# Problem with replacing selenium rectifier

#### Swaine1615

Joined Dec 24, 2011
10
I’m recapping a small preamp and want to replace the selenium rectifier. This is on an analog chime system that has small pickups on little chime bars that are struck by solenoids. The preamp is built around a 12AX7 tube. I finished the recap and verified all the resistors are in tolerance.

The transformer AC secondary is 120v and the rectifier is configured as a voltage doubler - so there are two rectifiers. The output voltage was 240VDC before changing out the seleniums. This yielded 106VDC on the plate of the first state and 148VDC on the plate of the second stage.

With 50 ohm 5W resistors in series with the diodes the output voltage was 322VDC. So I changed to 200 ohms and the output only dropped to 318VDC with plate voltages of 138 and 190 on first and second stages. I then went to 300 ohms and the voltages only dropped another volt.

Any suggestions? Everything I’ve read suggests that the series resistance shouldn’t have to be so large and it doesn’t look like further increases are going to help much to bring down the voltages to previous levels.

Joined Mar 10, 2018
4,057
Fast easy way is to use Zener diode in series with HV source. But
make sure you handle the power it has to dissipate.

For example a 100V zener with 100 mA going thru it has to get rid of 10W,
So heatsink a must. This is overkill but you get the idea -

https://www.digikey.com/catalog/en/partgroup/1n33-zener-diodes/41979

Regards, Dana.

Regards, Dana.

#### LesJones

Joined Jan 8, 2017
4,227
According to the data sheet for the 12AX7 (ECC83) the anode current for each section is going to be in the order of 1 mA so the total will be about 2 mA On the assumption that the valve is the only thing taking current from the HT rail then we need to drop about 80 volts at 2 mA which would make the resistance required 40 K. So each diode would require about 20 K in series with it. I practice it would be a bit less than this as most of the current will be drawn near the crest of the waveform. It is possible that the selenium rectifiers have increased their internal resistance with age so the design voltage could be higher than the 240 volt that you measured. If you have the schematic it may have voltage readings on it.

Les.

#### crutschow

Joined Mar 14, 2008
34,845
It isn't just the higher resistance of a selenium rectifier, it's their inherent voltage drop of perhaps 5 to 10V that you need to account for when changing to a silicon rectifier.
So, as Les noted you will need a much higher series resistance to generate that much drop if the load is only a couple mA.
Either that or use a series Zener of the appropriate voltage.

Way back when, not really knowing what I was doing where a little knowledge is a dangerous thing, I replace a plate-voltage selenium rectifier that had failed with a silicon rectifier in an old tube TV, without any other changes.
The TV then worked but had a bright edge along one side of the CRT display which was the edge of the scan.
I eventually realized it was likely due to the higher plate voltage from the silicon rectifier causing the horizontal scan to operate abnormally.

Last edited:

#### Swaine1615

Joined Dec 24, 2011
10
Thanks much all! I don’t have a schematic and have had no luck finding one. I even contacted the original company, Maas Rowe, which is still in business, but they never responded.

I like the thought that the rectifiers have increased in resistance. It doesn’t make sense to me to have to add something to drop the voltage because there is nothing else in the circuit I’m changing at this time other than the seleniums. So perhaps I’m good and just didn’t realize it.

So, with 50ohms in series with the diodes (smallest high wattage I have handy), I get about 1.8ma draw and there are no other loads. But only about .2ma goes to the first stage and 1.6ma to the second. The drop to the first stage is across two 220k resistors with a .1uf to ground between them. The drop to the second stage is only one 52K resistor. So, the math works, but a bit worried that the 1.6ma to the second stage is OK.
Thanks,
Steve

#### crutschow

Joined Mar 14, 2008
34,845
a bit worried that the 1.6ma to the second stage is OK.
That current at a plate voltage of 190V gives a plate dissipation of 304mW which is well within the 12AX7 1W per plate rating.

Last edited:

#### Swaine1615

Joined Dec 24, 2011
10
Excellent! Thanks for the help!

#### KeepItSimpleStupid

Joined Mar 4, 2014
5,088

#### crutschow

Joined Mar 14, 2008
34,845
After reading the reference from KISS, I would add about a ≈5k, 1/10W (or larger) resistor in series with the silicon diode to reduce the voltage by about 10V, bringing it closer to the selenium value, and also limiting the startup surge current.

#### MisterBill2

Joined Jan 23, 2018
19,466
You can also change the supply from a voltage doubler type to a bridge rectifier, which will provide a voltage closer to what you need. AND it will not make as much heat with a big resistor dropping voltage. Plus the regulation will be better and there will be less ripple.

#### Swaine1615

Joined Dec 24, 2011
10
Hi Crutschow,
Ok, now I’m confused again. The article suggests a series resistance of around 200ohms to drop the 10v. Can you help me understand why my situation is so unique that it requires 5K ohms?

Thanks MisterBill for the suggestion, but I’d prefer to try and keep the amp as close to the original as possible. I was hoping just to replace the caps and rectifier.
Thanks!
Steve.

#### crutschow

Joined Mar 14, 2008
34,845
The article suggests a series resistance of around 200ohms to drop the 10v. Can you help me understand why my situation is so unique that it requires 5K ohms?
Not unique, just specific.
You can't just look at the answer, you need to look at how the answer was derived (using Ohm's law).
The 200Ω was for 50mA of load current, which is 25 times larger than the 2mA you have.
Ohm's law states that the voltage drop across a resistor is equal to the current through it times its resistance, so to get 10V at 2mA requires 10V/2mA = 5kΩ. (or 25 x 200).

#### Bordodynov

Joined May 20, 2015
3,188
I've experimented with selenium rectifiers and even burned one. I disassembled it and found many square elements. You can simulate selenium diodes with high-voltage silicon diodes. They also consist of a series of sequentially connected diodes and many volts fall on them. The higher the voltage diodes, the greater the voltage drop.
See 2CL69
https://www.belchip.by/sitedocs/29495.pdf
Or you can take 17 low-power diodes (e.g. 1n4148) included in the series. This is the easiest way to do it.

#### MisterBill2

Joined Jan 23, 2018
19,466
I am wondering about the logic behind not wanting to change the power supply circuit, unless the whole thing is built up on a very tightly built printed circuit board. A bridge rectifier built of discrete diodes will take up far less space than even one selenium rectifier. And if any service would be needed in the future there are many silicon diodes that would be work as substitutes. And, if the circuit is a half-wave voltage doubler, the change would be even simpler, and only require one diode.

#### crutschow

Joined Mar 14, 2008
34,845
I've experimented with selenium rectifiers and even burned one.
As I recall, they give off a sort of rotten-egg smell when they fail and overheat.

#### Swaine1615

Joined Dec 24, 2011
10
Thanks MisterBill. Perhaps I’m just miss-understanding. Since this preamp uses a voltage doubler, wouldn’t I need to change out the transformer with twice the output to use a bridge rectifier and get the same output voltage as I had with the doubler? Or are you suggesting I configure the bridge as a doubler?

#### Bordodynov

Joined May 20, 2015
3,188
As I recall, they give off a sort of rotten-egg smell when they fail and overheat.
And so it was, and the wire insulation caught fire like a bickford cord.

#### MisterBill2

Joined Jan 23, 2018
19,466
Since the replacement rectifiers produce far more voltage the idea is that to get the original voltage, or something near it, with the far more efficient diodes, you would get the voltage that you need using a circuit that just delivered the peak voltage. The whole reason for my suggestion is that you would not need to use a different transformer voltage with a non-doubler circuit because the diodes are so much more efficient.

#### Swaine1615

Joined Dec 24, 2011
10
Ahhh. Got it, thanks.

#### Bernard

Joined Aug 7, 2008
5,784
A little off center but remember the 5" X 5" X 2 " selenium rectifiers used with alternators in the 1950s ?