bipolar transistor as cathode follower?

[imaradiostar]

Has anyone ever tried using a high voltage bipolar transistor as a cathode follower after a tube? I was thinking of trying this with a stripped down marshall style pre. I have some bipolars from computer power supplies that cross to an NTE379- datasheet here.

I'd imagine it shouldn't work too differently from the commonly discussed mosfet applications. Any thoughts?

jamie

[raiken]

Heh...technically, it's called an "emitter follower", not a cathode follower. If you use a MOSFET, it's called a "source follower".

All of them will work, although if you want to get closest to a tube, a high-voltage depletion-mode MOSFET source-follower would be the best.

"Normal" MOSFETs are enhancement-mode, which means they are normally "off" with no applied gate-source voltage. Depletion-mode MOSFETs are normally "on" with no applied gate-source voltage, like a tube would be if you had zero volts grid-to-cathode. You have to apply a negative voltage on the gate with respect to the source to turn them off, as you would in a tube with a negative grid-to-cathode voltage.

Randall Aiken

[imaradiostar]

Thanks Randall- I was thinking it was still called a cathode follower since it'd be tied to a tube but now that I think about it that's dumb. I figured the base and emitter want to be several volts apart which would make the thing bias up only a few volts different from a tube or mosfet in the same application. Maybe I'm not thinking clearly about this either.

The real question is would it sound different from a tube in this application? Considering the cathode follower in a typical marshall is driven into clipping I'm curious how it would effect the overdriven tone.

jamie

[raiken]

Thanks Randall- I was thinking it was still called a cathode follower since it'd be tied to a tube but now that I think about it that's dumb. I figured the base and emitter want to be several volts apart which would make the thing bias up only a few volts different from a tube or mosfet in the same application. Maybe I'm not thinking clearly about this either.

The real question is would it sound different from a tube in this application? Considering the cathode follower in a typical marshall is driven into clipping I'm curious how it would effect the overdriven tone.

jamie

The base will be at the voltage of the plate (assuming you build a DC-coupled emitter follower). The emitter will be around 0.7 volts below that (assuming a silicon junction). The collector current will then be set by the emitter resistor, so if you wanted say, 1mA, and your emitter voltage is 100VDC, you'd want a 100k emitter resistor. You'd have to make sure the transistor dissipation doesn't exceed max ratings, so multiply that 1mA by the collector-emitter voltage to get the quiescent dissipation. Also make sure the transistor can handle the peak voltage on the collector, with a suitable margin.

In linear mode, all of them will likely sound identical. If you drive it to clipping, however, it may have a different distortion characteristic. A few minutes with a Spice simulator should show you what it will do, or you can build one up and scope it out and listen to it for yourself.

RA

[loudthud]

That NTE379 has pretty high leakage current and like most high voltage bipolar transistors, the beta is not very high. But, those specs are worst case so it might work.

[jrfrond]

FYI- Marshall used an HV MOSFET up-front in the ValveState series.

[Enzo]

Actually his parts only CROSS to that NTE part.

What are the real part numbers on your transistors?

And if they are the main switching transistors, they are made intended for swithcing duties at high speed. They may well work at the low currents in our tube stuff, but are not intended for continuous duty.

[Steve Conner]

I'd check out smaller video amp transistors like the MPSA42 or the MJE340. You probably don't want to mess with big TO-247 cased switching transistors, they will have low gain and high capacitances.

The transistor should bias up just the same as a tube. The base draws current, whereas a tube grid doesn't, but in a direct coupled cathode follower the current can easily be supplied from B+ through the plate resistor. If the transistor is running at 1mA collector current and has a beta of 50, then it will only be 20 microamps anyway.

As for how it might affect the overdriven tone, why don't you try it and let us know

[R.G.]

Randall, Enzo, and Steve all have good answers. Loudthud is close to a good answer.

If you get a suitable transistor, an emitter follower with enough emitter resistance not to overload the plate that's driving it will be clean. It will, in fact be clean over a bigger range than the tube. So it is very unlikely to color the sound except by the coloration of driving whatever comes after it with a lower impedance, which presumably is what you're trying to do while you're doing this. Ditto a MOSFET. Both devices are not hamstrung with some of the voltage swing limitations of a triode.

But to do this well it is critical that the device have high enough input impedance not to load down the tube that drives it. This is a problem with all bipolars in this application because the currents are small and the voltages are large. Small currents puts the transistor well into its lowest current gain region unless the device is a low current device. Loud is right - with currents of 1ma or so, a big power device will have an even lower gain. The chances are right for it to overload the tube and sound terrible. This is one reason I have always championed MOSFETs for this application. They do not lose transconductance at lower currents in the same way that bipolars do.

Steve is correct - if you simply must do it, get a low current, high voltage bipolar.

The premium devices for these applications are the TO-92 MOSFETs from Supertex and Zetex which have quite small Ciss for a MOSFET and which can work in a 500V environment. They're about $0.50 each.

In all the versions of this I've built with MOSFETs the MOSFET is fast and linear enough that it does not change the tube's plate signal at all, as the tube is slower and has less voltage range than the MOSFET. All of the tone change comes from the MOSFET providing a lower impedance later on.