We're not overheating the cathode, because we'll be running it at its nominal voltage. There's really no way to tell whether this kind of overload will cause discoloration of the material. Of course, flaked off portions are easily distinguishable by the eye.
But then again, it does no harm to crack open a tube for examination after its spirit has been exiled to thermionic Valhalla by yours truly...
Today I 'm not going to set up the experiment, but perhaps in a few days. I'll keep you informed.
Jurgen
But then again, it does no harm to crack open a tube for examination after its spirit has been exiled to thermionic Valhalla by yours truly...
Today I 'm not going to set up the experiment, but perhaps in a few days. I'll keep you informed.
Jurgen
Connecting the grid to the plate effectively turns a triode into a diode. A small fraction of the electrons will be absorbed by the grid, but the majority will fly through and strike the plate. Any current draw from the cathode might cause some additional heating of the filament in a DHT, but that's not the dopic of this discussion.
Terman speaks
Here's an excerpt from "Electronic and Radio Engineering", by Frederick Terman, from pages 702 and 703:
"Voltage drop in the filament causes the negative half of the filament to supply more electrons to the plate than does the positive half; moreover, since the number of electrons drawn from any portion of the filament is proportional to the three-halves power of the electrostatic field, the extra current from the negative end of the filament is greater than the reduction in current from the positive end. This action tends to cause the space current to increase when current is flowing through the filament.”
In an AC-heated filament, this means that the overall cathode/plate current increases slightly as the AC filament voltage peaks in either direction (“when current is flowing in the filament”), due to the non-linear electrostatic effect. This will occur at a rate of 2 times the AC frequency (100 or 120 Hz). By itself, to the first order, this modulation of the plate current would be only an additive noise (hum) as opposed to a multiplicative effect. But as I stated earlier, since gm is crudely proportional to plate current, gm is also modulated at a 100 or 120 Hz rate. This implies that everything the tube will pass will be intermodulated by 100 or 120 Hz. Every musical spectral component will have twin skirts at 100/120 Hz, 200/240Hz, 300/360Hz, etc. This can be readily seen in spectrum analysis. I’ve seen them in my 60 Hz powered 833s.
To me, this effect implies that we should avoid line-frequency AC filament supplies for best performance. Either we should live with the one-sided aging of a DC-powered filament (perhaps switching filament polarities periodically as EC8010 suggests) or strive for ultrasonic heating. My ultrasonic pseudo-random noise idea above may have seemed outlandish, but for a group willing to go to great lengths to achieve ultimate performance, a group used to fancy magnetics and complex Pimm CCSs, this suggestion may be a real option. Getting rid of discrete frequencies from the amplifying environment, whether 60 Hz or 100 KHz, is probably worthwhile. I have a lot more work to do on this of course, and unfortunately little time at the moment.
BTW, Terman also talks about how the magnetic field generated by the substantial filament current can modulate plate currents. When the filament current peaks in either direction, plate current decreases slightly. This effect opposes the electrostatic effect discussed above, but probably not in any substantially compensatory way.
I hope no one feels that sacrificing a 300B is necessary. The literature is pretty solid on this.
Here's an excerpt from "Electronic and Radio Engineering", by Frederick Terman, from pages 702 and 703:
"Voltage drop in the filament causes the negative half of the filament to supply more electrons to the plate than does the positive half; moreover, since the number of electrons drawn from any portion of the filament is proportional to the three-halves power of the electrostatic field, the extra current from the negative end of the filament is greater than the reduction in current from the positive end. This action tends to cause the space current to increase when current is flowing through the filament.”
In an AC-heated filament, this means that the overall cathode/plate current increases slightly as the AC filament voltage peaks in either direction (“when current is flowing in the filament”), due to the non-linear electrostatic effect. This will occur at a rate of 2 times the AC frequency (100 or 120 Hz). By itself, to the first order, this modulation of the plate current would be only an additive noise (hum) as opposed to a multiplicative effect. But as I stated earlier, since gm is crudely proportional to plate current, gm is also modulated at a 100 or 120 Hz rate. This implies that everything the tube will pass will be intermodulated by 100 or 120 Hz. Every musical spectral component will have twin skirts at 100/120 Hz, 200/240Hz, 300/360Hz, etc. This can be readily seen in spectrum analysis. I’ve seen them in my 60 Hz powered 833s.
To me, this effect implies that we should avoid line-frequency AC filament supplies for best performance. Either we should live with the one-sided aging of a DC-powered filament (perhaps switching filament polarities periodically as EC8010 suggests) or strive for ultrasonic heating. My ultrasonic pseudo-random noise idea above may have seemed outlandish, but for a group willing to go to great lengths to achieve ultimate performance, a group used to fancy magnetics and complex Pimm CCSs, this suggestion may be a real option. Getting rid of discrete frequencies from the amplifying environment, whether 60 Hz or 100 KHz, is probably worthwhile. I have a lot more work to do on this of course, and unfortunately little time at the moment.
BTW, Terman also talks about how the magnetic field generated by the substantial filament current can modulate plate currents. When the filament current peaks in either direction, plate current decreases slightly. This effect opposes the electrostatic effect discussed above, but probably not in any substantially compensatory way.
I hope no one feels that sacrificing a 300B is necessary. The literature is pretty solid on this.
Terman indeed suggests what I have suggested here as well. But rest assured, the 300B's that I was considering to sacrifice, are ones that I consider unfit for amp duty. They are leftovers from pairs of which one had a filament failure. They are all from the same brand, and from a production run that was infamous for its filament failure. Therefore, I don't trust them in my amp but consider them to be perfect fodder for experiments like this.
But the weather is turning really well this weekend, so I doubt the experiment will be carried out very soon. That also means Brian has plenty of time to talk me out of this ;-)
But the weather is turning really well this weekend, so I doubt the experiment will be carried out very soon. That also means Brian has plenty of time to talk me out of this ;-)
I too have a few of those unpaired 300B's left around because their mates died of a premature filament failure. I use them for debugging a new design, and for those times that I feel like grossly exceeding the published ratings. It seems that the ones that lived this long are pretty rugged tubes, and don't sound half bad. I really don't mind running them at say 400 volts and 120mA with a 2.2K ohm load, and cranking rock music through them. I have seen 12 watts at 3% come out of these tubes. I have even experimented on the ones that only have half of a working filament left, they sound really bad though. Warning, when that loose piece of filament that is hanging around inside the tube decides to touch the plate, big sparks will fly around inside the tube, making life dificult for your speakers!
I would save the 300B's for some less than critical audio experiments, or find them a new home. It would seem that some cheaper DHT's could be found to sacrifice. I have a box full of DHT vacuum rectifiers somewhere in my warehouse, perhaps they could be tortured? I will look for them next time I go over there. I have some other random DHT's in there too. It would seem that the higher the filament voltage the better.
I would save the 300B's for some less than critical audio experiments, or find them a new home. It would seem that some cheaper DHT's could be found to sacrifice. I have a box full of DHT vacuum rectifiers somewhere in my warehouse, perhaps they could be tortured? I will look for them next time I go over there. I have some other random DHT's in there too. It would seem that the higher the filament voltage the better.
And for your output transformer and power supply as well... But that's the exact reason I don't want to use them for anything else than short-lived experiments. The filament problem with my tubes is structural: three out of three pairs had this failure within a few months of use, and I was not the only one having problems with these tubes. The leftovers just won the "who goes first" game and may fail any minute. That's why I consider these leftovers dangerous to use in any application which is not continuously supervised. An open filament is not dangerous by itself, but when the filament touches the grid or the plate, the amount of damage caused can exceed the price of a tube by a large factor. I prefer ruining a dodgy 300B for science over ruining a fine amp because of misplaced frugality...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.