mjc506 wrote: ↑Fri Apr 15, 2022 3:13 pmIf you've still got your 'start' directly connected to 12V, that would explain why it went in to run mode and wouldn't stop.
I see.
So that starts the inverter, not the control board.
See? Really obvious things once you know them, that are ambiguous before you know them.
Also bear in mind that with inputs disconnected (left floating) you'll be reading noise. Sometimes it'll look (to the main board) like all is ok on boot, other times it'll 'read' high currents, low HV, weird errors... With the main board powered but the inverter not powered (with 12V) you'll also get screwy inputs from internal resistances, capacitors charging/discharging...
Might be why my current sensors are angry occasionally. That or, if left at 0v, when it's expecting 1.65v center, must look like a maximum negative voltage.
No arguments that documentation is lacking. A symptom of lots of different projects and little dev time. And made worse of course by obsoleted hardware! But it should all still work!
*deletes long tangent about the importance of more people than just the developers contributing to open source projects*
...
Just to be pedantic and thorough (and also, 50% of these I don't actually know the answer to...):
[*]Make up a minimal set of connections between board and inverter (put 'start' on a momentary switch, or leave a flying lead so you can touch it to 12V
I've been using a flying lead out of laziness, and, leaving it connected. So, I'll just leave it disconnected until needed.
To start with, you want power and ground connections, and connect all the HW safeties to 12V.
Dumb question... What are the hardware safeties?
You mean the pre-charge? Just connect that terminal to ground instead of MNC (main contactor)?
[*]Get the inverter and main board powered up with 12V (a car battery is fine, especially if that's plugged in to a charger too just to keep it topped up during testing - a low 12V will make everything more difficult).
I got a 12v 5a power brick last week to replace my 1a bench supply. But I notice, presumably due to the polarity protection diode voltage drop, I only get 11.3v on stuff now. I am apparently making everything more difficult with low voltage. I'll grab a lead acid instead.
[*]Read values. You will get, generally, some noise on all the inputs.
Read which values specifically?
What inputs?
What should be expected for values or noise?
What am I doing with this information? What does it establish?
[*]As above, you don't need to, but I would be tempted to get an approximate calibration of the current sensors, if only to get rid of the warnings on the interface. But this may need some potential dividers etc (I don't have a gen2 board, but the schematics are on the wiki. The stm32 will be expecting an analogue voltage of 0-3.3V, with 1.65V resulting in a 0A reading. I think you would need to provide an external potential divider to convert the output from the current sensors to these levels. Plus it's generally more tricky to feed a fixed current through a sensor. But as above, not essential for sine firmware
I get the gist of it, but not in a way I'm confident to act on. Plus, I would like the current sensors to work, so, an actual solution would be best.
Digging up from this thread, Konstantin plagiarized Johannes' solution on his Gen2 board, but, that's a bit too complicated for me:
viewtopic.php?p=11441#p11441
Arber discusses it somewhat (centering on 2.5v instead of 1.65, corrected/discussed later), and verbally describes the process:
viewtopic.php?p=9031#p9031
Just below, Sirrocco links a calculator (referenced several times with annoyance in the future), that I'm sure works, but no one seemed to have success actually picking the right values with:
viewtopic.php?p=9043#p9043
People post the circuits they tried that didn't work or that they struggled with, but don't document their proper end result. So, if an end result was reached, everyone reinvents the wheel regardless.
Again I get the gist of it, but, I only somewhat know what an OpAmp is, I don't think I have any in my salvaged parts pins (maybe I do, I don't know the common part numbers), I don't know exactly what to order, and shipping is so expensive I kinda want to order all my next batch of electrics at the same time, not 2 op amps.
[*]You should now be getting pwm from the board pwm outputs. Ideally check with an oscilloscope, but a multimeter should be able to see 'something' happening (maybe also read frequency)
Well, something is happening.
However, my HVDC is fluttering all over the place. It was at 100v, but it's fluttering between 5v and 50v. This is right off the variac's supply's DC terminals.
It's only a 3amp variac, and my smoothing cap is only 47uF (what I had around), so, maybe this is suffocating the smoothing capacitor a bit? Shouldn't be, the inverter's main power cap should basically be in parallel with it, no?
... I connected a lightbulb in parallel with the HVDC output. It's on, and steady. But the same terminals it's connected to are wandering hugely.
[*]Send a 'stop' command
Well, fwiw, voltage stabalized back at 100v.
[*]Yes, that set of parameters does include udc offset and gain (which are hopefully pretty close to what you've got).
Indeed, they are.
[*]Now connect up the PWM outputs on the board to the PWM inputs on the inverter
Oops, had those from the start...
[*]Connect some HV (direct to the bus is fine, highly recommend some current limited supply - either a bench PSU or similar, or a battery with a filament bulb or oven element connected in series. You don't need massive voltage, 12V will do)
Oops, had it at 100v the whole time.
[*]The inverter should be making a whistling noise - this is the IGBTs switching
It might be so high pitch that I'm deaf to it (I've lost some upper range), but I don't hear anything.
[*]With nothing connected to the inverter, you should have 0 current on the HV line (any bulb shouldn't be lit). If you have significant current, you have shoot-through (+ve and -ve IGBTs turning on simultaneously) which should be impossible with the gen2 inverter...
My kill-a-watt knockoff says it went from ~5w on the variac input to 7w with no bulb on the HVDC, just the 3 on the phase outputs. So, sounds good.
[*]Motor next. Connect it up!
I'm too chicken, can't I test with the bulbs first and see them, I dunno, power up and flash around?
I'm not certain what use there was in having the bulbs in there. Just to be a litmus test for whether current would be flowing through them (bad) or not (good)?
I tried the first part of the motor procedure, spinning at 10hz with the lightbulbs... I could see a warbling voltage on the phase outputs, but no power was ever drawn to the bulbs.
[*]Power up, start, set fsplispnt/ampnom. You should be able to get the motor spinning by adjusting ampnom and fslipspnt - see the wiki. This may not work if your HV supply is too weak, but hopefully you'll at least be able to feel something while spinning the shaft by hand
...
Whelp, I lifted the motor up to the desk, removed the lightbulbs, and powered the HVDC off of a car battery (with narrow leads I could grab to disconnect as "fuses"). I can measure 12.6v on the HVDC inverter input terminals... but nothing happens. Nothing happens when it's supposed to be spinning at 10hz. No voltage on the phase outputs (so, it's not just that it can't move the motor, nadda is being sent). Twisted the shaft with my fingertips, can't detect anything different than off.
Tried setting my "udcmin" to 10v. No change.
Far as I can tell, nothing different with the setup than when I had the bulbs instead of the motor phases.
Reading ahead...
Following the wiki (
https://openinverter.org/wiki/Schematic ... _the_motor ), I'm not sure how to apply the context of the motor I'm using.
"'So for a DC bus voltage of 350V, a nameplate frequency of 60Hz and a nameplace voltage of 200V you get fweak=60Hz x (350V/1.41)/200V = 74.5Hz"
My test motor is the smaller of the two ACIMs from a forklift.
It says:
Now, my DC bus, is that my test bus? Do I have to keep changing this every time I change my test setup?
Else, let's say DC bus is 100V.
Nameplate frequency is 85hz.
Nameplate voltage is 33vac.
So...
"fweak" = 85hz x (100v / 1.41)/33 = 363.2hz?
... but do I care what my namplate voltage was? Since, I'll be running the forklift motor at, I dunno, at least 200v?
Likewise, do I care what nameplate frequency was?
Or, is this exactly the purpose of this configuration - that we'll be running motors at way higher voltages than spec and need to account for that?
... Doesn't really matter, since even the simpler 10hz spin doesn't work.
[*]Hopefully by this point, you'll have confirmed that the board and inverter are both working and capable of spinning a motor. And gained a bit of confidence?
Nope, never got a motor to spin.
Confidence? Ehn, just in the certainty of having less vague instructions, yes, that helps a bunch. Haven't made much progress otherwise.
I could rig the variac to spin the motor, but I'm still to chicken to try it, if the 12v didn't do anything.
Thanks, btw, you and others, for taking the time to assist.