see post 336 for circuit on which this current discussion is based
( but now experiementing with FET VAS )
Last edited:
Thanks Mike,
Great explanation..... it's all related to the gate drive, which does not load the first stage.
This is analogously exploited with bipolars by use of an emitter follower to drive the base.
Appreciated, I learn something new.... and am confirmed in my choice of bipolar for VAS!
Hugh
Great explanation..... it's all related to the gate drive, which does not load the first stage.
This is analogously exploited with bipolars by use of an emitter follower to drive the base.
Appreciated, I learn something new.... and am confirmed in my choice of bipolar for VAS!
Hugh
It is true. Using one BD140 as VAS transistor and dropping the OLG by increase feedback resistors
we do not need much or any compensation capacitance.
In my schematic below comp cap is smallest: 10p
Closed loop gain is set by 1500/100 Ohm and is Gain=16.
By further increase the 1500 Ohm feedback resistor
I wouldn't need the 10 pF cap.
Another way would be to increase R2 (100 Ohm) for even more degeneration of the BD140.
Attachments
Hi All,
Just a quick question....what is the official process for testing stability with a sig gen and cro? What is the process for testing capacitive loads?
With the bipolar vas and circuit in post 336 I see no signs of oscillation at any frequency or output voltage between 0 and 20kHz into a 7R resistor or 1uF capacitor. Just nice clean sine and/or square waves (sometimes a little ringing on the square wave, but not HF style stuff). No fuzziness or anything to indicate HF oscillation. Not like my old power follower, see here:
http://www.diyaudio.com/forums/pass-labs/183477-my-take-f4-advice-needed-7.html#post2518117
Just a quick question....what is the official process for testing stability with a sig gen and cro? What is the process for testing capacitive loads?
With the bipolar vas and circuit in post 336 I see no signs of oscillation at any frequency or output voltage between 0 and 20kHz into a 7R resistor or 1uF capacitor. Just nice clean sine and/or square waves (sometimes a little ringing on the square wave, but not HF style stuff). No fuzziness or anything to indicate HF oscillation. Not like my old power follower, see here:
http://www.diyaudio.com/forums/pass-labs/183477-my-take-f4-advice-needed-7.html#post2518117
If I remember, 2.2uF across 8 Ohm is a test for capacitive load.
You can start with lower value caps, say 100nF, and go up to 2.2uF
Actually the most challenging values can be 10 - 100nF so if we want to claim the design us unconditionally stable you would have to test 1n, 10nF 100nF & 1.0uF & 2.2uF in parallel with 8 ohms
I'm sure the higher values will cause ringing but should not oscillate.
Personally I think the 1.0 & 2.2uF is a bit crazy but it seems to be a standard test but in spice on this design I found the real killer value was around 100nF
I'm sure the higher values will cause ringing but should not oscillate.
Personally I think the 1.0 & 2.2uF is a bit crazy but it seems to be a standard test but in spice on this design I found the real killer value was around 100nF
Mike is absolutely right about it. Simulations has shown me that the most difficult load is around 1 - 220nF. To be sure i would recommend to test in 10nF steps.
And a Zobel network would be a good idear together with and output inductor (if you don't fix it in another way =O) ).
Stability testing results
Hello all,
Here are the results of a brief stability testing session.
Now, I'm not sure what to look for exactly, but my understanding is that we are looking for HF oscillation which will present itself as a very blurred waveform on my 40mHz scope. If this is the case then I think you will be quite happy. There are no HF (MHz range) oscillations.
However, there are some pretty terribly distorted waveforms with lots of ringing in the hundreds of kHz range. If this is what you mean by unstable then we do indeed have a problem.
Included in the file is the current schematic which sounds great at 0.5A of bias. Currently no compensation or zobel is employed.
By the way, pdf is best viewed at over 120% zoom.
Greg.
Hello all,
Here are the results of a brief stability testing session.
Now, I'm not sure what to look for exactly, but my understanding is that we are looking for HF oscillation which will present itself as a very blurred waveform on my 40mHz scope. If this is the case then I think you will be quite happy. There are no HF (MHz range) oscillations.
However, there are some pretty terribly distorted waveforms with lots of ringing in the hundreds of kHz range. If this is what you mean by unstable then we do indeed have a problem.
Included in the file is the current schematic which sounds great at 0.5A of bias. Currently no compensation or zobel is employed.
By the way, pdf is best viewed at over 120% zoom.
Greg.
Last edited:
Mmmmmm,
Well, it's pretty incredible that it can stand all that capacitance without any compensation without full scale oscillation - amazing !
However, it may be worth trying some stabilty creating measures just to see it if sounds any better.
But, as you already know, I would suggest non miller measures to avoid loosing any OLG.
Can't wait to build this one
mike
Well, it's pretty incredible that it can stand all that capacitance without any compensation without full scale oscillation - amazing !
However, it may be worth trying some stabilty creating measures just to see it if sounds any better.
But, as you already know, I would suggest non miller measures to avoid loosing any OLG.
Can't wait to build this one
mike
Phew! I was waiting to be laughed at - I take it you did look at the PDF? It does at least need a zobel, which I will add when I can be bothered. I will also try a few other minor things, but really I'm scared to mess with such a good formula.
I've just been listening to it again, and I have to say that there is no doubt in my mind that the bipolar vas is the way to go. It just feels right. You've seen the 100khz wave. It has speed and is well mannered. Brawn and docility.
I think 0.5 amps of bias and 15mA of VAS current is the sweet spot where vocals come to life and separate from the backing instruments. The sound is crisp and clear but at the same time fluid with no harshness or sterility. Beautiful.
Now let's talk vas transistors. Which one do you think might be the best here?
I can't wait for some other people to build it and provide their listening impressions.
I've just been listening to it again, and I have to say that there is no doubt in my mind that the bipolar vas is the way to go. It just feels right. You've seen the 100khz wave. It has speed and is well mannered. Brawn and docility.
I think 0.5 amps of bias and 15mA of VAS current is the sweet spot where vocals come to life and separate from the backing instruments. The sound is crisp and clear but at the same time fluid with no harshness or sterility. Beautiful.
Now let's talk vas transistors. Which one do you think might be the best here?
I can't wait for some other people to build it and provide their listening impressions.
swordfishy
congrats to the result of capacitive testing
You have a very good amplifier there
Will you add filter against capacitive loads?
For example: One bigger resistor 10 Ohm and wire wound inductor around it.
Like 10 turns. And put it in series with output.
Maybe you do this only if you need it.
congrats to the result of capacitive testing
You have a very good amplifier there
Will you add filter against capacitive loads?
For example: One bigger resistor 10 Ohm and wire wound inductor around it.
Like 10 turns. And put it in series with output.
Maybe you do this only if you need it.
Lineup: thank you - it was your design I will try the filter tomorrow. I don't think it needs it, but if it still sounds good I will use one.
Moschfet: yes I know the long wires are bad. I almost didn't post the pictures but I thought it would be good to illustrate just how stable this thing is. Even with the very poor layout it is stable and does a pretty nice square wave. On a well executed pcb it would do even better I suspect.
Miib: thanks. Going to look into that part now. I've never heard of it.
I will try the filter tomorrow. I don't think it needs it, but if it still sounds good I will use one.Moschfet: yes I know the long wires are bad. I almost didn't post the pictures but I thought it would be good to illustrate just how stable this thing is. Even with the very poor layout it is stable and does a pretty nice square wave. On a well executed pcb it would do even better I suspect.
Miib: thanks. Going to look into that part now. I've never heard of it.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.