• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Mosfet or other semiconductor substitute for cathodyne?

One of the primary ways to advance is to use the organizing idea of applying known principles to new applications. Tubelab and others have found that semiconductors, specifically some mosfets, can be used in special applications and will not affect the sound in a tube circuit. The best example is in source followers to drive tubes. There a numerous advantages there. I will let others more experienced to advance the argument.

I've recently been interested in phase inversion for push pull right before the finals. The cases where mosfets have been used successfully in tube circuits,( besides CCS) is where there is no gain in the stage. I'm thinking now that a mosfet may be a very good candidate for use in a cathodyne stage. That kind of stage also has no gain. Is there a case to be made for using them there in a tube circuit and not influencing the sound in a negative way? Inquiring minds want to know.
 
A power MOSFET is just fine as a split load, AKA "concertina" phase splitter. Somehow, "cathodyne" doesn't fit, where there is no cathode. ;)

You hit the nail, right on the head. To date, successful integration of FETs into tubed circuitry requires them to amplify current, not voltage. A voltage follower is such a situation and, being a "tricked out" voltage follower, a "concertina" phase splitter is too.

As a FET, like a pentode, can swing closer to the B+ rail than a triode, it can prove to be superior in the split load phase splitter role. Check this AA thread and all associated links out. A triode would not have worked.
 
Last edited:
Jeez Eli, I wish I'd seen that thread before. I spent a lot of energy to engineer a separate high voltage section for the concertina tube stage and already bought the parts for that section. Just preliminarily it looks like I may be able to use the B+ used for the rest of the amp for the concertina.

Actually I'm not put out that someone got to this idea before me. It is the nature of the world when there's billions of individuals inhabiting the planet that someone, probably many people, will get to an idea before you. At least it's nice to know that others are on the same wavelength as me.:D
 
Eli,
Finished reading that thread. It was very informative. Actually using a mosfet wouldn't have changed the B+ requirement for my needs. There really isn't a need, at least in my case, for a higher B+ for the concertina. All you have to do is forego direct coupling from the drive and its resolved.

Something that I found out when I was simulating it on LTspice - When you bias the cathode or source of the concertina at exactly 1/4 of the total voltage applied to that stage then the upper harmonic distortion just disappears! It's pretty remarkable. It seems the 1/4,1/2, and 1/4 rules isn't just a good way to load it for maximum signal output. At low to medium signal levels in there were not even any harmonics above the third. At higher signal levels, for example 2 volts rms in, there were no harmonics above the fifth for my simulation. You should try it yourself Eli, if you have access to LTspice.

I'm inclined to believe it's a real thing and would show up in a real circuit. I'm thinking that any nonlinearity in the transfer curves shows up as an additional interference between the split loads if those loads don't have equal and proportional "space" in which to operate. I'm speculating here.

The bottom line is that if you don't have the voltage necessary for the concertina to be biased at the 1/4 voltage level of the it total voltage, then it may be more important to give up direct coupling than to deviate from that biasing level. That's the scheme I'm aiming for. I going to capacitively couple it and do away with the complication of a high voltage supply for concertina. I don't need that high voltage for driving el84s or 6v6s. YMMV.
 
There is a sticky thread in the instruments and Amps forum....

http://www.diyaudio.com/forums/instruments-amps/190738-hundred-buck-amp-challenge.html

It was a challenge to see who could build the best guitar amp for less than $100. The thread got ugly several times, especially when a few vocal people protested the use of "sand" in a tube amp. After the formal challenge ended, several of us continued tweaking our amps and making new ones under new rules, basically.....rules, we don't need no stinkin rules, lets just make good cheap amps.

The mosfet PI, both split - load, and LTP were discussed during the challenge, but not used. Since the elimination of rules, mosfet PI's have appeared. They can be seen in a few designs starting with post # 1605.

The schematic of that amp as it exists today is included. The PI is a mosfet "cathodyne" using an LND150.

Yes, this is a low $$$$ guitar amp. Distortion in the PI and output tubes under high drive conditions is welcome, and encouraged. The same PI has been used in HiFi applications.

A split load PI, tube, mosfet, or Darlington BJT, works perfectly as long as the both loads are IDENTICAL for AC and DC currents, due to Kirchoff's law. Any imbalance in the loads from output tube grid current WILL cause distortion. For this reason a buffer stage between the PI and the output tubes should be included if this is a possibility.....Yes, mosfet followers are the hot ticket here too. An ordinary BJT will have some imbalance due to base current, but it can be managed if you need to use a BJT for some reason.
 

Attachments

  • GTA_4T_11-10.pdf
    22.6 KB · Views: 695
A split load PI, tube, mosfet, or Darlington BJT, works perfectly as long as the both loads are IDENTICAL for AC and DC currents, due to Kirchoff's law. Any imbalance in the loads from output tube grid current WILL cause distortion. For this reason a buffer stage between the PI and the output tubes should be included if this is a possibility.....Yes, mosfet followers are the hot ticket here too. An ordinary BJT will have some imbalance due to base current, but it can be managed if you need to use a BJT for some reason.

One nice aspect of a split load PI is that if you overload it and have grid current on the output it's the plate impedance that really changes, much more than the cathode impedance. If you have to have overload distortion then that's the preferable way. It will provide 2nd harmonic distortion rather that odd harmonics. So it's kind of benign. So if one goes cheap here it's not necessarily a bad thing, especially in an instrument amp. Even a hifi amp probably would have sound most would not find objectionable at low levels of overload.

I haven't yet read that thread, but will. I'm always up for seeing a good cat fight. Of course, that only goes so far, gets tiresome after a while, and isn't at all enjoyable if you're in it.
 
The schematic of that amp as it exists today is included. The PI is a mosfet "cathodyne" using an LND150.

I made a guitar amp a while back with an emitter follower after a gain stage driving a cheapo 10K:10K Edcor transformer as a phase inverter (sort of a tubelab meets Gibson topology). Might push the $100 amp to a $110 amp, but it works great.



You hit the nail, right on the head. To date, successful integration of FETs into tubed circuitry requires them to amplify current, not voltage.

Nonsense.

fvp5.gif
 

I finally got through the entire thread. I have to say it did get ugly there for a short bit with some attempted bullying by an individual who will go unnamed. (Not you, George.)

It is almost embarrassing how coincidentally close in time my own understanding of the ability to substitute mosfets came to the same understanding communicated by others. It feels a little bit like saying, "You know, I'm theorizing that there might be this universal force that holds us to the ground. I tentatively calling it gravity." I actually hadn't read any of those threads saying the same thing before I came to the same conclusion about cathodynes.

I really do believe that there might be an aspect of consciousness that not only exists within each individual but also in the space between all individuals and that is shared by everyone. Enough of the psychobabble, I guess. :D
 
universal force that holds us to the ground. I tentatively calling it gravity."

I just have to quote an old bumper sticker....

The world sucks, if it didn't we would all fall off.

it did get ugly there for a short bit

Let's just say that some of us got past the "sandophobia" stage earlier than others, and some haven't got there yet, and may not ever.

As a retired design engineer in a Motorola plant, I was taught to always choose the right component for the job, regardless of what it is or where it comes from.....That said I still got a lot of flak for specifying a competitor's semiconductor part for use in a Motorola product.

Sandophobia.....the sometimes irrational fear of silicon when it IS the right part for a given application.

In a "follower" circuit, the main criteria are high Gm and low capacitance from the input pin (grid or gate) to the B+ pin (drain or plate). This capacitance should remain constant over the range of voltages that the device will see in the given circuit. Not all mosfets pass the capacitance criteria.

A phase splitter is a bit more critical with respect to capacitance since a big fet can exhibit unequal phase response at high frequencies.
 
One of my pet peeves when looking at device characteristic charts are curves that appear to be manipulated to look better than they are. the http://www.mouser.com/ds/2/427/91106-60770.pdf seems like an example. I'm not saying it's at all a poor device, or couldn't be used, but you really can't get an intuitive sense of how the things will work just from looking at the Crss. I'm talking about what the chart shows on that device and the chart for this device, http://ixapps.ixys.com/DataSheet/DS98809C(IXTP-U-Y01N100D).pdf

I haven't yet come to any conclusion about which suitable device to use. One thing is certain: I'm making progress in reducing the number of valves in the device I'm trying to emulate. That amp had 16, yep 16, small tubes total for a stereo PPP. I've got mine down to 10 for the same functionality and estimated power. At this point I'm figuring on using a noval small signal tube like a 12ax7 in parallel mode since it's such a wimpy tube in gm. Then the mosfet split load PI driving the PPP stage. I will be using mosfet followers to increase the power of the 4 valve PPP stage to equal the 6 valve PPP in the original. I've gotten almost all the parts ordered. When I get closer to actual construction I'll try to start a thread just for the project.
 
sand works great with tubes provided it's in the right place.
voltage gain is poor use of silicon.
Current gain and concertina duties work amazingly well.
a BS107 fet works very well 200v 3 - 5 ma current N fet TO-92 package
that N100D fet could drive the next stage with 30 or more ma of current from a 12AX7 gain stage
 
Last edited:
I've moved on from the decision whether to use a fet for cathodyne duty to the decision of which fet to use. On it's face the capacitance values of the LND150 tubelab recommended is an order of magnitude better than any other fets I've seen. I'd really like to use it. The nearest value device in capacitance is the one smokingamp recommended, http://ixapps.ixys.com/DataSheet/DS98809C(IXTP-U-Y01N100D).pdf
It has approximately 10 times the values in capacitance of the LND.

The LND is only rated at 3/4 of a watt in the convenient TO-92 package. Originally I tossed it out of consideration for that reason. But I'm thinking now that even though I'm running 300 volts through that stage I can run well within half that 3/4 watt. I just have to have the resistance in each split load high enough that the power dissipation through those resistors take the load off the fet. So it looks like that will be the way I go. It will be driving into source followers, probably using the fets smoking amp recommended, so there shouldn't be much strain imposed on LND150 that's driving it.