Bipolar power amp with cascoded output, stability advice, please

Jump to Latest
#41
Hi Jens,
60dB open-loop gain matches my estimate. The gain is moderate due to the VAS loading resistors R123 and R125.

The closed-loop gain is 30dB. The frequency compensation needs to reduce the open-loop gain by 30dB. One pole at 20KHz will produce a unity-loop-gain frequency of 0.6MHz. This is well within the capabilities of modern transistors. The lead capacitor C107 should not be necessary.

I too prefer moderate open-loop gain and wide open-loop bandwidth, even if it means not winning the THD wars. 😉

Good luck!
Ed
 
#42
The compensation in the OP is fine, although I don't fully agree the approach.
Conventionally, we keep close loop -3dB bandwidth below 1MHz.
For modern output transistors, that have 30MHz fT, have no problem to withstand 3MHz. I believe 3MHz or so should be the new standard.
1st, please verify whether your amp is fine without cascode. Don't fix the compensation if it is not broken.
 
#43
EdGr said:
Look at the design of a good output stage. It has:

  • Pre-driver and driver transistors operate in class A. Only the output transistors run in class AB.
Ed
Click to expand...
Hi Ed,

Driver transistors in Class A. Would it not be well implemented by biasing them with a constant current (to opposite rail, ideally an opposite rail at higher supply level)? That way, they never switch off, and they are not throwing away valuable signal current to increase the voltage in the emitter resistor. I can afford two current sources 😎.

Best Regards

Jens
 
Last edited:
#44
jensting said:
Driver transistors in Class A. Would it not be better to bias them with a constant current (to opposite rail, ideally an opposite rail at higher supply level)?
Click to expand...
Just connecting a resistor between the emitters of the two driver transistors, without a center-tap to the output node, will act as a constant current source because the voltage across the resistor remains nearly constant.
Ed
 
  • Like
Reactions: jensting
#45
As simple as that 😎.

I did see a schematic for an Ampliwire II and noted that the driver emitter resistor was not connected to output, now I'm beginning to understand the interest. Interestingly, the team that did the design I was building, had an earlier amp which had two pairs of emitter resistors for the drivers, a normal pair connecting to output and a 10x bigger pair connecting to ground. But connecting together gets a fixed current nicely.

Best Regards

Jens
 
#46
Re the point of "missing" emitter resistors for the output transistors Q115 and Q118 in
schematic of post 1 please note that original Toshiba 2SC2565 and 2SA1095 are ring
emitter (multi emitter) types and so have integrated emitter resistors. I did not see this
addressed after a quick read through the thread.
 
#47
It’s not enough for thermal stability at elevated vce. It’s enough to equalize cells on the same die but they still require separate emitter resistors to parallel and for thermal feedback in most applications. MJ15024’s have internal ballasting too, but you never see those run without emitter resistors.
 
  • Like
Reactions: Boydk
#53
I

Post in thread 'The skinny on Sanken 2SA1216 Epitaxials'

One thing which has not been mentioned is that RET (and other similar technologies, for example see Sanyo 2SB817/2SD1047) have substantial emitter resistance 'built in' because of the many parallel transistor cells.
This is evident on the Vbe-Ic graphs in the datasheets, which show practically linear behavior once the Vbe=0.6V or so knee is reached. This means bias current setup is will vary compared to some established rules (see D. Self treatise on optimum bias). The 'internal emitter resistors' are sometimes sufficient even to operate the devices without external resistors while still...
 
#55
Well if you DO blow up otherwise perfectly good 2SC2565’s because of lack of emitter resistors (which will both invite thermal run away and make employing SOA protection impossible) don’t say you weren’t warned. You may not be able to obtain more. Those parts are NLA and if you do have real ones and not fakes consider yourself very very lucky. And I wouldn’t press that kind of luck.
 
#56
I am interested in getting an amplifier that's less at risk than the cascode design. So, I will be putting emitter resistors in. But I am interested in finding out about the options available and the design reasoning. So, the Douglas Self book goes on the reading list.

Best Regards

Jens
 
You must log in or register to reply here.
Top Bottom