Replacing pentode tubes with FETs?

[artofharmony]

I'm delving down a rabbit hole of building a tremolo based on the Vox AC15C1's power amp section (original thread and schematic here). I learned recently that replacing triodes with FETs is very simple, but I don't know what to do about pentodes. Anyone know would I handle translating the screen and suppressor functions to a FET-based circuit?

[R.G.]

I can't think of any way to do it directly with a modern FET. The screen grid is actually a second signal grid, but with very low "gain" compared with the signal grid. It modifies the actual transconductance (that is, current-through per unit of voltage change on the grid) of the tube directly.
In the infancy of MOSFETs, there were units that had a second gate that did this directly, but I haven't seen these in decades.  Maybe they still exist. Try searching for "dual gate MOSFET". Seems like their part numbers all started with "3N". They were very finicky as I (dimly)remember.
You're probably going to have to do this external to the FETs.

[artofharmony]

Thanks for that! Sounds like it's back to researching tubes lol

[R.G.]

I messed up a bit. Yes, dual gate MOSFETs are the answer. But they're not non-existent. Mouser has a couple of Toshiba RF MOSFETs, 3SK293 and 3SK294. They do the second-gate-modifies-gain trick. Toshiba may be getting out of the business, though, as both parts are shown as not recommended for new designs. They're about $0.50 each. There may be others out there.

[artofharmony]

Thanks again! I may just try throwing in some JFETs and see if it gets close enough.

[mozz]

Use the fets in cascode. 

[artofharmony]

I'd seen the term "cascode" floating around during my research, but hadn't paid much attention to it. After seeing it here, I found this article. Seems like that is a good solution! Thank y'all for the help! Getting closer and closer 👍

[PRR]

The screen grid is actually a second signal grid, but with very low "gain"

True, but here the only screen grids are Fixed Voltage. Here, "B+2".


In a FET the same function is baked into the crystal. You'd say "Idss"; Vg2 controls the zero-bias current in a vacuum, the crystal (and its process variance) set zero G1 bias current.

The trem works only on G1 bias wobbles.

The Gm/Vg1 of a FET is different from a vacuum tube (because no vacuum!). So it is about as "different" from a tube trem as any other simple scheme.

[merlinb]

I learned recently that replacing triodes with FETs is very simple, but I don't know what to do about pentodes. Anyone know would I handle translating the screen and suppressor functions to a FET-based circuit?

Do you actually need to translate the screen function? FET characterstics already look more like a pentode than a triode, and in most amps the screen voltage is fixed enough that there isn't much worth 'translating'...

[R.G.]

The dual gate MOSFETs don't do exactly the function of screen or suppressor grids, and the MOSFETs don't have exactly the same change in transconductance with - well anything, exactly, as pentodes. However, they are intended and designed for gain variation with changes in the G2 voltage. It's not a drop in replacement, but since at least one objective is tremolo, there's probably a place for them.

As Paul and Merlin are discussing, there isn't a drop-in replacement for a pentode by simply replacing the pentodes with FET-somethings. As I mentioned, it's very likely that the tremolo part of this will need to be emulated by other circuitry.

[BJF]

Hi There,

On various related

In preamp tube designs you'd choose a pentode in order to get higher gain, though sometimes it would be more economical to use two triodes in series as in two common cathode; with two stages following each other the resulting gain is the product of the amplification of both something that is also used in OP amp designs to increase performance. With the two series triodes there are losses in the interconnection but total gain can be very close to that of a pentode and mechanically a dual triode occupies the same space as a single pentode. Long term economy then usually favours dual triodes.

Due to the transfer characteristics of a pentode it high gain to be easily achieved while it will also produce more and sharper onset of overtones as compared with the triode . Dynamically, as in signal processing a dual triode has a slower and more gradual onset of distortion.

Some more famous small signal pentodes like the EF86 specifically designed for microphone  and pickup amplification would offer low noise and low hum levels along with considerable gain and the EF 86 has a special shield inside to reduce hum interference however as these tube ages
hum levels can reach well above that of a dual triode and a microphonic can be even worse.
This is one reason pentodes are seldom used in high gain tube circuits intended for diliberate distortion but MOS Fets can definitely be used in high gain distortions circuits.

In discrete transistor design one could argue J Fets can be used in most triode preamp circuits when voltages and currents are adjusted whereas MOSFET is better choice transitorising preamp pentode circuits and for low voltage such as 9 to 12Vs there are enhancement types that are suitable. Although design might rather start at what properties are needed for a specific amplifier stage.

In power amp designs CMOS make excellent performance and at low complexity in design and there are complementary units up to 200Ws
There are no similar JFET devices although you could make a power amp at considerable power using a couple of hundred small signal JFET
that would have to be close in specifications and that would mean 2SK types.

For preamp design with MOSFET just like with JFET aswell as with triode and pentodes the overtones produced are directly linked to the anode load or drain load but regardless the onset of distortion is sharper with MOSFEts and the amount of overtones much higher which really is obvious from the transfer graphs.

Looking at something like a phase shift oscillator gain needs to be larger than 28 times which is beyond a single JFET and MOSFET is the obvious choice and especially if power supply voltage is high and ruling out the use of bipolar transistor or OP amp and voltage regulators cannot be used and or overall production cost would be less....or really it is just a sales point that high voltage MOS Fets are used.

Kustom Amplifiers used a dual gate MOSFET in the preamp to create distortion, It was innovative at the time also because it involved deliberate preamp distortion that few other companies considered and it was different than fuzz circuits built into some other amps.
One could argue that the transistor used had something specific because it was a dual gate but probably more important it was an early MOS Fet and again the working point aswell as relative frequency compensations would have bigger influence most notably because the later versions had larger capacitors in decoupling making sound less defined.

For Pentodes the screen grid is really intended to get rid of a kink in response that the early tetrodes had. The intended function could be explained something like this that when electrons at too high speed bounce off the anode suppressor grid and screen grid will bounce back the electrons and moderate speed.

Screen grid can be connected to anode and make the pentode behave like triode but the anode can then no longer take as high voltage but is limited by the screen grid max values. In pentode operation the screen grid is typically a fixed voltage that should be within safe operation while anode voltage can be much higher. So for stability in some constructions with high anode voltage it can be wise to regulate screen grid voltage at
For screen grid appropriate voltage meaning that while anode could take up to 800V screen grid must remain at 300 V.
Screen grid is not intended to carry signal voltage but is typically regulated at a DC voltage.
There are special tubes with more than one control grid intended for AGC.

So building a push pul power amp that has LFO modulated bias could be done with bipolar transistors and if current consumption is not a problem and power is very modest two metal cans can drive a small transformer.
For the bipolar transistor the change in emitter current has a resolution of down to 3mV pretty precise meaning that if input current is expressed in  input voltage for instance by introducing a series resistant with base then change can be pretty precisely controlled in 3mV increments.
For vintage tremolo then a sine wave oscillator such as a phase shift osc can modulate and output need to be adapted to the bias to the bases
So if you have a bias trimpot providing current via resistor to the push pull transistors then modulation can be admitted right there and the modulating sine carrier will partly be cancelled because it appears in same phase at the transformer primary.

The first pedal I built for someone else was a tremolo pedal for a guy who played a Marshall half stack but that brought a Music Man amp for one song to get tremolo.

I made for this a circuit using two J Fets connected as common source but they shared source resistor and decoupling of source
One half passed the audio signal from gate to drain but the other half had its gate fed with a sine wave coming from a phase shift oscillator made with a bipolar transistor.


Have fun
Bjorn Juhl