On further reflection I think both of my feedback loops are modulating the cathode FET to compensate for the missing 15ish ma the screen driver is pulling away from the plate so the only real benefit from diverting those electrons through the output transformer would be power.
At that point in the load line the tube is conducting over 150ma I think, so 15ma is still significant though not hugely so.
It’s funny to think of G1 driving power to the speaker.
At that point in the load line the tube is conducting over 150ma I think, so 15ma is still significant though not hugely so.
It’s funny to think of G1 driving power to the speaker.
I've been looking at the 25v supply circuit and don't understand the Zener voltage notation of 10v on the 25v output side. Just an oversight?
I've been thinking about this: Imagine a low mu triode and a high mu triode connected as a Darlington device: both plates to the OPT, driver cathode to output grid and output cathode to ground. Bias voltage and input signal to input grid, of course. The low mu input triode would provide both grid current and negative feedback.
Yes, that’s fun to think about. Though if the anode on the power tube swings low it may cause a limiting effect on the driver tubes cathode follower action. Tubes can’t turn on as much as SS devices. That is to say they have a fairly large on resistance. Maybe the driver in your Darlington idea could be a pentode.
OK, not doing any regulating. It's for a catastrophic failure puting much more than 25v on the grid? Does the 28v-25v G/S difference represent a voltage drop of the FET? What happens if the G voltage rises to 38v? 35v out? I'm not versed in FET's, as you can probably tell. But if that's the case then the zener wouldn't be triggered. So what failure condition does it protect against?
FETs have what amounts to a capacitor for the gate input. The capacitors other lead in this case is the source pin. This capacitor has a max voltage it can withstand before damage, just like any capacitor. In the data sheet it’s the VGSmax. A zener with a value larger than the VGSon and smaller than the VGSmax between gate and source is common practice. So much so that this diode is sometimes included in the chip itself.
The reason the gate reference is 28v and the source output voltage is 25v is because this chip has a RDSon of a couple volts. So unlike tubes, Enhancement mode mosfets are “off” when the source and gate are the same voltage and need some positive gate voltage to become on.
The reason the gate reference is 28v and the source output voltage is 25v is because this chip has a RDSon of a couple volts. So unlike tubes, Enhancement mode mosfets are “off” when the source and gate are the same voltage and need some positive gate voltage to become on.
Yes, it would take a driver triode that can push some current at a very low plate voltage and still withstand high plate voltages 180 degrees later. 12B4A comes to mind. Could possibly work quite well together with a very high mu output tube that operates entirely in the A2 region but the output power would probably be limited compared to the same output tube driven by a mosfet follower.
Connecting the driver tubes plate to an UL tap on the OPT would probably work better but still provide some local feedback through the input triode.
This thread is interesting, any unconventional way of driving power tubes catches my attention.
I have a pair of LL1671/50mA interstage transformers somewhere, perhaps they could be wired as 4:1 and used for cathode drive?
I guess a transformer acts as a voltage source rather than a current source which should lessen the potential problem with current glitches when entering and leaving class A2.
I have a pair of LL1671/50mA interstage transformers somewhere, perhaps they could be wired as 4:1 and used for cathode drive?
I guess a transformer acts as a voltage source rather than a current source which should lessen the potential problem with current glitches when entering and leaving class A2.
Yes, TubeLab seems to have taken that path succesfully, turning cheap sweep tubes into good triodes. I believe the same thing could be done with high mu triodes as long as the feedback voltage divider from plate to grid is buffered with a Fet. The only problematic part is whan happens when the Fet between cathode and ground goes bad, which apparently happens. Blowing up a sweep tube or two wouldn't be the end of the world but I'm a bit reluctant to torture my rather rare and expensive DHTs.
Buffering the feedback to the grid is something I considered too. But I decided if the loop was going to include a fet I might as well just include the cathode fet in that loop instead.
I fused the B+ in the hopes that a failure would open it. If the FET shorts that will probably be ok. If it fails open I’m not sure.
I fused the B+ in the hopes that a failure would open it. If the FET shorts that will probably be ok. If it fails open I’m not sure.
When marrying tubes with SS devices keep in mind thermal stability. A decade and a half ago we actively discussed here using transistors and tubes together. I tried then a "darlington" made of a triode and a transistor. I gave up because it was hard to set optimal bias for both devices without running away with temperature.
What Doug did , is a different story. The FET is properly biased, thermal dependence of the current supplied to the tube is not so critical.
What Doug did , is a different story. The FET is properly biased, thermal dependence of the current supplied to the tube is not so critical.
So today I’m testing all my pin compatible tubes in the amp to see how they compare. I’ll not be changing the bias current though. This test will be power vs THD%. It will show differences in when the tubes enter A2 and how they behave in A2.
First up is a Shuguang EL34. You can see the point where the stage enters class A2 because of the bump in distortion.
First up is a Shuguang EL34. You can see the point where the stage enters class A2 because of the bump in distortion.