Tuesday, 10 March 2015

The MK87 Dreadnaught capacitor reformer - an explanation.

A terrific young lady called Elise has emailed me regarding my capacitor reformer. She'd like some more details on it's function, and, perhaps more importantly ... it's name.

WARNING. Do NOT try this at home. This project calls for some high voltages. It will floor you or kill you if you come into contact with it. It will not warn you, blow a fuse, nor pull your safety trip. It may not give you a second chance. RESPECT IT'S AUTHORITY.

Reforming capacitors carries the risk of explosion.

I will not, under any circumstances, accept any liability if you decide to re-create this project for yourself.

Now we've got that out of the way....

It's not a pretty looking thing. It's had many iterations, hence the state of the box. It used to have a 1uA meter movement (with adjustable shunts), but that suffered so many accidents, the needle ended up shaped like a banana, and eventually the meter went open circuit, due to a major overload. At this point I grafted in a digital meter, salvaged from scrap equipment. It directly reads 0- 999.9 uA, and is powered from a separate power supply (a redundant mobile phone charger)

The front panel also has the terminals for the capacitor under test, and a discharge switch.

On the side is mounted a selector switch, which switches in varying values of current limiting resistor. Nice labelling , eh?

The main power is supplied from a variable transformer (variac) which is built into by bench supply. It's variable from 0-350 VAC.

The sensible option would be to power the unit from a dedicated, isolated multi-tap transformer, but I don't have one...

It's wired up like this: -

So, what is this "reformation" anyway?

Well, when we're servicing or restoring a piece of ageing electronics, the electrolytic capacitors may have begun to lose their polarity. It's a chemical thing. My daughter would know. Anyway, the practical upshot is the capacitor will draw current if woken abruptly from it's slumber. (This even applied to "NOS" capacitors, even if they've never been used)

Say we have a 200uF 450V smoothing capacitor, and it's drawing 10mA at 400V, It's really pretending to be a 40K resistor, and it's going to start to dissipate power (Power = Voltage x current = 400 x 0.010 = 4 watts), and not only that but it's probably not being much of a capacitor either.

This power is going to be dissipated as heat. Now 4 watts doesn't sound like a lot, but this is inside a sealed can (usually aluminium) full of electrolytic gunge and tin foil. Pressure starts to build up inside, the resistance gets lower, more current is drawn, more heat is generated, and the whole situation just gets worse until the can explodes, and showers the workshop in bits of tin foil, and acrid smoke. If you ever experience this, just open a window with your now trembling hands, and leave for 5 minutes to allow the smoke to clear!

What the reforming process does, is let the plates regain their "form" or polarity. We need to gently apply power, and raise the voltage to the capacitors full rated voltage over a period of time. I usually do this for a period of a few hours, sometimes overnight.

So, having made our reformer, how do we use it?

Firstly we need to set our maximum voltage. Move the selector switch round to 0 ohms, and measure the voltage at the output terminals. Don't connect your capacitor as yet. Adjust the variac to give the required voltage (I usually use the rated voltage less 10% to allow for any variations in mains voltages).

Now switch off the supply and hit the discharge switch for 1 min. This will discharge C1. You can now connect the capacitor to be reformed to the terminals. Select the 12 megohm position, switch the discharge switch off, and switch the supply back on. The meter should show a few uA as the capacitor charges. Now decrease the resistor until the cap is drawing approximately 100uA and leave it there. Pop back in an hour or so, and the meter should have dropped again. Reduce the resistance again, and the meter will once again rise.

Repeat the above process until no improvement is made. Take your time. If the can is isolated, feel it to see if it's getting warm. It shouldn't. If it is, increase the resistance and just wait. Once the electrolytic has reformed, remove the supply, and hit the discharge switch. Wait for the capacitor to discharge.

What sort of leakage can we expect from an old electrolytic?

Good question. It depends on the value of capacitance; bigger values leak more. And voltage rating; Higher voltages *tend* to leak more. This is where experience is key. If your 22uF, 400V capacitor is drawing 1mA, forget it. If it's drawing 400uA it may be OK. If it's drawing 200uA or less, put  it back in the kit and switch on!

Modern capacitors have very low leakage values. Not so with the ancient stuff I deal with, but even 60 year old electrolytics that haven't seen power in decades can be reformed successfully. It's not always the case though. Some are beyond redemption, and replacement is the only option.

I sometimes choose to monitor the voltage across the capacitor during reforming, but we need to be aware that our meter will have some resistance of it's own, and will effect the reading on the current meter (it will increase slightly) and will drop the voltage across our capacitor slightly (as it discharges through the resistance of our meter). A good quality digital meter will have an internal resistance of 10 megohms. A moving coil meter , like an AVO 8, will have a varying resistance of 20 Kilohms per volt. Keep this in mind.

I'm lucky enough to have a capacitance meter (for measuring the actual value in uF) and an E.S.R. (equivalent series resistance) meter, and I'm always amazed how some of these old devices shape up after a successful reforming.

Now, here's the warning bit again....

A variac is NOT isolated from the live or line of your mains. Use an isolation transformer for your safety. Reforming a high voltage (more than 50v) electrolytic capacitor poses a very real shock hazard. It may shock or kill you. 

Be safe. There's only one of you, and I need all the readers I can get!

Oh yes... the name....

It was a joke. It started out as the dreadnaught , got blown up, became the dreadnaught MKII, then after so many failures and improvements, I lost count...

Friday, 6 March 2015

"Stupid" - The simple smart charger for lead acid batteries.

Now my good friend "Curly" has a caravan. It's a great little thing, but has the most bizarre electrics I've ever seen in a caravan. It's got AC mains, and AC 12V.... AC???!!! When mains isn't available , I think the idea was a relay clicked over to supply the internal lights from a battery someplace, thus switching from AC to DC. This battery was long gone, and most of the wiring with it.

Matt wanted to add a leisure battery back in to the electrics so he could run his LED lights and telly etc, without having to have a mains hook-up, and when mains wasn't present to charge the battery.

I took a look at the bizarre circuitry, and rapidly worked out we were going to need some for of charge controller, or "smart" charger to do the job.

Then I thought again. We've already got a nice 10A 12V AC transformer, and an automotive relay in this bizarre set-up, so can we re-arrange this to provide us some sensible charging?

Measuring the off-load voltage of our transformer showed that, after rectification we were nearing 18V. Not good for trying to charge our leisure battery for any length of time, it will simply out gas and dry out. Nah.

Some form of charge controller? Yeah. A pulse-width modulated thing, with possible arduino control and a sensible FET was considered. Then I got to simplifying. What we needed was a window comparator.

Now a window comparator circuit usually uses two op-amps as comparators, one to turn the relay on to start charging our battery when the battery volts are low, and one to turn the relay off to stop charging our battery, once the battery voltage has risen to a charged state. Now there will be much debate about sensing battery temperature, and constant current, baulk and float charge amongst you at this point, but let's keep this simple.

So, I start thinking about our window comparator, on at around 11.5V and off by 14.3V ... seems reasonable, a voltage reference, some potential dividers, a dual op-amp, a driver transistor and we're away .... but is there a simpler way?

How about just using a potential divider on the base of our transistor to switch the relay, and not using any comparators... well, that would work, but the hysterisis between the two required voltages might not be good enough and cause the thing to oscillate. I try it out. It oscillates. OK, so how about simply two transistors and a couple of steering diodes? Yep, that works... but can it be simpler still?

Looking at the circuit for a while, how about using the spare contact on my relay to switch in the second pot, altering the value of my comparator? Yep!

It's ended up like this:-

Now, it's not the easiest of things to set up. I connected Pin 30 of the relay to pin 87A of the relay and powered that point with a variable power supply to first set the minimum voltage (11.5V) using R2, then the maximum voltage using R1. Vary the power supply to the unit to see at what voltage the relay clicks over. There is some interaction between the pots, so it will take a few goes at adjustment to get the unit to achieve repeatable results.


Yes, it's simple. There are issues. If something fails, there's precious little to stop it boiling our battery dry. The voltage control does seem stable in use, however. If the battery does fail open circuit, or F1 blows, the unit will oscillate... but at least the buzzing will draw your attention to the fact the battery has failed! The charging voltage does not compensate for changes in temperature.

Don't try and use this on a sealed lead acid battery. Chances are if something does go wrong, the battery will not be able to vent quick enough. It could burst. I've seen SLA's that look like footballs because they have been abused. Don't do it. 

When charging lead acid batteries, hydrogen gas is given off. Do not charge any lead acid battery in a confined space, or worse, a sealed box. No naked lights. I've seen tops blown off batteries as something has ignited escaping hydrogen. It's not good. 

Univox SR-55 Analogue Drum machine repairs. Korg Mini-pops!

My very good friend, and confirmed member of the Taffia, has passed this in my direction.

"Just sort of humms, can you take a look?"

... of course I can.

Good grief, what a rats nest! Not a single integrated circuit! Discrete transistor logic. Simple power supply. A real treat to fathom out!

 Connecting to the workshop amplifier and switching on reveals not so much of a hum, as a scream. A little reminiscent of 1980's Casiotone PLL chord generators, only very loud. Listening carefully I can hear some very faint drum sounds though the noise. Given the age of the equipment (possibly early 70's) I was hoping for a simple power supply repair, but no, those 40 year old electrolytic's test superbly.

On the top of the unit is also the sequencer. It's job is to provide pulses, once per beat. It's a discrete shift register. The clock speed (and thus, tempo) is controlled via a pot on the front panel.

Each  pulse is sent to the switch board, which decides which pulses should go to which "instrument" when. It's a huge wiring nightmare. The rhythms are hard wired in this switch bank.

So our pulse has left the sequencer, been routed through the switch bank, and on to the diode matrix. This is used to turn each timed pulse into the instrument we hear. Each diode passes the pulse to a whole bunch of near resonant filters on the rear board. Each filter is tuned to a different frequency, and two are noise generators. One instrument can be several filters being triggered at once, each triggered by the pulse sent via it's own diode. All the filters are mixed together, buffered and output via a volume control. There's also a tone control, which is simply an adjustable high-cut filter. There's a start button, which stops and resets the sequencer, or starts it running.

It's a simply stunning piece of equipment. With any sort of equipment like this, it's important to take your time and work through it. It's won't be complicated. It may look like it, but there's no odd ICs, no firmware and no pre-programmed bits and bobs only the manufacturer knows how they work!

So, what's wrong with this one? One of the filters seems to be resonating all the time, and howling. After some time poking around with a scope, it transpires that there's an inductor open circuit in one of the filters, and is causing it to oscillate. I have no idea what value the inductor is , so it's un-wound and re-wound with new wire of the same gauge... all 1,400 turns! Replacing it and a leaky 2SC828 transistor near by, followed by a quick set up produces some improvement. Some instruments in some rhythms sound distinctly odd though... One diode in the matrix is short circuit... replacement restores operation ...

Now... where's the Jean Michel Jarre LP???!