Interstage trafo in vari-mu circuits

[NewYorkDave]

Or just use the xfmr primary as a centertapped choke and ignore the 2ndary--easy enough to set up for quick comparison with a switch.

Hammond makes an inexpensive 150H choke, but the DCR seems a bit too high for this purpose (something on the order of 3K).

 

[Larrchild]

Right!


You could dynamically load the unused secondary to shift the primary impedance to a Power Matched condition throughout the bias range too.


hehe.

 

[PRR]

> DCR seems a bit too high for this purpose (something on the order of 3K).

Too low. DCR is good. You idle at high current, but don't want to melt tubes: want low plate voltage. You want high plate voltage when GR-ed, for high grid voltages and low relative distortion. Since this is low-current, large DCR is mostly good.

A disadvantage is the huge common-mode shift flogging your next stage. For 3K working at 10mA-1mA it is only 27V, and a 100V long-tail can reject that. 3K working 30mA on down does get to be trouble; however the affordable Hammond may not stand 30mA.

Another disadvantage is that a large plate voltage shift as you start to go into GR means a very soft knee, unless you over-amplify the CV.

Anyway, two chokes will both flux-saturate, one CT choke is too advantageous to ignore. But they don't hardly make any CT chokes. OneElectron has a few dual-winding irons, but they are power amp para-feed jobs and may not have high Henries. Using half a transformer ensures double DCR of an "optimum" winding, but DCR is not all bad, and stock-part is good.

Between choke and transformer: choke reduces common-mode, transformer nearly eliminates it. OTOH some transformer faults vanish in choke use. Leakage inductance works with you, not against you.

I've played with loading the secondary while using the primary for both in and out. What happens is that all transformer faults appear in reverse: it won't roll-off highs because leakage inductance gets in the way of the load-down action. I think there was a broadcast limiter which did this, but it may have been a few hand-wound samples built in a down-under land.

BTW: the H-P 200AB has a nice fat wide-band 60K:600/150 transformer, and sad 200ABs can sometimes be found for the take-away. You also get enough power transformer for a 2-6V6 power amp (or similar), and a knob as big as your head.

 

[edanderson]

the edcor 1:1 transformers are wound bifilar, with both windings taken out the center tap at half the total turns count. it is easy and has good performance for 1:1 (cinemag and jensen do it also, among others). so if you connect primary and secondary together in series, you would have a choke with a DCR balanced center tap (since the two windings are identical). something like the xsm 15k:15k (EI625).

ed

 

[Kit]

kit, i have a edcor 10k:10k here and measured the inductance - it´s 8 henry per side, and i´m not sure if it works perfect in that application.

also the dc-symmetry of one one-coil center tab is not very good (120 vs 160 ohm)

Ah, yes. But................we dont have to use the center tap in the VCA stage. B+ can be fed from the junction of 2 resistors hanging on the plates.

Since I started this thread ive finished building the prototype: Its sitting on my bench right now, squishing audio as we speak/type.

Im not completely happy with the end result yet but Ill post some pictures and schem when Im 100% satisfied with the end result.

Thanks for your comments and help.

 

[Larrchild]

Ah, yes. But................we dont have to use the center tap in the VCA stage. B+ can be fed from the junction of 2 resistors hanging on the plates.

We discussed why that might not be best, a few pages back.

 

[Kit]

Hmmm....................where?

The way I see it, we idle the tube at high current. Dropping voltage over the plate resistors, keepin P within reasonable limits. DC resistance is large at idle, but with the non centertapped trafo in parallel with
the resistors, AC gain/resistance is low at GR. Look at the UA175.

 

[Larrchild]

If your configuration allows minimum plate voltage swing during GR, it's ok. If you are changing plate voltage a lot during GR, that's not as good.

Since bias changes plate current, and plate voltage will drop (or rise) on the plate as the current varies across the tube when a plate resistor is used, an inductor has less variation, and the sound is less "squishy".

Try to vary current only, if you can.

 

[PRR]

> B+ can be fed from the junction of 2 resistors hanging on the plates.

A center tap "forces" the two currents to cancel. 100-ohm differences between sides don't cause large current unbalance with >3K plate impedances.

Your plan lets 99% of the tubes' current unbalance flow in the winding. This is sure to be larger than the subtraction of the two currents.

There is also the large drop in resistors. This can be bad or good. I think it is both good and bad. The worst case seems to be when half-off, around 6dB GR. Signal level is high, but plate voltage has not risen near B+ as it will for large GR. On a soft-knee it may be no big deal, on a hard-knee you tend to have high THD as you go around the knee. If you are only peak-trimming, not crushing, this ~~6dB zone may be a much-heard zone.

Resistor drop also screws with the control voltage response. At low plate voltage, small CV starts to give some GR. But then the plate voltage starts rising, and it will take much more CV to get more GR. So such schemes "want to be" soft-knee, and it will take more oomph in the sidechain to get a stiff knee.

Bravo for actually building one. It is an incredibly inter-related system. Much more so than FET or LDR limiters. People need to do live experimentation.

 

[Larrchild]

To return to a little of the earlier discussion, if you do get it to sound as you like using the transformer for coupling, also switch to the use of the transformer as a center-tapped choke only and use capacitive coupling to the next stage.

That may sound more transparent in the end, and obtain a good current/voltage drop relationship on the plates of the mu tubes.

 

[analag]

Link

To further the discussion here is something I drew up in response to the interstage transformer problem. Obviously tubes can be used to provide constant voltage/variable current operation while also solving the problem of the lower B+ needed to power the remote cutoff tubes (6BC8) etc. The output transformer can be an Edcor XSM10K:600 this is how I use it (with excellent result) in my 660 type compressor.
This one can also be made to react much faster than the Altec 4xx design.

analag

 

[emrr]

I like the basic idea at a glance.

 

[Kit]

Happy holidays.

What I´ve got so far:

edit. schematics removed.

The transformers are edcors XSM,

Input: 10k:10k

Interstage: 10k:10k

Output: 10k:600

Sounds good, but it could sound even better if you help get rid of those nasty DC thumps.

I´ve tried a trimmer at the cathode(s) but that didnt help much.

Suggestions?

:thumb:

 

[Larrchild]

You could try making the 33k's adjustable since that establishes the "phantom center" point. Just a thought.

 

[Kit]

I´ll try that. Thanks.

But maybe I was wrong about the DC thumps.

I think the problem might be base "leakage" in Q1.

What happens is that because the gain is fairly large in Q1, the nonlinear regions of Vbe on-off, will mess up the effective threshold. It gives it a sort of soft knee behaviour, in a bad way.

Measurement confirms that Q1 is never really completely on during GR with average "tame" program material. The collector pulls the grid´s about 5-10V negative and then nothing more happens........................until a big transient comes along (like a heavy kickdrum). Q1 turns fully on, the grids swing down to ground and voila, massive GR and thump.

So, Q1 needs some more thinking and less gain. :roll:

Also, input headroom is a bit low. Might have to "up" the plate voltage in the VCA stage.

 

[Kit]

Hi Darius.

Yes, I did that. No luck.

I need to sort out the threshold issue in the sidechain first. I think much of the problems im experiencing lies there.

Thanks.

:sam:

 

[Kit]

Im dropping this project.

These ecc189´s I got my hands on are just killing me. Theres no way to trim the DC thump out of them. Not with the ones Ive got here, anyway.

Telefunken.....................bah.

They differ to much from triode/triode, across the curve.

Im removing the schem.

I am so mad right now.

 

[analag]

[quote author="Kit"]Im dropping this project.

These ecc189´s I got my hands on are just killing me. Theres no way to trim the DC thump out of them. Not with the ones Ive got here, anyway.

Telefunken.....................bah.

They differ to much from triode/triode, across the curve.

Im removing the schem.

I am so mad right now. [/quote]

Ha ha ha ha....don't let it beat you man. Think of yourself as a scientist. You must endeavor to persevere.

 

[Kit]

I like your attitude, Analag.

Its just that this project has already taken up way to much time.

And I dont want to waste anymore time on tubes that cant be made to work in this circuit.

Maybe later on I´ll by some 6386´s or PCC189´s, and give it another go.

But for now I´m done.

 

[analag]

If you parallel the GR tubes, thumping becomes less of an issue. And if you drive the cathodes with constant voltage, thumping can be further reduced.

analag