PRR176 vari-mu compressor - adapted for guitar?

[lion]

Turning to the board for advise once again.

Just recently stumbled across PRR's 176 vari-mu design https://web.archive.org/web/20040616095136/http://music-club.rutgers.edu/headfonz/Comp5/Comp5.html, and decided to give it a go - with the intention to try to adapt it for guitar.

I've built it with parts I had available - incl. the input and interstage transformers - and it's working. My complete circuit below. The yellow section is PRR's original design with output taken directly from the NE5532. The upper 2 triodes are Paul's suggested add-on for an all tube audio path. For pre-amplification I've set up a gain stage, using the other half of the NE5532, with a level control into the input xfm to control the amount of compression.



So far I like what it does, but there's a couple of issue and surely some refinements to be worked out for guitar use.

My first concern is the HV supply.

For the total power supply I happend to have something almost suitable at hand, with +/- rails, 12.6V heater DC and a B+ string. It was easy to add the 1.5V bias tap - and a couple of changes in the HV string made it useable, at least for testing. Focusing on the B+, here's just the HV part of my PSU:



I'm a bit puzzled I cannot get B+ voltage higher than 105V with the load of the tubes (unloaded there's about 195V into R21). Notice the B+ for the vari-mu tube needs regulation (took me a while to understand it's mandatory, not a "luxury"). I use two zeners I happend to have in series, giving me 94V - which I think is perfectly OK.

However PRR's circuit suggest a higher voltage of 140V for the output tube stage - doesn't need to be spot on I'm sure - but the 100V I'm able to squeeze out may not be ideal .

The output tubes are shown as a 12AX7 for the first triode and a 12AU7 for the CF stage. For obvious reasons I would prefer to use just one or the other - ie one tube. I've tried both 12AU7/12AX7 - and they both work, but the voltages I measure on either doesn't seem to fit at all with the indicated plate/cathode voltages on the orig schem (even considering the decreased B+, 100 vs 140V supply).

Here's my reading on the output:
using a 12AX7: 22.8V - 90mV - 22.5V
alternatively with a 12AU7: 11.5V - 160mV - 15,2V

And the vari-mu 12AU7 voltage:
Plates: 44.3 - 44.3V
Cathodes: 408 - 408mV

All readings with no input signal.

Maybe worth mentioning - for guitar use I'm probably more after nice colouration/bending (in addition to the actual compression) rather than strictly transparent/clean(ish) from the circuit. Which is allso my reason for running the all tube path instead of the simpler opamp out.

When I get the voltage supplies sorted, I plan to look into alternative/better transformers, as well as overall gain setting for best function/noise with a guitar signal.

Your thoughts, comments or advise will be much appreciated.

Erik

[PRR]

Gain of 122 from guitar to +/-15V opamp is surely too much. Standard Fender 12AX7 input is gain of 50 with a 300V supply! Make this gain trimmable. A 1Meg audio-taper pot, wired 2-terminal, in place of 1meg fixed. Then the other trim is not needed.

Your power supply has something majorly wrong. Dis-assemble and check step by step. 12V into 15:115V should give 92V AC each winding. Does it? (Do not get yourself killed!). Add diode, should get 90~V DC. Add some-uFd cap, should get 129V DC. Both windings stacked, 250+V DC. Loaded with R and Z, this will drop, but unless the transformer is wee-tiny it should not be far under 200V. 105V is so very wrong I can not guess what is screwed-up.

Skip the tube output stage for now. Consider it a joke (it may have been). Get signal off the make-up opamp.

[lion]

Thanks Paul

Concentrating on the supply first.

Dis-assemble and step by step check:

Wall wart output measured to 13V AC
13V into toroid 15V windings (in parallel) gives 78V AC out each winding - 2x78V windings in series = 156V AC measured,
Adding a diode bridge I get 140V DC,
and adding 100uF I read exactly 200V DC.

I measured all voltages both in the circuit (line broken between C21 and R21), and with the toroid coupled to another external diode bridge/reservoire cap - identical results.

The toroid I use is similar to this:
http://www.digikey.com/product-detail/en/acme-electric-amveco-actown/70003K/TE70003-ND/154798

Apparently my toroid delievers less than specs indicates - leaving only 200V (not 250+V) after rectification+ reservoire uF.

Furthermore
200V at C21 - BUT, as soon as line is reconnected to R21, and the rest of the circuit, the voltage drops to 105V at C21 - 100V at C23 - 94V at C24.

Observation. With the circuit complete, but zener shunt regulator disconnected - on power up the voltage rise to 180+V at C21, and then drops to 110V.

Confused!

Erik

[idiot savant]

Looks like you're overloading your power transformer. The little dinky 1.6VA bugger you linked to just wont cut it.

Look at it like this, wiring it the right way with 115VAC in and 15VAC out it's rated for 106ma. So if you wire it backwards, you have to account for the current in equal fashion. With 15VAC into the secondary, you should get 115VAC out, at 10-15ma-ish. Somewhere around there, but I'm just quickly eyeballing it.

I'm just going off of the schematics posted. But, it looks like the 12AU7 in the compressor needs about 10ma. The tube booster needs around that, or a little more.

Try disconnecting the tube booster section completely. Does your B+ go up?

You could rewire the booster section with a 12AX7 CF instead of the 12au7. Use a 100K resistor instead of 10K. That will reduce the current drawn by that whole stage. It will have less drive, and higher output impedance. Live-with-able maybe...?

Seems more logical that way anyways, otherwise you're sitting there with half of an AX7, and half of an AU7 unused.

[PRR]

> 1.6VA bugger

WHAT !?!?!

I do not remember yesterday, but I clearly intended ~~160V at C1 and 20mA in R5. This is 3.2 Watts DC. Conservatively doubling for AC VA, we want a 6 VA iron. Not a <2VA toy.

Clearly you set-aside the tube output. This wee iron *may* power the 12AU7 gain-cell. Re-wire the HV AC so you get 100V DC un-loaded. How bad does that sag and un-sag when the 12AU7 is connected and disconnected?

[lion]

Thanks gentlemen

<2VA toy!

DOH - I'm an IDIOT!

Strange thing. Before I read your last replies - I took the proto build for a test drive with my guitar setup, and at R&R soundlevels. It sounded pretty darn good, in fact I was blown away! The compression was like nothing I've heard from any (stompbox) compressor, lot of useful settings, and there was a nice colouration -  hard to describe, but so BEAUTIFUL!!! Worked with both the opamp output and the tube out, the later sounding absolutely incrediable (joke or not). Go figure! How could it ever perform like that with inadequate supply - not to mention the random audio xformers.

Bigger iron. I found another toroid hidding far back on a shelf - 15V/1A:220V.



Will the 2W zeners be happy here? I'm trying to understand/do the calculations around zener shunt regulators, but I'm struggling to fully understand it to be honest.

Voltages looks much better now. Circuit works on the bench, but I haven't played it properly yet. Wonder how it sound compared to the first proto. Anything "less"/different will be a disappointment. Fingers crossed 

I will start to look into the other issues you mentioned shortly.

Erik

[tubegeek]

DOH - I'm an IDIOT!

Evidence suggests otherwise.

Will the 2W zeners be happy here? I'm trying to understand/do the calculations around zener shunt regulators, but I'm struggling to fully understand it to be honest.

re: Zener regulator. Same idea as a gas-filled tube regulator except there is a much less onerous requirement for the conditions on the Zener/gas regulator.

With no load on the regulated voltage, the series resistor must pass enough current to drop the difference between the unregulated voltage and the regulated voltage.

The Zeners (still no load) must pass that same current.

Under load: the series resistor passes the same amount of current as before, because the voltage drop is the same. However, the current is now divided. Whatever current the load takes, is subtracted from the total current (no load value) to give the Zener current.

What you need to be sure of is that the no-load current is adequate to have the load current subtracted, and still leave some to go through the Zeners. In other words, size the resistor to drop the correct voltage required when passing slightly more current than the load will ever require. If the no-load current is too small, then the regulator will fail to control the regulated voltage when the load drops its greatest current. In other words, zero Zener current is a no-no.

Then, be sure (using Ohm's law and Power formula) do a calculation on each component (each Zener, and series resistor) to ensure adequate rating. For these calcs, the current through the Zeners would be the no-load current, and the voltage would be the Zener voltage. You do the series resistor value first: its current is found using the difference voltage, start with supply voltage before the resistor, minus the Zener string value.

I have made the same mistake, BTW, with an undersized transformer. You can only cheapskate but so far.

[lion]

Tubegeek - thanks a lot.

Your explanation on the zener regulator is somewhat clearer than what I've seen before - however, I still can't seem to get my old head around it. Starting already with your first sentences:

With no load on the regulated voltage, the series resistor must pass enough current to drop the difference between the unregulated voltage and the regulated voltage.

The Zeners (still no load) must pass that same current.

Under load: the series resistor passes the same amount of current as before, because the voltage drop is the same. However, the current is now divided. Whatever current the load takes, is subtracted from the total current (no load value) to give the Zener current.

Can you elaborate or reword, please? Maybe I can take it in step by step.

Parts of my doubts/confusion may be due to the use of more series resistor and series zener.

Erik

[tubegeek]

How about this:

All the current used by either the Zener or the circuit (the load) must pass through the 1K resistor.

When the circuit is not using any current, all the current goes through the Zeners. When the circuit is drawing current, some (less) goes through the Zeners. The current through the resistor - the total current - stays the same. We know this because the voltage drop across the resistor stays the same. The supply voltage (at the 47uF cap) is the larger voltage, and the regulated voltage (at the Zeners) is the smaller voltage.

So two conditions must be considered: what happens when NO current is used by the load and, what happens when the load is drawing its maximum? The total current value must be large enough to always send at least a small amount through the Zeners, or else they will fail to regulate the voltage. This value is determined by Ohm's law, using (the large voltage minus the Zener voltage) for V and 1K for R. If too large a resistor value were chosen, then there would be too little total current through the resistor to ensure that always some is available for the Zeners.

Note that (within their limits) the Zeners will draw whatever current value is needed to make up the correct total, depending on what the load is using.
Any help there?

[lion]

Thanks Tubegeek

I'm sure you have explained everything more than perfectly well - but I simply don't get it (actually I fear I'm loosing my sanity)!

Trying to match the theory with what I can measure, but I can't seem to make anything match up - and I feel stuck.

Here's where I am so far.

With R23 = 1K the voltage is 110V before the resistor and 98V after - that is unloaded (gain tube disconnected from supply).

Voltage drop across R23 = 12V. Current through the R23 is then 12/1000 = 12mA.

Now loaded (= gain tube plates connected to B+) the voltages reads 109V and 96V

Voltage drop across R23 = 13V , and current is then 13/1000 = 13mA.

Readings and calculated results doesn't make much sense to me (actually moving slightly in the WRONG direction unloaded vs loaded)!

Trying to do something. Upping R23 to 1k+2.2k = 3.2k the readings are 125V before the resistor and 96V after (unloaded)
Voltage drop 29V. Current 29/3200 = 9mA

And loaded, 124V and 96V. Voltage drop 28V. Current 28/3200 = 8.75mA

What am I missing here?

Erik

[PRR]

No-signal, vari-Mu 12AU7 should be pulling around 6mA. +/-30% variations are expectable. But I don't see that much in your numbers.

You seem to have 96V "unloaded or loaded", which suggests the regulator works, except we are not sure the 12AU7 is pulling its expected DC load.

As a test, make a 16K 1W dummy load resistor. (Two 33K 1/2W in parallel.) If you get around 96V with or without this hanging on the DC supply, we are more sure. (At maximum signal the 12AU7 suckage goes to near 0.2mA, which is so low we can just assume the zero-load condition.)

The real point is to NOT allow the supply voltage to vary as much as the tube current. A very saggy supply will do that, and the gain reduction is nearly negated. But a 10% (say 96V-86V) sag for a 10-to-1 change of current (6mA to 0.6mA) is not a big deal unless you do line-up for the BBC intercontinental links.

[lion]

Thanks paul, much appreciated.

With R23 = 1K
The regulated supply reads 96.3V with a 16K dummy load as suggested. Unloaded the figure goes up to 97.7V

Loaded by the vari-mu 12AU7 the voltage reads 96.6V

A couple of additional observations.

No signal input - B+ reads 96.6V. With input slammed and heavy compression the B+ stays almost rock solid - just a slight change/rise about 1 volt.

I have a "dB compression" meter applied, doesn't work perfect (mostly for show), but it sits across a 47R resistor from R32 pot wiper to ground on the vari-mu 12AU7 cathodes.

Checking the resistor it measures 48.5 ohms to be precise.

No signal - there's 0.207V across 48.5R = 4.27mA (close to your 6mA +/-30%).
On heavy compression the voltage drops to 0.019V = 0.39mA
Not bad - or?

Everything seems to work, and certainly sounds very good - different values for R23 (from 1K - 5.6) doesn't seem to make any/much difference in performance (at least on my bench).

My main concern with the zener regulation circuit (and optimal component values) is safety and reliability.

Erik

[PRR]

> Not bad - or?

Looks spot-on, considering my numbers are from the numeric idiot (SPICE) and yours are from some actual tube.

> safety and reliability.

Figure the voltage, thus current and heat, in the resistors and Zeners. At idle the 12AU7 takes current, at full squash it does not.

Or power-down with a voltmeter on B+ (to be sure it goes away), wear rubber shoes, put other hand in pocket, and 1-finger feel these parts for heat. Hot is OK, HOT! may not be.

[lion]

Thanks Paul

Looks spot-on, considering my numbers are from the numeric idiot (SPICE) and yours are from some actual tube.

Thus reassured I think I'm almost there

Figure the voltage, thus current and heat, in the resistors and Zeners.

The big 1K 3 or 5W I have should do fine, and the zeners are good for 35+ mA each (43 and 51V, both 2W).

No excessive heat after several hours at idle and heavy comp - nothing hot, just a bit warm.

Thanks for the assistance. Much appreciated.

Erik

[rankot]

Is it possible to make this kind of compressor using JFETs instead of tubes? 

[lion]

More simple, but along the same lines - I think.

http://aquataur.hilpold.net/aquataur/musicstuff/umble.html
Scroll down to fig. 1 about 2/3 down - with explanation and some details.

The circuit by itself:
http://aquataur.hilpold.net/aquataur/musicstuff/detail-compressor.pdf

I won't fail to mention I highly recommend the PRR176 vari-mu as shown - including the tube output (gain of driving opamp needs taming though). I'm over the moon with what it does and the colouration it adds. I only wish I had found it many years ago.

Erik

[rankot]

More simple, but along the same lines - I think.

http://aquataur.hilpold.net/aquataur/musicstuff/umble.html
Scroll down to fig. 1 about 2/3 down - with explanation and some details.

The circuit by itself:
http://aquataur.hilpold.net/aquataur/musicstuff/detail-compressor.pdf

It's interesting, but I had in mind Vari-Mu with JFETs.