I will study your changes @minek123I tried to adjust the TMC circuit, here are initial results:
Thd at 1kHz: 0.000050% (full power)
Thd at 20kHz: 0.019115%
Slew rate (estimate): 42 V/us
Phase Margin: 74
Gain Margin: 15
I also increased VAS current to 8.5mA.
Lineup, I hope you don't mind me messing with your amp. Please review and let us know if these changes make sense?
View attachment 1243481
View attachment 1243478
FFT looks good, with floor at -210 dB, and all harmonics below audible level of -120dB
View attachment 1243479
Have downloaded your schematic.
Why did you change the VAS current?
This is the squarewave with 200nF across load with the original Cello circuit
It is 2Vp-p and a Zobel filter in place.
It seems to be stable but with overshot.
It is 2Vp-p and a Zobel filter in place.
It seems to be stable but with overshot.
Reasons for lower LTP bias current are
- self heat up giving thermal drift
- lower base currents reduce dc offset errors
- decreased input noise current
Last edited:
Why did you change the VAS current?
1) See D. Self's book page 163, 6th edition
Optimal VAS current is determined to be between 6-10mA
2) See this thread https://www.diyaudio.com/community/threads/vas-requirements-for-mosfet-ops.404529/post-7484758
Same conclusion as above
3) From my experience, higher VAS current USUALLY, with some limitations/exceptions, makes amps
a) more stable
b) faster
In most of my simulations, I was able to get the best results with VAS current 10mA and higher.
4) Most of amps I've seen, with 1 output pair (even laterals), have VAS current over 6mA
Last edited:
Thanks to @minek123 I have redesigned Cello One.
It is among other things a completly new TMC compensation.
Also changed currect in LTP and VAS.
Added a Zobel filter.
The result is to find in my first post in this thread:
https://www.diyaudio.com/community/...att-with-tmc-and-laterals.406282/post-7526480
It is among other things a completly new TMC compensation.
Also changed currect in LTP and VAS.
Added a Zobel filter.
The result is to find in my first post in this thread:
https://www.diyaudio.com/community/...att-with-tmc-and-laterals.406282/post-7526480
Was the #62 waveform acquired at point T? If so, it may be a damped vibration (f=1/(2π√(5uH*200nF))=159kHz) due to LC resonance.It seems to be stable but with overshot.
To check stability, it is necessary to observe the ”OUT” point.
If the output hits the positive rail and becomes saturated during excessive input, excessive current can flow through U10 and the bases of U6.
I recommend adding an overcurrent protection circuit.
The resistor added to the emitter of U6 should be about half the value of R3.
Was the #62 waveform acquired at point T? If so, it may be a damped vibration (f=1/(2π√(5uH*200nF))=159kHz) due to LC resonance.
To check stability, it is necessary to observe the ”OUT” point.
It might be disputable, but my impression was that all sims should be done without Thiele LR network (especially squares).
Some amps don't sim/behave well without LR, and designers say that LR is REQUIRED for proper behavior; I wouldn't trust such
amps. In my eyes LR is only for 'emergencies' (like protecting amp from voltage feedback from speaker under certain conditions).
Amp should sim well without it, especially with simple speaker models (R only, or RC only, no L).
Last edited:
While at it, Baker clamp diode (BAV21) can be also added.If the output hits the positive rail and becomes saturated during excessive input, excessive current can flow through U10 and the bases of U6.
I recommend adding an overcurrent protection circuit.
The resistor added to the emitter of U6 should be about half the value of R3.
See https://www.diyaudio.com/community/threads/unusual-amp-from-1987.357369/post-7068145 for an example.
Please use scope time division like 10us/div or 20us/div. I do not understand why you use 200us/ div, such setting masks any details and shows nothing on stability.This is the squarewave with 200nF across load with the original Cello circuit
It is 2Vp-p and a Zobel filter in place.
It seems to be stable but with overshoot
Baker clamp diode should be next. Does it have to be BAV21 or can one use some other diode?
BAV21 is faster, but 1n914 or 1n4148 will also do.
BAV21 is widely available.
I leave building this amplifier to diyAudio members.
I can not build it.
Come on everyone lineup is just freestyling, cut him some slack, at least he was honest in post #1 putting out a clear disclaimer he's only simulating and not intending to build it.
Yes that will I do, in time.
Absolutely before making any PCB.
Was the #62 waveform acquired at point T? If so, it may be a damped vibration (f=1/(2π√(5uH*200nF))=159kHz) due to LC resonance.
To check stability, it is necessary to observe the ”OUT” point.
If the output hits the positive rail and becomes saturated during excessive input, excessive current can flow through U10 and the bases of U6.
I recommend adding an overcurrent protection circuit.
View attachment 1243670
The resistor added to the emitter of U6 should be about half the value of R3.
In one amplifier of mine (on the drawing board), I have no degeneration on the VAS and therefore use another method to limit the base current during clipping: add collector resistance to U10 (buffer / beta enhancer).
What do you consider a safe limit for base current on a 2SA1209 for example? Datasheet does not specify it.
Amp should sim well without it, especially with simple speaker models (R only, or RC only, no L).
I was wrong here. Actually when testing capacitive loads, RL Thiele network helps a lot.
Of course it would be better if amp can drive 1uF load without it, but it's rather not viable proposition for most of amps.
https://www.diyaudio.com/community/...load-capacitance-how-much.145249/post-1845424
https://www.analog.com/en/analog-di...to-avoid-instability-capacitive-loading.html#
Last edited:
H
HAYK
2m of low cost speaker wire I measure 400pF. PA amps are tested with 1uF, 100nF for HIFI.
- Home
- Amplifiers
- Solid State
- Cello One. Good Amplifier 15 Watt with TMC and Laterals