Crystalyte Controllers - Repair and Modification information

[mikereidis]

it is indeed an older V1 20A pedal first controller. i don't have one of those handy here to confirm what i suspect.

i think that the early pedal first controllers lack an item crucial to implementing a sensored controller in a simple way. they don't have a commutation chip to decode the rotor position and translate that to the drive signals. instead that is performed by some discrete analog chips and transistors.

the chip that i am thinking of is the C1246 chip. and it exists on the V1 20A and 40A controllers. it is in a surface mount SOIC-16 package. check to see if you have one of those on the board. if there is one than there is an easy place to connect the hall sensors to the controller and bypass the BEMF circuits. if it isn't there then the question of how to connect the hall sensors becomes much more complicated..

rick

Yes, the 1246 is there, underneath the board, properly wired to the hall sensor holes where a connector is on that immediate start board. I think both boards are virtually the same. Hmm, with immediate start controller, can it do pedal first if halls are disconnected ? If someone could confirm that works, then I'm pretty sure no jumpers are on the board to cut or install.

I'm too excited now and will try this mod tonight. Getting REALLY tired of turning bike down on a hill (When one of my batts BMS trips uphill) and swinging back up to restart.

[ngocthach1130]

Guys, I have the new white board V2 controller. I upgraded to irf4110 and 200v caps long ago. Recently i upped it to 100v battery source. Ran fine for a little while. The other day while up the hill, I popped a capacitor.
It was the C8 capacitor at the bottom edge of the board with the fets on the top edge. what is that capacitor and what is the value so i can replace it. Also a trace from the C12 capacitor leading to a surface mount part close to the fets is blown. It the copper trace broke but could be mended with a drop of solder. Controller still run...Weird. Should i be worry?

[mikereidis]

SolarBBQ:
This IS a Crystalyte controller, bought from ebikes 1 month ago. 72v, 20a, pedal first. Sort of an oddball in the lineup IMO, because it's a 72v unit but only rated for 20 amps ? Seems funny to have 12 4110 fets but only 20a limit.

These pedal first 20A controllers only have 6 4110s in them. I completely soldered my shunt and am seeing 55-60A spikes in the CA with 40ish sustained (which is all by battery pack is good for sustained wired 2p) and the controller is vibrating like crazy under full throttle until it gets up to speed. Don't expect it to last long, but I'm sure having fun in the meantime.

Mine has 12. 6 on top and 6 on bottom. Unless there are 6 other FET looking things there, but I doubt it, and board traces tell me they are probably FETs. I can't see numbers too easy. I'm wondering if there is a 35 or 40+a version of this board with nothing different except current shunt. 1 board makes for lower production and inventory costs, even if the extra FETs cost a few bucks in volume.

I'm still REALLY surprised a 20a controller has twelve 4110 FETs. I imagine it should be easy to run this at 30-40a with only a shunt mod.

[pwbset]

Mine has 12.

I got mine just this past May? Sure looks the same as yours. I most definitely only have 6 FETs.

DSCN3709.JPG (121.79 KiB) Viewed 2225 times

[mikereidis]

Mine has 12.

I got mine just this past May? Sure looks the same as yours. I most definitely only have 6 FETs.

I got mine one month ago from ebikes for $145 + gst + shipping. They didn't and still don't seem to have the immediate start version of this controller. C7220PF http://www.ebikes.ca/store/photos/C7220PF.jpg They seem to have some in stock again now: 8 allegedly.

Aren't these 6 additional FETs under the board? Yes, there is a voltage reg or something sandwiched on top of left FET and something on right FET.

Here's a side view. Screw on right is or first FET with VReg (?) sandwiched on top. That's the commutator chip you see under the board there.

Screw at left is for rightmost of these 6 FETS:

[mikereidis]

I connected the hall sensor lines, and powered the halls with 5 volts, since that's what the spec sheet seems to recommend.

Immediate start did not work. When pedal start engaged, the motor made a few noises but wouldn't spin. I disconnected the 5 volt supply and it returned to normal pedal first.

I'm thinking there must be (a) jumper(s) or resistors to cut or something. I'm thinking the motor 3 phases (at FET connection) are conditioned and fed to commutator. This would explain my 7 KMH pedal first start speed perhaps, since my motor should generate about 5 volts at that speed.

So I think I just have to interrupt the back EMF signals from the motor. Clearly, this board was designed to be usable for PF and IS. So I know this is probably possible, IMO.

[rkosiorek]

you probably have to remove the 6 resistors used for the BEMF sensing. they would be confusing the hall sensors. not sure but i would suspect that the resistors are the neat group of 6 resistors placed on the left edge of the board just below the big capacitor.

the hall sensors would also need a pull up resistor to the 12V supply going to the chip. on my IS controller these resistors are 18K.

since you are the only one with a board could you trace out the schematic of the pin connections going to pins 1,2,3,4,5 and 6 of the C1246 chip. pin 1 and 2 are used for the yellow hall, 3 and 4 for the green and 5, 6 for the blue.

rick

[fechter]

It's hard to see on the v.1 board I have, but it looks like the hall sensors go to pins 2,3, and 6 on the commutator chip. Pins 1, 4, and 5 seem to be tied to something else (?).

Correct about removing the BEMF sensing resistors. The hall signals need pull-up resistors to the Vcc of the commutator, which would be pin 8. I think it would work if they were pulled up by the +5v supply too.

Here's a picture of an immediate start 20 amp board (from somewhere else on the forum). This is what you want to copy...

20A controller top side.jpg (108.13 KiB) Viewed 2165 times

20A controller bottom side.jpg (96.67 KiB) Viewed 1850 times

[docnjoj]

Richard! What are the 2 empty sockets for? (white male 2 pin and 3 pin)
otherDoc

[rkosiorek]

the hall sensor schematic is identical to the one used for the 40A version. just the identifying Rx and Cx numbers are different. so here is a copy of the schematic for the hall inputs of the 40A version. the capacitor values are a guess and probably wrong. i also could not see the identifying numbers for the parts.

40A hall input section. same as 20A EXCEPT Rx and Cx numbers different. also Caps marked 104 are probably a lower value

V1 - HALL SENSOR INPUT.jpg (27.65 KiB) Viewed 2154 times

V1 - HALL SENSOR INPUT.pdf

40A hall input section. same as 20A EXCEPT Rx and Cx numbers different. also Caps marked 104 are probably a lower value

(12.26 KiB) Downloaded 126 times

pins 1,4 and 5 are connected to a voltage divider to set the trip point of the sensors. the inputs for each sensor go to a comparator built into the C1246

Here is the layout of the board connectors:

DSCN0912.JPG (94.93 KiB) Viewed 1836 times

and what i suspect are the BEMF resistor dividers. at least they look to be connected to the phase outputs.

bikedctrlr 004.jpg

(112.52 KiB) Downloaded 425 times

rick

[mikereidis]

you probably have to remove the 6 resistors used for the BEMF sensing. they would be confusing the hall sensors. not sure but i would suspect that the resistors are the neat group of 6 resistors placed on the left edge of the board just below the big capacitor.

the hall sensors would also need a pull up resistor to the 12V supply going to the chip. on my IS controller these resistors are 18K.

since you are the only one with a board could you trace out the schematic of the pin connections going to pins 1,2,3,4,5 and 6 of the C1246 chip. pin 1 and 2 are used for the yellow hall, 3 and 4 for the green and 5, 6 for the blue.

rick

Thanks Rick ! Yes, I suspected that neat group of 6 resistors. I will trace connections tonight.

I checked the hall sensor holes 2 days ago to confirm for myself it matches C1246 pinout:
A: goes to pins 1 and 4
B: to 3 and 6
C: to 2 and 5

So it appears to me it differs a bit from the schematic you show for 40a version. Instead of a voltage divider giving about 5 volts on negative side of comparators, these hall sensor lines seem to be setup in a differential mode. Perhaps this gives better EMF/noise immunity ? BTW, to hookup I used a few feet of cat5 network cable with 4 twisted pairs and each hall sensor pair is ground and sensor twisted.

Hall sensors need pullups ? If so, I'd think they'd be on the board. I will check.

If the immediate start version still has those BEMF sensing resistors, then I'd hope there was some way to disable them (or just disable their connection to C1246) without removing/destroying them. There MUST be a jumper somewhere or I won't be entirely happy. Would be nice, but not essential, if pedal first still worked if one or more of the hall sensors go bad.

[mikereidis]

It's hard to see on the v.1 board I have, but it looks like the hall sensors go to pins 2,3, and 6 on the commutator chip. Pins 1, 4, and 5 seem to be tied to something else (?).

Correct about removing the BEMF sensing resistors. The hall signals need pull-up resistors to the Vcc of the commutator, which would be pin 8. I think it would work if they were pulled up by the +5v supply too.

Here's a picture of an immediate start 20 amp board (from somewhere else on the forum). This is what you want to copy...

Thanks Fechter ! I note that the board you show has a somewhat different layout than mine, but all the major components are in about the same locations.

Should the phasing of the 3 hall sensor wires really matter for testing ? I'm trying to recall, but I think if I hooked them up wrong, it should still work, but just go backwards ? Or do I have to try all 6 combinations to see which one works for forward ?

[rkosiorek]

only one combination of hall and phase wires will work properly in one direction. and one will work in the opposite direction.

the other combinations will either not work at all or work very poorly drawing a lot of amps. the other combinations most likely will not start the motor at all unless the motor is first given a spin. it would be safest to test using a current limited power supply.

the input differential amplifiers/comparators are wired entirely differently to use for BEMF zero crossing detection than they would be for Hall sensors.

the hall sensors report a the actual position of the rotor. it is a logical function. so one side of the individual comparators would be tied to a reference voltage. the pulled up output of the hall sensor will be either above or below this reference and connected to the other side of the comparator. this signals an on or off condition. plain and simple. noise isolation is provided by the filter formed by the pull up resistor in series with the capacitor on each hall output. this is a digital approach.

BEMF detection is analog and a totaly different approach is used. the (-) input of the differential amp/comparator is tied to resistive divider that produces a floating "0" reference point that is half the voltage between the lowest and highest phase voltages. this is compared to a measured BEMF voltage. as this measured BEMF crosses the floating "0" the rotor position can be detected. there are several techniques to do this. each of these techniques uses a lot more parts than the Hall sensor input would. the wiring of the input amps is analog and specific to this arrangement and entirely different from the digital configuration used for hall sensor position reporting.

thinking that you can just simply attach digital hall sensors to inputs configured for analog BEMF detection is flawed and shows a complete lack of understanding how each technique works.

in simplest terms. all of the parts used for BEMF detection would have to be removed from the board. the inputs to the C1246 would have to be rewired to be something similar to the schematic that i supplied for the Hall sensor approach to work.

this is the reason why the boards look different. one method using the Hall sensors uses 6 resistors and 3 capacitors attached to the inputs of the C1246. the other uses a dozen or more resistors and a few more capacitors to connect the measured analog phase voltages to the same inputs of the C1246. that is why one board has a connector and a couple of parts and the other does not have the connector and a lot more passive parts.

if you want to use both on the same controller 2 different and distinct input sections would have to be designed. then some method would have to be made to select which of the 2 inputs is being used. jumper blocks? DIP switches? logic circuits? something has to be there to select which you use.

what some people have done is to build a separate analog BEMF detector circuit. this detection circuit is designed to output a digital on/off signal compatible with the Hall sensor inputs of the Immediate start controller. since the hall sensor input is digital this is easy to do.

what you are trying is fundamentally different. BEMF is an analog technique and trying to connect a digital signal to an analog input is much more complicated.

rick

[mikereidis]

only one combination of hall and phase wires will work properly in one direction. and one will work in the opposite direction.

thinking that you can just simply attach digital hall sensors to inputs configured for analog BEMF detection is flawed and shows a complete lack of understanding how each technique works.

in simplest terms. all of the parts used for BEMF detection would have to be removed from the board. the inputs to the C1246 would have to be rewired to be something similar to the schematic that i supplied for the Hall sensor approach to work.

this is the reason why the boards look different. one method using the Hall sensors uses 6 resistors and 3 capacitors attached to the inputs of the C1246. the other uses a dozen or more resistors and a few more capacitors to connect the measured analog phase voltages to the same inputs of the C1246. that is why one board has a connector and a couple of parts and the other does not have the connector and a lot more passive parts.

if you want to use both on the same controller 2 different and distinct input sections would have to be designed. then some method would have to be made to select which of the 2 inputs is being used. jumper blocks? DIP switches? logic circuits? something has to be there to select which you use.

rick

Thanks Rick. OK I'll try all 6 combos and see if it magically starts working.

I'm operating under the assumption the designers of these controllers did everything they could to keep the circuitry and board layout as close as possible for pedal first and immediate start models. Everybody does this when possible; reduces engineering, manufacturing, inventory and other costs a lot when different products are based on the same board and everyone is out to build cheaper products these days. I wouldn't be attempting this if I thought I'd have a HUGE amount of work to do. I'd buy a new controller otherwise. This mod attempt is partly for educational reasons; perhaps I'll end up with a controller that is more versatile than the stock controllers (pedal first and immediate start switch or "auto" selectable.

I have a HUGE background of hobby and professional experience with designing, building and repairing analog and digital electronics, so I'm completely up to the task. My professional experience switched to computers/software in 1989 and my eyes aren't as good for tight soldering spaces anymore, but I'm sure I can muddle through.

I DO think this conversion direction: PF to IS should be easier than the reverse (unless you use an external BackEMF to Hall sensors type output converter.) I think the back EMF detection is essentially digital (zero crossing ?), and the outputs from the C1246 are effectively digital, so I DO think this is all essentially digital: on or off. Yes, Back EMF is an analog sine wave, but one of the first things done to this analog signal is conversion to digital, with zero-crossing or peak detection or whatever.

The pictures in this post of mine: viewtopic.php?f=2&t=764&st=0&sk=t&sd=a&start=390#p93301 show what looks line 3 lines running from center tap of 6 voltage divider Back EMF resistors to area of c1246 and my "alleged hall sensor holes". I'll verify this and see if cutting the traces would help to remove Back EMF sensing.


I noticed last night that the voltages on the controller "hall sensor input" holes seemed to oscillate between 0.12 and 0.24 volts or so. I was thinking "confusion mode" (?). Later the voltage stayed steady around those voltages. I realized the second readings were with motor disconnected. I wonder if bike wobble was generating a small bit of AC voltage I was measuring. I've noticed before that it's virtually impossible to get accurate resistance readings from motor because just a bit of wobble causes the readings to bounce everywhere.

[pwbset]

Aren't these 6 additional FETs under the board?

Heatsink I believe. I'm looking forward to seeing if you have some success with your project.

DSCN3732.JPG (114.83 KiB) Viewed 2110 times

[mikereidis]

Aren't these 6 additional FETs under the board?

Heatsink I believe. I'm looking forward to seeing if you have some success with your project.

Ahh, OK, thanks.

I thought they looked a little flat and plain. The screws and the tight space fooled me.

So it has the minimum of 6 FETs, 3 high and 3 low. All is right with the world now.

[mikereidis]

I checked the hall sensor holes 2 days ago to confirm for myself it matches C1246 pinout:
A: goes to pins 1 and 4
B: to 3 and 6
C: to 2 and 5

So it appears to me it differs a bit from the schematic you show for 40a version. Instead of a voltage divider giving about 5 volts on negative side of comparators, these hall sensor lines seem to be setup in a differential mode. Perhaps this gives better EMF/noise immunity ? BTW, to hookup I used a few feet of cat5 network cable with 4 twisted pairs and each hall sensor pair is ground and sensor twisted.

Hall sensors need pullups ? If so, I'd think they'd be on the board. I will check.

If the immediate start version still has those BEMF sensing resistors, then I'd hope there was some way to disable them (or just disable their connection to C1246) without removing/destroying them. There MUST be a jumper somewhere or I won't be entirely happy. Would be nice, but not essential, if pedal first still worked if one or more of the hall sensors go bad.

Yes, halls need pullups. I tried some random 1-5K resistors pulled up to +5v but didn't work. Interesting observation: once I fixed one bad hall connection solder joint (insulation in hole ! My eyes aren't so good no more), then the pedal first still worked with the halls connected, with or without pullups. So at least it doesn't interfere (but does if one hall line breaks ?).

This didn't give me instant start however. I still think that no matter the combination, the motor should do something, even if just a growl or a bang or two when throttle twisted. Spinning the wheel a little will give all 6 combinations of 3 halls and at least one or two of those should make some noise.

Is pedal first minimum start speed enforced by a minimum AC voltage produced by motor ? Or does something measure pulse frequency to set the min start speed ? I think, at least, I should be able to replace BEMF sensing with hall sensing and have a hall driven pedal first setup. (I'm assuming I can make the phase offsets correct.) Once I do that, I can look for what to mod to reduce the start speed (to 1 KMH say, real easy to do) or eliminate the need for prior rotation.

I traced the connections to those 6 BEMF sensing resistors. The tap between them connects directly to the hall sensor holes and the comparator pins on c1246. Pins 1 and 4 are both negatives, 3 and 6 are both positives and 2 and 5 are one of each.

Each top resistor (I measured 38Kohms) goes directly to motor lead. Each bottom resistor (13Kohms) goes to negative bus/ground. So it seems like a simple 25% voltage divider, although probably more complex than that since 3 phase AC comes in here, and it's relative to all 3 phases, and not really relative to the negative bus/ground.

I could use the top resistors as pullups, and cut traces or whatever to disconnect them from motor. But it would be REALLY nice if I didn't have to cut components or traces, and still leave BEMF sensing there as a full time non-interfering thing, or switch selectable.

So I'm off to read C1246 app notes and learn what I can to go farther.

EDIT: Dagnab it ! The notes are in Japanese only: http://www.necel.com/cgi-bin/nesdis/o00 ... le=UPC1246

http://www.necel.com/nesdis/image/IEP-651.pdf
http://www.necel.com/nesdis/image/IEP-653.pdf

Datasheet is available in English but I already have that. Anyone know of English app-notes for this chip or very similar chip ?

[ejonesss]

The original 48V IS controller has an LVC at 29V.

The 33k and the R6(5.1k) will be removed to de-activate the LVC so that the controller will work from 24V battery to 48V battery for my AL1020 ebike.

i was wondering can the lvc be bypassed all together to allow 12 volt operation or even use of ni cads (witch has no problems with over discharging in fact requires complete discharging to prevent memory) just by jumping a wire from the bat+ to the resistor/diode junction? (basically shorting out both resistors)?

[fechter]

The original 48V IS controller has an LVC at 29V.

The 33k and the R6(5.1k) will be removed to de-activate the LVC so that the controller will work from 24V battery to 48V battery for my AL1020 ebike.

i was wondering can the lvc be bypassed all together to allow 12 volt operation or even use of ni cads (witch has no problems with over discharging in fact requires complete discharging to prevent memory) just by jumping a wire from the bat+ to the resistor/diode junction? (basically shorting out both resistors)?

Yikes! don't do that.

If you want to disable the LVC, you could disconnect either the diode or disconnect R6, whichever is easier.

12v might be pushing it since the voltage regulator is trying to make about 14v and needs a little more than that for voltage drop. It might work however. For 12v only, you could possibly bypass the voltage regulator.

[ejonesss]

i havent looked at the controller closely but the only voltage regulator there may be (typical of many electronics) is a 7805 to make the 5 volts for the logic power.

[fechter]

Right, there will be one of those too, but the FET gate drivers, PWM chip and commutator all run on 14v. 12v would be fine. In the normal application, there is a linear regulator to drop the battery voltage down to 14v, then the 7805 drops it down to 5v for the throttle and and some of the voltage references.

[ejonesss]

looking at the picture i could not find the diode you are talking about.

can i just connect pin 2 to ground (- terminal of battery)?

The original 48V IS controller has an LVC at 29V.

The 33k and the R6(5.1k) will be removed to de-activate the LVC so that the controller will work from 24V battery to 48V battery for my AL1020 ebike.

i was wondering can the lvc be bypassed all together to allow 12 volt operation or even use of ni cads (witch has no problems with over discharging in fact requires complete discharging to prevent memory) just by jumping a wire from the bat+ to the resistor/diode junction? (basically shorting out both resistors)?

Yikes! don't do that.

If you want to disable the LVC, you could disconnect either the diode or disconnect R6, whichever is easier.

12v might be pushing it since the voltage regulator is trying to make about 14v and needs a little more than that for voltage drop. It might work however. For 12v only, you could possibly bypass the voltage regulator.

[ejonesss]

looking at the picture i could not find the diode you are talking about.

can i just connect pin 2 to ground (- terminal of battery)?

The original 48V IS controller has an LVC at 29V.

The 33k and the R6(5.1k) will be removed to de-activate the LVC so that the controller will work from 24V battery to 48V battery for my AL1020 ebike.

i was wondering can the lvc be bypassed all together to allow 12 volt operation or even use of ni cads (witch has no problems with over discharging in fact requires complete discharging to prevent memory) just by jumping a wire from the bat+ to the resistor/diode junction? (basically shorting out both resistors)?

Yikes! don't do that.

If you want to disable the LVC, you could disconnect either the diode or disconnect R6, whichever is easier.

12v might be pushing it since the voltage regulator is trying to make about 14v and needs a little more than that for voltage drop. It might work however. For 12v only, you could possibly bypass the voltage regulator.

[fechter]

looking at the picture i could not find the diode you are talking about.

can i just connect pin 2 to ground (- terminal of battery)?

No, that will disable the current limiter too.

I think the diode is on the bottom or something. Just remove R6.

Brushless Controller disable LVC.jpg (87.5 KiB) Viewed 2130 times

[ejonesss]

what about applying the 29 volt lvc hack to the 300k resistor (looks like 1 volt per k of resistance so in theory it shourl be 9 or 10 volts )


the thing is the controller is still under warranty until november (witch fall will kick in and the weather will be too cold to ride).

if i can just jumper a resistor i can make a jumper on 2 alligator clips that i can clip onto the resistor and once the tests are done i can remove the clips and electricrider.com should not be able to detect tampering if anything goes wrong.

i did apply the abs wire for the regen so i can try out the regen braking and i simply took a piece of wire and poked it into the abs hole and did not solder it so if the controller fails requiring warranty work i can just pull the wire out and electricrider should not void the warranty.

at $250 a piece i would like to not have to pay a $75 diagnostic fee+ any repairs.

i want to find out how low i can take the voltage and still have hill climbing power because with lifepo4 batteries being so expensive i maybe able to trade off voltage for more amp hours.