Capacitor excited generator

[BurningTantalum]

I have a 2.5kVA capacitor excited petrol generator. It failed whilst in service powering a toaster during a power cut several years ago. I could not find the fault at the time, and was about to toss it out this week when I had a final attempt to find why there was no output.
I am not totally clear how this type works, and there seems to be a lot of confusion on the net about it.
There are two stator windings, in series for 240V output (parallel for 110V is optional via switch due to it being a UK building site unit.)
There is a 3rd stator winding with nothing but a 13uF cap across it.
The 2 pole rotor has a winding on each pole, commoned at one end and the other ends shorted with 3 series diodes each with a 20k resistor across.
Each rotor pole has a permanent magnet embedded near to one end.
I replaced the cap but it made no difference. The original tested as 10uF.
The stator windings are both identical at around 1.2R.
The cap winding is around 3R5.
All winding are clear of the frame on a 240V test.
The rotor windings both measure 12R and are approx 0.7mm dia (can't measure any wire easily without stripping the gen again.)
According to tables this would give about 300 metres of wire on each pole which seems a lot.
The diodes and resistors were originally found on a small pcb attached to the nylon coil former of the rotor. Some solder joints looked dodgy so I made a new pcb with new components.
Does anyone have any ideas, suggestions or experience of this type of unit?

[floobydust]

I see the fractured solder joints on the pictured diode board. Last time wasn't it a defective diode? https://www.eevblog.com/forum/repair/old-honda-generator/
I would repeat the rotor diode test and confirm they are working. The flux will not build up if there is an open-circuit with them.

[BurningTantalum]

Hi Flooby, Sorry I completely forgot that I had posted before, and it seems I have repeated the post from 18 months ago! The pictures are from the original paxolin pcb which I removed and replaced with a grp board and new diodes and resistors (which I didn't take a pic of.)
I fitted the new pcb and the genny still didn't work so I didn't post anything else on the subject, until I was about to toss it out this week.
I have just fired up the motor and measured around 1.5VAC across the capacitor winding with cap removed, and around 6.5VAC across the output of the stator winding(s). The RPM is correct at 3000 with a tacho.
I will remove the stator tomorrow to look where the cap winding actually sits, to clarify for myself how this works. The wire is a different size to the main winding so should be obvious, and when I had it dismantled I didn't think to look if it went down the slots or elsewhere.
I understand that generators can be troublesome if left standing for long periods and the residual magnetism decays, but this example failed in use.
BT

[floobydust]

I pick up old repair projects and shelve them all the time, when they are no fun. Then I forget and start over lol. I've learned to keep repair notes on my PC to not forget where I left off sometimes. Can't say it helps lol.

There's not a lot of parts to test - the diode board, flux capacitor aux, stator and rotor windings.
The capacitor and aux winding I believe make the system (stator) ferroresonant, that is the stator magnetic field once built up runs saturated which gives the voltage regulation.

With such low output voltages, you could try remagnetize the rotor- but that doesn't explain how it lost it or it was a defective diode board.

I'd test the windings looking for a ground fault or shorted turn. The inductance and resistance I guess we have no specs ?
Then I'd use a Ring Tester or the like to see if there is a shorted turn.

[BurningTantalum]

I'm glad I'm not alone!
My knowledge of rotating machines isn't too good, and only touched on with DC machines at college in an otherwise electronics course.
I have tested all of the windings with an insulation tester at 240V and they are infinite to the laminations.
They all look fine with no distress at all.
I double checked that I installed the diodes the correct way round when I made the grp pcb, which was some time after you spotted the cracked diode 18 months ago.
The inset magnets in the laminations of the rotor seem 'strong' and are N & S opposite.
I have just dismantled the thing again and noted that there are 30 slots. The main windings go through adjacent 5 slots, then the next 5 have the cap winding, then 5 with the main etc.
I have not heard of a Ring Tester, but assume it is a sort of 'Growler' for stators. Ah, I see it is not 'ring' as in circular stator but 'ring' as in resonance.
I will pause posts while I 'edificate' myself on testing windings.
At this point I am more interested in why it failed than getting it in service again.
Thanks for you interest, Flooby. I'll probably post again in 2 years when I have forgotten all of this.
BT

[floobydust]

You need this fixed for the upcoming zombie attack 
I think you can't put the rotor on a growler with the diode board connected, and it also would likely degauss the magnets. Nor the stator with the flux capacitor connected.
The original Blue Ring Tester design was by Bob Parker and offered at Dick Smith many years ago.
They seem to be an expensive bit of kit and I couldn't see any for sale at Altronics or Jaycar in its home Australia. Alltronics California seem to be making then over here. There's some on eBay as well.
Or if you can find someone with one in a repair shop or something.

[james_s]

I built one of those Dick Smith ring testers years ago from a kit. It's a really simple circuit, seems like a friend of mine built a copy of it from the schematic. I can take a picture of the PCB if the schematic is not still out there.

[floobydust]

Bob Parker 2nd Generation build



DiodeGoneWild Simple build


DiodeGoneWild 2nd generation with added comparator and for in-circuit


The schematics are out there, for the Bob Parker Blue and Blue2.
DiodeGoneWild also did two designs- for in-circuit and out of circuit testing (5V pulse can be too much for an IC). His first is build is quite simple.

I'm not sure if a ring tester will work with big iron and many turns?
I just tried a 120V/12V 12VA transformer and (my own homebrew) tester on the primary (0.1uF cap) it's about 10 rings at 485Hz (1.08H). Shorted secondary gives 9 rings at 1.7kHz (81.5mH), so there is a pass by that measure. Bob mentioned small iron transformers can be a problem.  I'll have to try test a car alternator stator/rotor sometime.

I remember as a kid going to a rewinding shop (guy worked there) and he was testing a large 3-phase stator, maybe 1m diameter.
He connected excitation and then tossed in a ball bearing. Sure enough it clung to the inside of the stator and started circling around. Any shorts and the ball bearing apparently does something odd.
These old tricks with big magnetics would help, but I haven't seen any for ferroresonant gear with a self-magnetizing rotor.
edit: added Bob's video

[BurningTantalum]

I miss-transcribed the detail of the windings -
The cap winding is in 5 adjacent slots and the 5 diametrically opposite. The main winding(s) are in 10 adjacent slots and diametrically opposite.
Thinking of the rotating ball bearing, and using a Growler on an armature, I supplied a low voltage AC to the capacitor winding ends and a steel rod rattled against each pole piece and each slot of the winding. I tried the same with each of the main windings and got some adjacent pole pieces that were 'dead', but not opposite each other.
I can conclude from this that something is amiss with the main windings that a simple low R test does not show up.
Enthusiasm has drained to zero so the whole caboodle is on the trailer awaitng transport to the dump. I may remove the GX160 as the type is one of the best I/C motors devised by humankind. They can be pull started by a toddler.
Thanks for the interesting links.
BT

[james_s]

You can probably sell it to someone else more motivated to repair. There's a guy named James Condon on youtube that buys broken generators and repairs them, there are probably some similar types in your part of the world.

[TheMG]

I doubt it would be anything to do with the main winding. If the winding was open circuit it would be obvious on the multimeter, if it was shorted it would be making the engine work really hard (and very likely cook the winding too).

I would still suspect the rotor. Could the diodes be breaking down at a higher voltage? Maybe there's an intermittent connection, which reads fine when not moving, but the centrifugal force when it's turning is causing it to go open circuit?

Not sure what effect (if any) having the diode board reversed would have, but might be worth investigating.

[james_s]

Yeah I'm also skeptical of there being a problem with the winding. If they're not leaking to the stator form or to each other then there isn't much else to go wrong with that. The only problem I've personally encountered with one of those generators is a bad capacitor.

[elecdonia]

Have you checked the capacitor for both uF and ESR?

Also many rotors have a couple of small permanent magnets embedded into their pole pieces for initial excitation. But not all of them.

Without any permanent magnets the residual magnetism could decay, especially if a diode had previously shorted out. I think the rotor could be remagnetised by disconnecting one end of the rotor winding from the diodes and then apply DC current across the two disconnected terminals (1 is coil wire, 2 is the terminal of the disconnected diode assembly)  with the DC +/- polarity which causes the diodes to conduct. The other end of the rotor coil remains connected to the diodes. The goal is to cause current to flow through both the rotor winding and the diodes in the same direction as it would flow in a functional generator. I’m thinking a current of about 10 amps for a few seconds would be sufficient.

In the case of a rotor with embedded permanent magnets then the polarity of the diodes must permit current to flow through the rotor coil in the proper direction which causes magnetism to increase in the same north/south direction which matches the north/south poles of the embedded permanent magnets. Otherwise the rotor coil will buck the north/south magnetic field from the permanent magnets. In other words we need “positive feedback” to cause the current flowing through the rotor winding to boost the residual magnetism. To identify north/south magnetic polarity you could use a traditional hiking compass. If you suspect the new diode PC boards might not be connected to the rotor winding with the same polarity as the original diodes, then this step is really important.

Disclaimer: I’ve only personally repaired one “brushless” generator. It had other issues, but no problems with rotor, diodes, capacitor, or stator. I was surprised by how well it worked after I fixed its other faults.

[elecdonia]

After re-reading this entire topic a couple of times I saw these items:

-    The rotor has permanent magnets embedded into the pole pieces
-    DC resistance of windings on both rotor and stator are OK
-    Rotor PC board contains 3 diodes connected in series
-    The initial failure (loss of output power under load) was likely caused by solder connections on the old
           diode PC board going open circuit while the generator was running

Based on this I’ll revise my suggestions:

1) Disconnect one end of the rotor winding from the diode PC board
2) Use a multimeter in “diode check” mode to confirm that all 3 diodes are connected for same anode/cathode polarity. In the forward direction (conducting) the total voltage drop across 3 diodes should be about .7V x 3 = 2.1V. The other +/- direction should have a resistance equal to 3x the value of each resistor
3) Identify the north/south poles of the permanent magnets
4) Use a variable DC bench power supply to apply a small amount of current to the rotor (< 1 amp). This current must flow through the rotor winding and through the diodes
5) There should be current flow for only one polarity of applied DC voltage. Select the polarity for which there is current flow
6) Gradually increase the rotor current from 0 up to perhaps 1 to 5 amps. The magnetism strength at the rotor pole pieces should gradually increase with current
7) The north/south polarity of the magnetic field must not reverse as the current increases from 0 amps. The north/south orientation of the magnetism must stay the same as the north/south orientation of the permanent magnets
8 ) If applying current causes the north/south magnetic poles to reverse this indicates the diode PC board is connected to rotor winding with wrong +/- polarity. Therefore you must reverse (swap) the connection of the 2 wires from the rotor winding to the diode PC board


Summary: If the diodes are connected properly the magnetic field strength of the rotor will steadily increase with the amount of applied DC current and will always have the same north/south orientation.

[BurningTantalum]

This morning I knocked the rotor off the Honda shaft and put the remains on the trailer. Now I am all fired up again! The diodes are the correct way round - I removed the pcb to check a few days ago. I have photos of the old paxolin pcb and it matches (worth a check of course, as per good fault finding technique.) The diodes fitted by me were 1N5403 brand new old stock, as were the 20k resistors. I did think of passing a current through the rotor winding and checking magnetising effect... Oh, all right I will have a look. The permanent magnets obviously haven't moved and are opposite in sense, testing with a bar magnet from a kid's construction toy.
All the capacitors that I have tried checked as OK uF and ESR OK
Bearing in mind it did fail in service at well below its rated output, and there appeared to be damage on the diode pcb (but this may have been caused by the difficulty in removing it as it was gummed in its slot with varnish) Occam would suggest the rotor as a cause... but there is definitely something not right with the main windings.
There is no chance of finding spare parts here in Australia as this was a UK building site genny with switchable 240v/110v outputs and sockets as per regs.
BT

[floobydust]

I thought the problem might be the rotor windings or the new diode board has an issue. The vibration, heat, PIV is high stress for the diodes which is typically what I've replaced.
edit: 1N5403 is an oddball 300V part and possibly too low voltage. I use 600-1,000 PIV.

Big generator rotor tests are mostly what we know but there is the RSO (Recurrent Surge Oscillograph) Test. It seems to be TDR fed into each winding end and comparing the reflections. I don't think the windings are wound symmetrical enough for that technique? and there is mention some shorts can appear only when the rotor is spinning or at operating voltage/temperature. A bit more info for big stuff: https://www.sidewindersllc.com/generator-rotor-testing/
I think my idea for using a Ring Tester has troubles if it's a short between many winding turns, not a single turn short. The Q can still be up there.

The stator magnetic field having some dead spots could be normal if there is a magnetic shunt. But I have never looked that hard at their construction. I guess it's a guessing game - replace armature, then the stator.

I would be tempted to disconnect the diodes and inject AC voltage, say 24VAC to each rotor winding (on the bench) and see what the current is like. But at some point it would weaken the magnets. It should work as an electromagnet?
Same for the stator - run it backwards and inject AC voltage say 24VAC at its output, see what the resulting current is like and flux capacitor voltage too.

I keep wondering if your original diode failure didn't wreck something else, or polarity got flipped somewhere. Maybe injecting DC to the rotor windings could verify that the (magnetic) polarity is correct wrt windings and magnets. The connection of the windings to the diode board might be reversed.

[BurningTantalum]

OK, what I did before I read Flooby's last post:
Aligned the laminated poles E - W on the bench.
I took a hiking compass and found which perm magnet it was attracted to. Marked this N.
Took my kid's bar magnet and found which end the compass was attracted to. Marked this N.
Confirmed that the gen magnet and the bar magnet repelled each other.
Checked the compass was not attracted to the laminations (the perm magnets are towards one end.)
Connected a variable current limited PSU to the windings - +ve to the cathode end of the diode board, -ve to the anode end.
The diodes would be reverse biased but the current was flowing in the same direction as if they were conducting, but was flowing through the PSU instead.
At around 100mA I checked with the compass and it was strongly attracted to the laminations on the pole with the N magnet.
This seems to confirm that the connections and sense are correct (?)
BT