Jaakko > From the UA article: http://www.uaudio.com/webzine/february/text/content4.html
"Variable-mu tubes have the ability to reduce their mu (gain), as the input signal increases."
Am I going mad or is this pure bullshit? I always thought remote cutoff triodes just acted more linearily in a vari-mu compressor?
It is poorly written.
It isn't "input signal", it is bias current, which we can control with bias voltage or other techniques.
All tubes (and transistors) will reduce gain when bias current is reduced.
The problem with "regular" tubes is that you have to reduce current a lot to reduce gain a little. To reduce gain 10:1 (20dB gain reduction) you generally have to reduce current about 100:1.
Note that an ideal linear tube would have NO reduction of gain at any current level. (And would be useless as a variable-gain device!) A tube where gain drops 10:1 when current drops 100:1 is considered very linear.
If a device's gain varies with current, it will distort. (The tops of the wave get bigger than the bottom of the wave.) However two such devices can be worked in push-pull, so that when one side's current (and gain) goes up, the other side's current (and gain) goes down. This cancels the distortion. The cancellation is not exact and leaves an odd residual distortion. However push-pull is usually the best topology when you need large current swings.
In a compressor/limiter, this means that at the highest signal levels, the tube is starved. It distorts. It helps only a little to start from a very "rich" high-current condition: quadrupling the no-GR idle current from 2mA to 8mA or using four tubes per side would give you only about 6dB more margin before distortion, except the higher current forces lower load impedance (resistor or transformer) so you typically don't even gain that much.
This is where "mu" comes in. In a triode, as current is reduced, transconductance (Gm; gain voltage/current) is reduced, but plate resistance increases. Plate resistance affects gain. If you could work a triode into an infinite resistance, gain would not vary as Gm (10:1 gain for 100:1 current) but rather as Mu. And in the most linear triodes, Mu does not drop until the tube is totally starved for current. Using a low load resistance makes Mu moot. In a Pentode, Mu is always too high to matter. Or you can find non-linear tubes where Mu does drop with current, and the increasing plate resistance does not fight the falling Gm so much. ("Vari-Mu" is a nice name but not really how tube compressors work: the change of Mu in vari-Mu tubes is not enough gain reduction for our purposes.)
In a Remote Cutoff tube (triode or pentode), to get 10:1 (20dB) gain reduction, the bias current only has to go down about 10:1 or 20:1, instead of around 50:1 or 100:1 as in a "linear" tube. In deep gain reduction, the current is small but not as small as a straight linear tube.
In bipolar transistors, gain is exactly proportional to bias current. In "linear" amps, transistors have more distortion than straight tubes (however they also give more gain and love feedback, so the total design can give lower numeric distortion than any tube). In gain control use, they are more like remote cutoff tubes, though the difference in gain/current makes them a very different animal from any tube.
In an ideal world, there isn't a whole lot of difference between using single pairs or tubes or transistors. Either one can give about 120dB total dynamic range. In compressor limiter use, subtract the maximum gain reduction you want: for 30dB max GR, your output signal to noise ratio is about 120-30= 90dB. However to do this with tubes requires a very "rich" design: lots of current and super good transformers. In transistors it is trivial. And because transistors are made in both polarities and can be very exactly matched, you can use a second pair plus some external gain and get a GR cell where S/N is extrememly large; always limited by external parts.
"Variable-mu tubes have the ability to reduce their mu (gain), as the input signal increases."
Am I going mad or is this pure bullshit? I always thought remote cutoff triodes just acted more linearily in a vari-mu compressor?
It is poorly written.
It isn't "input signal", it is bias current, which we can control with bias voltage or other techniques.
All tubes (and transistors) will reduce gain when bias current is reduced.
The problem with "regular" tubes is that you have to reduce current a lot to reduce gain a little. To reduce gain 10:1 (20dB gain reduction) you generally have to reduce current about 100:1.
Note that an ideal linear tube would have NO reduction of gain at any current level. (And would be useless as a variable-gain device!) A tube where gain drops 10:1 when current drops 100:1 is considered very linear.
If a device's gain varies with current, it will distort. (The tops of the wave get bigger than the bottom of the wave.) However two such devices can be worked in push-pull, so that when one side's current (and gain) goes up, the other side's current (and gain) goes down. This cancels the distortion. The cancellation is not exact and leaves an odd residual distortion. However push-pull is usually the best topology when you need large current swings.
In a compressor/limiter, this means that at the highest signal levels, the tube is starved. It distorts. It helps only a little to start from a very "rich" high-current condition: quadrupling the no-GR idle current from 2mA to 8mA or using four tubes per side would give you only about 6dB more margin before distortion, except the higher current forces lower load impedance (resistor or transformer) so you typically don't even gain that much.
This is where "mu" comes in. In a triode, as current is reduced, transconductance (Gm; gain voltage/current) is reduced, but plate resistance increases. Plate resistance affects gain. If you could work a triode into an infinite resistance, gain would not vary as Gm (10:1 gain for 100:1 current) but rather as Mu. And in the most linear triodes, Mu does not drop until the tube is totally starved for current. Using a low load resistance makes Mu moot. In a Pentode, Mu is always too high to matter. Or you can find non-linear tubes where Mu does drop with current, and the increasing plate resistance does not fight the falling Gm so much. ("Vari-Mu" is a nice name but not really how tube compressors work: the change of Mu in vari-Mu tubes is not enough gain reduction for our purposes.)
In a Remote Cutoff tube (triode or pentode), to get 10:1 (20dB) gain reduction, the bias current only has to go down about 10:1 or 20:1, instead of around 50:1 or 100:1 as in a "linear" tube. In deep gain reduction, the current is small but not as small as a straight linear tube.
In bipolar transistors, gain is exactly proportional to bias current. In "linear" amps, transistors have more distortion than straight tubes (however they also give more gain and love feedback, so the total design can give lower numeric distortion than any tube). In gain control use, they are more like remote cutoff tubes, though the difference in gain/current makes them a very different animal from any tube.
In an ideal world, there isn't a whole lot of difference between using single pairs or tubes or transistors. Either one can give about 120dB total dynamic range. In compressor limiter use, subtract the maximum gain reduction you want: for 30dB max GR, your output signal to noise ratio is about 120-30= 90dB. However to do this with tubes requires a very "rich" design: lots of current and super good transformers. In transistors it is trivial. And because transistors are made in both polarities and can be very exactly matched, you can use a second pair plus some external gain and get a GR cell where S/N is extrememly large; always limited by external parts.
![Electronics Soldering Iron Kit, [Upgraded] Soldering Iron 110V 90W LCD Digital Portable Soldering Kit 180-480℃(356-896℉), Welding Tool with ON/OFF Switch, Auto-sleep, Thermostatic Design](https://m.media-amazon.com/images/I/41gRDnlyfJS._SL500_.jpg)
