CA Speed all whacked with Infineon XC846 72V Controller

[GCinDC]

Re: Infineon XC846 72V 45 Amp Controller

i just bought one of these controllers from nicobie (who got it from keywin) and i'm having trouble getting the CA to work with it.

the speedometer/odometer are off. it goes up to 20 mph fine, very accurate compared to my bike speedometer, but as soon as i accelerate over 20mph, it starts going down, and by the time i'm at 30mph, the speedometer reads zero. i could live w/o a speedometer per se, but not the related odometer and other data...

when i first got the CA (with a different controller) i modified the wheel size and poles. justin told me the gm kits have 23 poles, and that the CA only goes up to 14, so that i should set it to 14 and THEN modify the wheel size to 14/23 (of say 26"). this alteration worked fine with the earlier 72V 30A controller (which unfortunately fried at 80A). however, with THAT setting 14/23 the wheel size, the CA speed read 99mph at 20mph! so i lowered the wheel size down again by a third!

anyone have any ideas? am i opening this controller this weekend??

thanks!

greg

 

[Mike1]

Do you know where the speed wire is connected inside the controller?
Keywin supplies his controller with a blue speed wire but this is no good for the CA. Search Philf's posts, he explained it some time ago.
The SP wire from the CA needs to go straight onto one of the hall signal wires.

 

[GCinDC]

Phew, so there's a resolution..

Before I open the controller, what do I need to know? I recently opened my fried one and couldn't remove the board, even after unscrewing the fet screws, which i can't imagine ever having to put back in! Am I going to have to somehow hold the plates that the screws attached to in place with tweezers...?

thanks!

greg

 

[nicobie]

Hi Greg,

Sorry you are having trouble with the controller. I don't think I ever hooked it up to a CA as I didn't like the way it controlled my 600w BMC. Even if I had, the problem you are having wouldn't have shown up as my BMC motor needs the external magnet triggered probe equipped CA.

To get to the board just take off the end plates and the board will slide out. Be sure to take pics of where the CA wires are attached. You might send pics to Justin at ebikes.ca maybe he can tell you if they are in the right place.

If all else fails, just send it back to me and I will gladly refund the purchase price.

Nick

 

[icecube57]

Tap a hall wire..internall or externally not that sorry useless speed sensor wire. Ive been through the same thing. I just moved the wire and soldered it to the hall directly on the underside of the board.

 

[Mike1]

Remove the four end plate screws (the end with the wires coming out) and remove the five heatsink screws from the side. The heatsink will tend to stick to the inside of the case so it can sometimes take a little jiggling to break it free (some leverage with a small screwdriver may be required). The board should then slide out. Follow the hall wires to the board and solder the SP wire from the CA to any one of the hall inputs (this controller is well documented in the technical reference area).

 

[GCinDC]

cool, thanks guys. great explanation. and now i have courage...

no worries, nick. the price was right and you were honest about it. it's like this with everything...

just wondering if it's worth it to make this flash usb cable while i'm at it... or wire up the ebrake. just found out that just cuts the controller, doesn't slow me down. anyone suggest any other mods while i'm in there.

oh, like putting more solder on the shunt? i read about that today...

 

[GCinDC]

okay, the plot thickens. the top photo is the knuckles from an earlier post with the CA pinout. mine is below. totally different colors. black in place of red!

anyway, what do you guys think about the isense soldered to the shunt?
and the digital and analog connected tot he same spot?

so should i pull of the isense wire and solder it to one of the hall spots close to the right side of the board?

 

[GCinDC]

ps. should i solder the yellow isense wire to the bottom of the yellow, green or blue, shown in the bottom right of the image. does it matter, as long as it's not the red or black? could it be on the same signal as the orange hall wire?

 

[Mike1]

Don't make any assumptions about the wire colours. There are a few different versions of the pigtail connector. The ISense is the current sense and should be left on the top end of the shunt. Trace the wire that goes to SP on the CA, that one should go to any of the hall connections.

Both GNDs should be connected to the bottom of the shunt.


I would wait until you have a fully working CA calibrated to the existing shunt before soldering it down any, you need to know where you are starting from to avoid possible damage to the controller, motor, or battery.

 

[GCinDC]

Okay, the colors are irrelevant. That's fine.

Vbatt, Hall & EBrake all match knuckles photo.

But is the "CA SP" the same as the ISense in the pinout diagram? If so, that one is soldered to the shunt. You're saying I should leave it there.

And remove the Digital & Analog wires and put them on the opposite end of the shunt?

Now I'm scared.

<dashes off to technical reference area to search for documentation..>

 

[Mike1]

Both GNDs stay at the bottom of the shunt. This is connected to the board GND. Best to keep the GND wires together and as close to the shunt as possible.
ISense stays at the top of the shunt. This senses the voltage across the shunt and tells the CA how much current is flowing.
The CA SP is the one labelled HALL. It's labelled SP inside the CA.

 

[icecube57]

weird... My Isense is on the the top of the shunt. I actually have my ground tied together but attached on the other end of the board by the halls somewhere. This is where knuckles and keywin put their grounds at. This may account for the "accurate reading" but "idle current drifting" after the controller has been under a HEAVY load it doesnt display the correct idle current load and its almost always negative but the CA consumed amps is always right within 1%.

 

[GCinDC]

This is where knuckles and keywin put their grounds at.

Not according to knuckle's photo. the black digital grnd wire looks like it's going over toward the shunt. and the analog gnd is on the bottom of the board, at the bottom of the shunt, as in (from tech ref area Infineon Controller Hardware Settings):

i see a d This may account for the "accurate reading" but "idle current drifting" after the controller has been under a HEAVY load it doesnt display the correct idle current load and its almost always negative but the CA consumed amps is always right within 1%.

"accurate reading" of which? amps or mph?

Both GNDs stay at the bottom of the shunt....

Mike1, so you're saying I should remove the two GND's from the right side of the board, near the halls, and wire them both to the bottom of the board, bottom of the shunt?

And I should leave the one wire going to a hall, and the others, going to Vbatt & ebrake should be fine, no?

thanks, guys

 

[Mike1]

Yes, move the GNDs to the bottom of the shunt. The ISense GND has to be there and after reading the 18FET thread it appears that current thinking is that both GNDs should be close together.

I don't really know much about the Ebrake line from the CA (is this the version 1 CA?) I know about the throttle over-ride on the version 2 but don't know if this is the same thing.

 

[GCinDC]

okay, will do. thanks for following up. i just don't want to damage anything.

since the wires are now too short, i'll have to splice in extensions. could i not just solder in ONE extention and splice/solder both GND wires to it?

 

[Mike1]

No! The reason there are two GND wires is that as the CA draws power there is a voltage drop across the supply wires (both positive and GND wires). This voltage drop would affect the very small voltage that the ISense detects across the shunt. They must have their own wires all the way to the board.

 

[icecube57]

My CA was accurate with my Amp reading with the I Sense being at the top of the shunt and the ground being across the board by the halls. But the idle current was never right. It was correct on first boot with a cold controller but after riding it around a bit i found it was always drifing around. The other guy mentioned that the ground needs to be on the bottom part of the shunt. This would probably correct the wandering reading. Even when its wandering around it still was pretty accurate to less than 1% on my CA.

 

[icecube57]

Funny that some areas on the board say twist them both together and find a ground to solder it on to and some others say that you have to go to put it specifically on the digital ground and the analog ground.

This is my first time reading about the bottom of the shunt for placing the CA ground wires. I think Doc Knuckles or methods or Dmnum need to comment and set thing straight.

 

[GCinDC]

soldering operation underway. big day for honey-dos, slow day for ebike. should be able to test it tomorrow.

but for those following this thread, i thought you'd be interested in a response i got by PM (and i hope he doesn't mind my sharing):

Hi Greg!

Let me say up front that I have YET to find a truly reliable speed signal from the Infineon controllers, as shipped. I have played with a number of different mods in an attempt to overcome this, but have to confess... I haven't found a perfect solution.

At first, it seemed that simply dividing the signal that the controller offered up on its blue wire (in order to get it down to where a device like the Cycle Analyst could count it) was an obvious fix. The signal on this wire appears to follow each PHASE of commutation, rather than each CYCLE. Therefore, there are six times as many pulses (for a given pole count) than the CA expects. A 23 pole motor, like a GM or Nine Continent, will produce 138 pulses per rotation... more than you can even program into the CA to measure (the latest firmware allows a maxium value of 99, and besides - the interrupt service routine doesn't really want to handle 6 times as many events as it was designed for).

I built a circuit that was supposed to GUARANTEE that we're measuring real wheel rotation by picking off the 1st and 4th phases of commuation and reporting ONE event each time this is cycled. The circuit works brilliantly - EXCEPT with an Infineon that has regen enabled. The Infineon, for whatever reason, creates a lot of noise on the Hall lines when in regen mode. Consequently, your apparent speed/distance appears to INCREASE when you hit the brakes. I haven't completely gotten my head around WHY this happens yet.

I haven't spent a lot of time on this problem, but intend to come back to it. I have a lot of other pet peeves about speed/distance measurement on e-bikes that I keep intending to document, but need to spend some more time collecting the empirical data to support my assertions

Cheers!
Phil

 

[GCinDC]

okay, i spliced in extensions for the Analog & Digital GND wires and soldered them to the bottom of the shunt (bottom of the board). this is what you suggested I do, right Mike?


sadly, the speedometer reading on the CA performs as before (in lifted wheel test):
accurate up to 20MPH
as wheel accelerates above that speed, the speedometer GOES DOWN, and at top speed, full throttle reads 0.00 MPH.

the Amps reading looks pretty accurate. Full throttle w/ hard brake got it up to 30A.

now that i'm in a testing mode, what else should i try?

did someone suggest putting a GND on another HALL?

in knuckles pic, he shows the analog GND going to the bottom of the board, bottom of the shunt.
in the top view of the controller, he indicates the digital GND (black) but it's not clear where it's going. is it on the shunt too, on the top side?

somebody help!

 

[dnmun]

you should not run the motor at WOT and hold the brakes on to test the controller. that is bad luck. everybody does it. if the wheel stops during this WOT, then it can overload the FETs. imho.

by the bottom side of the shunt, we meant electrically.

'bottom' side of the shunt is the side closest to the battery, and top of the shunt is separated by the delta V across the shunt which the CA uses to calculate the current flowing through shunt.

where you soldered is the bottom of the shunt, but it could be on the other side of the pcb where the black wire from the battery is attached at the 'bottom of the shunt'. where you soldered it is ok though, since it is on the battery side of the shunt.

i don't know why the speedo reading is dropping off though. if a search turns up nothing, maybe ask justin.

 

[Mike1]

So where is your SP wire connected now?

 

[GCinDC]

I don't know about the SP, but the ISense is where I left it, on the top of the shunt.

 

[GCinDC]

I PM'd Justin last night. No word yet. Hopefully he's had some experience with these controllers.. We'll see.

you should not run the motor at WOT and hold the brakes on to test the controller. that is bad luck. everybody does it. if the wheel stops during this WOT, then it can overload the FETs. imho.

note taken. the wheel didn't stop. i just slowed it down with the brakes to see some AMPS. <phew>

by the bottom side of the shunt, we meant electrically. 'bottom' side of the shunt is the side closest to the battery, and top of the shunt is separated by the delta V across the shunt which the CA uses to calculate the current flowing through shunt.

the good news is that two weeks ago, i had no idea what a shunt is. so i'm learning something.

what i don't understand now is how such metal (m shaped) wires would have a delta V across it unless it's resistive, in which case what's the point of soldering it up (to increase it's conductivity)... in any case, if it does have a delta V across it, i'd have to imagine that the delta V would be less between two relatively close spots along the shunt (like the peaks of the M's) than two further spots (like the feet of the M).


My M shunts are pinched together and somewhat soldered up (connected at the peaks), so could this cause problems with the V delta reading? should i try to solder the wires at the base of the M, to get a larger delta V reading?

and please forgive me, if it's not obvious, that i'm a total hack...

from the technical reference area i found this. it is current wisdom, no? everything seems (to me) right. the only other thing i could do is move the Isense wire to the base of the M or the bottom of the board.

1) Vbatt - This is the battery (+) voltage supplied to the CA unit. The CA only draws a few millamps of current.
The CA should draw full battery voltage (current) from the switchable "ignition" wire AND NOT the main positive (thick) power wire.
The CA unit draws power from this tap and also measures the battery voltage.
A fuse or poly fuse is reccommend on this connection point.

2) Digital Gnd - This is the battery ground (-) for the CA unit.
It can tap into ground anywhere along the gound bus or other ground wires.
It should NOT tap into the high side of the controller shunt, however.

3) Analog Gnd - This connection should be as close to the ground (-) shunt side as possible.
It is used in conjunction with the ISense to measure the millivolt drop across the shunt.
It can be attached from the top or the bottom of the pcb.

4) ISense - This connection should be as close to the high (+) shunt side as possible.
It is used in conjunction with the Analog Gnd to measure the millivolt drop across the shunt.
It can be attached from the top or the bottom of the pcb.

Note: The mV drop across the shunt is an indirect measurement of the current thru the shunt.

5) Hall - This connection is to the amplified hall signal (the blue wire).
It provides a signal for every hall transition (of one motor hall sensor) during rotor rotation.
The CA unit uses the number of hall transitions (per time) to calculate rotor rpm.

6) Ebrake - This is the same connection as the switch-type handle bar ebrake.
IT IS NOT the high voltage ebrake wire used in conjunction with brake light systems.
This is used by the CA unit to increase the LVC above the pre-programmed LVC of the Infineon.