Tvr, I have been designing complementary differential jfet input stages for about 40 years now. NO, they are NOT mosfets. Mosfets are fragile, not jfets. One of my clients just bought 120,000 of them for future products. We use them at a rate of many hundreds per month or more. My other client has 50,000 in storage. My greatest competitor, Ayre, as about 1 million in reserve. I have thousands, myself. We use them every day.
Anatoliy, I've done it once or twice. 😀 And my tube power amps have almost universally used feedback as well.
Quite the opposite- I was pointing out that this was well-known at that time in general electronics, and was a non-issue in commercial hifi amps shortly thereafter.
I would like to pose a thought problem. What is 'propagation delay'? Are we referring to CLOSED LOOP propagation delay, or OPEN LOOP propagation delay?
Second, think of a black box with a bandwidth of 10 Hz, before it starts to roll off at 6dB/oct. What is ITS propagation delay?
My best guess is that it is 1/8 of .1 sec or 1/80 of a second, or 12.5 milliseconds.
It seems a lot to me. Where did I go wrong?
Because a typical op amp should have a similar OPEN LOOP delay.
SY, PMA, T, Wave? Show me my error, please. 'Enquiring minds need to know'.
Second, think of a black box with a bandwidth of 10 Hz, before it starts to roll off at 6dB/oct. What is ITS propagation delay?
My best guess is that it is 1/8 of .1 sec or 1/80 of a second, or 12.5 milliseconds.
It seems a lot to me. Where did I go wrong?
Because a typical op amp should have a similar OPEN LOOP delay.
SY, PMA, T, Wave? Show me my error, please. 'Enquiring minds need to know'.
Zero interest in that. There's plenty of guys out there (John included) who design excellent SS amps. Excellent tube amps are quite a bit rarer.
a ill formed question - 10 Hz corner frequency is only part of the equation - gain, how much is to be used to meet Av requirement, what’s "left over" as loop gain?
today common SS audio amp numbers are near 1 MHz loop gain intercept for ~160 ns delay - where single pole compensation, conservative phase margin good separation from higher order poles (mostly output stage) is employed
audio frequency "amp delay" can be harder to pin down with high gain circuits using 2 pole compensation - formally the output can just keep accelerating with constant input error signal within the linear 2-pole gain region
"newer" decades old parts like Mosfet, Ring Emitter output Q are probably limited by common "audio" layout practice, and the package parasitics before the "ultimate" limits of ft, transit time terms of the die themselves are a hard limit
if any competent engineer understanding the issues really thought "amp delay" was a problem then why aren’t they using IXYS Linear Mosfets in RF power packaging? http://www.ixysrf.com/pdf/diodes/app_notes/kelvin_lead.pdf
today common SS audio amp numbers are near 1 MHz loop gain intercept for ~160 ns delay - where single pole compensation, conservative phase margin good separation from higher order poles (mostly output stage) is employed
audio frequency "amp delay" can be harder to pin down with high gain circuits using 2 pole compensation - formally the output can just keep accelerating with constant input error signal within the linear 2-pole gain region
"newer" decades old parts like Mosfet, Ring Emitter output Q are probably limited by common "audio" layout practice, and the package parasitics before the "ultimate" limits of ft, transit time terms of the die themselves are a hard limit
if any competent engineer understanding the issues really thought "amp delay" was a problem then why aren’t they using IXYS Linear Mosfets in RF power packaging? http://www.ixysrf.com/pdf/diodes/app_notes/kelvin_lead.pdf
Last edited:
As already pointed , propagation delay in Bjts is extremely short ,
less than 10pS is a typical value and is indicative of the intrinsical
speed of the bipolar devices if they werent "infected" by parasistic
caps and resistors that are an inherent byproduct of the device geometry
and principle.
As for what is commonly understood as delay propagation,
it is the time it take for a device output to reach 90% of its
definitive value , wich is not a propagation delay but rather
a bad interpretation of the device slew rate.
The chicken before the egg. Before you can LOWER the delay, you have to use FEEDBACK to FORCE the input stage to compensate. IF the input stage is NOT perfect, then it will distort, and interesting things will happen. That IS the difference between open loop and closed loop delay.
There is a difference between the 'delay' added by a single-pole low pass filter, and the real delay added by a transmission line or other propagation mechanism. You can undo the first by inserting a complementary filter. You can't undo the second without violating causality, as you would require output to occur before input. Don't get the two confused.
- Status
- Not open for further replies.