GroupDIY Bridge Compressor.

[DaveP]

This takes over from the Group DIY Varimu thread.

I've now had a chance to try out some designs on the bench and I feel fairly confident about getting a decent Bridge compressor up and running.  This will be based on the Collins 26W but it will not need the fancy interstage to work.

Just for the record, the 26C and 26W use a variable (tube) resistance in a bridge circuit to control the compression.  The nearest equivalent to this is an LA-2A which also has a variable resistance to control the compression, albeit via an EL panel.  The advantage of this approach is that there are no thumps and no problems with matching Varimu tubes or keeping them balanced.

The disadvantage is that there is only about 20dB compression available, if you want more than that, then this comp is not for you.  The audio diodes are reversed to provide a +ve DC CV and the control tube is biased to -30vDC.  As the +ve DC rises it begins to cancel the -ve CV and turns on the tube.  When the Tube is fully on its internal resistance balances the opposite leg of the bridge and the signal is progressively cancelled and reduced.  This is an elegant and distortionless way to obtain compression.

Now I need some guidance:

The existing tubes, namely 6N7's and 6F6's draw hefty heater currents, so the total draw is 3.6A which is a lot, namely 23W of heat!  I realise the 6N7's are nice vintage tubes and very reasonably priced, we could even use one for the output stage, instead of the 6F6's. 

The 6N7's are very linear and with a mu of 36 are not easy to find a sub (6SN7GT are only 20 mu).
6DJ8 are possible but more than double the 6N7 price.  Any thoughts?

Next question is what input and output levels are required?  As most people will use a mic pre, do we need low mV sensitivity or an input TX?  Similarly, do we really need 28dBm output levels that will require padding down?

Over to you.

best
DaveP

 

[1sound]

Russian made 6N1P's are kind of equivalent to the 6DJ8 (with mu of 35) and in current production....  They're also reasonably consistent in quality from what I've got here...

 

[Joechris]

Since most of the AD interfaces out there has a max input of 20 to 22 db into a 10k load, I think that would be OK output wise. A option for line level bal transformer input, and a low level unbal for instruments would be great. But if I must choose one...bal line. Look forward to this, thanks.

 

[Kingston]

Russian made 6N1P's are kind of equivalent to the 6DJ8 (with mu of 35) and in current production....  They're also reasonably consistent in quality from what I've got here...

No they are not an equivalent, even if the russian Ebay stores keep repeating it over and over. The have higher plate resistance (rp) and won't drive things like a 6DJ8/ECC88 would, and heater current is also nearly doubled. It's a good tube nevertheless, but a unique one with no direct equivalent.

[edit]

forgot to mention that 6N23P is actually a direct equivalent, in fact a clone of ECC88 (and hence very nearly a 6DJ8  too).

 

[1sound]

Sorry and thanks,

I forgot to mention the higher plate resistance or the extra heater current (which I figured Dave might look up before taking "my" word for it, even though total heater current draw has been "forgotten" here in the past ;D )...  But to my defense, I did say "kind of" an equivalent .  Sometimes in tube land, sort of close (in the same galaxy) is all you get these days...

Again, sorry if that last post was "misleading".

 

[PRR]

> 6N7's are very linear and with a mu of 36 are not easy to find a sub
> 6DJ8 are possible but more than double the 6N7 price.  Any thoughts?

Mu is NOT the big deal.

There is nothing like 6N6. What other single-bottle will make 10 Watts?

6DJ7 is so non-linear with bias-change that you will be sorry.

6J6 is the closest "small modern" tube. It very likely is a pint-size 6N7. In this economy a basket of 6J6 may cost less than their shipping.

Oddly, the "Collins 26W" I found shows 5814 as the choke. This is a fancy 12AU7. It would be worth trying both tubes. I suspect 6N6 is better but had gone out of production before the 26W finished its run.

> distortionless way to obtain compression

Plate resistance changes with signal. 2nd harmonic. Fat blunt 6N6 in push-pull at high supply voltage may be fairly clean.

> will not need the fancy interstage

Interesting.

 

[JohnRoberts]

distortionless compression... ?

I don't know how to do that...

JR

 

[lassoharp]

The 6N7's are very linear and with a mu of 36 are not easy to find a sub (6SN7GT are only 20 mu).
6DJ8 are possible but more than double the 6N7 price.  Any thoughts?

http://www.mif.pg.gda.pl/homepages/frank/sheets/127/6/6A6.pdf


6A6 if link doesn't work right.  Looks like same tube. Leave DJ8 to the crowd that loves to pay through the nose.


J5s are a less $$ alternative to SN7s for an alternate output driver.  6F6s are ran as triodes so less output power than BA6A or SA39 is doing.  A 600r ladder attenuator should cover it.  We can always build a permanent H pad in front of ladder if needed.  Looked at another way, even the smaller output stage of the 436 requires an attenuator so it's already a given for DAW use.  Also, 6V6 is uber common sub for the 6F6.

Most all modern PTs have  fairly hefty filament windings but the OT for the bigger tubes will be more $$.  If circuit doesn't require big tube characteristics to function correctly then smaller power output section would make good sense.  OT & PT cost look like biggest cost difference factors.

The disadvantage is that there is only about 20dB compression available

I don't see this as an issue for typical use in, say a vocal chain for example.  I rarely ever dip beyond 4 or 6db on the two vari-mus I have.

Looking forward to seeing what you came up with! 

 

[PRR]

> do we really need 28dBm output

Output power is dominated by a need for BIG current to charge the time cap at rapid rate.

 

[DaveP]

Thanks to everyone for their input, its given me plenty to chew on.
best
Davep

 

[ruffrecords]

Russian made 6N1P's are kind of equivalent to the 6DJ8 (with mu of 35) and in current production....  They're also reasonably consistent in quality from what I've got here...

No they are not. I have tested many of them and they bear very little resemblance to the 6DJ8. The 6N23 is pretty close to a 6DJ8 from the ones I have measured but they are quite pricey now.Try the 6922 instead, still in production and not too expensive from EH.

Cheers

Ian

 

[DaveP]

Ok, been chewing things over now,

The 26W is a non feedback design so they used linear tubes to keep distortion down the the minimum, even though the push pull design would cancel.

I think I will stick with the 6N7 because at $5.50 it is good value for money, 10W if you want it.  The plate lines are so laid back you can easily operate on the linear sections.  I will also try the 6J6 at $3 and see if it makes the grade.  The 6N6G is priced out at $28.50.

Can someone check my maths now:-

This is to work out the CV cap charging current required.

Amps=Coulombs/sec  and Coulombs =Capacitance x Volts/sec

therefore charging current Amps= Capacitance x Volts/sec

Using the 26W as an example with -30V and 1mS and 0.5uF

we get Amps = 0.000,000,5 x 30/0.001

which is 15mA does this seem reasonable?  or should I use 63% of 30V which would be 9.4mA

From this formula you can see that the CV cap makes a big difference to the required charging current, using a 0.25uF and 2mS spec (like the 23C) and 63% would need only 2.3mA.
This makes a difference to the design of the CV circuit.

best
DaveP

 

[Jean Clochet]

Post removed.

 

[DaveP]

Hi Jean,

This is the original circuit I'm working on.

I am also trying to use neglected tubes that don't cost as much as the popular tubes without compromising the performance.

Lassoharp and EMMR have told us how good these compressors are and probably some of that is down to the tubes used, fortunately the 6N7's are cheap enough, only problem is the heater current but that is not too expensive to supply.  PRR also likes these tubes and based on the spec and the one I have tested, I do too.

Anyway two more nights in the lab to come so I hope to make some more progress this week.

Must repeat, I need someone to agree my maths and reasoning to make continued progress.

US. brains should now have had their coffee and coming on line!

best
DaveP

 

[Jean Clochet]

Post removed.

 

[Jonte Knif]

Just for the record, the 26C and 26W use a variable (tube) resistance in a bridge circuit to control the compression.  The nearest equivalent to this is an LA-2A which also has a variable resistance to control the compression, albeit via an EL panel.  The advantage of this approach is that there are no thumps and no problems with matching Varimu tubes or keeping them balanced.

Did you mean that this compressor would not suffer from thumps? I don't think that is quite correct, because there is DC voltage present (unlike in optos), and if tubes don't track thumbs will occur.

 

[emrr]

I've never been able to make one thump under any condition.

Look forward to seeing what you work out. 

 

[DaveP]

Jonte,

My understanding is that the thumping in a varimu comes from the ac signal not tracking properly, particularly at low frequencies.  However this bridge relies on imbalance for the signal to pass, its only when the signal is completely balanced and cancelled that no signal passes.  On this basis the dc can play no part as it is out of balance most of the time.

best
DaveP

 

[Nishmaster]

I was under the impression that DC control voltage feedthrough was what was responsible for thump primarily. If the DC is cancelled/nulled properly (through whatever means, this varies by topology), then no thump.

I'm not the most experience guy around here though, there may be brainier heads that have a better answer.

-Matt

 

[Jonte Knif]

Correct. And as far as I can see, it will happen in this bridge topology too, but perhaps to much lesser extent. I frankly have to simulate things to properly find out what kind of difference there is.
The thumping problem *not* happening in the bridge design really sounds too much of a free lunch to me, let's see if my intuition failed this time.

Totally irrelevant: Why are resistor values in kilos marked with an M in the schematic? What is the historical reason for this?