I seem to be hitting some sort of unknown limiter with battery current at 150 amps. In the second to last test in the video below (at 8:55 in), the thing just stops at 150 battery amps. Motor amp limit is 320, battery amp limit is 220 and absolute max is 380:
I realize I'm over-currenting things in the last test; I won't run the motor at more than 280 amps; but this mystery battery limit needs to be solved.
We were able to make 650hp with the first version (powered by an APD ESC, hitting a consistent 13.8 kW using the exact same motor, and a slightly bigger impeller), so I have a good feel for what this thing should be able to do:
Thanks in advance!
There is only so much torque this motor can produce. The torque needed to spin the impeller faster needs to be higher and that can be achieved with more gearing. These inrunners want to spin fast and compensate for torque via RPM. At a certain point you can not increase torque by upping the Amps. BTW, 440 Absolute Max is really the very edge you should use. The motor seams to be dialled in very well. If you can push 400A with 440 Absolute Max everything seams to be good. You pushed about 10KW in that test. Changing the gearing will allow you to go higher in ERPM and duty cycle and push a lot more. At the moment you stall the motor by pushing it to it's toque limit. So you can't achieve higher duty cycles. Instead of using 1.22:1 you could use 1:1,5
Battery Amps will the go up and you will probably be able to push 20KW.
Thanks Frank. I have an assortment of smaller motor pulleys on order. I don't think we're torque limited because I was able to get over 150 battery amps before I doubled observer gain. But I had current spikes over 160 battery amps. Once I doulbled observer gain, the spikes were gone, but we seem to be limited to 150 battery amps.
I have screen shots, but I can't insert images here. I'll have to email them to you.
In short, doubling the observer gain seems to have caused this. Any idea why?
At 400A your motor will generate X Nm torque and that torque will not increase right now. You can change the gearing and then get higher motor RPM and therefore higher duty cycle and therefore higher battery amps and higher input and output power. I'm not sure if your belt and pulley is strong enough to push 20KW. It looks a bit small to my eye.
I'll do some comparative testing. First thing I want to do (while waiting for the pullies) is to try BLDC mode. I also need to back motor current off. 280-300 should be enough provided I can get duty cycle back up to where it was with the other ESC. I'm not confident in that belt either, but the blower's made for that belt - the T5 profile is machined into the impeller shaft. The belt has steel wires inside and is supposed to be good for 40,000 rpm max and 15kW. Sort of hard to believe. So I bought a sheet of polycarbonate to make a blast shield. Sometimes I wonder how I get myself into these things...
This is an interesting project. Please keep us updated. I hope you get it sorted out. I'd suggest trying a different firmware(hardware no limits firmware). You'll have to retune the motor of course(unless you backup your current configuration and use same firmware but check box that says"show non default firmwares" and use the no limit firmware). Benjamin released a tuning video a while back. I have it downloaded but not sure if I can post it here. He was tuning a paramotor using rt data to remove oscillations and tweaking the motor parameters. It helped me tune a large out runner up to 280amps with no oscillations. Here's a link to the video https://cloud.vedder.se/index.php/s/jpm9Soiq68HBbdw. Saw your video, tune looks good with little to no oscillations. Not sure if you tried the no limits firmware yet.
Thanks. I'll check it out. I definitely want to get this sorted. I'm not using the "no limits" firmware - I'll give that a shot.
that would only allow you to go higher in amps but you really should not. 440A absolute max is the limit really.
Yeah, like Frank said I wouldn't go over 440amps. I've had issues in the past where settings wouldn't stick but reflashing firmware seemed to fix the issue. I always use no limits firmware and haven't had a problem with motor configs since.
That video is PURE GOLD. Thanks. He really should post it - after turning up the volume on it, of course.
The no limits firmware is a great idea to try too. Anyone know what the VESC 75/300 firmware limits are?
BTW, I need to "tune" the setup with pulley ratios so I don't need to push that much current through the motor. Realistically, at 54-56 volts under load, I really don't want to feed the motor more than about 280-300 amps - that 's the point of saturation for this motor. The problem is getting the duty cycle up. Once those other pulleys get here, I can see if the 150 battery amp limit is a real thing or not. In the meantime, I bought a bunch of lexan (polycarbonate) sheet to make a blast housing in case this thing decides to explode mechanically.
I watched your video. Wouldn't it be safer to position your work table perpendicular to how it is now? Although when shrapnel starts to fly, maybe not much benefit?
Believe it or not, there isn't room to do that and not be in the line of fire. The camera is in the doorway to the garage. If I turned sideways, I'd be in about the same relationship with the thing as I am now (which actually isn't in line - I'm a couple of feet offset), but without the welding table, laptop and acrylic to protect me. Since we're also building a much bigger unit, I figured $100 for a polycarbonate/plywood blast shield was cheaper than fixing the car or myself should things go wrong (again).
Did you figure out what the problem was?
If the motor current isn't the limit and the full input current isn't reached a few things can be the cause:
Hi Benjamin,
No, I haven't. As a Ukrainian-American, I've been preoccupied the last few days. Now that I'm feeling a little more positive about the situation, I'm currently starting to test again. I've taken Frank's advice and changed the pulley ratio (before it was 22 tooth motor pulley/18 tooth impeller pulley) - it's now 14 tooth motor/18 tooth impeller. That should at least let me get the duty cycle back up under heavy load.
I can confirm that it's none of the 4 issues you've listed. My battery cutoff voltage is set to start at 13.6 volts and completely cutoff at 12 volts. It's been cold here and our tests are generally no more than a few seconds at a time. The Trampa 75/300 VESC is sitting on a large chunk of aluminum (2-3 kg) that stays at around 10* C. We never saw less than 56 volts in testing. I'll do some more testing today and report back.
Just got back in from the garage. We're making some progress. And it's starting to look like Frank was right (sorry for doubting you, Frank). With the new gearing, running the compressor into choke (i.e. max load) we were able to hold 9.1 kW and hit 18,900 impeller rpm (72,917 eRPM to be specific; 6 pole motor = 24,305 motor rpm * pulley ratio of .778 = 18,909 impeller RPM). Voltage was pretty solid at 58.5 volts (my batteries are monsters - at 60lbs (27 kg) they should be - unloaded voltage was just under 62). Battery current was holding steady at 155-156 amps. The motor current was the limiting factor - I had it set at 300 motor amps, and it got there and just sat there. Duty cycle was also pretty stable at 61%. The only thing that got above ambient was the belt. As I would expect pushing 9kW of energy through it. And it was just a little warm to the touch (maybe 20-25* C).
I'm not too concerned about the performance with the compressor in choke - it'll never run there on the car; so this is just a durability test. Just out of curiosity, what's the max I should run the motor current at? At 300 motor amps, it's a little high for the motor. At 95% duty cycle, however, that would put us at 285 motor amps, which is about it's rated max. Obviously, the best thing to do would be to find the optimal pulley ratio that's the best compromise on the running car. But that I need to test on the dyno - testing on a public street is far too dangerous. I probably should also machine the bore on the 16 tooth pulley and see if that gives us better on car performance; but once again, right now it looks like the 14 tooth pulley is our friend (also Frank's suggestion). Especially since there's so little voltage drop.
Ok, so a bit more progress. I tried running with the restrictor in place and started getting current spikes again. So I upped the observer gain from 350 to 500. Frankly, I wish I knew the observer gain was the culprit from the beginning, but instead I threw down $200 for more heavy long cables for the motor. Well, I suppose long motor cables are better than long battery cables anyway.
Upping the observer gain to 500 allowed me to hit 90,000 eRPM without spikes. I chickened out there. There's an issue with the drive belt - it's an off shore belt and something's happening to it at high speed. I want to take it out before it fails and examine it very carefully to determine any potential failure modes before it lets go at high rpm. I'm also going to try to get a real Gates belt in this size. BTW, with the restrictor plate in place it was only pulling about 4.5 kW. Huge difference in load on the motor.
Is it possible increasing observer gain lessens motor torque? I ask because before I was stuck at 150 battery amps and before raising observer gain I was able to hit 160 battery amps. Why not just set observer gain to some really high number? What are the downsides?
I've taken the drive apart, cleaned up the belt (there were some imperfections in between the teeth) and checked everything out. Now I'm not able to get over 95,000 eRPM no matter what the observer gain is set to. I've re-run the resistance, inductance and flux linkage identification; no difference. I'm getting those current spikes again. I've seen the suggestion in this thread: https://vesc-project.com/node/1968 Any ideas?
I'm not sure why the currents get instable at 95K erpm. Maybe the switching frequency needs to be higher. You can try to go to 40Khz.
Motor Settings>> FOC>>Advanced >> Zero Vector Frequency.
Setting ZVF to 40khz and observer gain to 600 after re-running motor id seems to have worked. Interestingly, initially it didn't, but observer gain was 700... I tried 50khz, but that didn't work either. I think re-starting the VESC might have worked at 40 khz initially - read on. I've managed to hit just under 125,000 eRPM at 94.4% duty cycle with a 1" (25.4mm) restrictor in the outlet. It's moving so much air through that small hole that it's starting to lift a car battery, 2 36" long 2" x 10" pieces of lumber (.9m x 50mm x 250mm), a 20 lb (~9 kg) shot bag and my polycarbonate/plywood blast shield that weighs about 25 lbs (11kg), as well as itself. This is at 11.3 kW. The most I've seen out of this motor before is 13.8 kW. The only thing that's concerning now is that when it hits a fault now, I have to re-boot the VESC before it'll run the motor again.
Once the motor is dialed in, you probably won't face OC faults any more. Nice to see this thing finally at 95% Duty ! You learned a lot on the way and that is the valuable part.
Wha are you using for Fault Stop Time? The default is 500 ms.
General => Advanced => Fault Stop Time
Frank - yeah, at this point it runs and starts well (probably better than the BLDC ESCs I've used in the past); there's still a little shakiness in the current at max eRPM, but nothing that would qualify as spikes. I need to make the looooong motor cables and re-dial it in (which is probably nothing more than re-running id, and playing a bit with observer gain) and install it.
j.konrad - I'm using the default stop time. Before it would reset on its own just fine. Now it's exactly like another user who was trying to hit a high eRPM discovered - the need to power cycle the VESC. What difference would fault stop time make? Would I want to raise or lower it?
I'd also like to make a suggestion in the setup wizard. It might be helpful to ask what eRPM you're expecting to reach, and adjust the frequency accordingly. Or just set the default to 40kHz. For that matter, what's the issue with just setting it higher as default? Say 50kHz. I'm just trying to understand this better.
The higher the switching frequency, the greater the losses (and consequent heating) in the power semiconductors. As with everything in engineering, there are tradeoffs.
I would think that longer battery cables would be preferable to longer motor cables, but neither is desirable. Either way, I would twist the cables together as tightly as practical. Although this degrades their ability to dissipate heat, it decreases inductance and radiated EMI.
The Fault Stop Time is just how long a fault must be cleared before the VESC allows operation to resume. I don't know enough about the VESC to say if some "hypercritical fault" would require a reboot by design, or not. If the microcontroller is somehow "latching up", that's very bad.
Ok, this is frustratring. Today I tried to run it up again, and I got current spikes around 110,000 eRPM. It's just not repeatable (after the initial "wake up" run, it wouldn't even start up again with the same settings I had before). The only difference is I charged the batteries - took them from 59 volts to 62 volts. I tried all kinds of frequencies - 30kHz, 36kHz, 38kHz, 40kHz, 42kHz, 50khz, and 60kHz. The motor wouldn't even start at 60kHz, and I got the least current spikes at 40kHz. That seems to be the sweet spot. I've also tried various observer gains with those different frequencies, and 600 appears to be the sweet spot for that. Is this as good as it gets for this setup? Anything else I can try? Should I try BLDC mode? I'm at a loss.
Was the motor and VESC and wire cold? Temperature changes can have such an effect. Any other changes?
They were similar temp wise. Maybe a little warmer - a few degrees. No other changes. In fact, everything was left in place and untouched since the last test.
You can try to lower or increase the resistance value by +-5% and see if that has an effect.
Are there any firmware updates slated to come out soon? Honestly, I've invested so much time into trying to get this to work and TBH, my youtube revenue is dying on the vine as a result; I may have to move on to the bigger unit (MGM ESC & LMT motor) and see if there are any improvements in the VESC world in the meantime. I don't know what else to do. I can't play with this for weeks more in hopes that it might work - besides the $$ loss it's destroying my credibility. I may also look around for another motor with fewer poles that's sensored that might be easier for the VESC to manage.
Just the lack of repeatability is an issue - remember, we're dealing with real cars here and failure to perform or not being sure if it'll work could potentially have disastrous consequences.
I've also experienced very inconsistent behavior with VESC Tool 3.01. Have tried under Windows, Android and two versions of Zorin (where the executable won't even start up).
I think many of my problems stem from the Input App. Under "App to Use", I would like to use "ADC and UART". The ADC (connected to a variable voltage source) seems to work consistently. But trying to issue commands (speed or duty cycle) from the VESC Tool results in only a single "tick" from the motor and no movement. However, if I choose UART alone as the App to Use, it seems to work fine.
I'd recommend trying it with the ADC as the command input via a voltage source/potentiometer.
Hi Alex, we should have a zoom meeting with Vedder and try out a few things. I guess something needs to be tweaked to take the long cables into account.
Changing the resistance slightly does nothing at speed, only just when starting. The inductance is much more important at speed.
I cannot fix a problem that I cannot reproduce with the setups I have access to. Controlling these electric motors in all configurations in all conditions is an extremely challenging task, and if you make an unusual setup with super long cables, which probably have more inductance than the motor itself, you can expect there to be additional problems. It could also be something related to the motor as I have never tested a similar motor, and motors are vastly different from each other. I'm willing to help and try to figure out how to approach a setup like the one you have, but just stating "fix the firmware" will not get you anywhere. It could be a simple fix, but it can also be a long process to get it to work reliably in that configuration if it is possible at all. I have some ideas on where to start, but if you are out of patience and need to move on asap with your youtube career I suggest that you put this to the side for now.
The purpose of the youtube revenue is not to establish a "career" but to help offset some of the many thousands of dollars I have tied up in these things. Without that, I couldn't justify the money I'm spending on them - three weeks ago, the video revenue hit an all time high of just over $ 1,000 for the previous 28 days. Today, after several videos trying to get the VESC setup to work, it's down to $ 382.12. The effect is real. The ESC (in this case, VESC) is just one part of the entire system. Albeit a troublesome part. Here are my conclusions regarding VESC so far:
1. It's finicky. A change in an environmental variable (temperature or in my last tests, voltage) can be the difference between it working brilliantly or not at all. Automotive EFI systems typically deal with these issues starting with the ideal gas law and then compensate based on sensor location (heat soak), afterstart conditions, fuel wall wetting vs temp corrections (both coolant and charge temps), ignition advance vs voltage corrections, etc. I don't know what VESC does with the brushless motor analogs of these situations.Temperature swings in particular are much greater in an automotive environment than say, an electric skateboard. This was the purpose of me asking if there's new firmware in the works.
2. In my application, the major benefit of VESC is the excellent current limits; this paper clarified why: https://www.infineon.com/dgdl/Infineon-Motor_power_tool_Block_Commutatio... The differences in efficiency and speed control aren't a big deal in my case, since space/weight isn't as much of an issue in a full size car.
3. Practical installation in a car necessitates long cables somewhere in the system. This is proving to be the most challenging aspect of getting any ESC to work well.
3. The most likely immediate solution is to take some computational load off the VESC. That would probably involve a different motor - one with fewer poles (less eRPM), and with position sensors to make startup easier, particularly with the long cables. But such a change would mean more downtime and money; hard to justify right now when I have a beast of a motor/ESC already sitting here ready to go - a 40kW, 50,000 rpm, sensored setup - not that that's guaranteed to work, but it's something worth trying since I've exhausted the variables I'm aware of in VESC.
Benjamin - I do have OBS screen records (with narration so I could remember what I was doing/thinking at the time) and simultaneous video of the unit itself for my own edification for these latest tests. I'm happy to sync those up and upload them as unlisted videos if you'd like to see what I've tried and what the results were if that helps. Just let me know.
Screw it - I just ordered a sensored Castle Creations 4 pole motor. I want to see if that works better.
Ho about a larger motor, and use maybe 3:1 up gearing to get the RPM on the turbo ?
I recommend installing a temperature sensor on the motor. The vesc seems to do well with temperature compensation. I installed one on a large inrunner to protect the motor with good results. The motor runs well now, cold or hot and it's protected from overheating. I hope you get this sorted. Please keep us updated with your results. I'm curious to what you find out, It may help others in the same situation.
I haven't given up on this, but in the interim I did take the path of least resistance and put an 800 amp MGM ESC and 30100 LMT motor on the other electric turbo, strapped it to the car and took it to the dyno. FWIW, the setup just worked. And it pulled a rather astounding 35kW continuous (~675 amps at 52 volts), and made over 700 flywheel hp (800-850 is the target, and I see that as completely possible with some minor tweaking):
https://youtu.be/PjOJU02FIGw
I will get back to this one; I think the VESC is having issues with sync - both at low rpm with the longer cables (i.e. startup wasn't repeatable from day to day), and with the relatively high pole count of the TP Power motor at high rpm (6 poles). To solve both issues, I have a Castle Creations sensored 4 pole motor sitting next to me to try out when I can - it might not be for a little while as I want to get the other setup to the drag strip. I don't expect this one to make the same power as the MGM/LMT combo, but it should still support 600-650hp, and is much cheaper and smaller. Hopefully the different motor solves the issues the VESC was having.