SOLVED! Infineon Cutting out Question

[matt_in_mtl]

- There is also the good old shunt soldering method that would also help reduce the cutoffs of course.
actually I tried this, and it did not resolve the problem. Jeremy Harris suggested that this may be due to the shunt operating as an inductor instead of a simple resistor at these large fast changing currrents. The other solutions give much better results

- Adding some extra capacitance in parallel with "C20" may also help reduce sporadic overcurrent cutoffs by filtering out the spikes.
C20 is unmarked, so we don't know what value it is. My solution was to increase R43, it is 1k standard, I have increased it to 47K, which has effectively filtered out all the current spikes, but if the current is high long enough it will still trip the cutoff. My controller still has stock fets, so I don't want to increase the cutoff to high. I am not convinced if this is the best solution though.

Does this sound right, guys?

And Matt - maybe you could change the thread tittle to reflet it's "SOLVED" status?
good idea

 

[matt_in_mtl]

Well... that was fun. I found 30amp phase to be a little anemic, and it was only drawing 8amps from the battery, so I upped the battery current to 40 amp and set the phase to 75 amp ( note this is with stock fets ) it was a lot better, and there was NO cutting out... except after a couple of blocks of medium uphill at low speeds pulling approximately 13amps from the battery... the controller died.. looks like a low side fet blew. So... 75 amp phase is too high for the stock 6 fet.

 

[ZapPat]

- There is also the good old shunt soldering method that would also help reduce the cutoffs of course.
actually I tried this, and it did not resolve the problem. Jeremy Harris suggested that this may be due to the shunt operating as an inductor instead of a simple resistor at these large fast changing currrents. The other solutions give much better results

I just soldered up my 206's shunt and it has cured my own cutoff problem... and more than doubled my current draw too! (up to 85A peak on the CA). I did no other mods to my controller, exept change the caps because they had almost exploded (and tripled their values so they hopefully don't pop again). It also seems like gwhy succesfully used this along with changing the controller type to a "EB212" to compensate for the lower shunt value.

Of course I'm using a 9C hub with this controller, not an outrunner like you are so my peak currents going into the motor aren't as high as in your system. So even though soldering the shunt might not have worked for you, it might well work for others using hubs also (which have higher resistance and inductance). Anyways, that little shunt needed to be soldered, because it was obvious that it was getting too hot at 35A with so much resistance.

- Adding some extra capacitance in parallel with "C20" may also help reduce sporadic overcurrent cutoffs by filtering out the spikes.
C20 is unmarked, so we don't know what value it is. My solution was to increase R43, it is 1k standard, I have increased it to 47K, which has effectively filtered out all the current spikes, but if the current is high long enough it will still trip the cutoff. My controller still has stock fets, so I don't want to increase the cutoff to high. I am not convinced if this is the best solution though.

That's a good way to change the RC time constant - and yes it's true that all this does is delay the reaction time, not lower it.

 

[ZapPat]

Well... that was fun. I found 30amp phase to be a little anemic, and it was only drawing 8amps from the battery, so I upped the battery current to 40 amp and set the phase to 75 amp ( note this is with stock fets ) it was a lot better, and there was NO cutting out... except after a couple of blocks of medium uphill at low speeds pulling approximately 13amps from the battery... the controller died.. looks like a low side fet blew. So... 75 amp phase is too high for the stock 6 fet.

Unmodified 6FET controller @ 40A battery = sure death!!

IIf you want to beef it up good you'll need: Very thin kapton /mica, low ESR caps, good FETs (3006 = 60V, 4110 = 100V), good wires... and patience! Have fun!

Pat

 

[matt_in_mtl]

Unmodified 6FET controller @ 40A battery = sure death!!

Pat

Yeah, apparently to be fair I knew I was pushing it, and am not too surprised that it blew. Also, I was nowhere near 40A from the battery, more like 20A peak, I suspect it was hitting 75amp phase current limit which really hurt it. I have opened it up, and the fet that blew also melted it's little bolt insulator... not sure which happened first though, the melting causing a short or the heating/blowing causing the melting. Also, after looking up the datasheet for the fets turns out they are only rated for max 60A continuous (I thought they were rated for 75A continuous)... oops!

Glad to hear you solved your cutting out problems by modding the shunt. I suspect you are right that it worked because you are using a hub motor. I was able to scope the shunt voltage hitting 0.6V even after modifying it.

-matt

 

[gwhy!]

Unmodified 6FET controller @ 40A battery = sure death!!

Pat

Yeah, apparently to be fair I knew I was pushing it, and am not too surprised that it blew. Also, I was nowhere near 40A from the battery, more like 20A peak, I suspect it was hitting 75amp phase current limit which really hurt it. I have opened it up, and the fet that blew also melted it's little bolt insulator... not sure which happened first though, the melting causing a short or the heating/blowing causing the melting. Also, after looking up the datasheet for the fets turns out they are only rated for max 60A continuous (I thought they were rated for 75A continuous)... oops!

Glad to hear you solved your cutting out problems by modding the shunt. I suspect you are right that it worked because you are using a hub motor. I was able to scope the shunt voltage hitting 0.6V even after modifying it.

-matt

Hi Matt,
The stock fets are a bit rubbish on these controllers My thinking is that the insulator melted due to the heat this fet can generate ( High Rds value ) which then reduced the thermal contact to the heatsink which made the fet even hotter and so on ( so probebly nothing to do with excessive current in the true sense of the word), The case that these controllers are in also are not the best for moving the thermals about. I have re-housed my 6fet controllers into a similar size box with better heatsinking.

 

[kZs0lt]

Hi guys,

I just came across this thread, I see the members have just solved a critical problem with the little 6 FET controller.

I am about ordering a little 6FET controller myself as well from Keywin, are there any other issues I should know about?
I am asking this because once I burned myself with getting a controller from him which turned out to be a slow start one

I will fit it with 4110 FETs, and want to be able to run it 40-50 battery amps max, but more like 20-30 normally. And phase current as high as possible.
My 12 FET controller is just too big

I have the usb wiring harness for programming, it was for the 12 FET 846 infineon controller, I think it can be adapted to these.
So I can run these higher currents with programming as a higher PCB, and/or shunt soldering.

What should I specifically ask for from keywin. Do I need to specify the chip 116/846, instant start, not pedal first sensorless...
Any advice is welcomed.

I appreciate your great work in solving this issue.

Zsolt

 

[gwhy!]

Hi guys,

I just came across this thread, I see the members have just solved a critical problem with the little 6 FET controller.

I am about ordering a little 6FET controller myself as well from Keywin, are there any other issues I should know about?
I am asking this because once I burned myself with getting a controller from him which turned out to be a slow start one

I will fit it with 4110 FETs, and want to be able to run it 40-50 battery amps max, but more like 20-30 normally. And phase current as high as possible.
My 12 FET controller is just too big

I have the usb wiring harness for programming, it was for the 12 FET 846 infineon controller, I think it can be adapted to these.
So I can run these higher currents with programming as a higher PCB, and/or shunt soldering.

What should I specifically ask for from keywin. Do I need to specify the chip 116/846, instant start, not pedal first sensorless...
Any advice is welcomed.

I appreciate your great work in solving this issue.

Zsolt

Hi Zsolt,
When I ordered my latest 2 6fets controllers from E-crazyman ( keywin ?) I didnt specifiy anything, I recieved the newer 116 instant start controllers, I HAVE NOT confirmed this but the pedal first can be disabled/enabled in the software should that be a problem ( it wasnt for the 2 of mine ). Your programming lead will be ok to use but may need a slight mod. If your new controller cuts out when rolling on the throttle then in the software select the next board type up ( 12fet ) and halve the Shunt value this will then give you the indicated motor/phase current set in the software and try again, If still cuts out then set 18fet and reduce the shunt value again, and so on till it don't cut out. I have only tried this upto the 18fet settings but I think the newer software supports 24fet .
Hope this helps.

 

[hydro-one]

Why not just order a 6fet from justin ebikes.ca. already has 4030fets (?) or good ones anyways. i soldered the shunt and now get over 5000w during hard acceleration on the CA (calibrated pretty close) 60v. My 14g battery wire gets warm. but controller stays cool!!! Am i just lucky???

this controller rocks with the good fets!!!!

mike

 

[kZs0lt]

Hi Zsolt,
When I ordered my latest 2 6fets controllers from E-crazyman ( keywin ?) I didnt specifiy anything, I recieved the newer 116 instant start controllers, I HAVE NOT confirmed this but the pedal first can be disabled/enabled in the software should that be a problem ( it wasnt for the 2 of mine ). Your programming lead will be ok to use but may need a slight mod. If your new controller cuts out when rolling on the throttle then in the software select the next board type up ( 12fet ) and halve the Shunt value this will then give you the indicated motor/phase current set in the software and try again, If still cuts out then set 18fet and reduce the shunt value again, and so on till it don't cut out. I have only tried this upto the 18fet settings but I think the newer software supports 24fet .
Hope this helps.

Thanks for your input gwhy!
I am ordering a 6 FET controller and a separate PCB board just in case I screw up something
Oh, I didn't knew that these are not pedal first or instant start, and they can do both. My 846 infineon is not pedal first, but slow start and not instant start. I came to hate that delayed throttle response.

Why not just order a 6fet from justin ebikes.ca. already has 4030fets (?) or good ones anyways. i soldered the shunt and now get over 5000w during hard acceleration on the CA (calibrated pretty close) 60v. My 14g battery wire gets warm. but controller stays cool!!! Am i just lucky???

this controller rocks with the good fets!!!!

mike

That controller is a great value indeed, but I can come out cheaper from Keywin with a little labor.

I see that soldering up the shunts is a widespread hobby . Wouldn't that affect the accuracy of the CA? It may be a quick solution, but maybe not the best. They use constantan/manganin for a good reason in shunts instead of lead. I'd rather double up, or shorter the shunt. But however it may have little significance, because the solder has a much lower resistivity and it's temperature dependance may not show significant shift.

zsolt

 

[ZapPat]

I see that soldering up the shunts is a widespread hobby . Wouldn't that affect the accuracy of the CA? It may be a quick solution, but maybe not the best. They use constantan/manganin for a good reason in shunts instead of lead. I'd rather double up, or shorter the shunt. But however it may have little significance, because the solder has a much lower resistivity and it's temperature dependance may not show significant shift.

It might actually not be such a bad thing to have the shunt's resistance go up with temperature... when your controller gets hot then it will pull less current, thus protecting it a bit (hot FETs can't take so much current).

Of course on the other hand if you use your controller's internal shunt to measure your current, then you couldn't rely on your Ah readings as much as it drifts.

 

[hydro-one]

I calibrated it by watching wat goes back into the lipo's after a run....also i noticed when its really hot in there, the zero amps drifts a few tenths into the negative, so the temp does affect the soldered shunt (lots of solder on there its just over 1 miliohm as far as i can tell. I can tell you the power is IMPRESSIVE!!

 

[Arlo1]

Cool guys

 

[Arlo1]

Ok so.... I am thinking the cutting out thing is the controller trying to save it self when massive over currents happen. This seems to be a common thing with low resistance and more importanly low inductance motors. It will be interesting to see what the best way to proceed with all if this is.... I have started to test air core inductors wound in series with my X5 motor to see what I can see with my oscilloscope. I will be doing a lot more tests asap.. But here is the link http://endless-sphere.com/forums/viewtopic.php?f=30&t=29852

 

[mdd0127]

I don't think that this issue is fully solved. It seems like Arlo was really onto something in his last post.

For those thinking about using a 12 fet with a 3220, some more research might be in order.

I had good result on the bench at first with my new lyen modded 4110 12 fet and a sensored 4t 3220 and for the first few runs, it did ok on the bike too. I had the bike geared too high so acceleration wasn't too great but between 15 and 45ish was pretty fun and the controller was pulling 40 battery amps, measured on the CA through a calibrated shunt. I got a new drive sprocket and reduced the gearing to a 30mph top speed and went on a run and the controller started cutting out when I rolled on the throttle. At first, I could roll it on slower and get up to top speed but after a few runs up and down the block, movement of the throttle would just make the motor spin up for a second then shut down. I took the controller apart and everything looked and smelled perfectly fine inside. No fets were shorted and there was no evidence of anything overheating. The controller got warm to the touch while it worked but I could hold my hand on it so it wasn't too hot. So I put it back together and thought I'd reprogram it, maybe drop the lvc and current limits a little in case they're off or something......I tried everything, including block time, using up to an eb-36 board setting, every combination of phase currents, etc. I've thoroughly checked the motor wiring, throttle, monitored for voltage stability during the cutouts, checked battery connections, etc. At the time of the pics, everything was hooked up with crappy 30A crimp automotive connectors and 45A andersons but it's all been changed to 8 gauge wire with big castle bullets since.....and the controller still cuts out.

So with everything I could think of done in the programming and after checking out everything else on the bike out, it was time to go inside the controller and do some work. I've wanted to do some real upgrades to it for a while now but haven't really had the time. Rather than mess with it for much longer, I decided to give it everything I've got and just build a good controller out of it. I started by adding a 500 ohm resistor across c20. Tested it....still cut out. Then I increased r43 in six stages all the way up to 47k....tested it....still cut out. The next step was to remove QF1. That's a tiny little booger! The controller still cut out when rolling on the throttle with very conservative settings so I took it apart again, pulled the fets from the board and tested them, tested the caps(as best as I could), added a couple of extra shunts, modified the board to accept some really nice 8 gauge power wires. I left the ten...ish gauge phase wires because I didn't really have an appropriate replacement and the wires going into the 3220 aren't any bigger. I also ran lines in to the speed switch taps. And best of all, I went nuts with the soldering iron and solder wick and really beefed up the board traces. I noticed two traces on top that I haven't seen anyone else pay attention to that feed power to two of the fets so I built them up a bit too by drilling a small hole next to the fet leg and running a few strands of the braid up through it, then soldering it to the board on top.

I set the programming as an eb-36(4 shunts right?), set battery current to 30 and phase to 75 and took it for a spin. All of a sudden it's definitely a much more interesting controller, almost lifting the front wheel before.......it cuts out and I have to let off the throttle to reset it and ease it back on. So far, in a few runs, I did one 2600 watt pull from 10 to 30 that was pretty cool but the cut out thing is terrible.

Anyway, here are some pics I took of the bike before I started today. I wanted to get some in case I got too fed up with the controller and decided not to mess with putting it back together! I also swore I'd never build a bike with a bunch of duct tape but there's wood under it at least and it's only temporary until I build a nice battery box for it :wink:

If someone's already figured out the cutout thing and I've just missed it, please let me know. I've searched and soldered my eyes out.

Should I even bother putting QF1 back in there? It obviously doesn't have anything to do with the problems I'm experiencing.

Here are some pics:

 

[Burtie]

mdd0127 wrote:
...I got a new drive sprocket and reduced the gearing to a 30mph top speed and went on a run and the controller started cutting out...

Sounds like the problem only started when you changed the sprocket to achieve lower gearing. The lower gearing would not cause this, so something else must have changed at the same time...

Are you sure your rotary encoder didn't get accidentaly moved or damaged in some way during this operation?

 

[mdd0127]

I pulled the chain off and tested the encoder position by checking no load amp draw and it's pulling 3.00 amps spinning the reduction drive at full no load speed. Without the drive attached it's around 1.4 so that seems right. I'm going to try shifting it a tiny bit each way tonight if I'm able to. Since the e-bike thing is so experimental, we've been taking on landscaping work to make ends meet....woo hoo....

If I had my way though, I'd probably be in the shop 14 hours a day so it's probably good that I'm getting some sun and exercise! Drives me nuts to be pulling weeds when I've got controller issues to sort.

Whatever the problem is, it doesn't have anything to do with QF1 because even with it removed, it still cuts out. I'm convinced that it's something in the programming. Even with all of the explanations of the parameters online, I'm still not sure we really know what some of the fields do....then I'm not sure on top of that if the hacked versions even have correct translations.

 

[John in CR]

1.6A just to cover wind resistance of the stationary wheel and the chain and reduction(s). About 50W. That's interesting and the first time I've seen someone quantify it, thanks.

I ran into the same cutout problem with a shunt modded 9fet 4110 controller, but it was launching one of my hard to drive speed hubs that killed a Methods 18fet controller at nearly the same power that killed the 18fet. Needless to say it only took a few resets on flat ground launches before the 9fet controller fizzled. It took being stuck behind too slow traffic going up a mountain to kill the bullet proof 18fet.

We need to find and do a group buy of big toroidal cores so we can wind some minimal loss coils to protect our controllers from these low resistance wound motors. Count me in for at least a dozen if the price is right. In the meantime, I think you need more controller for your 3220.

 

[Thud]

md,
Interesting you are running with the QF1 removed. I have attepted it twice now wih the outrunners & the net result was 2 blown controllers in less than a minute each. I can only speculate your lower pole count is helping.

what is the spec on your motor? tuns/termination.

I have found I can get the 12fet units working acceptably, but they really perfer my motors in Wye over the delta stockers.
I also really hate all the bad JuJu with trying to fool the brains with programming with different boards. It has never yeilded a net gain in my experiance. the throttle cutting allways reapears & at the same level of performance.

I would rather let the brain deal with the reality of the fet count & goof the hardware to get the results I am looking for.
on one set up i was getting a throttle cut out not under load but with higher rpm's
It would pull alright but when it was near top speed it would cut-out...upshift & it would pull again untill unloaded & cut again (??? i can only speculate it was some frequency/feedback related anomoly)

which is what you describe is happening with the lower gearing you changed to.
Since you have allready lowered the shunt resistance, I really have no idea what to suggest other than my steps to finding the limits of a particular controller.

I just program for the board I'm running, set the battery current to some thing you think will be your max amp draw at full speed, unloaded (I use 50 or 60 with the stock shunt) then set a low phase amp & test ride the set up....then I only adjust the Phase amps settings untill I either get the accelration rates I expect, or the throttle starts cutting out.

If the controller is still running cool & I think it is still too weak....only then will I start to goof the shunt values.
I will just solder them up a little & drop the phase settings to start the test riding stepping up proccess again.

I have nothing other than the "seat of the pants" to go by but I seem to get better performance from the 3077's than the 4110's...to be fair I blew all my 4110 loaded controllers to smitherines in the testing proccess....the 3066's i tested with also ran well....but one of those just stopped working on the bench for no clear reason.....& the other was a victom of the QF1ectomy I subjected it to.
definatly the bigest step was getting the hall placment at a correct 120deg spacing & clean signals from them (isololated from the motor phase wires as much as possible)
T

 

[mdd0127]

Well, I'm done landscaping for the day and as soon as my fingernails feel like they're going to stay on and I stop hallucinating from the heat, I'm going to figure this thing out. I've been really thinking hard about the whole process and I'm convinced it's something stupid. I loosened up the slipper clutch last thing last night to where it slips pretty easily and it still cut out pretty easily....even when only pulling 550 watts..... It's almost like the throttle is bad but only when loaded. On the bench, it puts out smooth clean voltage up to 4.2 ish volts but it sure seems like the control section of the board is starving for power or something when the fets start pulling hard. I need to figure out a way to load it up on the bench so I can watch what's going on. Other than that, it looks like I need to put an HV160 and RC CA on there and move on with life......start saving up for a sevcon....

I'm either going to fix it or blow it up. I'll post pics........

 

[Arlo1]

Video is always fun. I dont think sevcon will happen unless someone can get you into a programer for it cheep. I blew up lots of bits today too, if anyone knows how the 5 volt regulator works on a 6 fet let me know.

 

[mdd0127]

Well, I don't know what to think. I've done everything I can think of to get this controller to quit cutting out and it cuts out at around 22-27A every time. Since my last post, I tested the throttle again, replaced it with another brand new known working one, rewired the battery pack with 8 gauge and 6.5mm bullets, 120A breaker, and 100A shunt, set the lvc to 19V, tried every different combination of programming parameters, reinstalled QF1(not easy), reset and tested the sensor position many times, in many different positions, and tried a different 12fet controller with a sensorless module. The sensored 12-fet spins the motor up smoothly with a very low no load draw until it cuts out but the sensored/sensorless one won't even start the motor with the halls plugged in. Without the halls plugged in, it starts pretty well for the most part and runs fine except for a full throttle at full speed cutout here and there. I'm really confused. Logic says that since the sensored/sensorless controller won't run it with the sensors that something's wrong with my sensor module......but, it spools up smoothly on the sensored only controller.....

It's weird that it's cutting out at roughly the same amperage.....

 

[Arlo1]

Well, I don't know what to think. I've done everything I can think of to get this controller to quit cutting out and it cuts out at around 22-27A every time. Since my last post, I tested the throttle again, replaced it with another brand new known working one, rewired the battery pack with 8 gauge and 6.5mm bullets, 120A breaker, and 100A shunt, set the lvc to 19V, tried every different combination of programming parameters, reinstalled QF1(not easy), reset and tested the sensor position many times, in many different positions, and tried a different 12fet controller with a sensorless module. The sensored 12-fet spins the motor up smoothly with a very low no load draw until it cuts out but the sensored/sensorless one won't even start the motor with the halls plugged in. Without the halls plugged in, it starts pretty well for the most part and runs fine except for a full throttle at full speed cutout here and there. I'm really confused. Logic says that since the sensored/sensorless controller won't run it with the sensors that something's wrong with my sensor module......but, it spools up smoothly on the sensored only controller.....

It's weird that it's cutting out at roughly the same amperage.....

Maybe you have arcing someware in the motor like a winding with insulation pealed off. Other wise can you moniter the amps with a shunt or something on the phase side ? If so try all three phase wires.

EDIT: I experianced this once and it was a bad conecton in the phase wires!!!

 

[Burtie]

You mention a 120A breaker in the battery connection. Not clear if this Is it just a switch, or some electro- mechanical over current protection device?

If this component is inductive, it could likely trigger the premature controller shutdowns.

Burtie

 

[mdd0127]

I added the breaker as a safety measure. The controller cuts out without it inline. I don't think the issue is with the motor because I got the sensorless 12-fet to run it at 65 amps pretty reliably. Startup sync sucks and it will still cut out here and there at full throttle when cruising though.

When I got the Astro, one of the phase wires was almost shorted to one of the screws holding the endcap on. It had rubbed through the insulation during shipping so I fixed that before ever firing up the motor. All three phases have the same resistance when measured at the fets with the controller attached and nothing is shorted to ground.

I might mess with it more today.......or I might just go on a pedal ride and forget about electric bikes for a while.

Why is so hard to get a reliable, high powered controller?

Has anyone looked into grafting a Mamba Max brain onto an HV160 to make a sensored HV160?