Why no feed forward Vari-mu?

[strangeandbouncy]

Hi Guys,


I was just wondering if there are any vari-mu feed-foreward comps? (maybe there are?). Come to think of it, why are optical comps feed forward( maybe not?), and not feedback?


    Kindest regards,


      ANdyP

 

[abbey road d enfer]

Opto devices are largely nonlinear; the use of feedback tends to give them a more predictable behaviour. Vari-mu tubes (and FET's) are more linear but still have large dispersion and drift that make them difficult to use in open-loop mode.
Only the log-antilog Blackmer VCA has the db/volt control law that makes it suitable for accurate feedforward mode.

 

[strangeandbouncy]

Hi Abbey Road De Enfer,


    Thank you for such a succinct explanation!



    Kindest regards,



        ANdyP

 

[emrr]

As has been mentioned, you could reconfigure a Fairchild type circuit to be feed-forward.  I've not heard of anyone actually trying it and commenting on the result.  hint hint. 

 

[Kingston]

I'll do it for my PM670 when it's ready to roll. I already have the relay connections drawn if it sounds good for including as a feedforward/feedback front panel switch.

[edit]

to recap, in a fairchild type circuit, connect balanced input directly to sidechain amp input transformer and disconnect makeup amp output from sidechain input.

I guess we could even have a kind of feedforward/feedback hybrid if we don't disconnect makeup amp output from sidechain in?

 

[PRR]

> Vari-mu tubes (and FET's) are more linear but still have large dispersion

They are not linear in the right way. The dB/Volt is almost linear; exponential. Rectifiers are essentially linear. So we need a LOG function. The easiest way to get that, with just tubes, is feedback around a dB/Volt stage. And as long as you are doing that, it is a feedback limiter anyway.

Same with opto, which loosely follow an exponential; in fact I think that is how my concert limiter works (I forget). Wrap NFB around the LED/opto and it does what you want. Wire two matched LED/optos together, the second one does what the first one is forced to do. Switch some connections, you tap the signal "before" the program-path loss-cell. But it is still a feedback system.

Oh yeah, great dispersion from one device to the next! (I think I spent a season sorting photoresistors into sorta-match bins.)

Any curve can be approximated with linear over some range. There is an Australian "cathode follower" feedforward limiter which:

1) uses cathode feedback to linearize the tube's exponential
2) approximates quite well over a 10dB range but goes way off course beyond ~~20dB

As you say, the excellent LOG-conformity of BJTs (or linear PWM like my Fostex or some broadcast program processors) is what makes feedforward practical.

And I bet, for small amounts of limiting, FB and FF give the same result. The difference is that useful FF is necessarily a very precise processor, while useful FB can be implemented with fairly sloppy techniques. We praised FF, not because it was FF, but because it had very low error. (And because you can push it insanely far without NFB instability.) When that got boring, we swung back to FB, calling the errors "flavor".

Now ANdyP wants to mix-and-match the loose linearity of tube/opto control with the loose action of feed-forward. My suspicion is:

1) it won't work well, incredibly soft knee and erratic "limiting" level
2) by the time you make it useful, you will have either NFB or a super-computer
3) you lost the sweet simplicity and natural action

The FET seems more suited for feedforward. It is an exponential, but silicon FETs follow square-law very-very closely over a wide range. Much better than an exponential approximation to real tubes and photoresistors. Therefore a sharp log-law conversion between linear rectifier and gate ought to give a total response so predictable that feedforward "works". But much of the point of FET limiters is their simplicity. Stick a log-law in the works, you are as complex as a Blackmer but not quite so precise.

 

[strangeandbouncy]

Hi PRR,


  as ever a most eloquent missive! Loads more facts to chew over. Actually ANdyP doesn't want to mix and match, merely curious! I have a 2 3/4 yr old boy who has just started asking "Why?", and it must have rubbed off! ( Whilst he is a child genius(Of Course! aren't they all?), his questions are more often than not about steam engines, tractors and Fire engines, not the niceties of compression!)

  I will definitely try swapping the control amp source(when I finally get around to building) Actually, having a soft knee over a small range of compression could be Hella Cewl, so who knows?


    Kindest regards,


        ANdyP

 

[ioaudio]

german standard compressor is feedforward
big pdf! - http://www.irt.de/IRT/publikationen/braunbuch/U73_U83.PDF

 

[strangeandbouncy]

Thanks Ioaudio!


  that is by far the clearest schematic for the u73 I have ever seen. I have a u74 in buts, and a strange u73 type limiter with eq on it, that has only a high impedance output. I think i might build a pair of compressor/limiters out of them now! I hadn't noticed that they were feed-forward design.


    Cool!


    ANdyP

 

[Michael Tibes]

I have a u74 in buts, and a strange u73 type limiter with eq on it, that has only a high impedance output.

I'm just curious, do you have any documentation on the weird U73? And did you get the U74 to work? They're probably the predecessors of the SSL talkback unit  Both share quite some important parts, but with the U74 being halfway transitorized and the input tranny being different it looks like quite some work to create a stereo unit out of both.

Michael

 

[strangeandbouncy]

Hi Michael,


  yes, I did have the u74 working for a while, I disconnected the mic pre. It always caused me major hum problems, but sounded fantastic! Somewhere I have the schematics for both. Can't remember the limiter/eq number just offhand.


    I know it'll be loads of work. They both use different tubes in the sidechain amp too!



    kindest regards,



      ANdyP

 

[strangeandbouncy]

Michael,


  it's an FB76 (filter/begrenzer), iirc.



      Same size as u73


    ANdyP

 

[PRR]

> german standard compressor is feedforward

Interesting.

The large G2 resistor semi-linearizes the V/I curve. The total range is only 16dB (albeit with very tight specs). Lots of trims. Klirrfaktor 2%-3% at 16dB GR is fine but not "Wow!" like the better bipolar VCAs manage.

Well, there ya go. Ready to clone. Five audio irons. The essential ones are 1:1 or could be 10K-split:10KCT; they seem to like this T188. The choke-coupled output is a frill, the whole output stage after (76) could be anything you want.

 

[strangeandbouncy]

Hi PRR,

  i was led to believe the T188 is higher impedance than that, more like 100k:100k. I am probably wrong tho'! I also remember reading somewhere that they had a range of 20dB gain reduction


    I used to use a pair of u73's. They were awesome, but could never really be considered as a matched pair.


    Kindest regards,



        ANdyP

 

[ruairioflaherty]

I could be on crack (or just very very tired) but I remember vaguely that the last time we discussed tubes and FF and FB that Jacob mentioned that his Vari Mu has a feed forward sidechain. A quick search confirms...

http://www.groupdiy.com/index.php?topic=631.msg11518#msg11518

(very interesting thread too with contributions from GML amongst others!).

Cheers,
Ruairi

 

[strangeandbouncy]

Hi Ruairi,


    Thanks for the link! That actually answers loads of questions I had, but hadn't got round to asking!


  Kindest regards,



      ANdyP

 

[ruairioflaherty]

You're welcome Andy.  I'm very interested in your progress and questions on this one.I find your curiousity and have a go attitude great.  One question for you, is there a reason you're not interested in building Analag's Poorman 660 with the SCampboard for the fairchild time constants?  Seems like it's well documented..

Cheers,
Ruairi

 

[strangeandbouncy]

I would LOVE to build the Analag poorman,


  but I wasn't around whilst it was born, and all the documentation has since been removed. I missed the boat for pcb's as well. I respect Analag's stance entirely. I wasn't here, so that's that! I am not about to start badgering people. I have been seriously impressed by SuperMagoo's very home-made limiter, and figure that I might do something similar, and learn far more than just putting a pcb project together. I perversely enjoy the detective work, and deciding WHAT questions to ask to maximize my education at ( I hope!) the minimum input from the great and the good, who so generously heap their knowledge on us mere mortals.

. . . . So many Vari-mu's, So little time(and budget . . . . .)


    kindest regards,



      ANdyP

 

[ruairioflaherty]

Let me help you with that....email sent!

 

[strangeandbouncy]

Riuari,


    Thank you very much!



  ANdyP