Federal AM864/U attack/release mod

[thomasdf]

Ok, thank you.
Do you know how much voltage is going thru C1 and R1? On the manual, I read 0V, but when I see the size of C1 I have a doubt...
I am asking this because I need to figure what rotary switch I should use.

Also, I realise by sneaking the Fed+ website that adding a plate balance trim could be a smart move...

 

[DaveP]

Thomas,

You can get an idea of the voltages on C1 by checking the data on the 6SK7 tube, (Frank's site) better still, mock an AM864 up in "birdsnest" format and check it with various tones from a sig gen and music.

I have built a few vari-mu's now and can say from experience that as you reduce the value of C1, the comp usually becomes more unstable and starts to motorboat as stated by others earlier.  The reason is not hard to understand, you have in essence a simple AC rectifier with C1, just like any power supply.  If the cap gets smaller then more AC ripple appears on C1 to get amplified and fed back, hence it becomes an oscillator.  Bigger C1 means better rectifier and less feedback.  You can get away with a lower value C1 with very careful layout and perfect wiring dress, but this was a US army unit with a rough service life so the safety margin had to be wider than for say a home mainland broadcast transmitter.  (repairs in the field)

This type of comp is a "tone box" and will not produce the same DC  for equal inputs of various frequencies, therefore the compression will be different on different parts of the musical spectrum, this is on top of the limited frequency response of the amp as well, hence tone box.

best
DaveP

 

[thomasdf]

Hello Dave

Thanks for that detailled answer.
I know that trials and errors are often more musical in the end than an pen and a paper So I was thinking of swapping capacitors anyway to find out what sounds best to my ears. I will aslo check the voltages in C1 and R1.

My only concern now is that "tone box" that you are talking about. I used a vintage Federal comp in a recording studio for quite a while and it worked great on bass and some vocals, I didn't remember it to be narrow or much limited in terms of bandwidth.
Since I am building the unit from scratch with a new output tube and modern / new transformers, do you think that it will still act as a "tone box"? What mod could I do to the output stage in order to have a better overall frequency response? I don't mind the compression being non linear depending on the frequency spectrum, but I don't want my unit to lack bass or top end. I am not aiming at a perfectly flat response, but something useable as an insert on a lead vocal track or bass track.

I read somewhere that a 15H inductor could be used to avoid motorboating when playing around with R1/C1. I also know that the shortest the time response will be, the more it will overload the output stage but it's fine since I don't want extremely fast times... Just something a bit more groovy than the stock 2 seconds.

Thanks again for your time and precious pieces of advice!

Thomas

 

[emrr]

To get the same result you need identically band-limited transformers.  The same result may not matter!  Somewhere here I think I posted response plots of a Federal.  Clearly limited response, though not to the ear without clear A/B comparisons.  The ear does lie and fill in gaps. 

 

[DaveP]

Yes, as Doug says, do not use the highest quality transformers to match the AM864 you already have.

High quality TX's will produce more DC/Compression at the low bass and high treble ends of the spectrum, this may not be what you want!

If you needed to modify the tone to match the original, you could use smaller coupling caps to reduce the bass, there are lots of ways to restrict the frequency response, it's not so easy to extend it!!

Experiment and play around, is the best way to learn and make something you will be confident and happy with, that's what we all did!

best
DaveP

 

[thomasdf]

Thank you all for your feedback. I am not trying to match the original Federal so it's fine I'd just like to get a cool tube comp that would give me interesting results on lead vocals for instance.

I'll go with Edcors I think because they are really cheap. The only problem seems to be the input transformer needing a center tap. What is your experience with that? Can I use a non tapped transformer? (Such as the Edcors). And if so, how what should I change in the schematics?

 

[DaveP]

You can use an artificial centre-tap with two identical resistors, so if the secondary is a nominal 50k you can use 2x 24k 1% metal film types.

Best to use mu-metal covered IPT's  for noise reasons.  A coil is a great device for picking up magnetic fields (guitar pick-ups) and there could be EM junk from the power supply  popping up anywhere in a new design.

I have found this IPT to be very useful. http://uk.rs-online.com/web/p/audio-transformers/2106352/

best
DaveP

 

[thomasdf]

Thank you for the link, but 50K seems way to much? I am looking for a 600R / 10K Xformer. Sowter has a perfect one... But it's a little bit too expensive for that project right now... But maybe I'll end up buying them.

Do you think I can get away with Edcor + shielding? I was even thinking of building a grounded metal case to but around them. But maybe it's too much trouble for 30 dollars per transformer.

 

[DaveP]

50k is nothing! most grid resistors are 470k or 1M. 

I've tried the edcor 10k/15k IPT's and it's harder to make a good shield than you think, probably needs 2 or three layers of steel to be effective
DaveP

 

[DaveP]

I've found a formula in Morgan Jones book that illustrates the motor-boating issue very well.

Vripple ~= time of half cycle x current/C in Farads.  This is a rectifier calculation but it's still valid for vari-mu comps.

If we take say -30 volts on the grids and a 1meg bleed resistor on C1 then the current will be 30uA.

Taking 50Hz as a benchmark, its half cycle is 0.01Sec

say 1uF for C1

Then Vripple = 0.01 x 0.00003/0.000001

which equals 300mV

If C1 was 0.1uF then Vripple would be 3V

If the frequency was 5Hz (motor boat territory) then the ripple would also be 3V

Probably none of my figures are accurate but the principle is true whatever the real figures are.

Hope that helps
DaveP

 

[thomasdf]

That's very interested indeed. With a bit of maths I guess I can still use my comp as a cool looking ghetto oscillator if it is not working properly haha
... I will try to avoid that  ;D