If running directly heated cathodes on DC, can they be wired in series? (4x or 6x 833a) For the purposes of wire gauge, rectifier, smoothing, transformer, etc. the thing to do would be wire them in series, for 10A at 40 or 60v respectively. If that is done, would that not cause a problem of a hotspot on the negative end since the cathode current of all 4 or 6 tubes would have to glow through each filament? It seems like this would be a great way to ruin tubes, but maybe there is a way to negate the problem? Even if polarity is switched periodically, I would think the tubes at each end of the series would still see far more wear than the middle. Maybe a compromise of sets of two in series for 20v at 20 or 30 amp could be done? Otherwise it's 10v at 40 or 60 amps, at which point I may end up using buss bars for the filament wiring, which might look cool, but is a bit of a pain...
The 833A is a transmitting triode. Go to the tube manuals and look this tube up. There are specific instructions on how to run the filament for longest life. I have worked with medium wave (AM) transmitters in the 1 and 5 kw range that use this tube. I suggest the parallel connection and a center tapped filament transformer for AC operation. These tubes are expensive, and maximizing reliable service was necessary in commercial use to minimize outages.
DC operation in parallel is probably OK but regular switching of the polarity of the DC filament voltage may be a good idea. Do not connect directly heated tubes up like Christmas tree lights....
DC operation in parallel is probably OK but regular switching of the polarity of the DC filament voltage may be a good idea. Do not connect directly heated tubes up like Christmas tree lights....
Well, exactly this has been done in portable battery powered radios with battery tubes and the option of mains supply. In this mode all filaments were series connected and fed from the mains via a rectifier, a filter capacitor and a dropping resistor. There were additional resistors form each tube's negative filament pin to ground to take over plate and screen currents.
Yes, I know I'm talking about small tubes here that cannot be compared to big lamps as a 833A...
Best regards!
As already mentioned a move, series string 833 filaments is not a good idea.
Series string filaments of tubes in radios were usually done with tubes that had
indirect heated Cathodes. Then there was no bias issue.
There Is a bias issue when the tube is a Direct Heated Triode, like the 833 (DHT).
An 833 has a u = 38. If the bottom tube has 0-10V, and the top tube has 30-40V, the centers of those filaments are 5V and 35V. That is a 30V difference for 4 tubes. For 6 tubes it is 50V.
u x Vg:
38 x 30V = 1140V.
38 x 50V = 1900V.
Even if the load resistance = the 833 plate resistance, (gain = 1/2 x u) that gives:
520V for 4 tubes
950V for 6 tubes
Yes, you have to bias each tube separately.
And that means you have to have 4 bias taps or pots.
And for RC coupling from the driver, you need 4 coupling caps.
For Interstage transformers, you have to have 4 secondaries, and apply the 4 different bias voltages to the bottom of the 4 secondaries.
Series string filaments of tubes in radios were usually done with tubes that had
indirect heated Cathodes. Then there was no bias issue.
There Is a bias issue when the tube is a Direct Heated Triode, like the 833 (DHT).
An 833 has a u = 38. If the bottom tube has 0-10V, and the top tube has 30-40V, the centers of those filaments are 5V and 35V. That is a 30V difference for 4 tubes. For 6 tubes it is 50V.
u x Vg:
38 x 30V = 1140V.
38 x 50V = 1900V.
Even if the load resistance = the 833 plate resistance, (gain = 1/2 x u) that gives:
520V for 4 tubes
950V for 6 tubes
Yes, you have to bias each tube separately.
And that means you have to have 4 bias taps or pots.
And for RC coupling from the driver, you need 4 coupling caps.
For Interstage transformers, you have to have 4 secondaries, and apply the 4 different bias voltages to the bottom of the 4 secondaries.
Wall-power kitchen radios were (after the very first) all indirect heat, yes.
Battery-power "beach radios" were often directly-heated tubes. Naturally wired parallel across the 1.5V A battery, but sometimes there was a wall-power option where a (massive!) switch or plug put them in series. Another design ran all the filaments series on a 9V filament batt. The RF/IF/1stAudio stages would "self-bias" with 10Meg gridleaks returned to the more-negative side of the filament. On battery the power tube would back-bias with a resistor in the B- return; on wall-power it could get bias from the filament string. However tiny-tube bias in a radio can be simpler (or more crude) than we can reasonably do on a hulking 833.
Ok, so it's fairly well established that at higher voltages and currents, series connection is not really an option. Rmb mentioned ac with a center tapped transformer, and whilst many methods of reducing hum are discussed in the heater wiring sticky thread, it is mentioned that the higher current makes hum elimination harder. Nazaroo gave a circuit specifically for large DHTs, but then stated that it might as well be dc with a polarity reversal switch, because "why have hum if you don't have to?" So did most larger transmitters use ac? I know WLW used dc on the filaments of their 500kW transmitter (something like 30v at a thousand amps or something ridiculous), supplied by massive motor generators. They only switched to ac when tubes with three filaments each were used, so that 3 phase could be used to somehow make the hum cancel.
If you balance the ends of the filaments with regard to the center, the hum can be reduced. i.e., a center tap on the winding would be at or close to the center.
But consider a wire (filament) that has lots of AC current through it. It is an electromagnet with a field that has 120 alternations per second.
If that "electromagnet" filament is closer to one side of the steel plate than the other side (steel plate 833, not carbon plate 833), it will be attracted to one side, and will vibrate (move). That changes the distance between filament, grid, and plate.
What might you expect might happen?
Hum?
But consider a wire (filament) that has lots of AC current through it. It is an electromagnet with a field that has 120 alternations per second.
If that "electromagnet" filament is closer to one side of the steel plate than the other side (steel plate 833, not carbon plate 833), it will be attracted to one side, and will vibrate (move). That changes the distance between filament, grid, and plate.
What might you expect might happen?
Hum?
Good call. Carbon anodes are not only non-magnetic, but I believe there are less secondary electrons when the "target" is Carbon, versus a steel "target".
It is primarily the direct heated filament that vibrates.
The filament is a little like the string on a musical instrument.
The steel plate is generally a little more geometrically stable object than the filament is.
It is primarily the direct heated filament that vibrates.
The filament is a little like the string on a musical instrument.
The steel plate is generally a little more geometrically stable object than the filament is.
I would expect the filament to vibrate more, as it is quite soft and springy, especially when hot. From what I have gathered the carbon anodes are more durable as well, and while I won't be running them hard, if I decide to I can be more confident. They are good for 1600W audio in push pull (per pair) at 3kV with convection cooling, and I need 2100W. So with 4 I'll have 1100 more watts than needed and they won't be working too hard. If I use 2 I'd have to push them to 2100W (still at 3kV), but I'd use forced air cooling.
You're going to spend the money anyway elsewhere in the project, just DC heat them from individual windings/rectified supplies.
Even if you're running the 833 low and hot (maybe 750V/500mA), you'll have positive grid bias and you'll really want to keep an eye on the current that the tube is drawing, which isn't so easy if the filaments of each tube are tied together.
The 10V/10A requirement is a total pain in the *** though!
Even if you're running the 833 low and hot (maybe 750V/500mA), you'll have positive grid bias and you'll really want to keep an eye on the current that the tube is drawing, which isn't so easy if the filaments of each tube are tied together.
The 10V/10A requirement is a total pain in the *** though!
Could I not use a single dc filament supply, then have a current meter on both the positive and negative leads of each tube, the difference in current being the cathode current? Would plate current be an accurate measure of cathode current? Also on the tube datsheet for 3kV plate, grid bias is recommended to be -70V, not positive.
I don't see how 10V 10A is any more of a pain than 5V at 14.5A for a 4-400A, or 7.5V at 21A for a 4-1000A.
I don't see how 10V 10A is any more of a pain than 5V at 14.5A for a 4-400A, or 7.5V at 21A for a 4-1000A.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.