Federal AM864/U attack/release mod

[AusTex64]

Posted this under the Dynamics build section, maybe I should have put it here, my bad.

I've seached high and low on the net for a definitite, specific approach to modifying the Federal AM864/U to vary attack and release times. As I understand, the stock 1uf cap paralleled with a 2M resistor give the user a 2 sec release, and the 47K resistor in series with C1 sets the attack time at 47ms.

I'm gonna try using the stock 1uf cap with a 22K resistor in series with a 50K pot for attack (22ms to 72ms) and 100M resistor in series with a 2M pot for release (100ms to 2100ms). That way I can still get the stock TC's and more options as well. I've reduced the 47K attack resistor to 23K by adding another 47K in parallel with no ill effect. I'm more concerned with lowering the release to 100ms. Your thoughts? Thanks!

 

[emrr]

the ? is always 2 sec release by what definition? 

So you've successfully halved the attack time, OK.  I'd try shorting the attack R to see what happens, play with it by ear with the stock release time, decide on a pot or stepped switch value along with a fixed R if necessary for a minimum attack time longer than zero.

Release is as much a matter of what you like combined with how short you can make it before it no longer limits and instead just overloads the output stage.  The shorter you make it, the higher output will be, the less headroom there is.

Blow some sh*t up and see if it catches on fire, I say, no one here really knows, and if they do, that's how they figured it out.  That's how I do everything in electronics, guided by years of experience in grokking what will be a definite bad idea.  I dial in every limiter timing selection mod by ear with a pot on clip leads, and search out the abuse points, decide if they are worth keeping or avoiding.  Math is so cold, sterile, and soundless in this case.  It's more a simple predictor than final arbiter. 

 

[AusTex64]

Doug, thanks for your reply and comments. Here's where I got the info on the limiter timing: http://groupdiy.com/index.php?topic=43763.0

I'm gonna take your advice and experiment with pots and resistors to see what works and sounds good. Just wanted to see if I was obviously violating the vari mu speed limit before I started...

 

[mjrippe]

I can tell you from experience that too small a cap will cause oscillation and "motorboating" particularly with heavy limiting.  You'll know it when you see it!

 

[Matt C]

been staring at this circuit a lot the last few days, planning on building one and trying to suss out what all the pieces are doing (I'm new to compressor circuits).  anyone willing to type up a quick explanation of how the attack/release circuit works in general terms?  seems simple enough but I just can't SEE it.  honestly the whole 6SQ7 "constant output control" section is still a mystery to me.

 

[mjrippe]

anyone willing to type up a quick explanation of how the attack/release circuit works in general terms?  seems simple enough but I just can't SEE it.  honestly the whole 6SQ7 "constant output control" section is still a mystery to me.

If you download the entire manual (first hit on Google) it has a great section on theory of operation.

 

[Matt C]

perfect, I love it when they do that.  thanks

 

[AusTex64]

Hey Matt,

See this: http://groupdiy.com/index.php?topic=43763.0

And this is handy too: http://hughestech.com/rc_calculator/

Robert Mokry
Austin, TX

 

[AusTex64]

I can tell you from experience that too small a cap will cause oscillation and "motorboating" particularly with heavy limiting.  You'll know it when you see it!

I'm going to try using the stock 1uf cap and varying the attack and release resistances to get different time constants, per Doug's instuctions. Will report back on my findings.

One thing I have going for me - it seems I got extraordinarily lucky on my pair of 6SK7's, they seem very well balanced (no humming or weird noises with deep limiting).

 

[PRR]

> the 47K resistor in series with C1 sets the attack time

The 47K is *also* in series with the 6SQ6 stage's output impedance (say 22K) and C6 (being 0.01uFd, it can't pump-up 1uFd C1 in one bang).

 

[AusTex64]

> the 47K resistor in series with C1 sets the attack time

The 47K is *also* in series with the 6SQ6 stage's output impedance (say 22K) and C6 (being 0.01uFd, it can't pump-up 1uFd C1 in one bang).

Afraid I don't fully understand the implications of this...

 

[Matt C]

> the 47K resistor in series with C1 sets the attack time

The 47K is *also* in series with the 6SQ6 stage's output impedance (say 22K) and C6 (being 0.01uFd, it can't pump-up 1uFd C1 in one bang).

I'm in the same boat, not fully understanding these implications, but doesn't C1's charging current come from the rectifier section of the 6SQ7?  not from the triode's plate (through the 0.01uF cap)?

 

[PRR]

> doesn't C1's charging current come from the rectifier section of the 6SQ7?  not from the triode's plate (through the 0.01uF cap)?

And what feeds the rectifier? It doesn't make power on its own.

C1 is charged-up (negative) through the 0.01u and the triode plate-circuit (as well as whatever you have at R11).

If R11 is zero, there's still about 22K triode-amp output impedance (27K||Rp). Plus 0.01u which means C1 can't be full-charged in one gulp, it has to ratchet-up charge-pump.

Here we face the basic limit of simple circuit. It does work well enough for the intended function, minimizing audible clipping and splatter in a transmitter. If you want LOW attack times, you need a LOT more power..... see the Fairchild 660 with its 10 Watt rectifier driver.

Or you could just build this, apply transients and bursts, watch C1 and see how fast it charges-up with various mods.

BTW, if you want fast attack *and* release, just reduce C1.

 

[Matt C]

hmmm, I've got more studying to do.

another thing that is confusing me about that 6SQ7 circuit is that it looks like the triode is totally unbiased.  maybe it just doesn't matter since we're only worried about the loud peaks, but wouldn't it make more sense to tie the bottom of that 500k pot to ground?  or does that just over-complicate things because then the "current control" pot would be adjusting the bias for the rectifier AND the triode?

 

[PRR]

> I've got more studying to do.

You won't find quite this thing in any textbook.

Take Doug's advice. Blow some sh*t up and see if it catches on fire.

If you lack his years of experience in grokking what will be a definite bad idea, then you can delay the smoke-test by doing math. I'm guessing the supply voltage is around 225V? If the tube could actually go short, 225V in 27K is 2 Watts. A 1W resistor will smoke slowly, but we are a long way away from burning the house down. More likely the 6SQ7 triode (like a 12AX7) won't conduct more (even with zero bias!) than like a 60K 40K resistor. Guess low, 40K. So what is 225V across 27K+40K? It is 0.755 Watts total, 0.45W in tube, 0.31W in resistor. IIRC, the tube is good for a whole Watt and you can still get 1/2W resistors.

me> what feeds the rectifier?

My eyes were full of pine-chips. C1 does _not_ charge from the rectifier. The triode plate makes up and down swings, rectifier clamps the up-swings, down-swings are passed from triode plate through C6 and R11 to C1. The re-distribution of charge even for a simple tone-burst is a complicated thing to figure by hand. Charge transfer for speech/music is worse. You can ask a fast computer to make numbers of uncertain meaning, or you can jerry-rig it, *listen*, and know what it sounds like.

 

[AusTex64]

Reporting back. I'm going with a 22K resistor in series with a 50K linear pot to replace R11 for attack. 200K with 2M linear pot for release, only because I would like to offer future users the option of the factory time constants - not sure I'll ever use a 2 second release. Anything faster for attack or release sounds like doo doo to me. This is with the stock 1uF mica cap. I'm guessing that cigarette pack cap might have something to do with the sound/funk of this limiter, but haven't tried another cap to confirm. Do have a 1uF 600V Sprague gumdrop standing by though, might have to do a test. Also gonna add a 500 ohm pot for meter zero on the front panel, just to be elegant. And a Bourns 600 ohm T pad attenuator for the output. And maybe some curb feelers and wire rims too!

Kinda ready to button this thing up and make some music though....

 

[mjrippe]

And maybe some curb feelers and wire rims too!

One word - Spinners!

Seriously though, thanks for sharing your solution and the reasoning behind it.

 

[AusTex64]

So here's a pic of my Federal AM864/U with the mods completed. Gonna get two more DakaWare knobs to replace the chickenheads, but otherwise I'm done. Finally. Gotta say I'm really liking the ability to dial in the attack and release, it makes this unit much more flexible for me.

Thank for your help and guidance.

 

[thomasdf]

I'd be interested in reading a sum up of the mods you've done, and how you achieved. After these few years, do you still find them useful?

I am about to DIY one of these, and I am thinking of using a rotary switch with different capacitors value for C1. If I understood correctly that would act as a attack AND release presest selector? I'd use 4 positions for instances, with different values of C1 and / or R1 choosen by ear.

On that blog, the guys says he ended up using .1uF instead of the stock 1uF! http://akavalve.blogspot.fr/2009/07/federal-tube-limiter-release-time.html

 

[AusTex64]

Thomas,

In post #15 in this thread, I detailed all the mods on my AM864. Please have a look at that.

Yes, you can use multiple caps for different time constants. I think that's how the Schmidlin guy did it. I was looking for something a bit more simple I could apply to a stock unit. When building from scratch, the options are wide open.

Robert Mokry
Austin, TX