Dodgy Condensers -Citroen GS

schlitzaugen said:
Well, he's using this microphone:


You can of course replace the points with a Hall effect trigger and get rid of both the points and condenser. Relatively cheap, easy to install, set and forget. Takes out the wear in the dizzy shaft square sector too, but not the wear in the bushings.
Click to expand...
Thankyou everyone for the response - particularly Schlitzaugen. I hadn't through about the details of capacitors since I was playing with electronics and old radios when I was at school, so really appreciated the detailed explanations. :cheers:

Curiosity got the better of me, so I pulled apart a NOS .27uf ducellier that failed very soon after fitting. This is what I found:
1649564708752.png

And from the other side:
1649565011603.png

It is very obvious what has happened here - everything except the copper spacer between the can and the bottom end of the foil core is shiny and covered in oil. The copper spacer on the can side was dry and had a think layer of corrosion on it.

Next step was to test the capacitor it's self - a healthy .31uf
1649565208951.png

So out with the autosol and off with the corrosion:
1649565287631.png

Back together with smear of silicon grease on each side of the copper plate - tested again - ta da! back to a healthy 0.31uf inside the can.
1649565390628.png


So what happened here? It looks like the copper plate was entirely dry on the outside allowing a layer of oxide to form. The rubber top was quite hard, but I doubt this was a factor, as the copper plate is slightly domed, presumably to work as a spring.

Who knows how long this will last - I am nervous about was whether there is any risk of chemical reaction between the silicon grease and the oil or aluminum, simply because I have no idea about the chemistry. I guess only time will tell.

On thing that did strike me was how good this design is from a vibration perspective - no hard connections and everything held together and in place with flexible materials (rubber, paper, copper sheet). Fingers crossed this one will now be good for many years.
 
Are you sure the 'doming' isn't swelling? The corrosion build up on the inside of electrolytics is commonly checked by looking for such swelling.
 
The Gonz said:
Are you sure the 'doming' isn't swelling? The corrosion build up on the inside of electrolytics is commonly checked by looking for such swelling.
Click to expand...
yep, the corrosion was minimal - 5min of hand rubbing with autosol and it polished off. I didn't unwrap the core of course, but I suspect it is a plastic film type. Definitely not electrolytic - you can see the aluminum foil at both ends.

Got fond memories of electrolyics flying across the room in high school electronics after being connected reverse polarity (by others, of course!)
 
andrewj said:
yep, the corrosion was minimal - 5min of hand rubbing with autosol and it polished off. I didn't unwrap the core of course, but I suspect it is a plastic film type. Definitely not electrolytic - you can see the aluminum foil at both ends.

Got fond memories of electrolyics flying across the room in high school electronics after being connected reverse polarity (by others, of course!)
Click to expand...
Yep, reverse V cap fireworks in the lab, a favourite pastime. 😂
 
Or, at a certain tertiary institution, a small voltage electro cap soldered on to the end of a 240v cord, hung out the window down from the second floor, and the switch flicked on. Wasn't me - I was an observer.
 
Ahem.

No.

Your capacitor shows typical signs of internal DC leakage. It is a paper in oil type and the oil infused paper is the dielectric. What happened was the oil leaked out of the paper and this changed the capacitor value. It reads a "healthy" 31nF because of the way your meter measures capacitance. It applies a small voltage and it counts how long it takes for the capacitor to fully charge. Because the capacitor leaks some of that voltage across the dielectric, it takes longer to charge, which makes the meter think it is a higher capacitance. I can show you boxes of old capacitors I pulled out of old valve gear that do exactly that. They read slightly higher for value, but with a leakage tester they don't even hold charge at minimal voltage. They are basically resistors by now. So is yours. Put it back in your car and you'll see.

Have a watch of those videos and you'll see the device you need to check for leakage.

Corrosion is also a bad sign. It means the capacitor isn't air tight anymore and water can get in. I guess the leaking oil has damaged the rubber and leaked out and then water got in.
 
Caps are not a serviceable part. They dry out, become electrically ( and otherwise) leaky and are difficult to test properly without the right gear. Unless things have changed, they are unrepairable.
 
Ok, curiosity got the better of me, so I took some more measurements before trying on the car (it has a SEV distributor, so swapping the condenser is not straight forward).

So:
Checking resistance and off the 2Mohm scale.
Applying and removing 12V with the little repco digital multimeter, so the capacitor discharges through the volt meter and any internal leakage resistance:
Disconnection without a condenser - 12-0v close enough to instantaneous. So no lag in the meter.
A "thought to be good condenser" gives 12-1v decay in approx 1 sec. The "repaired" one gives about 2 sec.

By my calculations, 1000rmp x 4 cylinders = 66.7 sparks per second = 15mS, or 150mS @100rpm. So even if the voltmeter is very high resistance, it suggests the decay time should not be a big issue.

Is it possible that the resistance of the condenser drops with higher instantaneous voltages or high dv/dt on the actual circuit? I guess the only way to know for sure is to test in the car.
 
Have look at this picture:


What you have measured with the meter on the resistor scale is meaningless.

A leakage capacitor tester applies a voltage on the capacitor to fully charge it and measures the leakage current at the same time. Given the current can be in the microampere range, these devices are very sensitive. Mine for instance is a valve unit and it has a magic eye that shows as the capacitor is trying to charge and getting there or not.

The magic eye is supposed to close relatively quickly and stay closed. A good capacitor is charging so fast, you don't even see when the eye has closed and it stays firmly closed.

A bad capacitor will struggle to get the eye closed and you can see the sectors moving in and out without ever closing. A really bad capacitor will have the eye open solid.

As you go up in voltage you can see a bad capacitor shows the eye closing slower and slower until at some voltage it will just struggle and not close at all.

The voltage is switched manually so you can see what happens at each step.

The capacitor dielectric breaks down with voltage rise.

Your capacitor might be okay-ish at 12V but the voltage spikes across the points are going to be wildly higher. I mean hundreds of volts if not thousands. You capacitor will basically be a shunt on the points hence you will have poor or no spark at all because the current in the coil primary will never go down to zero and then rise again to create the inductive magnetic field that generates the secondary magnetic field you need to fire the plugs.

If you wanted to run a better (not good enough but will give you an idea) test of your capacitor I suggest you connect your meter across the capacitor on voltage scale and apply a voltage across, then short the capacitor, and repeat if possible with a frequency close to that in real life you have calculated. To see what happens an analog meter would be best, but the inertia of the pointer will be an impediment. An oscilloscope would be better. My money sez you will see a nearly constant voltage (okay, you might see some mild ripple) because your capacitor won't charge and discharge as it is supposed to do, which would show like a lot of spikes.

Curious to hear the results of your tests.
 
Hi Schlitzaugen

I had no idea that magic eyes are still being manufactured and used. I never cease to be amazed with what was achieved with valve technology which was nothing more than carefully combined metal and glass.

So if I get what you are saying, as the breakdown of the dielectric is a function of V and dv/dt, these capacitors start to act as much as a voltage regulator than a capacitor at higher frequencies. Again, out of curiosity, I get the dielectric breakdown being a function of voltage, with migration of ions being depending on the strength of the field. But what about dv/dt – do you know what the mechanism there is?

I don’t have any real electronic gear -i.e. frequency generator or oscilloscope, but I am thinking I could mount a distributor in a lathe with a digital speed read out, set up a coil and see how the distance the HV spark will jump varies with speed. Have also got a big fat AVO meter with massive inertia and damping. This should give a nice average voltage reading across the capacitor.

Just had a 4 post hoist arrive, which will distract me for a week or so, so will do some more experimentation after that.

Cheers,
Andrew
 
Massive inertia in the pointer needle is not what you want, damping is what you want, but the two are tied. That was just posted as a thought experiment more than anything. The test in real life won't really work. An oscilloscope is what you want.

But.

Your idea with the dizzy in the lathe will work. I have done that as well to test an electronic ignition I built. I actually rigged an aluminium bar with spark plugs in it. I think it is a bit too much trouble though, especially since you will have to hang onto the dizzy body for dear life if you don't find a way to clamp it somehow and then you need to perfectly center it, etc. Much easier to just stick that capacitor on your car and see what happens.

And there is another reason. In the car, driving hard uphill you will place a load on all the components in the ignition system and they have to withstand that for your car to work properly. That ain't gonna happen in your lathe. Your test in the lathe is equivalent to revving the car in your driveway.

As for the dielectric, it is both voltage and frequency that kills them (apart from old age). Modern electronics don't need high voltage capacitors save for smps (and more recently variable frequency drives) and even then, you won't see capacitors rated at over 450V or so (you might see in VFDs). These are the capacitors that work hard their entire life and if you look online at some serious manufacturers you will see they have entire lines of capacitors designed to take high ripple and high voltage. Back in valve times, these capacitors were rather poor, paper in oil or wax and such. The most spectacular failure I have seen was on my own lathe where the starting capacitor literally pushed its guts out like a snail's eyes. The lathe dates back to the eighties, and I suspect this was the original capacitor. Not a bad run, but it shows the weak spot of these older designs.

The development and spread of SMPS however has reignited the interest in designing good high voltage capacitors (and high temperature) and we do have some pretty good stuff from the usual suspects. But even these have a rather low life expectancy under harsh conditions. Have a look and you will see the life expectancy at 85 deg is say 10000 hours whilst at 105deg is 2000. That's what high ripple and high voltage will do because the capacitor has to charge and discharge many times a second at whatever voltage it has to work. This creates heat. The higher the frequency, the more heat. The higher the voltage swing, the more heat. And it builds up.

Low voltage capacitors (say 50V and below) don't work as hard and they usually die of old age rather than work fatigue.

I am not sure magic eyes are still being manufactured (I think some are) but mine is a NOS I found rather cheap right here in Oz at some chap. They are not that rare, but mostly come from overseas. The advantage of valves is twofold. One is they are very sensitive and two they are very robust. Mosfets are currently giving them a run for their money but MOSFETS are hard to drive in every application and have some capacitive impedance in the junctions so you have to take some care in the topology design. Valves are fine as long as you don't expose them to high DC on the grid. Which is what happens when coupling capacitors start leaking DC and turn the driven valve up too high, eventually burning out either the filament, the cathode connector wire or both. Or if these don't go, you might take out the power transformer or the output (speaker) transformer and then it's game over. That is literally the only Achilles' heel of valves. Take care of the capacitors and your valve gear will probably outlive everybody (which most of it already did).

Back to your car and its capacitor, what temperatures do you expect under the hood? Pretty high, and that is before we factor in the capacitor's own heating due to its function. Different perspective, right? And I would put money your capacitor is not a high quality one like the ones I mentioned above with a rated life expectancy of 10000 hours at "normal" temperatures. I would say even if works you might want to have a spare on hand especially on a long trip.

One last very important thing about capacitors. Old capacitors were much more tolerant to overvoltage, probably having been designed for a time when even the grid was just approximately 240 (or whatever) volts. Not so with modern capacitors. You exceed their rated voltage, they go poof pretty much instantly. You have been warned.
 
"The advantage of valves is twofold. One is they are very sensitive and two they are very robust. Mosfets are currently giving them a run for their money but MOSFETS are hard to drive in every application and have some capacitive impedance in the junctions so you have to take some care in the topology design. Valves are fine as long as you don't expose them to high DC on the grid. Which is what happens when coupling capacitors start leaking DC and turn the driven valve up too high, eventually burning out either the filament, the cathode connector wire or both. Or if these don't go, you might take out the power transformer or the output (speaker) transformer and then it's game over. That is literally the only Achilles' heel of valves."

Just relating an old comment from a cousin of mine who worked as an electronic technician at the local ABC transmitter station years ago, about the time when they changed from valve technology to solid state. He said he hated storms and had to standby on hand to fix the ouput stages when the aerial had strikes, it being the highest lightning arrestor around. :rolleyes:

The 'new' solid state units were just replaced when they had a hit, in a hurry. But in his experience the valve units often would just glow a bit all over and then keep going !! Much more reliable in that situation of a bit of electricity floating around upstairs.:cool:
Jaahn
.
 
Ok, being a grey and rainy day today I decided to do some "science" with a selection of GS condensers, and a modern film capacitors from the local electronics shop.

The lathe gives smooth control from around 100rpm to 3000 rpm - not the full range of the engine, but enough to see what's going on.

Here is the setup -
The can with the bare wire is the condenser I tried repairing, while the other cans are a "known good" SEV unit out of the Orange Peril, a random NOS one, and the new yellow film capacitors are in series as they where a little low on voltage withstand.

At the bottom of the video you can see my very sophisticated high voltage meter - a plug lead positioned near the negative plug and held in place by some crimpers.

To cut a long story short, all the capacitors gave a reliable arc at a 4mm gap at from 100rpm to 1000rpm. From 1000rpm onward, this could be increased to around 7mm.

I could not discern any difference between all of the condensers tested, which suggests the repair on the original can has indeed been successful - at least at low temperatures. Any leakage should have manifested as reduced performance at low revs, but this was not observable.

I also tested mounting the capacitor on the distributor, rather than remotely on the coil to see if inductance of leads made a difference, and no, there did not seem to be any change in the distance required for a reliable arc. This is consistent with experience on the last raid, where new condensers mounted near the coil could often get a dead 2cv going in a matter of minutes.

Next experiment will be a long term one (hopefully) - mount an 80cent 630VDC polyester capacitor near the coil and see how long it lasts.

Cheers,
Andrew
 
andrewj said:
Ok, being a grey and rainy day today I decided to do some "science" with a selection of GS condensers, and a modern film capacitors from the local electronics shop.

The lathe gives smooth control from around 100rpm to 3000 rpm - not the full range of the engine, but enough to see what's going on.

Here is the setup -
The can with the bare wire is the condenser I tried repairing, while the other cans are a "known good" SEV unit out of the Orange Peril, a random NOS one, and the new yellow film capacitors are in series as they where a little low on voltage withstand.

At the bottom of the video you can see my very sophisticated high voltage meter - a plug lead positioned near the negative plug and held in place by some crimpers.

To cut a long story short, all the capacitors gave a reliable arc at a 4mm gap at from 100rpm to 1000rpm. From 1000rpm onward, this could be increased to around 7mm.

I could not discern any difference between all of the condensers tested, which suggests the repair on the original can has indeed been successful - at least at low temperatures. Any leakage should have manifested as reduced performance at low revs, but this was not observable.

I also tested mounting the capacitor on the distributor, rather than remotely on the coil to see if inductance of leads made a difference, and no, there did not seem to be any change in the distance required for a reliable arc. This is consistent with experience on the last raid, where new condensers mounted near the coil could often get a dead 2cv going in a matter of minutes.

Next experiment will be a long term one (hopefully) - mount an 80cent 630VDC polyester capacitor near the coil and see how long it lasts.

Cheers,
Andrew
Click to expand...
 
The distributor only turns at 1/2 crank speed on the engine, so wouldn't your 3K rpm on the lathe be the same as 6K rpm on the engine?
 
  • Like
Reactions: cjl
Incidentally I found an old ignition capacitor in one of my boxes of crap. Don't know anything about it (I suspect it served on a R12 in the distant past), but I said, hey let's test it. So I whacked it on the capacitor tester and it doesn't hold a charge even at 20V.

Maybe I should mail it to you and see what you can do with it.

But like I said the ultimate test is what these do on the engine under load and at temperature.
 
Hi Greenpeace,

The GS this condenser came from is a 1220, which is marked red from 6500rpm on the tacho in the car - GS's rev very nicely :cool:

Hi Schlitzaugen,

I doubt the technique I used would work on a leaky capacitor - it would seem the problem with the one I fixed was an open circuit, due to internal corrosion on the copper between the capacitor and the can, as opposed to degraded dielectric.

As you suggest, the ultimate test will be in the car.

Cheers,
Andrew
 
andrewj said:
Hi Greenpeace,

The GS this condenser came from is a 1220, which is marked red from 6500rpm on the tacho in the car - GS's rev very nicely :cool:
Click to expand...
Yes I owned a 1220 a number of years ago. But I guess unless it's a race car, testing the spark to 6K should cover 99.9% of driving?
 
Greenpeace said:
Yes I owned a 1220 a number of years ago. But I guess unless it's a race car, testing the spark to 6K should cover 99.9% of driving?
Click to expand...
Indeed, and more to the point, this particular experiment suggests that performance of the capacitors at low rpm is more interesting than at high rpm.
 
Just ran across this & 22/27 microfarads is about right. 6,12 or 24V doesn't matter. Generally NOS are fine it is the Chinese & Indian repros that are next to useless some bad from new!

We had same issues with WW2 Jeeps
& found VW BEETLE Bosch [being the operative word] ones worked & fitted fine, with no failures
 
You must log in or register to reply here.
Top