What is causing this voltage spike?

[BillTheBuilder]

I have a 2-stroke engine that has a charging coil that is powered from a flywheel attached to the engine crankshaft. The original design had a charge controller for an SLA battery. For weight saving, I am replacing the SLA battery with a LiFePO4 battery. For the charger, I opted for a solar battery charger as I could set the output voltage, and it supports the constant-current/constant-voltage charge profile the new battery needs.


Knowns:
-The charge coil outputs ~4 Voc when the engine is idle (2k rpm), 16-18 Voc when mid-throttle(5-6k rpm), and reaches around 27 Voc at max throttle (8k rpm)
-Isc is .43A at any rpm, which says to me that the coil output at max rpm should be around ~10w.
-The controller on the charger pcb is a CN3795. The spec sheet states the input voltage is max 30v, with the absolute max being 32v.
-The voltage output of the battery charger is set to 14.2v at charge termination.

I chose this controller as it is my (probably poor) understanding that an MPPT controller set to a certain voltage, here 18v, will "drag down" the input voltage close to the set MPPT voltage. In my mind, I was banking on this keeping the board from reaching an overvolt condition, pulling the voltage down from 27v max to the 18v needed. However, once I connect the pcb to the charging coil, I no longer read 4v at idle but instead I read 40v! At first, I thought it was due to my poor soldering job on the first board, but improved soldering on the second board showed the same 40v at Vin at engine idle. Both boards burned out shortly after. My guess is that having an MPP set point is causing this, and tonight I ordered this board which has the CN3765 controller and lacks the MPP set point. Is there something else that is causing the voltage to spike that high?

You can see the splatter on the pins that were exposed to the high voltage:

 

[Bluejets]

Best you provide a circuit diagram.
I suspect the previous original setup had some load attached and kept the voltage at a required level.
Not uncommon for open circuit coils such as this to go to rather high voltages unloaded.
Early ('60-70's) motorcyles had a rather large zener attached to avoid blowing headlamp bulbs.

 

[BillTheBuilder]

Best you provide a circuit diagram.
I suspect the previous original setup had some load attached and kept the voltage at a required level.
Not uncommon for open circuit coils such as this to go to rather high voltages unloaded.
Early ('60-70's) motorcyles had a rather large zener attached to avoid blowing headlamp bulbs.

Looking at the board, it seems the manufacturer simply followed the circuit diagram in the spec sheet. Pic attached.

The charger for the original SLA setup did warn to not connect it to the charging coil unless a battery was also attached. The only load is the engine starter. Once the battery was fully charged, wouldn't the input voltage then spike back up to 40v?

Putting a zener regulator across the input did cross my mind, but I was hoping there was another reason for the voltage spike so I could avoid the added labor. Plus, it feels wasteful to burn off that much power. Since the coil seems to output 10w max, would that mean I'd have to go for a 10w zener, or would a 5w zener suffice? I was considering a 1n5361 27v/5W zener coupled with a 27 ohm/1W resistor, but this was when I guessed a max of 32v coming from the coil, not 40v.

 

[BillTheBuilder]

Here's a pic of the original SLA charger. Black is ground. Red is to the battery. Blue is to the charge coil. All sealed in epoxy. Writing on the side says "128 35/21" Since I no longer need it, I'm considering tearing it open. What would work best?

 

[BillTheBuilder]

I took the plunge and tore apart the regulator meant for an SLA battery. I learned that it is simply three 1N5337B 4.7v 5W zener diodes in series, putting the breakdown voltage at ~14.1v. There is no resistor to limit the current, so I can see why this board would burn itself out were it not first attached to a battery before being connected to the coil. Unfortunately, this would seem to always push 14.1v to the battery, even if it was full, and eventually cause damage.

I'm figuring that an 82ohm 5W resistor and an 18v 5w zener should put enough load on the charging coil to where the voltage shouldn't swing as high as 40v with no load. Depending on how high the voltage swings, I may be able to lower the resistor value. My concern is the resistor absorbing too much power and not enough going to the charging board for the battery. Am I on the right path?

 

[Bluejets]

What happens if you put the original board back in circuit.
No idea why you seem to need to go to 18V.
The voltage you see is simply open circuit voltage and should not take much effort to get it under control.

 

[BillTheBuilder]

What happens if you put the original board back in circuit.
No idea why you seem to need to go to 18V.
The voltage you see is simply open circuit voltage and should not take much effort to get it under control.

When I put the original board back in, the voltage from the charging coil spikes up to 40v at idle. When the board is disconnected, my multimeter reads 4v coming from the charging coil at idle.
I selected 18v at first since the first board I found that supported charging a small lifepo4 battery was set to 18v. Since the board's max input voltage is 28v, I can instead pick a 27v zener. What would you suggest?

 

[Bluejets]

Sounds like the original zeners are kaput.
Your measurements appear to be incorrectly done.
Suggest replacing what was already there.

 

[BillTheBuilder]

Sounds like the original zeners are kaput.
Your measurements appear to be incorrectly done.
Suggest replacing what was already there.

I'm not sure how I'm measuring incorrectly. With no load connected and multimeter probes connected to ground and to the charging coil, I register 4v at engine idle. With the probes in the same location, I attached the battery charging board. The voltage spikes to 40v for about 5 seconds then drops back down to 4v after the board burns out. Does that sound correct? The zeners I pictured were what I was using the board to replace.

 

[Bluejets]

You don't say exactly how you come to measure these voltages.
That was my point.
So when you arrive with unexpected voltage readings, one simply assumes not done correctly.