyway please give me a little more time and patience that I conclude this version to final product and than we can certainly discuss about group buy.
One very nice and prosperous day to all of you, regards Andrej
Maybe the thermal track diode causes this distortions, at 1 mA first stage bias, also diode bias current, dV/dI curve is higly nonlinear. Global feedback can compensate this behaviour to some extent but not to ppm level.
If you really believe this (did you ?), why to be interested in current feedback amp ? Classical ones (voltage) have better P.S. ripple rejection.
Yes the CMMR of CF is poor. However the linearity of ic/ib is unmatched by any other device inclusive triodes.
Hahfran, on my side, i consider slew rate is the most important thing, and the first one i look in a IC data-sheet. OMHO, the strict minimum is 20 time the values you mention. It is the key for low transient IM, flat phase response and transparent reproduction in a closed loop amp. Insufficient open loop bandwidth is the reason why CR has bad reputation.
We discovered that 40 years ago and i had made a lot of studies about that. I was dedicated on CF in the same time than Analog Design.
My last work was converting a still very fast classical amp in SSA. With a same amp, the difference is obvious, looking at numbers and they correlate with my ears.
Google translate this (clic on the images to magnify them).
www.esperado.fr - Le crescendo revisité
220V/µs -> 1200V/µs
And a distortion.divided by ~10.
A last thing we have to consider. Supposing you have a 10v/µs preamp with 1V rms output. If your amp is a 100W 8 ohms one, you'll have 28v out at full power.
So you need 280V/µ to be able to follow the spec. 340V/µs for 150W, 400V/µs for 200w etc...
More power, more open-loop bandwidth, more and more difficult to reach.
Last edited:
I don't doubt that not at all. However there is no source which can reproduce slew rates you demand. The max frequency a CD can reproduce ( sampling theorem) is 22 kHz and now assume 10 watts into 8 ohms that yields 9Volts peak and the slew rate is 2*pi*22000/sec * 9 Volts= 1243400 Volts/sec or 1.24 Volts/ µsec
even at 100 Watts 4 Volts/µsec is enough.
even at 100 Watts 4 Volts/µsec is enough.
Its a shame...the circuit at least in simulation can be tweaked to yield a constant idle current from junction temp 20 degree to 130 degrees...Maybe the thermal track diode causes this distortions, at 1 mA first stage bias, also diode bias current, dV/dI curve is higly nonlinear. Global feedback can compensate this behaviour to some extent but not to ppm level.
wow.
But the distortion level is outright too much.
That alone is not bad, but the harmonics spectrum is poor as odd harmonics dominate . That will not sound good.
Thus next attempt coming soon.
Yes that is open loop bandwidth. A technical aspect. But for the aspect of audible quality what is not recorded cannot be amplified. And should not. Thus for the audible impression the slew rate of the amp is sufficient if it is 10 V/µsec at full output. Btw. which speaker can reproduce 10 Volts/µsec?
You find that in step response. The fastest tweeters need 18 µsec for max excursion.
If you build an amplifier without loop feedback I think your augument may have some truth.
For an amp with loop feedback speed is absolutely essential to achieve a good sound - if you don't quite get that yet you should try building and listening to a few fast & slow loop FB amps and then making comments based on ur own experience.
For a fast amp - the SSA is a very very good contender but as stated earlier perhaps not ideal for a beginner - Fetzilla may be easier even if not quite as fast.
good luck !
For an amp with loop feedback speed is absolutely essential to achieve a good sound - if you don't quite get that yet you should try building and listening to a few fast & slow loop FB amps and then making comments based on ur own experience.
For a fast amp - the SSA is a very very good contender but as stated earlier perhaps not ideal for a beginner - Fetzilla may be easier even if not quite as fast.
good luck !
Topic for a new thread:
Basicly, a no feedback performance depends of of the linearity of the components involved as the dv/di is much lower the in an Open loop design.
The bandwidth and gain of closedloop design does not evolve a lot more, only of it uses several gain stages or take the feedback before the output stage.
So: to lower thd we need to use the design principles from Open loop design, or in other words improve the linearity of each stage. Basicly a proberly design closed loop design should on paper performe better than a Open loop design if there is taken Care of the stability.
Basicly, a no feedback performance depends of of the linearity of the components involved as the dv/di is much lower the in an Open loop design.
The bandwidth and gain of closedloop design does not evolve a lot more, only of it uses several gain stages or take the feedback before the output stage.
So: to lower thd we need to use the design principles from Open loop design, or in other words improve the linearity of each stage. Basicly a proberly design closed loop design should on paper performe better than a Open loop design if there is taken Care of the stability.
The problem that most older amps had a very low open loop cutoff frequency In closed loop an upper cutoff of some 100 kHz is achieved because the feedback factor is very high.
That is the open loop amplification factor is very high and so is the gain bandwidth product. Assume the amplification factor is 1000 and the upper cutoff is 1000 Hz then the gainbandwidth is 1 mHz. The feedback reduces amplification to 10 but does not affect the gain*bandwidth so the upper cutoff becomes 100 kHz. But the feedback speed is still determined by 1 kHz open loop, thus such amps produce TIM and other nasty stuff.
Bode diagrams become important lead lag compensation and so on.
Thus what matters is not the slew rate rather it is the open loop cutoff frequency. This must be at least as high as the cutoff of the source.
IT is the heart of the topic, on the contrary. Why did this SSA topography sound so good ? Because it use CF ( and the elegant way this feed back is mixed with the input signal).
It means higher slewrate,= less distortion, and no active stage in the feedback path = less distortion.
Again it is not about the bandwitch we need for your ears, it is about transient distortion in closed loop systems.
To begin to experiment that, just compare (listen) a strait line with a 10V/µs OP amp ( gain 10 + 11/10 bridge), inside a high end hifi system on one side, and do the same with some 500 or 1000V/µs video OP amp, on an other, you will be surprised.
To make your mind about things like that, I suggest you use your ears. Then to try to correlate your experiences with mathematics. Because there is a lot of factors we don't think about at first sight, or we do not know how to measure.
My assertions about slew rates lie on >40 years of experience and experiments, both as an electronic designer and as a sound engineer. And this is well known nowadays, even in AES; Ask Nelson Pass or Douglas Self.
Last edited:
Or Bob Cordell: CordellAudio.com - Another View of TIM
or Walt Jung: http://waltjung.org/PDFs/SID_TIM_TAA77_P1.pdf
or Walt Jung: http://waltjung.org/PDFs/SID_TIM_TAA77_P1.pdf
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.