[Electronpusher0]

I have just come across this thread.
I recently decided I needed a capacitor reformer and decided to use the circuit Radio Constructor circuit as a basis, but like Symon I wanted added meters.

My wish list was:
• It should use as many components from the junk box as possible
• Since the objective is to prevent the bang at switch on by reforming the oxide layer it is not necessary to use the exact rated voltage but a minimum of four voltage settings should be available.
• Voltages below 60V are not needed as I have a bench power supply that goes up to 60V that I can use.
• The voltages should be derived either by transformer taps or by doubling as I want to minimise the heat and wanted to avoid dropper chains.
• A current meter with ranges of 10mA, 1mA and 100uA is desirable to indicate progress and final leakage but the flashing neon is unnecessary.
• Volt meters to indicate the HT volts and test capacitor volts are desirable.

The biggest problem was the transformer, while I do have some transformers in the junk box from old valve tape recorders I have stripped for parts, I want to keep them for audio projects.
I did however have a couple of spare transformers with dual 12V secondaries, one with dual 120V primaries so decided to use the old trick of running them back-to-back with the secondaries connected together and the dual 120V windings providing isolated high voltage outputs – 120V or 240V if connected in series.
Additionally if the “output” transformer is under-run by feeding its (series connected 12V) 24V winding with 12V from the “input” transformer I have 60V available from one of the 120V windings.
Finally I decided I could either simply rectify or double the 240V output.
After rectification I now had the following voltages available
• 82V
• 165V
• 330V
• 650V

All the above are actual measured voltages, off load, from the finished unit.
For the Doubling circuit I used a trick I discovered when I was a design engineer some 40 years ago, I do not claim originality but have not seen it used elsewhere.
See the pdf of the (hand drawn I am afraid) schematic..
It uses four diodes connected in a standard bridge configuration but with two reservoir capacitors connected in series (plus voltage sharing resistors). One end of the transformer winding is connected to the bridge circuit as normal. The other end goes to the common connection of a simple changeover switch. The changeover switch connects the other end of transformer winding either to the diodes as a standard bridge, or to the junction of the reservoir capacitors to form a doubler circuit. When used in the doubler configuration the two unused diodes are reverse biased and essentially out of circuit.
The transformer tap selection and Bridge/Doubler switching I combined into a 3 pole 4 way rotary switch to make the voltage selection easier.

For the voltage measurement I used two, modified, three digit DVM modules.
The basic module is available from the well-known auction site and is described as “Three Wire LED Panel Meter Mini Digital Voltmeter DC 0V To 99.9V”. They cost around two or three pounds. They have three wires, Black is common 0V, Red is a power input for the meter and takes 4 to 30V, the yellow wire is the measuring input, 0-100V.
As supplied the meter has a relatively low input impedance, I wanted to connect a meter directly across the capacitor under test so needed to minimise the current drawn by the meter. The IC used in these modules has an input sensitivity of 2V for full scale and uses a potential divider to scale the input to 100V. To increase the input impedance it is necessary to change the lower resistor of the divider from 2K to 220K. Additional resistors can now be added in series with the yellow lead, two 47M ohm and two 10M ohm in series to give a full range sensitivity of 999V. This will give a current drain of about 1uA per 100V which will be indicated on the meter, at 650V with no capacitor it will give a “standing” reading of 6.5uA. I decided that the advantage of seeing the capacitor voltage outweighed the disadvantage of the standing current.
I also cut the track to pin 3 of the display which is the decimal point LED.
I attach photos to show which resistor needs to be changed. This a surface mount resistor and quite fiddly to change, I managed to fit a standard resistor rather than SM types.
The meter supply volts I derived by half wave rectifying one of the 12V secondaries on the input transformer, while not strictly necessary, I also added a 12V regulator to stabilise the supply.
Since I was not using the neon indicator I decided to only use one series current limiting resistor (in the original design the series resistor is chosen to drop about 60V at the desired final leakage current which would result in the neon going out). I chose to use 81K made of three 27K, 2W resistors in series. This gives a maximum available current at 650V of 8mA.
For the current meter I used a movement from an old cheap analogue multimeter, this has an f.s.d of 41.1uA and a resistance of 2K ohm. Shunts were calculated to give the required ranges. The current selection switching is slightly complex due the fact that I only had break-before-make switches to hand. If make-before-break are available it would make the switching much easier.

Finally a biased switch was used to switch the plus end of the capacitor under test from the reforming circuit to a load consisting of two 2K2 2W resistors in series to discharge it.

I have successfully used this to reform the 16uF capacitors on my Mullard High Speed Valve Tester (Although I did later change them anyway)

I attach a pdf of the final circuit and some photos of the finished unit. Sorry about the quality of the hand drawn schematic, I drew it for myself, not for sharing. If anyone is interested I will redraw it on the computer and post it.
Peter