Confusion about Vari Mu published attack times

I read one of PRR's posts that minimum attack time is calculated by multiplying the side-chain capacitor by the overall resistance in the rectifier path - including the plate resistance of the driver tube and rectifier diode and any resistance after the rectifier.

Take for instance the Collins 26U
Attack time is published in the manual as fast as 0.5 millisecond or 0.0005s
The driver tubes are 6v6. Rectifier is 6AL5.

To achieve this attack time the combined amplifier/rectifier impedance must be 500 ohms. Surely this is a mistake in calculation.

The plate resistance of a 6v6 can be as low as 2000 ohms. I dont know what the plate resistance of the 6AL5 would be - anyone know?

What am I missing here??

The plate resistance of 6AL5 is around 200 Ohms at moderate currents. It is very easy to read that from datasheet.

It is much harder to guess the plate impedance of that particular limiter, because the 6V6 is working under feedback loop and in _pentode mode_ which means that open loop impedance is very high without any load connected. At least I'm not able to estimate how much FB there is or how big is the impedance reflected from the secondary (which determines the open loop plate load impedance)

One could also look at the 6V6 "Impedance" more from the point of view "how much current it can supply". If there is a lot of feedback that is going to determine the max current to the release cap.

Any how, total of 500 Ohms is certainly possible.

So, the attack time can be well below 1ms.

OTOH manufacturers have certainly used figures quite liberally, so it might be more in this case.

-Jonte

Yeh I missed the feedback. I got mixed up with the Collins 26u output and 26J output (no feedback) - I thought they were the same.
So even if you changed the RC network in the 26J to be the same as the 26U - the minimum attack time would not be as fast because there is no negative feedback in the driver amplifier??
So with the 26J - if the 56k resistor after the rectifier was changed to a variable pot and the cap to 1uf (like in the 26U) what would the minimum attack time be?? Please show me your calculations.

So even if you changed the RC network in the 26J to be the same as the 26U - the minimum attack time would not be as fast because there is no negative feedback in the driver amplifier??

Click to expand...

That's the way I see it. In the J model the plate impedance is higher, how much, I can not say, because the schematic I found does not have voltages or trannie ratios shown.

So with the 26J - if the 56k resistor after the rectifier was changed to a variable pot and the cap to 1uf (like in the 26U) what would the minimum attack time be?? Please show me your calculations

Click to expand...

Why would you change the release cap from .22 to 1 if you want fast attacks?

The resistor bypassed and a plate load impedance of, well lets take a familiar value of 5k from datasheet, the 6AL5 is insignificant and we get:

5k*0.22u=1.1ms
(or then 5k*1u=5ms)

But now we are loading the output stage a lot, so waveform will be shaped because of the cap charging current, and the resulting attack time and waveform will be different. I'm not capable of estimating the result, but you should expect to see some soft clipping and somewhat longer actual attack. Perhaps. Might sound good. Anything that happens at sub 1ms timescale is pretty insignificant and compression by default is wave shaping anyway. I dunno.

BTW, from the J model manual we see that the attack time was indeed calculated simply 56k*0.22u=ca.11ms.

BTW 2, we "measured" the attack and release times of UA fairchild plugin. Release was pretty much right, but what the heck, attack is about 3ms to 10dB limiting. Is this cheating or what? Perhaps plugin users don't actually know how freaking fast 660 is.

-Jonte

Thanks Jonte thats great.
So what we are talking about here is reflected load resistance rather than plate resistance per se?

Take for instance another example - on the slower end of the scale - the Federal AM864.
Attack time is published as 0.05 second.
The 47k resistor after the rectifier limits the speed of attack anyway. If there was no resistor there and the 6sk7 tubes are perfectly matched what is the minimum attack time.
Impedance of 6SQ7 driver amplifier and 6h6(negligeable). 27k??
So theoretical attack time of .027s???
The plate resistance of 6SQ7 is very high - on data sheet it is about 60k.

Original Collins output transformer is 8K:600.  I think the manual (online) may list the feedback amount. 

So what we are talking about here is reflected load resistance rather than plate resistance per se?

Click to expand...

Yes.

There is a thread about the Federal AM864. The published attack time makes sense, because the "detector" circuit gets input from one half only. The 47k resistor will probably dominate the resistance in SC circuit. I can not answer to your other questions, sorry, I'm just not very familiar with that topology and don't have time to study it right now.

Original Collins output transformer is 8K:600.  I think the manual (online) may list the feedback amount. 

Click to expand...

I could not find the FB amount from the 26U manual. Too bad. It can not be very high though because the gain from 12AU7 isn't stellar. Could we perhaps throw a figure of 20dB max in the air? In that case the impedance from plate would be 400 Ohms.

Otherwise very good manual to read through if dealing with any kind of Vari Mu:s.

-Jonte

I have 24 dB in my head, but it may be wrong.