Grin Tech 12FET Throttle Rollback/Surge Issue

[amberwolf]

AFAIK this 12FET is from the generation just prior to what they sell now. It hasn't yet been opened or programmed, etc., so it just has whatever the default settings they ship with.

It's from this post:
http://www.endless-sphere.com/forums/viewtopic.php?p=747654#p747654

I mentioned this problem (at some length) in my CrazyBike2 build/evolution thread, here:
http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=12500&p=779515
http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=12500&p=781325
http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=12500&p=820114
Other mentions are here:
http://www.endless-sphere.com/forums/viewtopic.php?p=747862#p747862
At one point I mistakenly thought the CA was causing the problems:
http://www.endless-sphere.com/forums/viewtopic.php?p=748492#p748492
but I was very wrong. I don't know why I didn't have problems at that time, but did before and later.
http://www.endless-sphere.com/forums/viewtopic.php?p=748896#p748896



The basic issue is that regardless of method used to set a constant throttle voltage input directly to the 12FET's throttle connection, and regardless of which motor is being controlled by the 12FET, speed does not remain remotely constant even on level ground with no wind. Speed decreases by at least 1MPH, then increases by at least 1-2MPH, and cycles intermittently (not on a specific time or frequency), over several seconds to nearly a minute (varies).

It *will not* simply let me ride at a constant or even near-constant speed, unlike any other controller I've ever used.

AFAICT using a meter, no voltage variation from the throttle itself occurs while it's set at any particular point, so it's not being commanded from the throttle itself--it's something in how the controller works.

The throttle is run direct, not thru a Cycle Analyst. The only CA connection to the controller I had (which I don't now) was for the speedo sense wire and ground.

I can see the power dropping dramatically, but not completely (despite what I posted in the above links), when the controller rolls back output. If I am at say 500W battery draw, it'll drop to 200W suddenly, until the speed drops far enough, and hten it will surge back to well above 500W (maybe 800-900W, depending on conditions at the time) till it reaches it's overspeed, then it will roll back, and cycle like that.

It doesn't go to the same speed, either under or over, each time, but rather oscillates around and each peak may get higher and higher successively, until at some point it reverses the trend. Same for rollback--it may drop to lower and lower speeds, until at some point it reverse the trend. But the trend is just a trend--it will sometimes still peak lower somehwere during a rising trend, etc.


Fixes tried so far, with identical results for each, were:
--Zombiess Throttle Tamer, to try to smooth out possible throttle voltage variations (even though I dind't see any variations, it dind't mean they weren't there, perhaps).
--Gripping the bar and the thumb throttle so tightly it hurt, to guarantee that my fingers weren't moving the throttle and causing it to happen.
--Tightening the grip against the throttle body hard enough for friction to hold the throttle in one physical position unless I drag it back to wehre it was (defeating it's internal spring, effectively).
--Uisng a switch to flip over from thumb throttle's signal line to a resistive divider instead, to input a specific voltage to hold a speed.
--Using a potentiometer as voltage-divided throttle.
--Using a different throttle, on hte left bar, wiht my left hand instead of right.


More specific behavior, to edit (for clarity and updated info) and summarize from the first linked post:

When using this 12FET to control the HSR3548 in 20" wheel on the rear, I put throttle at WOT from a stop, and it accelerates as desired (with or without using *both* front and rear for the first few seconds to get going quicker, with front 9C/6FET on a separate throttle), all the way up to my normal riding "top speed", which is just shy of 20MPH, exactly as I want. (top speed of the system is probably a lot higher than that, but I don't use it under any normal conditions).

Then it creeps up to say, 20.4MPH, which I could live with--I don't think the police would care about half a MPH; my speedo in the CA probably isn't even accurate to that since it's just set for basic diameter of a 20" tire. But as soon as it reaches that, maybe a second or two later, power cuts way way down, perhaps in half.

Then it goes back up just a little in power but not enough to hold the speed, and coasts down to as low as 18.5MPH.

Then it applies another long burst of power, this time going up to as high as 20.9MPH.

Then power cuts again, and eventually cycles back on and goes even higher the third time, up to 21.5MPH before it cuts back.

Then it goes back to the first cycle again to not quite 20MPH, or maybe to the second cycle where it's around 20.4MPH.

That set of cycles will repeat constantly as long as I hold any throttle position, under any road conditions or wind (the actual speeds may vary, but the cycle itself does not).

If I immediately re-ride the same stretch of road in the opposite direction, it doesn't repeat the variations in reverse, but they instead happen in the same "random" manner they did the first time, implying they have nothing to do with the terrain. Same is true of re-riding it in the same direction.



The problem also occurs to a lesser degree (lower variation, mostly, and far lesser "kick" from the acceleration) when using hte 12FET to control the front 9C 2806 26" wheel. But it still happens in the same way.



What I'm looking for are potential ways I can deal with this issue.

Most likely, I will have to reprogram the controller, assuming it is simply a parameter in it that is not correctly set up.

Presently I do not have any way to try reprogramming the controller (my USB link from Lyen is somewhere in one of the sheds at the house (assuming it's not stolen with other stuff by looters), inaccessible until the house is repaired and I can move back in, probably at least a month and a half or two months from now).


If it is not a programmable function, then I may also be able to physically alter the controller, but I would guess that if it's not a setting change that it is something in the way this firmware works. If that's the case, I would love to find someone willing to trade an either older or newer Grin Tech 12FET for this one, as long as that one is working normally, doens't need repair, is reliable and doesnt' have this problem--meaning it will actually *hold* a particular speed when commanded to a particular throttle voltage, assuming flat terrain and either no wind or constant wind. (basically, under a constant load).

 

[chvidgov.bc.ca]

.

 

[teklektik]

So - to ask the obvious about this tale of woe that began back in May: What did Grin Tech have to say?
It's hard to believe they did not just exchange it.

 

[amberwolf]

I haven't wanted to use up their time on it--at first it wasnt' a big deal, and I figured that it was just a programming issue, and I could just find my USB adapter and see about using XPD or similar to see about finding optimal settings for my uses anyway, which might well fix the problem in the process.

But I haven't been able to go thru my stuff at my house in the sheds for a number of reasons, and now, because of some of those same reasons plus others like looters, all that stuff is now completely buried under everything else I have tried to salvage from the rest of the surviving stuff that hadn't made it to sheds yet, and it's going to be at least a month or more (probably a lot more) at this point before I can dig thru any of it to find the adapter.

Of course, I can always order another one...and will probably end up doing just that, since I don't even know for sure that the one I had survived the fire and made it out into the sheds, or if it was still there after the looters got into the sheds, too.

And the controller does work, and reliably, except for that "small" thing--which probably wouldn't even bother anyone else. I've had more than one person basically say that I'd "get used to it", which is always an answer that *really* irks me--because one shouldn't have to get used to something working in a way you don't want it to, instead of fixing it so it works the way you do want it to. (I already have to live with a lot of things, little and big, that I can't do anything about...so the less of those frustrations I have, the better for what little sanity I still have. :lol


But anyway...the point was that I had intended to try to fix it myself first, and didn't want to waste Grin's time on it--they have a whole lot of stuff to do to stay afloat and expand and create new things, and not nearly enough time to do it in as it is. Plus it's sometimes harder to communicate with them than I'd like, which also deterred me a bit--I know it is because of their limited time to get stuff done, so the less I interrupt them, the more they can help others that *can't* deal with things themselves.


So since I haven't been able to deal with it myself, yet, I thought I'd ask here first for other options or ideas, if anyone has any.



EDIT: Also, I guess I also haven't asked Grin yet cuz I'm stubborn....

 

[nuleeconverted]

Why not open it up and have a look-see? It ain't gunna be a Jack-in-the-box puking its guts out in your face ordeal IF you discharge the main bus bar caps before poking around. Measure with a yolt meter to dispel all fears. Then deal with the Silicon heat sink grease. It's so bleeping messy, but easily cleaned up with 91% alcohol. Round up the usual suspects! Intermittents can be very challenging even for seasoned half blind ol' fogies. Here's a couple of SWAGs:
- semi-functioning FET(s) due to associated drive circuitry including a bent 12-20v localized Al electrolytic bypass cap (three total, one for each phase) [I had pulled too hard on the red (+)Vbatt wire that resulting in snapping one of these caps to one side. I have a spare and can cheerfully mail it off to you if that's the issue. ],
- previously pinched wiring that looks superficially O.K.

I could talk my brother into selling one *cheep* recently produced, 12 FET, non-programmable, "almost virgin" Justin brand controller. (The un-programmable "feature" made it useless, that's why it collects dust).

Good Hunting, and Good Luck!

 

[amberwolf]

I doubt that it is a defect like those, simply because as long as it is still under acceleration (and thus high current draw) it doesn't exhibit the problem--it is only after it has reached the target speed set by the throttle, and power would have dropped anyway, that it will begin doing it, AFAICT.

So, for instance, if I'm going up a hill, it won't do it until after I've topped the hill and gotten back to flat terrain where it will then begin the issue as power drops.

So I'd expect a problem with a FET or wiring (power wiring, anyway) or a power filter cap to show up during the *high* power usage, and not only after power usage drops as speed setpoint is reached.

FWIW, I still need to put it up on blocks and see if the issue is replicable without a load on the wheel. I'm pretty sure I tried it before but I cannot remember what the results were, and I cna't find the post about it in my CB2 thread or in my notes here, so I probalby never wrote it up for whatever reason (interrupted?).


Also, the effect is less noticeable on the 26" 9C 2806 than on the 20" HSR3548--but I don't know exactly why.



Anyway, at some point I probalby will open it--but I can't risk that until after I have ensured I have a backup controller working on the bike (I have a couple at the apartment that need connectors changed over to match what's on the bike, and hall/phase order figured out for this motor, but haven't yet had enough time to finish dealing with it).

Since this bike *is* my transportation for my 20-mile-round-trip work commute, and everything else, it has to stay working--I can't pedal anywhere near that distance even on a regular bike...much less this beast. (which is why I put two motors and controllers on it to start with, this past summer when I got it going again...but I am down to one motor right now and that's bad enough). At the apartment I just don't have really enough space to keep multiple bikes even if I had another working one (Delta Tripper the trike works, but doesnt' fit thru a doorway as narrow as the apartment's without disassembling parts of it every time I have to go in or out, and if I can't take it inside I can't have it there). So till I get back to the house, whenever they finish fixing it, eventually, I'm kinda stuck with just CrazyBike2.


I do still have the 6FET on there as an emergency backup controller, but it doesn't provide enough current for startups to let me accelerate from a complete stop at intersections fast enough to keep cars behind me from wanting to run me over. The 12FET *barely* does that, and some (very few) are still impatient enough to zoom around me into the path of oncoming traffic long enough to pass me *in the intersection* before they screech into my lane in front of me.



Anyway, I was hoping someone might have an idea what program parameters (if any) might cause the issue it has. It sounds like so far that's a negative, though I'll still wait a bit to see if anyone else pops in with ideas.

If it is a hardware issue, other than something in the MCU itself, I might be able to fix it myself if it doesn't involve SMT work--but I gotta get a second 12FET working as backup before I open it up to check anything out.

 

[bigmoose]

Amber it sounds like a PID loop "hunting" for speed control in the firmware. Whether that feature is in the firmware, I have no idea. As you said, your bike is very heavy compared with what is "average." Try it with the wheel off the ground. If unloaded RPM is constant, then you have another data point. If it is in the firmware, then you have a "feature" and not a "failure."

 

[amberwolf]

Well, hopefully I can remember to try the offground test today while doing the frame repairs for this problem:
http://www.endless-sphere.com/forums/viewtopic.php?f=2&t=12500&p=827313#p827313

I also got the halls to fix the other motor, and can retest with taht one after the repairs, too.

With luck I can also test up a spare controller or two of the ones I brought with me to the apartment, and then I can open this one up and check for physical faults (though I expect it is a programming thing, though whether or not it's user-alterable I dunno yet).

 

[amberwolf]

Crossposted to the CrazyBike2 thread:

Something I keep forgetting to post, about that controller throttle rollback/surge issue:

During the rollback portion, it is definitely actively braking the wheel, not just lowering the power level output. It's not going into regen, though, cuz there is zero regen current measured by the CA if I don't use regen braking manually at any point in a ride.


I can feel the braking, each time it peaks in speed and then begins to slow--it's not the same as if I just cut throttle back manually--that just coasts down to the new set speed, whereas the problematic behavior is definitely actively braking the wheel down below whatever setpoint it thinks there is.


A version of the problem that doesnt' happen very often (relatively) but does at least once every other ride or so, sometimes less, is that it will suddenly decide after braking/rollback down to around 1MPH less than where it is actually set at, it will keep that as it's new setpoint. So if I have the throttle physically friction-clamped in place to give me ~19.5MPH, and it does it's surge/rollback thing for a while, someitmes it will roll down to 18.5MPH or less, and then instead of ever surging back up to and past the original 19.5MPH, it will now surge/rollback around the new center setpoint of ~18.5MPH, going up to around 1MPH above and below that point. If I use the ebrake to cut out the controller, and slow to a stop, without touching the throttle, then when I relase the brake and finally re-reach the original 19.5MPH setpoint it will now go back to cycling around that one. So it is definitely not that the throttle slipped off it's original point and went down--it's that the controller "forgets" what it was set to, and just starts doing something else. But it is consistent in that it is almost always ~1MPH less than the original setpoint.

If I were on a long enough run without any stops, I could find out if it would continue to dro in 1MPH increments like that, but I only get at most a mile before I have to stop again, and usually only half a mile. On extremely rare occasions I'll reach a light when it turns green and not have to slow or stop, but at most that buys me a half a mile. I think I'd need several miles without stopping or slowing to see how this behavior might continue.


I have yet to do the off-ground speed test, because I have to mvoe the speed sensor off hte back wheal onto the front one, and the wiring is inside the "fairing" so it isn't a simple job. It also takes a while to ge thte magnet and sensor lined up right on the rear wheel cuz of the way the frame is made back there, and I don't like havingt to redo it. So...eventually I'll get that test done. :/

 

[deepfraught]

I am reading this thread with a new MAC 8T with 12fet Infinion kit from EM3EV in current limit mode in all 3 limit positions it surges from 24kph to 31kph. Max limit simply passes the speed range quicker so the stuttering is rapid/shorter/less drawn out than in lowest power position.
I'm into the 3rd week of commuting in the new year on it. I'm searching for other MAC 8T with CA users to find what their CA settings are, Paul is looking into giving me the list of what the CA settings should be to confirm my figures when he gets free.

I'm using thumb throttle, go to WOT, no problem up to 24kph where surge begins, continues through to 31kph and ceases above that. In this time the current limit hovers around 700-800, but does change.

Reading this again something seems not right that it's the same speed affected repeatedly, so back to the doc to see what I'm missing.

 

[amberwolf]

If yours is always at a particular speed, and involves a Cycle Analyst doing limiting or any kind of controlling (not just monitoring via the shunt), then I doubt it is a related problem, given that it doesnt' matter what speed I'm trying to hold, and I don't even ahve to have a CA hooked up to the bike to get this problem.

You need to disconnect the CA from the system to eliminate that, first. If it still does it, and you find that it will also do it at other speeds, then maybe we should compare our controllers to see if they are the same hardware/firmware and figure out the problem.

If yours is fixed by disconnecting the CA, you'll want to start your own thread about your specific problem (or in one of the other CA troubleshooting threads about similar issues, if there are any), so we can help you narrow down the exact issue and fix it.



I still need to move my wheel sensor to the front and reset the CA's wheel size to do the off-ground unloaded test on the 9C (or get the rear motor finished and reinstalled--either one is about as much work).

Today's excuses: Meant to do that during the "vacation" I got this week since I couldn't use it to move back into the house (not ready yet like it was supposed to be, and for which the vacation had been planned). But I forgot until late day before yesterday, and was already tired, then work needed me to come in in place of someone else that hadn't shown up, and today is planned to help someone else (who hasn't called me yet this morning, half the day gone I could've used to do this if I'd known I'd've had the time, but couldn't start it without knowing I'd have time to finish it and ensure the bike is all functional for work tomorrow, cuz sometimes stuff goes wrong and takes a long time to fix). Then I work the next five days, so that's out.

I think I am probably not going to be able to do this stuff till I am back at the house for a while, and have a second one up and running as a spare, just in case.

 

[deepfraught]

Thanks I have a CA bypass cable set I will try over the weekend.

I read your fire tragedy a while back, I hope the repaired place works out and you can settle down again.

 

[amberwolf]

FWIW, the surge/rolback issue hasn't changed.

A few things changed on the bike that made no difference include removing the phase and hall and power connnectors and soldering all these directly, so taht eliminates those as a possible resistive/etc problem.

The Grin 12FET is still running the front 9C at this time. Since I mostly use the rear HSR3548 it doesn't bother me much anymore, but whenever I forget about it and try to cruise with just the front 9C, I am quckly reminded of why I don't use it for much more than extra power for startup from a stop and hillclimbing....


Someday I hope to have time and energy at the same time to work on the problem.