Solar direct drive kayak

Hey Will,

What a shame!
I wish I could help you more.
I know enough to understand what you're talking about but not to help you.

So, what's next?

 

Supercaps! That is what's next!
I think I have figured out why the little buck failed.
Yesterday I had the motor back on the test bench with the wiring the same as on race day.
The only difference was that I now had a stack of supercapacitors and a 3s Lipo connected in parallel to the MPPT battery terminals. It was late afternoon and overcast, so the panel only generated around 20W.
Once I had the motor running I disconnected the Lipo to measure the reserve endurance of the supercaps. I noticed a significant lag in the MPPT's response to a sudden voltage drop over the caps. On race day the buck would have been much quicker to respond and was therefore left to supply the full current demanded by the motor.

These are 4F caps rated at 5.5V. Stacking them 2s4p gives an equivalent of 8F, 11V. Not much in terms of reserve power but enough to keep the motor going when passing under a bridge so that the prop does not become a drag brake.

 

Will, your problem is the MPPT. Every time you use the ESP the MPPT will see another battery capacity/SOC and adjust acordingly. You are basicly confusing the MPPT. That happens because your current ESP is not a PWM that rapidly shuts power on and off but a VFD (you switched motors). The elegant solution is to find a ESP with variable voltage input and drop the MPPT. The brutal solution is to replace the MPPT with a dumb buck-boost like in your previous setup. As I said you don't actually need the MPPT's functions because you are not chargeing batteries.
Search for a ESP that covers your panels output and motor input voltage range, that will eliminate one conversion step.

 

You would think that some supercaps will provide enough of a reference voltage for the MPPT to use. The MPPT supplier did mention that the Victron controllers are way faster in how they respond to changing loads. Even with the LiPo battery attached, the controller would take its time before it starts to help out with the load.

I just did a number of different tests with the supercaps. The big voltage drop seems to only occur at start-up. It would fall to 6V and the motor would struggle along at low rpm. Then at some point the MPPT would wake up and start charging, very slowly though, 0.2A at first. Rpm would gradually climb with a lot of 'hunting'. Once the caps approach the charge limit (set at 11V), I can start applying some load to the motor and the MPPT will do a good job of maintaining the 11V at the output. This is with the ESC set at full power from the start.
I also used a 20A buck to get the voltage within safe limits for the ESC. Instant power and no voltage drop at the output despite not using any supercaps. I definitely need to do some proper sea-trials with this setup.
I had the buck set at 25V out, gave the ESC full throttle and applied enough load to draw >200W from the panel. This got it down to Vmpp of 31V. Any more load and the panel voltage would drop right down to 12V but recover instantly as soon as I ease the load.

 

More tests and more tests, always varying some variable until a pattern starts to emerge.
For reasons yet unknown to me, there is a marked difference in the MPPT's response time depending on voltage difference between panel and set output.
I had re-arranged the supercaps 3s3p, so that I can charge up to 16.5V instead of just 11V. I had the 3s Lipo still connected in parallel but with a diode so that it can only deliver current if voltage dips below 11V. It obviously could not take on any charge.
After initial connection the caps charged up to 16.5 and if I gave power to the motor with the MPPT still charging the caps, there would be a voltage dip down to 11V, the battery would help out for a short while until the MPPT starts to deliver enough current to get the caps back to 16.5V. At that stage I could apply varying loads and vary the ESC speed without any issues. The moment I stopped the motor the caps would fall down to 11V and no charge would appear from the MPPT. Starting up the motor from there relied entirely on the battery.

I then switched to a single 18V 50W panel, specifically to see how the MPPT deals with loads that demand more than the panel's Impp.
Same setup with the caps and battery as before, good initial charge but no voltage drop as the motor started. And no loss of charge on the caps after the motor stopped. Ditto with the battery removed, supercaps worked well on their own. Power on, power off, load or no load, the caps stayed at the set charge limit except when the load got too big and the MPPT had to start reducing output voltage in order to boost the current. As expected. With load eased, the voltage would recover immediately.
My first thought was that the smart panel might be delivering charge in pulse form and that this confused the controller.

The next test was to put two 50W panels in series, with a Voc of 42V. No charging, exactly the same problem as with the smart panel (Voc = 35V).

So if it is a voltage issue, then the MPPT should behave well with the smart panel if the output voltage is higher. Caps arranged in 5s2p for 27.5V. Loaded with an initial charge of 24V from the lipo's and bingo, problem solved!

 

It all came together today. 6kts!
Still lots of issues to address, leaking gear-case, vibrating drive shaft.
At least the concept has now proven itself and it will be worth my while spending more time and effort to refine it.

 

I'am glad to hear, keep working on it.