Viitalahde
Well-known member
The title is a little misleading. I've not designed a vari-moo. I just want to sum up some things and see if I've digested a number of issues right. I want to design my own vari-mu, and this sum-up and the following discussion might be of help to someone.
If you find bad mistakes in my findings (very likely!), please notify and I'll edit this post. I will have a couple of (?)'s where I feel I have not understood it all correctly. This is all coming out of my freshly stuffed memory.
I have not designed a single tube gain stage, and I can't figure out a harder way than to design one that does variable gain, but hey, I'm not scared of some elbow grease. :razz: I do own a vari-mu which I've built, and calibrated several times, so the issues in cathode/anode balances and tube mis-matches are familiar to me.
Possible problems (and solutions) with vari-mu compressors
Audible thumping and distortion when the push-pull stages don't cancel out the CV correctly
This can be solved with good matching of the triode halves, regulated plate voltage (?) balancing the anodes & cathodes correctly and/or paralleling several triodes for evening out the tube mis-matches. There seems to be a trick with regulating the cathode bias voltage, too, We'll see if Analag chimes in on this.
Coupling the variable gain stage to the next stage
Capacitive coupling, as in most Altecs, STA-Levels and Langevins, might result in odd action when the wandering plate voltage charges & disc-charges the coupling capacitors.
A better approach is an interstage transformer, which results in whole lot of new problems.
The transformer should be able to withstand some imbalanced DC (possibly gapped), and, depending on the plate resistors, have a high inductance. High inductance means high winding capacitance, and HF loss is possible. Varying DC in the transformer changes the inductance and might give droop in both ends of frequency response during gain reduction.
The transformer ratio depends on the plate resistors and the triodes in the variable gain stage want to see a load of something like 3x - 5x their output impedance. Too small = distortion, too high = reduces gain (?).
The plate voltage can be driven through the center tap of the transformer, or through the centre of the plate resistors. Normal plate resistors give a more predictable range for possible output impedances of the variable gain stage. You can also accomplish this by driving the B+ through the CT and shunting a resistor accross the primaries (=plates).
Sidechain
A normal rectification diode (SS or tube) fed back from the output and smoothed out might not be beefy enough for good compression. Solutions are separate sidechain amps or a killer output stage as in Fairchild.
If the remote cutoff tubes are paralleled, even more power is asked from the sidechain as there will be more grids to feed.
...
This is where my memory ends and I'll go and read more. Please chime in!
If you find bad mistakes in my findings (very likely!), please notify and I'll edit this post. I will have a couple of (?)'s where I feel I have not understood it all correctly. This is all coming out of my freshly stuffed memory.
I have not designed a single tube gain stage, and I can't figure out a harder way than to design one that does variable gain, but hey, I'm not scared of some elbow grease. :razz: I do own a vari-mu which I've built, and calibrated several times, so the issues in cathode/anode balances and tube mis-matches are familiar to me.
Possible problems (and solutions) with vari-mu compressors
Audible thumping and distortion when the push-pull stages don't cancel out the CV correctly
This can be solved with good matching of the triode halves, regulated plate voltage (?) balancing the anodes & cathodes correctly and/or paralleling several triodes for evening out the tube mis-matches. There seems to be a trick with regulating the cathode bias voltage, too, We'll see if Analag chimes in on this.
Coupling the variable gain stage to the next stage
Capacitive coupling, as in most Altecs, STA-Levels and Langevins, might result in odd action when the wandering plate voltage charges & disc-charges the coupling capacitors.
A better approach is an interstage transformer, which results in whole lot of new problems.
The transformer should be able to withstand some imbalanced DC (possibly gapped), and, depending on the plate resistors, have a high inductance. High inductance means high winding capacitance, and HF loss is possible. Varying DC in the transformer changes the inductance and might give droop in both ends of frequency response during gain reduction.
The transformer ratio depends on the plate resistors and the triodes in the variable gain stage want to see a load of something like 3x - 5x their output impedance. Too small = distortion, too high = reduces gain (?).
The plate voltage can be driven through the center tap of the transformer, or through the centre of the plate resistors. Normal plate resistors give a more predictable range for possible output impedances of the variable gain stage. You can also accomplish this by driving the B+ through the CT and shunting a resistor accross the primaries (=plates).
Sidechain
A normal rectification diode (SS or tube) fed back from the output and smoothed out might not be beefy enough for good compression. Solutions are separate sidechain amps or a killer output stage as in Fairchild.
If the remote cutoff tubes are paralleled, even more power is asked from the sidechain as there will be more grids to feed.
...
This is where my memory ends and I'll go and read more. Please chime in!
