diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Solid State (http://www.diyaudio.com/forums/solid-state/)
-   -   Amp design attempt number 2 (simpler) (http://www.diyaudio.com/forums/solid-state/236002-amp-design-attempt-number-2-simpler.html)

mcd99uk 14th May 2013 11:08 PM

Amp design attempt number 2 (No longer simpler)
 
2 Attachment(s)
Hello all,

Decided something was fundamentally flawed with the last effort. Found the excess phase was at the non inverting input. Also, finding a satisfactory way of loading the VAS so standard miller comp could be used failed. Even though i think that amp was a failed idea I have learnt a lot.

Have gone for something simpler this time. I have an obsession with the HEC at the moment, so forgive that ;) Cascodes were banned and I found that mixing TPC and TMC using an additional resistor works perfectly. I wish I could find where I read that trick so I could thank the person that suggested it. I was amazed at the effect it had on this amp, at least in simulation. Perfect square wave output without the need for a cap across the FB resistor.

I won't bore you with THD spec's or 'owt like that. All I would like to know is whether the design looks reasonable and whether the phase plot looks ok as far as stability is concerned.

manso 15th May 2013 04:08 PM

This is a current feedback type circuit, miller compensation very rarely can be used to compensate this type. Compensation should be to ground or you have your work cut out for using other compensation schemes. If you use a feedback capacitor you are sure to make it unstable, rule no1 with current feedback (youd have to change the noise gain of the circuit for it to be at least viable and that would worsen other parameters such as noise and offset).

Where is attempt no1 ??

overall feedback 15th May 2013 05:30 PM

Hello,
I am afraid that I am interested on THD and other specs...:) Suppose lower line presents phase shift it is still not clear without grid. Scale is huge too. Design reflects imagination, I like that. Ok, how it looks on sim?

Mark Johnson 15th May 2013 06:15 PM

I don't think it's a good idea to use the LTSPICE "default diode" for D5 & D6; much better to download a .MODEL of the actual diode you actually plan to use (1N914B??). I'm also concerned that C4 may be waaaay too big. See Cordell's book pp. 199-200 "The Speedup Capacitor".

mcd99uk 15th May 2013 08:40 PM

Thank you for taking interest in my creations
 
6 Attachment(s)
Manso, I have found the miller compensation appears to work well in this case. I'll have to do some research on CFB amp compensation. And yes a capacitor across the feedback resistor did result in oscillation so I quickly looked for other ways to cure overshoot. Offset is something I am a little concerned about. I know its a result of the input stage and at the moment the servo is taking care of it but I wonder if there is some other way.??


mark, I have read the section you mention and understand what is being said. The counter point to that was by Rod Elliot who reckoned it was safe to make it very large. However, making it much smaller or removing it all together makes very little difference to the THD. I think I'll take it out all together. Diode choice for D5,6 is critical to THD I have found. Leakage appears adversely affect the circuit THD.

Overall feedback, here's some numbers for you

All into 8R load
20Khz THD +/-2V out = 0.000064%
20Khz THD +/-32V out = 0.001782%
1Kz THD +/- 2V out = 0.000005%
1Kz THD +/- 32V out = 0.000132%

Also, here are some FFTs, square wave and clipping plots etc.

Edit: As far as I can tell the phase margin is 52 degrees with 10dB gain margin

Edit: First attempt... http://www.diyaudio.com/forums/solid...le-ef-amp.html

Mark Johnson 15th May 2013 09:52 PM

Are you attempting to estimate gain margin and phase margin, from simulations of the closed loop amplifier? If so, that's not the right way to do it; you need to open the feedback loop and measure the gain and phase of (Vout/Vin) at the unity gain crossover frequency of the open loop amplifier ... including the gain setting voltage divider resistors (R45 and R46 of your schematic).

Please see Figure 4.7 (p.88) of Bob Cordell's book. Notice that (a) at all audio frequencies, there is no feedback; therefore the loop is open. This is because L1+C1 form a lowpass filter that filters out anything above 0.001 Hz; (b) the gain setting resistors R1 and R2 are included in the simulation; (c) the normal amplifier input port is connected to AC ground, and the test input signal is injected via Cc into the other input port for this simulation; (d) the VAS is an inverting stage, so Bob has chosen to label the IPS input pins in a way that accounts for the VAS inversion. Don't let this confuse you.

On another topic, I think I see a LOT of emitter followers + source followers in cascade (series), in your schematic. How much phase shift do each of them add, around the crossover frequency (~ 5 MHz)?

mcd99uk 15th May 2013 09:57 PM

Thank you mark, I will be trying this in a while. Got the book open and reading.

mcd99uk 15th May 2013 11:08 PM

1 Attachment(s)
I have some strange shaped graph. It appears at 3Mhz (unity gain) the total phase at the output is +82 degrees. At the non inverting input the phase at 3MHz is +5 degrees. Most of the changes in phase happen at the ips and the vas. After Q11 not a lot changes.

How do you decipher this mass of traces into something meaningful?

I have attached plot corresponding to where the AC meter is attached in Fig 4.7.

overall feedback 16th May 2013 05:09 PM

Quote:

Originally Posted by mcd99uk (Post 3492644)
I have some strange shaped graph. It appears at 3Mhz (unity gain) the total phase at the output is +82 degrees. At the non inverting input the phase at 3MHz is +5 degrees. Most of the changes in phase happen at the ips and the vas. After Q11 not a lot changes.

How do you decipher this mass of traces into something meaningful?

I have attached plot corresponding to where the AC meter is attached in Fig 4.7.


Can you make some measurements after change R40 value to about 0.3-0.1 ohm?

mcd99uk 16th May 2013 05:58 PM

Yes, no problem. What would like measured?


All times are GMT. The time now is 05:22 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2