Varimu "design" help

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

mich

Well-known member
Joined
Oct 31, 2004
Messages
449
Location
Israel
Ive been playing with this after doing a few 436 type circuit,and looking at the AM864 circuit diagram.
wanted to hear how pentodes sounds like in the varimu action.

i dont need it to do super fast limitiing,and simpler is better..built it and it works nice and sounds interestingly good...

any comments or sugestions on how to improve things (or how to not burn the house down)will be great..

6sk7varimu.jpg


 
Looking at big picture, and no small details, I would do away with the filament scheme first.  Everyone complains about hum in the Federal, and this begs for it.  Also, I have never seen use of a pot on side chain grid.  The normal method is to vary bias to affect threshold and ratio.  I imagine your time constants may change drastically in practice, once on the bench.  This side chain is fully PP, and will undo the polarity weighting of the Federal.  Be sure you want that, if playing in Federal territory.  I think it's most of the charm. 
 
The trade-off I've discovered during my varimu build is sizing the sidechain components to get the flexibility in attack and release that you want. The smaller the capacitor (C5), the faster possible attack, but the size of the highest release resistance is limited by the tubes max grid leak resistor (anything too big, and your grid voltage starts to climb). So if you want a sidechain capable of slow release, you want a larger C5; fast attack you want a small C5. Looking over the various Vari-mu schematics, 4.7uf is large, but you say you don't want super fast, so maybe you are looking for slower release...
Also, I'm not sure if you need D3/D4.
 
thanks for the replies guys,

once on the bench

- its allready on the bench and working ,but since its my first times of trying things "alone" and not just following schematics - im looking for improvments or things that ive done wrong.

I would do away with the filament scheme first

actually its one of the quietest builds ive had...no hum what so ever...and its running on AC filaments.

if playing in Federal territory

not really.only the input tubes are the same.

the size of the highest release resistance is limited by the tubes max grid leak resistor

for 6SK7 max. grid is 2M i think, so 1 Meg. for  2 tubes.that should give realese of 4.7 seconds with this cap.and 50 mS with a 10k resistor.


for calculating attack we need to know source resistance- wich i think is about 12k?

and will yield 56 mS?
ive puted in a big cap - because with smaller values i get a shift at the beguining of the wave and my tubes are very well matched at various grid voltages...why is that?




 
with a 1k tone burst , the shape gets out of symetry at the attack - instead of "closing in" from both extremes, both sides goes up(or down) for the first mSeconds

-like the zero point shifts up and then gradually goes back to place..not your clasic compresion wave shape...

as the cap gets bigger (and attack slower) - it gets more symetric.

5.3k for the 12BH7 plate is true - so 2 plates ~ 50 mS..

D3/D4 does seem to be unnecessary.
 
Well one half of the 12BH7 handles positive peaks the other negative peaks. So the time it would take to respond to a peak either pos or neg would be half that.
 
I was going to try a circuit like this but drive the rectifier with cathode followers to get a lower drive impedance. How would this work - good attack time?
 
That's what got added to the WE 1126C.  Lost the 6J5's and added 6SN7, with second stage cathode follower driving side chain.
 
That's what got added to the WE 1126C.  Lost the 6J5's and added 6SN7, with second stage cathode follower driving side chain.

- with a transformer driving the rectifier?got a schematic?
i tried to do it with "only" 6 transformers for a stereo unit(and all tube).

i guess that if it was that easy,there woudnt be a need for the fairchild's 100 tubes and transformers...

also,the 6SK7 cutoff point is at -35V...
 
> it was that easy,there woudnt be a need for the fairchild's 100 tubes

You have attack times of 25-50mS.

If you are limiting to avoid distortion, distortion will be audible on each attack.

If you drive a film-sound light-valve, even 1mS transients sound awful. If you are cutting a disk and the cutter even grazes the next groove, the cut is ruined. Then it is worth BIG troouble and cost for a strong sidechain.

OTOH, the simpler Federal limiters seem to be for communications speech where bits of distortion are a mild annoyance, and equipment simplicity (cost and maintenance) is a virtue.

There's many ways to make limiters. This is neither "best" nor "simplest", but perhaps just-right for many modern studio uses.

An asymmetric jump on attack is typical for VCAs with im-perfect balance. Transistor VCAs may be trimmed for fairly low jump. Tubes never match this good, even with trims. It may look awful on 'scope; how does it sound with your system?

> 6SK7 cutoff point is at -35V...

Yes, but at 100V G2 you can get a LOT of gain reduction at -10V. You would never usefully force 6SK7 to -35V in audio; the plate current would be too low for audio loads. (6SK7 works with tuned-circuit RF/IF loads which can have a very high AC/DC impedance ratio, 500K:500 or better.)
 
The 1126 looks a lot like the Gates SA-39.  No transformer driving the SC rectifier.  They simply added CF drive in the last version. 
 
how does it sound with your system?

not distorted , but not very clean..

so this jump has nothing to do with the sidechain/timming network and is only tube imbalance?whats the cure?paralleling tubes?
 
unfortunately you have to buy quite a few to get 2 that work really well together. I got 2 pairs out of 16 tubes recently


 
Im thinking,why should it make a difference with the matching of the tubes? one less variable?(the second grid)
 
Back
Top