I was thinking about current sharing in parallel damper diodes (or diode wired triodes for that matter). Not being able to find a rule of thumb I decided to crank up LTspice and see what I could see as it were.
To simulate fairly mismatched diodes I used two somewhat similar triodes in the form of 12AU7 and 6SN7. As a first pass I ran a single 12AU7 and divided current by voltage drop and got an effective resistance of 2K. This is no where near the data sheet Rp but I just ran with it. Now with the two "diodes" in parallel with a 2K resistor in each plate I got a 10% current difference v.s. 32% with none (1 ohm) and 27% with 200 ohm resistors.
I suppose the 200 ohm would be "safe" but the 2K seems really solid given that the simulation was with totally different triodes.
Does anyone have a rule of thumb that they use?
To simulate fairly mismatched diodes I used two somewhat similar triodes in the form of 12AU7 and 6SN7. As a first pass I ran a single 12AU7 and divided current by voltage drop and got an effective resistance of 2K. This is no where near the data sheet Rp but I just ran with it. Now with the two "diodes" in parallel with a 2K resistor in each plate I got a 10% current difference v.s. 32% with none (1 ohm) and 27% with 200 ohm resistors.
I suppose the 200 ohm would be "safe" but the 2K seems really solid given that the simulation was with totally different triodes.
Does anyone have a rule of thumb that they use?
Your Thread title says "Parallel Diodes".
Did you mean Parallel Triodes?
Or are you just trying to model 'Parallel Diodes' by using 'Parallel Triodes' that have a resistor from the grids to the plates?
For Damper Diodes, start with the "static" resistance of the diode (data sheet voltage drop/current).
The dynamic impedance may possibly be lower than that static resistance.
Consider those values, when you pick the value of the series resistor in each plate of the damper diodes.
Did you mean Parallel Triodes?
Or are you just trying to model 'Parallel Diodes' by using 'Parallel Triodes' that have a resistor from the grids to the plates?
For Damper Diodes, start with the "static" resistance of the diode (data sheet voltage drop/current).
The dynamic impedance may possibly be lower than that static resistance.
Consider those values, when you pick the value of the series resistor in each plate of the damper diodes.
In reality, I've found using 27R or 47R per plate works fine to prevent hogging.
Also, why not tie grid to cathode instead of to plate through 4k7?
Here's LTSpice model data for a 6D22S damper tube
* 6D22S rectifier
.SUBCKT 6D22S 1 2
b1 1 2 i = 6.77E-3 * uramp (V (1,2)) ** 1.2646
Cpk 1 2 10P
.ends 6D22S
Also, why not tie grid to cathode instead of to plate through 4k7?
Here's LTSpice model data for a 6D22S damper tube
* 6D22S rectifier
.SUBCKT 6D22S 1 2
b1 1 2 i = 6.77E-3 * uramp (V (1,2)) ** 1.2646
Cpk 1 2 10P
.ends 6D22S
In the parenthetical I mention "or diode wired triodes for that matter" as I assume the issues are the same. I used the triodes because it was easier to figure out two similar triodes than two similar diodes in my model database and I was considering using a diode wired 12AU7 as a soft start element in a SS rectified B+.
I tied the grid to the plate as a positive grid should turn it on harder than tying it to the cathode. The resistor is of course there to limit grid current.
Since the tube is a dual triode I could parallel the sections. However if I re-bias the load a little bit I could get the total current under 20mA and then go ahead and just make it a simple pass regulator since it is there already. 🙂
I tied the grid to the plate as a positive grid should turn it on harder than tying it to the cathode. The resistor is of course there to limit grid current.
Since the tube is a dual triode I could parallel the sections. However if I re-bias the load a little bit I could get the total current under 20mA and then go ahead and just make it a simple pass regulator since it is there already. 🙂
You could take the practical approach and test the dc voltage with a known load when just using one 'diode' at a time, then pad in some series resistance to equalise any differences. You may want to confirm that mains voltage doesn't change between measurements.
I did static voltage drop testing on a large batch of 5U4 of unknown history - each diode's drop can certainly change with use, but also the static current can change at different voltage drops, so just balancing on one static measurement may be misleading.
I did static voltage drop testing on a large batch of 5U4 of unknown history - each diode's drop can certainly change with use, but also the static current can change at different voltage drops, so just balancing on one static measurement may be misleading.
When it comes to regulator service the 12AU7 doesn't seem great (not too surprising) but the 6EM7 shows a lot of promise. Just at quick non-optimized run...
The heater winding on the PT I have in mind may not be quite up to it but I have a couple other filament trannies laying around. Hmmmm... Might be worth a few kicks.
The heater winding on the PT I have in mind may not be quite up to it but I have a couple other filament trannies laying around. Hmmmm... Might be worth a few kicks.
If I do use a regulator like that above would I want to connect the output to the first stage directly, to an RC decoupling network at the first stage, or should I use a final reservoir capacitor at the output of the regulator?