It has to do with the bad placement of the NRF module and a weak signal. This was highlighted on the old thread
A higher switching frequency seems to make me trigger my max current fault code. Should I just up the max from 170 amps?
How much of a higher freq? Have you tried reducing the observer gain value like @diavolescaCZ?
30 then 40. Tried gains from 2-5. Gains seemed to have no noticeable effect so I left it at 3.
Try backing down to 35kHz. Was there a particular reason for going to 40kHz?
Just trying stuffs. 30 was still throwing over current fault codes so went back to 20. Works fine now just wanted to add my results. Unless, I should be doing it differently.
Nope youāre doing fine. Just trying to extract details
Sorry guys, donāt have the habit of checking this forum - it sounds like 30kHz switching frequency is the recommended setting for the LittleFOCer? Iāll have to give that a try with my Funwheel hub, letās see if that is compatible with the balance app, Iāve been using 20kHz so far. The MCU has limited processing power, so doing 50% more motor calculations might not leave much room for 1kHz balancingā¦ But the BalancePro/Floatwheel guys all use 25kHz so at least I know that should be safe. Will report back!
Good news for balance vehicle users: 30kHz appears to perform better than 20kHz. Didnāt do a full ride yet, but tested the basics: low speed/ high speed and transitioning to/from sensorless with and without high load (aka hill). It all behaves nicely and slightly better than with 20kHz. Also at very low speeds the noise is more subtle (maybe cause Iām old, some young riders need to weigh in here). Overall itās a keeper! Thanks @shaman
The kind of feedback like that from @diavolescaCZ, @bj97301, and @surfdado are great for developing some documentation on tuning. The little FOCer can be fine tuned to fit your ride so you can squeeze out every bit of capability.
Speaking of parameters nobody ever seems to touch: Deadtime compensation - all the Funwheel/Floatwheel guys are using 0.12us. No idea how that value came about, but the help says: āCompensation for dead time distortion. Makes some difference at low speedā.
At very low speed my hub happens to be at its worst, so anyone got suggestions for how to tune this? Is it safe to play with this value a bit?
Well Iām glad to help. Iāll try to set up data logging and a motor temp sensor this weekend. Iāve got a short maybe 200m long hill with a 15% grade to go up and down. Iāll see how my motor does. It doesnāt have that much torque.
The compensation parameter is simply some math that adjusts certain values to compensate for the actual deadtime delay the hardware is using. You can try increasing or decreasing it in very small increments if you want. It does affect detected values though.
LOL, so I had my 11yr old try out 30kHz. He finds that noise more noticeable/annoying than with 20kHz. He has the same complaint about 25kHz. His comment: āmaybe Iāll like it better when Iām olderāā¦
So yeah that makes sense. The switching frequency of VESC based controllers is actually an āeffectiveā switching frequency. Really the PWM switching frequency of each MOSFET is half that. Itās just done in a clever way(too long to explain here) that functionally doubles that frequency.
But this means the audible frequencies are actually around half that of the FOC switching frequency. So when set to 20kHz(default), youāll hear something around 10kHz and some of itās harmonics. This is a mid-range frequency and can be heard by most. However increasing the FOC frequency to 30kHz causes an audible frequency of around 15kHz, which is higher pitch and could easily be perceived as more annoying.
So all your kid has to do is listen to Metallica at full blast for 10 years and then heāll be fine with it.
Lol, that would work - or he just integrates a BT-speaker into his Funwheel (heās been asking me to do that for a while now) which would also overpower any motor noise
Is like a small village with roads and buildings
Nice pics man!