GroupDIY Bridge Compressor.

[DaveP]

Rodney & Erik,

Thanks for the encouragement, I'll do my best.

Re V8 the 6H6, in theory any diode will do (as PRR has said before) but there is a voltage drop with that tube that may or may not be significant to the timing, so I decided to play safe.  I want to keep as much of the original vibe of the amp as possible.  I have already ignored all the original transformers which may give wonderful clarity or the kiss of death, who knows, we shall suck it and see, as we say over here.

baadcode, The resistance of the 6H6 is ~800 ohms and the 6AL5 is ~200, the first attack time is all system based, no resistor at all so you can see why I was reluctant to use a 6AL5 for 7 pin

Thanks
DaveP

 

[gemini86]

So maybe some series resistance would be all that's needed (like a sag resistor in those guitar amp recto kits)

 

[emrr]

Voodoo injection: I've seen many more bad 6al5 than 6h6.  Can't say why, personal experience. 

 

[mikeyB]

The 6AL5s seem to be getting a bit pricey from usual sources!

 

[lassoharp]

The 6AL5s seem to be getting a bit pricey from usual sources!

Been about a 4 to 6 dollar tube for a long time in the US.  Maybe a reason to avoid "doing exact as original" when alternatives are available.  Tell the gouging vendors (who may lurk as guests) "No thank you" if it gets out of hand.  Last summer one US vendor who had supplied numerous popular tubes for 1/5 of price of the bad guys suddenly decided to sell those same tubes as "audio tubes"  :
and increased prices 5 to 10 times as high literally overnight! 

 

[DaveP]

I buy all my tubes from Jim Cross at http://www.vacuumtubesinc.com/
They are $3.00 there.
He ships same day too.
Tip.  If buying from UK its best to keep the value under £18 to stay under the customs radar.

best
DaveP

 

[DaveP]

I have run into a serious snag so I'm going back to the drawing board.

The problem is that an input overload drives the sidechain so hard that it generates sufficient DC to make the bridge tube grid go positive.  This causes massive oscillation.  I am not happy about producing a design that does this and it would make it very tricky for less experienced guys to build and set-up.

A vari-mu design in the same situation would simply drive a tube more negative into cut-off, this design turns the tube on more and more.

This 26W design is also quite tricky to balance, hence the use of a special quality 5814A tube.  It manages to balance high frequencies quite well but when it drops down to 100Hz and below, its becomes more difficult to balance.

I have now been working on this for 2 months so I'm a bit disappointed at this outcome.

I am going to investigate the 26C design now as it operates on a different basis and may be easier to bring to a successful outcome.  I know that will please at least one guy on this forum.

best
DaveP

 

[emrr]

i've never thought of such a thing, and it may have an obvious flaw, but could you not Zener limit the side chain?  Or similar?

 

[gemini86]

would a diode on the grids cathode to ground keep it from going positive?  Seems there has to be some way of preventing it.

 

[DaveP]

i've never thought of such a thing, and it may have an obvious flaw, but could you not Zener limit the side chain?  Or similar?

would a diode on the grids cathode to ground keep it from going positive?  Seems there has to be some way of preventing it.

These are good ideas and I'll give them a try.  My OPT does not have the taps of the original so I guess there is too much output coming direct from the plates.  This does not cure the low frequency balance problems though.  I have found it very difficult to provide perfectly equal but opposite phase signals to the bridge.

Looking at the 26C, it operates on a different principle.  The bridge tube runs as a normal hot bias and peak signals turn the grids negative and begin to uncouple the transformers.

I have an interstage that I did not end up using in my BA-6A project that I recently tested for T6 in the rectifier circuit of the 26C and it worked very well.  I know it won't have any bass response if I use it in the bridge circuit as the inductance is too low, however if I were to drive it from a cathode follower inserted after the input triode, I could drive it from 200 ohms which would give the required bass, even from a low inductance transformer.

Any thoughts

best
DaveP

 

[emrr]

Even Gates Radio stuck with gain reduction elements between interstage transformers through the 70's, probably for reasons of thump removal.

If you can translate down in Z to get that interstage to work full range, it may solve that problem. 

 

[PRR]

> an input overload drives the sidechain so hard ... to make the bridge tube grid go positive

Last paragraph of this post. It's a 4-quadrant multiplier. It's gonna do that.

> would a diode on the grids cathode to ground

The tube has a grid-cathode diode. Maybe just need some series resistance.

However the problem isn't exactly "positive grid". The problem is that we are balancing a tube against a resistor network, but the tube's minimum resistance is ill-defined. Limit the control voltage, trim the resistor to within 3% (for 30db GR before freak-out), then have to change the tube.....

 

[emrr]

I am reminded that max GR on 26C is quoted at 13dB, because the input stage overloads at that point.  Input and output level set on a 26C is very touchy in a modern setting.  And never touched in a broadcast setting. 

 

[DaveP]

> an input overload drives the sidechain so hard ... to make the bridge tube grid go positive

Last paragraph of this post. It's a 4-quadrant multiplier. It's gonna do that.

> would a diode on the grids cathode to ground

The tube has a grid-cathode diode. Maybe just need some series resistance.

However the problem isn't exactly "positive grid". The problem is that we are balancing a tube against a resistor network, but the tube's minimum resistance is ill-defined. Limit the control voltage, trim the resistor to within 3% (for 30db GR before freak-out), then have to change the tube.....

PRR,

I'd forgotten that you had predicted the overload scenario!

There are some good fixes emerging, I think a zener on the CV might help as long as noise is not a problem.  The attack resistors act as grid stoppers on all but the initial fastest setting which is the one I'm testing at the moment.  I'll test all these options when I get back to work on monday.

best
DaveP

 

[emrr]

Zener with a parallel pfd range cap maybe.  Have seen pfd at least on CV at end of chain several places. 

The 26C method is pretty elegant too; just make the input stage blow up before the CV gets too high, then no one will go there, at least without being well warned. 

 

[tardishead]

I built a Collins 26w about 5 years ago. I have built loads of tube compressors over the years some successful some not so. But my 26w clone is is the best.
I use it all the time and it has stayed in its prototype chassis because I am too scared to rehouse it in case anything goes wrong.
It has a huge sound probably to do with the old octals.
You can smash it with level and it just deals with it.
It's super fast. The fastest transients are nothing for this baby.
In fact I really have to control myself because I always get carried away.
IIRC the output transformer taps are necessary to reduce cv voltage and lower impedance even further for the fast attack time. It's hard to get transformers with these specs especially modern ones.
The interstage is a nightmare I tried 3 before I got a rather expensive one from sowter. The inductance is a major issue. It's essential
I would love to see this project come to life especially if it involves cheaper parts
If you need me to check things on my unit let me know

 

[emrr]

The 26C was built, if you missed that thread.  I have it here for a final power supply refinement. 

 

[emrr]

Dave, working on an original Collins 26W, must be last generation as the gain reduction tube is 5965 rather than 5814A.  I've not seen that schematic; it's probably no different. 

 

[DaveP]

I don't see how that tube can work?

The cut-off is all wrong and occurs too early.  It should be set up to run a 12AU7 family tube.

Has someone fitted the wrong tube?

DaveP

 

[emrr]

No, that's what's screened in there as a label.  It's original.  12AV7, 12AZ7 and 6829 are listed as subs for 5965 in the books. 

Remember 6N7GT/G was the 1st tube they used in that position.