I have a device with a half wave selenium rectifier. It is fed from a transformer which it says is 145VAC and is followed by a 20uF capacitor. It says the voltage at the capacitor should be 145VDC. This is followed by a 1k resistor and another 20uF. It says the voltage across that second capacitor should be 125VDC.

Would it be OK to replace the selenium rectifier with a silicon one?

 

With a silicon diode the voltage peak across the first capacitor would be just over 200 volts. So you would need to make sure that the voltage rating of the capacitors was high enough. Also make sure that the extra voltage did not damage the circuit it was supplying.

Les.

 

The Selenium-Rectifier has a lot of Internal-Resistance as well as a substantial Forward-Voltage-Drop.
This "old-style" "Power-Supply-System" depends upon this Resistance for Filtering.

You basically have 2 RC Filters in series.

When You change-over to a Silicon-Rectifier You will be eliminating the Resistor in the first Filter,
causing it to no longer act as the same type of RC-Filter that the Circuit was designed with,
this may, or may-not, allow an excessive amount of Noise to be allowed though to the Output.

Adding "enough" Resistance, ahead of your new Silicon-Rectifier,
to cause the Output-Voltage to sag down to
the originally designed operating-Voltage while under normal operating Loads,
should resolve this dilemma, as far as the Filtering part goes.

But, the possible PEAK-Voltages will still be in the ~200-Volt range.
This means that BOTH Capacitors must be replaced with higher-Voltage parts,
preferably ~30% to ~50% higher than the expected new ~200-Volt Peak-Voltage.
And, it wouldn't hurt to increase the Capacitance maybe ~10% to ~20% while you're at it.

Note: The added Resistor is likely to dissipate quite a lot of HEAT,
I can't do any calculations on how much heat because I
don't know the normal Current being drawn on this section of the Power-Supply.

Note-2: If this is a Tube-Circuit of some kind, some Tube-Circuits will tolerate
higher than supposedly "normal" Voltages without any significant problems,
but the surrounding Capacitors may not, and could overheat and explode.

Note-3: In any case, it's a good idea to replace most of the Capacitors with
their modern, and far superior, equivalents.
Generally, You should not mess with Mica-Capacitors,
as they generally don't go bad, and quite often are selected, "critical tuning" parts.

Note-4: There are definitely better ways of creating a "Power-Supply".
You could use a Bridge-Rectifier and an actual Solid-State-Voltage-Regulator-Circuit.
.
.
.

 

I have done that: Replaced selenium rectifiers with silicon diodes in an amplifier power supply. The DC voltage was higher, and I was fortunate that it was still within the ratings of the filter capacitors.
BUT there was an "interesting problem" that developed, which was a soft 120 hZ buzz in the output of the amplifier. NOT a hum, but a buzz. The solution was an additional stage of input filtering, 1 100 ohm series resistor and the a 0.1 mfd capacitor to the supply common. It seems that my diodes did not switch on and off in their conduction as smoothly as the old selenium rectifiers. So there was a very sharp voltage step of about half a volt in the delivered DC output.

DO NOT replace the single diode with a bridge device unless your system can work with a higher input voltage.

 

It is valve equipment - a Heathkit grid dip oscillator. I will in any case be replacing the two 20uF capacitors (new ones are 22uF 250V) as they are getting on a bit now.

After I made this post I realised that I could use LTspice to see what would happen. This seems to show the output voltage would be very similar to that obtained from the selenium rectifier. Did I do something with the simulation?

 

Hi Albert,
LTS expects Vpeak for its AC voltage source ie: 145*1.414 =205Vpk

E

 

Indeed! Just replacing the rectifier will increase the voltage a bit..

 

Hi Albert,
LTS expects Vpeak for its AC voltage source ie: 145*1.414 =205Vpk

E

Yes, of course it does. I always forget that. Thanks.

 

Hi A,
Consider a 5V to 10V drop across a 150Vac Selenium rectifier section.
It's a 1V drop/plate section.
Use a chain of 1N4007 diodes.
E

 

Having corrected the input voltage in the simulation, I find that a resistor of 470Ω in series with the diode brings the voltage back to as expected and adds back in the missing filter section. What wattage rating should this resistor be?

 

Also keep in mind that a Selenium rectifier can handle more abuse than a Silicon one - Over Temperature / Current, Rev Voltage...

 

Hi A,
The only problem with a series resistor is that the voltage drop is proportional to the current drawn.
Do have a value for the inrush current and the operating current of the device?
E

 

The value I am working with is that there is 20V across the 1k resistor, so 20mA. That's how I came up with 6k load resistor (120V @ 20mA). I plan on using a 1N4007 which should easily take the strain.

 

Given that the stated application is in fact a tube type Grid Dip Meter, the DC current load will be rather constant, and so the voltage drop across a series resistor will likewise not change very much. So a half-watt resistor will be an ample rating. Consider that the operation was satisfactory with that old rectifier that probably had a similar forward resistance until it failed.

 

Given that the stated application is in fact a tube type Grid Dip Meter, the DC current load will be rather constant, and so the voltage drop across a series resistor will likewise change very much. So a half-watt resistor will be an ample rating. Consider that the operation was satisfactory with that old rectifier that probably had a similar forward resistance until it failed.

Excellent, thanks.

 

Paper that discusses figuring the voltage drop resistor and shows a few pictures modifying it neatly for a collectable radio.
Microsoft Word - Replacing the Selenium Rectifier.doc