Version 2 Crystalyte Controller information

rsisson

100 W
Joined
Oct 18, 2007
Messages
124
Apparently my crash did more damage than I thought. Got the hall effect wires fixed powered it up and it seemed ok, motor runs...

Put it on the Ground, apply power and THUNK on power up, and nothing more. It just goes THUNK, and then nothing...

Ok, this is obviously NOT a warranty issue, so I open the controller. Oh, Oh, I fried two surface mount resistors and a fet or so... but good...

Whats interesting is my board looks NOTHING like the boards posted earlier.

Looks like it has room for another set of FETs... no connectors... and lots of other changes ....

I have ordered a NEW controller as I can't see well enought to work on things like this anymore... who might be able to repair this one for me?
 

Attachments

  • PA270018.JPG
    PA270018.JPG
    96.4 KB · Views: 15,444
  • PA270019.JPG
    PA270019.JPG
    68.5 KB · Views: 15,377
Hay rsisson, welcome to the club!

I've had 1 Motor blow, and 2 controller blows since august.

My bikes ran a week since I paid for it! :roll:

Same style controller as yours though.

(By the way, 2010 sucked, don't watch it.)
 
here are the circuits for the 15V regulators used in the Crystalyte controllers. first the one used in the original analog controller.

15V REGULATOR.jpg

and now the switching regulator used in the newer V2 controller

Low Voltage Switching Supply.jpg

hope this helps.

rick
 
Managed to get the controller working - two of the mosfet insulating washers needed to be replaced.

Here are the controller codes from Justin if anyone needs them.

V2 Controller Indicator Status Instruction

 Indicator light off: Normal working
 Indicator light on: EABS braking status
 Indicator light on for 0.5 second - flash 1 time - off for 1 second: Standby status
 Indicator light on for 0.5 second - flash 2 times – off for 1 second: Brake signal
 Indicator light on for 0.5 second - flash 3 times – off for 1 second: MOS damage
 Indicator light on for 0.5 second - flash 4 times – off for 1 second: Immediate start at high speed protection
 Indicator light on for 0.5 second - flash 5 times – off for 1 second: Electric current failure
 Indicator light on for 0.5 second - flash 6 times – off for 1 second: Power supply’s low voltage protection
 Indicator light on for 0.5 second - flash 7 times – off for 1 second: Hall effect signal failure
 Indicator light on for 0.5 second - flash 8 times – off for 1 second: No throttle signal
 
The first lot of the new controllers I recieved all had only 6 mosfets of 4310 type, all failed with mosfet failure with some other components also damaged. I would see as a warranty issue as 6 fets is not what the board is designed to use. Perhaps some cost cutting by using only 6 fets, with 12 fets is no problems. The board is designed for 50amp, that means 25amp through each fet pair legs, the controllers that I had fail with single fets the high amp leg of a fet ,near to it, the thin section of solder had blown ( not designed to handle over 25amps). Unfortunately customers have been used as beta testers with the 6 fet controllers, they have now gone to 12 fet versions which should have been used from the beginning.
 
getadirtbike said:
Hi,

I started reading through the thread that you started about reverse engineering the Crystalyte controller but I seem to have lost my way there somewhere.

I'm just after a quick and simple answer to a couple of questions.

Firstly,

Exploded caps_2.jpg

Now I would assume that every part of the controllers is important.

I had this happen to one of my 12 mosfet 4310 controllers the other night which was relatvely new. I was running at 40A and 72V which I thought was relatively safe. Surprisingly the controller still works fine. I can't notice any difference in performance, no difference in the take off, and no difference anywhere else throughout the power delivery. The 2 large caps were actually detached and floating around inside the housing... along with a few bits of solder.

Anyway, I changed the controller and put another one new one in there....I rode it for about 5 minutes and then took it out of the case. the exact same thing had happened, but the controller was still running fine.

So my questions are..... what the hell are those Caps for?

and if the controller still works, why are the in there?

also, what will happen if I just throw out the caps and keep using the controller?

I haven't noticed any difference in the operation of the bike.... so I'm a llittle confused.

I'm a mech engineer so if you can answer in laymans terms I'd greatly appreciate it.

You're the first person I thought of when I saw this thing so I'm hoping that you can help.

Regards,

John

Yikes! That picture should go in the "failed components" thread.

Those are your main capacitors.

Don't run the controller without them! If it still works, that's amazing.
It's possible you have a good enough connection to your batteries that they are performing that function and preventing the controller from blowing.

The voltage spikes from the motor windings get circulated to the capacitors during the switching cycle. I the caps are gone, the voltage spikes can easily exceed the rating of the FETs, causing them to blow instantly.

I'm not sure what the smaller one is. That one bothers me a bit, but if it's in parallel with the big ones, then it would be understandable.

The kind of explosion like that can happen in several ways:

Reverse polarity; obviously not the case here.

Overvoltage; Can't read the numbers on the side of the caps, but I assume they are over 100v rated, so probably not the case either.

Overdissipation; If the ESR of the capacitor is rather high, and there are large currents passing through it (definitely at 40A), then the capacitor can overheat and explode. I bet that's what happened here.

The solution is to use capacitors with a lower ESR, more capacitors, or both. By using more capacitors, the ESR of the group is lower, and the heat is spread out over a larger area, which reduces heating. I'd recommend both if you have room to stuff more caps inside the case.

From Mouser, something like a 647-UPW2C221MHD, 647-UPW2C331MRD, or 647-UPW2C471MRD might be good choices. Not sure which one will physically fit in the spot, so check dimensions before ordering.
 
My board is a factory 4110 72v50a unit from ebikes.ca. The printed version info on the board is "RFW-4850-V06", dated "07-05-11". I'm going to guess that that means November 5, 2007. All directions are with the FETs oriented up, or north.

The board I have uses at least seven parallel filter caps: three small ceramic and three huge electrolytic, and at least one small electrolytic. Tow of the electrolytic ones are 470uF, one is 1000uF, and all are rated for 100v. The little ceramic ones are only marked "0. 1 K 100" The small electrolytic, which fechter mentioned and is shown exploded above, is 100v 10uF. Again, these are all in parallel, spanning battery positive and battery ground.

There are five smaller caps scattered about. I have no idea what they're for, but four of them are rated for 50v, one for 16v. The little 16v 220uF oddball is right by the status indicator LED, a 47uF one is by each phase, and a 220uF is roughly half way between the shunt and indicator LED.

I'm trying to decipher the solder points of various connectors, since I tore out some wires accidentally while adding copper wire to the buses. So, I'll post what I know, and ask you to fill in the blanks.

The on/off switch connects the battery positive to the "L" solder point, which then drives the low voltage portion of the board. Thus, it needs to be able to handle all the current the low voltage regulator circuit will draw... I suspect that that is far less than an amp.

The on/off light is wired from this same L solder point to battery ground, at the "D-" point. Thus, when battery power is applied and the on/off switch is closed, it is exposed to full battery voltage and lights up. Inside the light's shrinkwrap is a tiny PCB. No idea what's on it... most likely either nothing or a tiny step-down resistor.

The throttle connects to a +5V, a ground, and the signal (normally green wire) connects to "SP". On my unit, the GND of choice was one in a vertical row to the far right, and the +5V was directly to the right of the Hall effect sensor solder pads.

The Hall sensors are soldered to a vertical row of pads .5" to the right of the microcontroller. From top to bottom, the pads are marked W, V, U, +5V, and GND. The last two are self-explanatory, while the first three are signal. On my unit, the top "W" pad was soldered to a blue wire. A brown and yellow wire attached each to the other two signal pads, but I did not see which is which. From the other pics in this thread, it appears to be brown to V and yellow to U.

The CycleAnalyst plug has six wires. The red and black are battery voltage. The red is soldered to the aforementioned L point so that the CA can detect when the controller is turned off (and frantically save data, hopefully finishing before the filter caps are drained). Blue and white are for measuring voltage drop across the shunt, and are soldered to D- and D7 respectively. That is, the white wire is practically soldered to the north side of the shunt. The remaining wires, yellow and green, are for Hall sensing and throttle override, respectively. The yellow is, I believe, connected to the HM pad to the left of the aforementioned row of grounds on the right side of the board. This would be a decoded output from the microcontroller. (I'm emailing Justin on this to be sure. It's the only CA wire that was torn). The green throttle override wire is soldered to the XS+ pad, directly left of the SP where the throttle signal connects.

I have no idea where the ebrake lines connect, since there is no such connector on ebikes.ca's controllers. Meh.

And a close-up of the relevant part of the board:
 

Attachments

  • 0419080206.jpg
    0419080206.jpg
    280 KB · Views: 7,870
Hey,

That's my blown controller that Fecther has posted the images of. I have 2 of those now that are identical. One of them was only used for about 5 minutes before that happpened.

The blown Caps were floating around inside the housing but the controllers still work fine. I'm not noticing any difference in performance.

Being a mech engineer rather than an elecrical engineer, not a lot of the dissections of these controllers makes sense to me.

All I want to know is..... What will happen if I continue to use the controllers without the Caps in the circuit? As I mentioned earlier, I have 2 controllers like this and both are working just fine.

Looking forward to hearing the answer to this.
 
Like fechter said, you'll get a voltage spike out of the motor which will fry the FETs. Not fun. Try to replace those caps asap!
 
Well, ignoring everyones advise (as I do). I took the bike out again running it flat out for about 45 minutes... And by flat out I mean that the CA was reading about 3500W for most of the time. The controller isn't even really mounted in a place wher eit gets decent air flow. IT got too hot to hold your hand on on the side that the heat sinks are on, but it's still functioning fine.

So what do I do next? Well.... I'm going to thrash it has hard as I can until something else stops working. As a mech engineer I design everything so it has at least one purpose... sometimes more than one purpose. I can't make head nor tail of an electronic circuit, but I figure that if it's working without them, then what exactly am I trying to fix?

Plus it saves me about 30 grams in weight.... ha ha

Once the thing fries I'll send some photos.
 
You must have a very low impedance path to the batteries and very low ESR batteries to take up the spikes.
If one of your battery connections ever develops some resistance, you'll be more likely to blow something.

Anyway, capacitors are fairly cheap and easy to replace, especially compared to the FETs.

What kind of batteries are you running?
 
I'm running a 72V LiPo pack and drawing 40-50Amps. I don't have any info on the battery specs though. Sorry.

I know that replacing the capacitors is cheap and easy..... but blowing up the controller is cheaper and easier.... actually ... not cheaper... but easier.... and more fun.
 
One would think there would be some kind of transient spike protector that a person could put inline (outside of the controller).

I'm understanding the surges come from the motor, and not the battery?

Fechter - what do you think is causing most of this and what could be done to avoid it (besides rebuilding the controller w 4110s)??

Reckon a fuse is too slow, have seen some fast react breakers at Mouser - any ideas?

DK
 
Deepkimchi said:
One would think there would be some kind of transient spike protector that a person could put inline (outside of the controller).

I'm understanding the surges come from the motor, and not the battery?

Fechter - what do you think is causing most of this and what could be done to avoid it (besides rebuilding the controller w 4110s)??

Reckon a fuse is too slow, have seen some fast react breakers at Mouser - any ideas?

DK

That's exactly what a filter capacitor is, a transient spike protector. If the capacitor was located outside the controller, it would be too far away to effectively and reliably level off spikes before they hit the FETs.

A fuse would be ridiculously slow, and any breakers that would protect in such an instance would trip constantly under normal operation. Not much fun.

The best way to prevent this is to beef up the capacitors. Add more or larger big catalytic ones for the big spikes and small ultra-low ESR ceramic caps for the high frequency little spikes, as close to the FETs as you can get them, in parallel over the battery positive and ground buses.

I know that replacing the capacitors is cheap and easy..... but blowing up the controller is cheaper and easier.... actually ... not cheaper... but easier.... and more fun.

:shock: This is a very scary statement. I don't think blowing a controller is very fun. Replacing or repairing them can be a real pain in the rear. Every time a FET blows, there's a chance of it taking other things with it, with this chance increasing the more power involved in the setup. A cascading failure can easily take out other FETs, or the low voltage regulator, in which case you can get battery voltage on the low voltage side of the controller. If that happens, your controller is beyond repair.
 
[quoteIf that happens, your controller is beyond repair.][/quote]

I don't plan on repairing the controller. I plan on seeing what it takes to blow it up.

I've run the bike flat out at over 70km/h, ridden it on rough terrain, up steep hills, over jumps, heated it up to the point where I can't hold my hand on it, but she still goes.

Can anyone tell me how I can achieve this "power spike" that everyone's talking about by means of riding the bike? I'm sure there are other ways of doing it but I'd like to do it while putting the bike through it's intended use. Not by some sort of simulated power spike... if you get what I mean.

Now don't I remember a wise man from here saying................ "One test is worth a thousand opinions" ?

Do I need more tests or more opinions?? More tests I think.

Thanks for the info guys
 
arg. too bad i live too far away from you getadirbike. otherwise i would rescue those controller from you and repair them. I need some cheap, used crystalyte controller right now. They must've been something very terrible in their past life to deserve such torture. I hope you do put them out of their misery soon.
 
lazarus2405 said:
That's exactly what a filter capacitor is, a transient spike protector. If the capacitor was located outside the controller, it would be too far away to effectively and reliably level off spikes before they hit the FETs. 2 inches outside the case would make that much difference? Also, seems like if the protector is between the source of the transient and the FET it would be more effective.

The best way to prevent this is to beef up the capacitors. Add more or larger big catalytic ones for the big spikes and small ultra-low ESR ceramic caps for the high frequency little spikes, as close to the FETs as you can get them, in parallel over the battery positive and ground buses. I thought the motor was the source of the transients from previous discussions.


My suggestion would be : identify the source of what is blowing the controllers. Identify the type: long time high current overload or short transient spike (voltage or current). Identify what type of normal pulse goes to the motor from the controller (in order not to interfere with it). Then develop an inline device that could handle the transient (problem).

Getadirtbike - if you blow the controller please note exactly the conditions prior to its demise. (Long hill, sudden full throttle, etc)

DK
 
pic for ebrake location from previous post this thread
 

Attachments

  • connections.jpg
    connections.jpg
    71.7 KB · Views: 7,290
one way you might be able to blow your controller ( glad to help !! no problem)
have the motor running at full load ( eg up a hill drawing max amps) then while under load make the motor stop very quickly ( instantaneously if possible )but still while under load ( i.e. throttle on full ), a stick thrown into the spokes, or running into a brick wall should suffice........
otherwise just pull the thick motor wire connection apart while under load that will give a decent voltage spike, seems a main cause of controller failure is a bad connection on those thick wires or the connector comes apart while motor is under load
 
John (Getadirtbike..sorry to out you, but "getadirtbike" is too long and cumbersome..so's "solarbbq" Brett ;) )

If you want I can replace the caps (but I guess you can solder).

We could also try hooking up my old cro and see if we can spot the spikes, and their level. Running at freewheel speed should be constant enough for the cro to get a lock - whether it would show the full level would just depend on how short they are, it's only a 15mhz cro.


I guess it's research for you, but almost deliberately killing the controller is just plain wrong!!

Cheers

Mark
 
****Surprise******

My controller went pop. not through riding it flat out, or overheating, but from what seems to be a power spike. My mate was standing there touching the throttle and using the torque to lift the front wheel. He had both feet on the ground and was kinda rocking it backward and forward using the throttle.

At least I know what the Capacitors do now :)

p.s. Here's a photo that I took in the loungeroom this morning. There's one missing..... it's still on the bike

DSCF6892.JPG
 
Looks like they're having an orgy. Perhaps they'll reproduce if you leave them alone.

When the controller is operating against a stalled motor at low duty cycle will be when the spikes are the greatest. At full speed and 100% duty cycle, the capacitors aren't doing anything.
 
does anyone know the location of low voltage cut off resistors on v2 controller?
Had a controller sent back to me from a customer thought i would try repair it, measured resistances of high amp side of fets while still in place, if a pair was ok resistance was about 7kohm i think from memory, one pair has a lower resistance so thought that might be where to look. Found a very small break in solder on bottom of board from middle leg of a fet from excess current, interestingly the fet where solder blew was ok, it was the other one that had blown in that pair. So that controller now working. Have another one with a few resistors on board blown might be too far gone. These ones were early version with 6 lots of 4310fets only, v2 crystalyte, the white coloured board ones. White colour is good for seeing which tracks got hot too, they turn brown.
 
Justin (the guy who designed the CA) at ebikes.ca and Kenny from Crystalyte both agree that the resistor divider is made up of two resistors labeled R15 and R16. kenny also added a fragment of a schematic with some hand written notes for the resistor values. the schematic is almost illegible and half is in Chinese, but the labeled numbers he provided agree with the information from Justin. i reproduced the information that i was given.

LVC.jpg

i have ne of the controllers from Crystalyte on its way to me. should be here in a couple of days. at least then i will be able to see if the info given matches the controller or if they have gone to another change.

rick
 
Another little part of the circuitry. here is the schematic for the Low Voltage Regulator. it uses a switching regulator to step down the battery to 15V and then a 78L05 regulator to give 5V to the Cypress chip.

View attachment 1

Rick
 

Attachments

  • LOW VOLTAGE SWITCHING SUPPLY.pdf
    16.2 KB · Views: 190
Back
Top