Real time measuring is not a problem anymore.

Our FlySky i6 RX/TX has dead cheap telemetry modules for RPM (magnetic or optic), temperature and voltage.

True, brushless be weird and that goes double for the ESCs I am using - the set-up program is the strangest that I have used! There could be problems with the incorrect drive frequency and timing, more things to check out. I found that when the motors are turning at minimum speed the ESC and wiring also gets hot:





The ammeter was showing exactly one amp during the test, so much of the 45 watts of input power was wasted as heat - its time to have another crack at the programming.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

your ammeter might well be smoothing a whole bunch. The RMS current might be 1A but you might be getting big peaks perhaps.
_________________
Mechanical engineers build weapons, civil engineers build targets

It could be smoothing out the current, but wouldn't the RMS value be the one to use for calculating the power anyway? I just checked the timing (set to neutral) and the frequency (8KHz), which are the recommended values for the motor. The other thing I can try is a new VESC controller, it might work better with the NEU motor.

I found a different sort of problem; some idiot miscalculated the motor KV / gear reduction / voltage formula and the top speed of the bot is a ludicrous 31KPH . After looking at all the options, adding an extra stage to the gearbox and making it 36:1 is the best for cost, weight and speed. Its not so good for complexity but you can't have everything.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Keep in mind a brushless motor at its slowest speed is acting much more like a stepper motor than a traditional brushed motor.

It is highly likely that it will use a significant amount of current when going slow with any amount of torque. Unlike a brushed motor which will just stall and you can increase current to get it moving again. The brushless in sensorless is likely to try and hold its position with a certain torque as if it loses this position it will have to try and restart. This is likely why its using 45W just turning slowly.
_________________
Steven Martin
Twisted Constructions
http://www.botbitz.com

Would that still apply at around 7,000 RPM? The motor was always around that speed during testing. It got hotter as the speed increased too.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

ESC getting hot at minimum throttle could be switching losses? If it's so slow that the fets are barely switching on before the pwm pulse is over then you'll get very poor efficiency for 90% of the cycle.
_________________
Online store - www.ranglebox.com
Pulsar and friends.

It could be that, but the motor & ESC both get hotter at higher speeds. I also have the switching frequency set as low as possible, which is supposed to minimise FET heating.

It could be that the ESC is just a piece of crap, but its an up-market Czech MGM Compro



Specifications here: http://tinyurl.com/j6vzxyx

I recall another overheating issue with really low inductance 'coreless' motors, but the motor here has a sizeable steel armature and is a smaller version of Mr Mangle's weapon motor, which doesn't heat up at lower loads.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

RE RMS, could be. I think I should have said aliasing, like your meter might be way under reporting short high current spikes.
Do you have any kind of scope? you could look at voltage drop in a wire or across a current shunt to see a realtime plot of the current draw.

Failing that if you run it for X minutes and see how many w/h you put back into the battery it'd give you a dodgy approximation of the w/h and hence w you are pulling.
Probably around 70-80% efficiency on the battery charge is my guess but you can look it up lipo is normally pretty good.
_________________
Mechanical engineers build weapons, civil engineers build targets

OK, scoping it sounds like a good idea - will need to find a current shunt though. I am reading up on the VESC set-up and have also found a Castle HV ESC to trial against the MGM unit. The Castle ESC has logging, so perhaps it's data will show something useful.

The MGM and the VESC both have inputs for sensored motors - I wonder how hard it is to add sensors to the back of the motor? The main thing I have NFI about is how to position the hall sensors relative to each phase winding.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

With the VESC you may need to watch your E-rpm they max out at about 100,000 e-rpm.

7000 * howevermany poles (Glen help)

It seems like the timing or something is off if its using that much current with no load even at high speed.

Steve
_________________
Steven Martin
Twisted Constructions
http://www.botbitz.com

Its an 8 pole motor, so that would be 56,000 RPM. The top speed of the motor on 48V is 32,000 RPM and under load its going to be less, so that's all good.

I checked the timing and it's zero degrees, what the manufacturer recommends. I'm zipping out to Bunnings to buy a cheap current shunt so I can see what's really going on with the current.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

Update:

I got a 100A shunt from Jaycar (tip: they are selling off their larger 200A shunts for $4, a real bargain!). The voltage across the shunt didn't show anything exciting, just low level noise and around 1 to 1.5 amps.

Substituting the Castle ESC made a huge difference; almost no heating in the motor or ESC, even at higher RPM and a much longer run time and the current draw was around 50% less. Bottom line, it's definitely the ESC, not the motor.

There are some other differences: the Castle ESC starts the motor at a much lower RPM and when the MGM ESCstarts the motor, it sounds like a stalled stepper motor for a second.
_________________
Australian 2015 Featherweight champion
UK 2016 Gladiator champion

I recall sending off an annoyed email about one of the Jaycar current shunts (Many years ago, maybe it's fixed now). Something to do with the resistance stamped on there being wrong by a factor of 10 or 20... I don't remember any more that than.

It's still useful, but calibrate it first off a constant current supply to know what you are reading.
_________________
They say that he crossed the fine line, from insanity to genius.

I use an Alegro 200A current sensor from Element 14

http://au.element14.com/allegro-microsystems/acs758ecb-200u-pff-t/current-sensor-200a-uni/dp/2057428

There is a library for it in Eagle that allows you to make cheap PCB's to mount it on. Last time I played with it, piping the data into a Teensy/Arduino was dead simple (then displaying the value on a serial 4 digit display). I was looking at using them as a short circuit/current limiting feedback circuit for BotBitz 85A's.
_________________
https://www.halfdonethings.com/