Upgrading generator control board

[Jester]

I'm attempting to repair and possibly upgrade the control board for a 7kW Onan generator.

The problem... The generator shuts down on over-voltage just as it starts to output voltage. I scoped the output and indeed the output very quickly climbs past 135V before being shut down.

The way it works (my best guess after poking around the board and capturing wave forms:

A dedicated winding on the generator (referred to as quadrature winding) drives a two SCR/two diode rectifier that is used to control the field. Best guess is that the uC monitors the output voltage and either stops or delays the turn on time of the gate pulses when the voltage is too high, and turns them on sooner in the cycle when the output voltage is too low.

I tried disconnecting the field drive from the controller and substituted in a DC supply and I can control the output voltage by varying the field voltage as expected.

Apparently these control boards have a history of failing, especially this SCR based portion of the circuit. I'm speculating it's the transients that do them in or or possible just degrade the breakdown voltage over time.

Both the SCR as well as the diode (2 of each) are D2PAK:

SCR is TN 1215800G           12A, 800V
Diode is  ISL9R18120S3      18A, 1200V

If I can replace these devices with similar parts (with same package) with better ratings it just might live longer next time. Any suggestions for upgraded SCR / diode

I beeped out the circuit and it appears to be as follows:

[Renate]

I have some (limited) experience with Onan generators.
The ones that I know have three windings on the stator: the main output, the ("quadrature") exciter and a pokey (1 Amp) battery charging winding.
The exciter winding goes to a full wave bridge and a big cap for ~150 VDC.
The field coil is PWM'ed in the kilohertz.
There is also "field coil flash" when it starts to throw 12 VDC to the field coil to get things started.

Yes, this is different than yours.

It's amazing how difficult they make it to get a grip on this stuff.
It's all proprietary, potted regulators.
They can't give you a general idea how the field coil is wired in.
They tell you polarity but not which side is switched.

I was looking at one point to have a µP controlled throttle.

A friend's generator fried out the quadrature winding from an internal short.
Sure, you can have it rewound for $500 or so.
I took it apart, ripped out the fried winding.
I ran an isolation transformer from the main output to the exciter input.

[moffy]

This might do for the SCR: TN1625-1000G :- 16A/1000V but 25mA trigger current instead of the 15mA for the original. You would need to check that it triggers correctly.
https://www.digikey.com.au/en/products/detail/stmicroelectronics/TN1625-1000G-TR/686419
It's D2PAK, but since they are surface mount with probably limited heatsinking it might be heat that is killing the devices not voltage or current.

P.S. If there are brushes involved in the field coil connection, check their condition and the slip ring in case they are inducing arcing.

[Jester]

I have some (limited) experience with Onan generators.
The ones that I know have three windings on the stator: the main output, the ("quadrature") exciter and a pokey (1 Amp) battery charging winding.
The exciter winding goes to a full wave bridge and a big cap for ~150 VDC.
The field coil is PWM'ed in the kilohertz.
There is also "field coil flash" when it starts to throw 12 VDC to the field coil to get things started.

Yes, this is different than yours.

It's amazing how difficult they make it to get a grip on this stuff.
It's all proprietary, potted regulators.
They can't give you a general idea how the field coil is wired in.
They tell you polarity but not which side is switched.

I was looking at one point to have a µP controlled throttle.

A friend's generator fried out the quadrature winding from an internal short.
Sure, you can have it rewound for $500 or so.
I took it apart, ripped out the fried winding.
I ran an isolation transformer from the main output to the exciter input.

Yes they have made an effort to make it difficult to repair, and they charge $2k for the replacement board (plenty in stock for quick delivery;-) )

DMM diode check of the SCR appears okay, however might be breaking down with higher voltage.

Your description matches how mine works, except no big capacitor.

I'm going to do another test and capture the field signal after the initial 12V blip.

Images of preliminary testing below.....

The last image is the interesting one. This test is with a heavy load and the generator NOT shutting down with error code 12 as it normally does with no or light load. I think this load is the sweet spot (enough to not trigger the over-voltage shut down but demonstrate the field drive problem. It looks like the field drive is only occurring during one half of the cycle (this makes sense the uC is trying to reduce the field drive current). However during the other half of the cycle it looks like the SCR is breaking down and turning on as soon as the voltage is too high. Looks to be about 400Vp-p!

If replacing the SCR's doesn't work, I'm half tempted to make a little circuit to monitor the output voltage and drive the field with a PWM based 60Vdc supply.

[Jester]

This might do for the SCR: TN1625-1000G :- 16A/1000V but 25mA trigger current instead of the 15mA for the original. You would need to check that it triggers correctly.
https://www.digikey.com.au/en/products/detail/stmicroelectronics/TN1625-1000G-TR/686419
It's D2PAK, but since they are surface mount with probably limited heatsinking it might be heat that is killing the devices not voltage or current.

P.S. If there are brushes involved in the field coil connection, check their condition and the slip ring in case they are inducing arcing.

There are brushes for the field. I cleaned them up with scotch-bright. I think they are okayish now.

Probably one of the following:
1) SCR is breaking down when voltage from quadrature winding builds up.
2) Brush arcing noise is getting back and false triggering the SCR's
3) Reference signal to uC is noisy and so uC is not acting as it should.

[Renate]

I don't see a free-wheeling diode in your diagram. I mean, there has to be one.

[Jester]

I don't see a free-wheeling diode in your diagram. I mean, there has to be one.

Yes there is (CR8), I added to my messy sketch.

Also updated post #3 with test results, the last one is interesting.

[ Specified attachment is not available ]

[Renate]

I'm getting all confused by your plots.
Is the field minus connected to ground?
There has to be a path for the 12 V "field flash".
There must be another diode or something.

Is the spikey last plot really showing only one SCR firing? At 60 Hz and not 120 Hz?

Can you start this under a decent load, then drop the load and still have it run?
Is this a case of dynamic response or that the field is always getting too much current under low demand.

[Jester]

I'm getting all confused by your plots.
Is the field minus connected to ground?
There has to be a path for the 12 V "field flash".
There must be another diode or something.

Is the spikey last plot really showing only one SCR firing? At 60 Hz and not 120 Hz?

Can you start this under a decent load, then drop the load and still have it run?
Is this a case of dynamic response or that the field is always getting too much current under low demand.

The field signal is the blue trace, the yellow signal is the generator output. first image external 12V, 2nd image external 36V (both with no load). final image control board driving the field with 20A load. Scope “Gnd” is referenced to negative side of field.

Spiky pulses are every 16ms

Generator will not start with no load or light load, it immediately shuts down with over voltage error and I have confirmed the output quickly rises to 135V before shutting down. If I start with the heavy load and then reduce it it shuts down with the same error.


I don’t have the schematic, so I don’t know what device is being used for the momentary 12V field flash but that aspect seems fine, I have captured that event and it looks normal.

Looks like the field is always excessive with light or no load.

[Renate]

Yeah, but it's 60 Hz full wave so it should be spikes ever 8 ms.
With a load is it doing that?

[Jester]

Yeah, but it's 60 Hz full wave so it should be spikes ever 8 ms.
With a load is it doing that?

I agree and that's exactly my point, it should be short duration pulses every 8 ms. My interpretation is that the controller knows the field voltage is too high so it's trying to turn both SCRs off, and it's succeeding with one of them, but not the other. The bad one is contributing too much (about the equivalent of what's required for a 20A load), that's why it runs when it's heavily loaded.

I know for no load:
 with 12 V field I get 42 V out
 with 36 V field I get 107 Vout
 so if the field voltage was about 40 V , I would expect to get 120 V out

the spiky waveform (400Vpp) is who knows what average but definitely more than 40V!

I have captured the output on start-up with no load and its >135Vrms and climbing when the controller shuts it down.

I'm thinking one of the SCRs is breaking down and conducting while the controller is driving the gate off.

I'm going to order the SCRs and swap them out (not easy), and see what happens.

The following waveform is the field from the controller with a failed start-up and shut-down on over voltage error code 12

It looks like once the quadrature voltage drops to about 150Vpp during the shut-down, that both SCRs start doing what they are supposed to do, short blips every 8ms