Selenium rectifiers to diodes

[TrevorG3VLF]

When valve rectifiers are replaced by silicon diodes, a resistor should be placed in series. This extends the current pulse and reduces dissipation in the transformer and reservoir capacitor and result in the correct voltage.

Old selenium rectifiers go high forward resistance and heat up and smell of garlick. Should a resistance be placed in series with a replacement diode or will a [B]NEW[B] selenium rectifier be a match for a diode?

[kalee20]

Silicon diodes have very low forward resistance, though do impose a voltage drop of about 0.8V. This is negligible for HT rectification, but for low voltage (LT battery eliminators etc) it could be significant and you may get a LOWER voltage by replacing selenium rectifiers with silicon.

Personally, for HT, I would put a resistor in series with a replacement silicon diode or even a NOS selenium rectifier, of such a value that the HT voltage is about right with the mains set to specified input voltage (using a Variac).

[Herald1360]

Presumably a NEW selenium rectifier will be a drop in replacement for an OLD one. No resistor needed. A replacement of a NEW selenium rectifier by a silicon diode would require a resistor.

[GMB]

I think you need to clarify what is meant by "a NEW selenium rectifier".

Do you mean new as in "made last week" or new as in "not been used yet".

The key thing with using selenium is that it must not get too hot. I am fairly sure the failure mode that leads to awful fumes involves them initially getting very hot. Selenium was supposed to have a self-healing aspect when it came to impulse damage. Aged selenium rectifiers I suspect are likely to have a higher resistance then brand new which could lead to thermal disaster depending on the current. You have to keep checking very carefully.

[TrevorG3VLF]

By new I meant a rectifier to new specification. I have details of a host of valve rectifiers but have not seen similar details for selenium rectifiers.
I assume there is some voltage drop or they would not have fins or a metal heat sink.
Old selenium rectifiers go high resistance in the forward direction and leak a little in the reverse direction. In the few HT cases I have dealt with, I have left the selenium rectifier in place and connected a diode in parallel. No need to construct a terminal board. Check the temperature.

The latest one I have looked at was a high voltage EHT stick rectifier in a scope. This was replaced by a few 1N4007 diodes in a ball point pen tube.

[paulsherwin]

They do have a significant forward resistance even when new.

[Craig Sawyers]

Originally Posted by GMB

I think you need to clarify what is meant by "a NEW selenium rectifier".

Do you mean new as in "made last week" or new as in "not been used yet".

You can still buy new selenium rectifiers from https://www.gdrectifiers.co.uk/about_us

They made custom multi-plate rectifiers for me the restoration of the Lorenz SZ42 cipher machine at Bletchley Park. They even sprayed them the same colour as the originals (red).

The main difference is that original rectifiers use steel plates, and GD manufacture them on Aluminium. These have better long term reliability, and a lower voltage drop.

Craig

[Techman]

Originally Posted by TrevorG3VLF

I have left the selenium rectifier in place and connected a diode in parallel.

This is actually very bad practice. You're still left with the reverse leakage of the old rectifier, and the parallel added forward pass of the old rectifier added to the already low forward resistance of the new diode is going to give HT even higher than with just the diode in circuit, and particularly if no series resistor has been added. I've had to deal with stuff that's passed through here where someone in the past has done this, and what with blown electrolytics, cooked valves and a mains transformer stood in a pool of melted wax - not good. Also, what if the original rectifier suddenly fails?

However, it's your choice to do it your way, and if you think you've got away with it, then fair play to you!

[greg_simons]

Originally Posted by Craig Sawyers

You can still buy new selenium rectifiers from https://www.gdrectifiers.co.uk/about_us

They made custom multi-plate rectifiers for me the restoration of the Lorenz SZ42 cipher machine at Bletchley Park. They even sprayed them the same colour as the originals (red).

The main difference is that original rectifiers use steel plates, and GD manufacture them on Aluminium. These have better long term reliability, and a lower voltage drop.

Craig

Those look ideal for my 2000 tripler rebuild, might've saved me some trouble there. thanks.
Greg.

[AC/HL]

Like the cat's whisker before, it wouldn't have been used if it hadn't been the only choice. You could say that the modern diode has to be degraded to imitate one, usually by adding internal resistance.

[TrevorG3VLF]

Originally Posted by Techman

This is actually very bad practice. You're still left with the reverse leakage of the old rectifier, and the parallel added forward pass of the old rectifier added to the already low forward resistance of the new diode is going to give HT even higher than with just the diode in circuit, and particularly if no series resistor has been added. I've had to deal with stuff that's passed through here where someone in the past has done this, and what with blown electrolytics, cooked valves and a mains transformer stood in a pool of melted wax - not good. Also, what if the original rectifier suddenly fails?

However, it's your choice to do it your way, and if you think you've got away with it, then fair play to you!

If as I suspect, when a Si diode is in parallel with a Se diode, the effect will be that of the Si diode which has a voltage drop of up to 1V. The Se diode will have no effect on the output voltage.
The electrolytics will be overvoltaged by both types of diode until the rest of the valves have warmed up. If the valves are cooked due to over voltage, that brings me back to my original question of what the voltage drop of the Se diode is and whether a series resistance is desirable. Reverse leakage will show up as high temperature in the Se diode.
My bodges are for my own use and I would not do the parallel trick for any equipment which is not supervised by me. I have little experience in this area, perhaps I have been lucky. I have very little time and so I go for the quickest and easiest way, I doubt if I will do many more radio repairs.

[Sideband]

I saw this once in an old TV where the selenium rec had been paralleled with a BY100 silicon and no series resistor. Bulging and weeping smoothers and a severely over-run line output stage was the result...….I was surprised that the smoothers hadn't actually blown violently but they were rated at 350V which must have helped. I can't remember if I actually fixed the set or wrote it off...….

I wasn't aware that new selenium rectifiers were still available? There may be some new old stock ones about but I wouldn't trust them. Besides the smell when they fail has to be witnessed to be believed....!

[Herald1360]

If the old Se rectifier is to be left in situ, just add another tuppence worth of Si in series with the parallel pair. Won't matter what state the Se job is in then.

[GMB]

The thing about selenium is they have a very low PIV rating so you usually find them in stacks. Add in a 1V forward voltage drop - per unit - and you can see why they drop a few volts. But they have a lowish maximum temperature rating so it's thermal problems that get them.

I have a suspicion that they are much more reliable than people think. Indeed their claim to fame was extreme long life and surge immunity. It may be that the only thing that degrades them is getting rusty (as old ones used steel plates).

If you believe their P.R. then the reason for the smelly catastrophic failures was typically electrolytic capacitor failure resulting in the rectifier overheating - the one thing they cannot stand.

[G6Tanuki]

The bad thing about leaving the old Se rectifier in parallel with the new diode is when the old Se rect ages more, and decides to go short-circuit.

Why leave it there, when you can easily hoick it out and fit a nice 3-way tagstrip to take your new diode and the obligatory series-resistor?

[One potential issue with this: generic silicon power-rectifiers are 'slow switchers' due to charge-storage effects in the semiconductor junction - this can cause significant RF/MF-noise-generation which is something you don't want around a receiver. Modern Schottky-diodes are no more-expensive and their lack of charge-storage effects makes them a lot quieter]

[Herald1360]

1000V Schottkys are a bit thin on the ground..... fast recovery Si does the job OK.

[TrevorG3VLF]

Originally Posted by Herald1360

1000V Schottkys are a bit thin on the ground..... fast recovery Si does the job OK.

I have used RAS310AF diodes since I bought a cupfull at a radio rally many years ago. They are supposed to be suitable for series connection.
I have also used 1N4007 without problems. Latterly several in series for a scope rectifier.
The UF4007 is a fast diode which produces little noise. This may become my standard replacement.

The RAS310AF will need a new mortgage but the 4007s are cheap as chips.