openinverter forum

Ev8 wrote: ↑Wed Jan 18, 2023 9:50 pm Just got back from little logging, some of them might not be much use as it looks like a couple were cut short thanks to wifi going out, but i hope there is some data Pete can use for his wizardry

Results:

MG1 - log 1 : Lq=1.44mH, Ld=0.72mH, λ=112.4mWb
MG1 - log 2 : Lq=0.90mH, Ld=0.33mH, λ=52.0mWb
MG1 - log 3 : Lq=1.07mH, Ld=?, λ=58.7mWb
MG1 - log 4 : Lq=0.87mH, Ld=0.4mH, λ=58.0mWb

MG2 - log 1 : Lq=1.49mH, Ld=0.68mH, λ=106.9mWb
MG2 - log 2 : Lq=0.83mH, Ld=0.42mH, λ=60.0mWb
MG2 - log 3 : Lq=0.90mH, Ld=0.42mH, λ=59.5mWb
MG2 - log 4 : Lq=1.19mH, Ld=?, λ=58.9mWb

Now that's a bit confusing

Is it possible that some of the logs got mislabelled? The first of each set look like MG2 while the rest look like MG1

How do these compare with your meter measurements?

The unknowns are where the calculator came back with a value of zero because it couldn't find a minima - not too sure why?

Could be worth doing a log while holding stationary on the brakes and blipping the throttle a bit to calculate Rs (only useful to improve the accuracy of simulations).

Hmm I can see more logs in my future, it is possible that I was confused somewhere with 2 Wi-Fi adapters and 2 boards to connect to!

Is it possible that the WiFi is losing the signal and then auto-reconnecting to the wrong one (or the one with a stronger signal)?

Possibly I thought I had auto connect disabled but maybe not

It's also possible that the the logs are fine and the calculator doesn't work!

I also need to re do my meter measurements as I never did it properly in the first place!

I think I’d lean towards me making a mistake over your calculator being wrong!

The first couple of logs I did were much lighter throttle. Maybe I need to include something else in my he logs to help me differentiate

Hmm..

MG1 - log 1 : 216A peak
MG1 - log 2 : 170A peak
MG1 - log 3 : 94A peak
MG1 - log 4 : 216A peak

MG2 - log 1 : 220A peak
MG2 - log 2 : 210A peak
MG2 - log 3 : 75A peak
MG2 - log 4 : 213A peak

Does that make any sense, would plots help?

A few were lower for the first half but then with a large peak in the second half.

I think I just need to start again, I’m confused now!

Right hopefully these make more sense..... they are a little limited as i did them on the way home in traffic so range of speeds and currents is fairly small

MG1_1: Lq=1.04mH, Ld=0.22mH (ish- struggled to tune), λ=57.5mWb
MG1_2: Lq=0.95mH, Ld=0.34mH, λ=57.3mWb

MG2_1: Lq=1.62mH, Ld=0.58mH, λ=100.4mWb
MG2_2: Lq=1.56mH, Ld=0.59mH, λ=99.3mWb

They look reasonably consistent and the numbers look believable

cool, thanks pete, so now i'll go play with the simulator with these figures and maybe change some parameters, what is the effect of having the wrong lq-ld? set in the firmware?

In theory it would drop peak power a little as it would miss the optimum Ld/Lq raio for MTPA, it might also go into fw a little earlier. In practise you probably have a better idea than I do!

Haven't managed to get much more done on the MTPV stuff, tried to have a look last night but just couldn't get into it. Going to try and do a bit on it later but otherwise will be tomorrow.

No rush Pete the work you’ve done to date is amazing and much appreciated! Wish I was able to contribute so much

Thanks but what I'm doing is easy compared to welding/machining up a dual motor RWD compatible drive unit out of bits of FWD transmission, getting it all aligned, installed in a car, wired up and all working! It's the projects like yours that show what's actually possible.

Well if you ever need help with anything like that let me know!

Well a couple of weeks have gone by and I've been struggling to get a good tune with the new data Petes calculator extracted from my logs,
im running into unwanted regen when i lift off and brake above base frequency, like the good old days!

things seem to run fine in the simulator but not on the road.. ive attached current paramters, and previous ones which ran ok,
the majot diffrence rreally seems to be the that the data on motor inductances and fluxlinkage seems to suggest lower FWmax than i'd been using previoulsy, the isssue being that they seem like they should be right as above base frequency feels much smoother and has less audible torque ripple with lower fw current....

if anyones got any ideas that would be great to hear (yes syncof is well tuned)

previous good params

based on this data

MG1_: Lq=0.95mH, Ld=0.34mH, λ=57.3mWb

MG2_1: Lq=1.62mH, Ld=0.58mH, λ=100.4mWb

Ev8 wrote: ↑Mon Feb 06, 2023 9:04 am Well a couple of weeks have gone by and I've been struggling to get a good tune with the new data Petes calculator extracted from my logs,
im running into unwanted regen when i lift off and brake above base frequency, like the good old days!

things seem to run fine in the simulator but not on the road.. ive attached current paramters, and previous ones which ran ok,
the majot diffrence rreally seems to be the that the data on motor inductances and fluxlinkage seems to suggest lower FWmax than i'd been using previoulsy, the isssue being that they seem like they should be right as above base frequency feels much smoother and has less audible torque ripple with lower fw current....

if anyones got any ideas that would be great to hear (yes syncof is well tuned)

Could be a few reasons, but sounds like the control loops are struggling to keep control of Iq and Id in FW. Difficult to pick the most likely as there's not a lot of data to go on so in no real order:

1. Calculated values are wrong1 - the calculator just works out the best average value for each parameter. Unfortunately the motor parameters do change with load current so the Lq, Ld, FL that's most appropriate at low load may not work well at high load (we might be getting to the point of tune where this is stating to matter). One approach to test is to do a log containing the operating area that is causing problems, feed just that section through the calculator to see if it comes up with different answers.
2. Calculated values are wrong2 - I don't remember working out Rs for your motor. When analysing Bigpie's data I noticed that this can make quite a difference if it is wrong, particularly to Ld. If you do a log while the motor is locked and the throttle is blipped a couple of times then the calculator can use this to work out Rs and this value can then be used to improve the accuracy of the calculations for Lq, Ld and FL from other logs.
3. Calculated values are wrong3 - The calculator may just be coming up with the wrong values, it seems to agree with physical measurements but that's no guarantee.
4. Over-modulation - This will have the effect of reducing control loop gains at the limit which could limit how well the loops can maintain control of Iq and Id. A simple test would be to reduce/disable over-modulation to see if the symptoms improve. If they do then upping the loop gains, or increasing fwmargin, may help get control back.

I'm also not clear from your above post which set of parameters feel smoother?

Unfortunately I think we have now got most of the low hanging fruit in terms of performance gains. To improve things further needs better data which we are still short of.

Will do some logs against the handbrake later,

Ironically the current Params with unwanted regen drive smoother under full and part throttle!

ok a couple of stationary logs, rotor probably crept a bit, these are 5th gear and wheels chocked because it push forward with the handbrake on as hard as i can pull it!

Ev8 wrote: ↑Mon Feb 06, 2023 1:00 pm ok a couple of stationary logs, rotor probably crept a bit, these are 5th gear and wheels chocked because it push forward with the handbrake on as hard as i can pull it!

You only need to blip the throttle!

New figures:
MG1: Lq=0.97mH, Ld=0.28mH, λ=57.2mWb, Rs 102.6mR

MG2: Lq=1.58mH, Ld=0.54mH, λ=100.0mWb, Rs 98.5mR

So a bit of a reduction in Ld which will give a higher FWmax (around 5% on Mg2, 20% on MG1), might be enough to make a difference?

Lol thats me always going to 11