Version 2 Crystalyte Controller information

[lazarus2405]

Bleh.... because on my unit I have error code 5. Whatever the heck it means...

Time to look for shorts on the low voltage parts... *sigh*

Update: After changing fuses, and finding that I had a bad connection in the holder, it seems that the controller has changed its mind from "Electric current failure" to "MOS damage". It seems that that error code refers to a "volts but no amps" situation, like Link experienced, caused by a high resistance connection.

So, "MOS damage". I suspect that it is detecting a short somewhere in my high voltage side, most likely a high resistance path (since I didn't find any obvious ones).

So, can anyone tell me what I should expect resistance values to be? B+ to B- should be no connection. What about between each battery pole and any phase, or between two phases, and between various FET legs?

 

[fechter]

If the power switch is off, from B+ to B-, it will just be reading the main capacitors, which should start out low, and rise to very high or open.

From either battery wire to any phase wire, it should be open. Any continuity indicates a shorted FET.

 

[lazarus2405]

Worked it out. I had a low side FET's gate leg shorted to the phase by a little errant solder. I cleaned it out and the error message changed to the LVC. Good news, since I was testing with a 28v pack.

After more troubleshooting I've found that I have one dead or disconnected Hall sensor, with two good ones. This made the motor start when the throttle was jerked, and then inexplicably die, and momentarily jerk to a start again if the throttle was returned to zero and then turned again.

Into the motor we go again...

 

[Deepkimchi]

Quick crazy question - for the LVC cutoff, believe it was R16.

What would happen if one took the surface mounted resistor out, soldered in 2 wires, run the wires to some kind of 3 way selector switch mounted to the controller which had the proper value resistors for 36V, 48V and 72V?

I want an award... :lol:

DK

 

[ngocthach1130]

Or you run it to a potentiometer and have a variable LVC. Resolution of the LVC depend on the resolution of the potentiometer and the range of the resistance.

 

[Deepkimchi]

I think the average consumer and non-electronic types would prefer the 3 click switch on the side of the controller marked (in tiny white letters) 36V 48V 72V.

DK

 

[Dee Jay]

I think the average consumer and non-electronic types would prefer the 3 click switch on the side of the controller marked (in tiny white letters) 36V 48V 72V.

DK

:wink: J

 

[ngocthach1130]

:lol: i forgot about the whole consumer simplicity thing. I was thinking along the line of potentiometer because we tinker with our controller alot.

 

[lazarus2405]

I think the average consumer and non-electronic types would prefer the 3 click switch on the side of the controller marked (in tiny white letters) 36V 48V 72V.

DK

Don't forget 24v!

 

[fechter]

My understanding is that it *might* run a BMC, but that others have had problems with the v2 units. There's some sort of software limitation that these units have with respect to high commutation frequencies. Since the BMCs are geared and run at a much higher RPM, they have had problems with these units. Not a hardware thing at all, but an example of Crystalyte shortsightedness.

I suspect that the issue is more of a BMC thing. the controller is designed for neutrally timed motors. the placement of the hall sensors on the BMC suggests that the Hall sensor timing is advanced a few degrees. that would explain why the motor runs fine at low rom but develops a timing problem at high rpm. but this is an untested theory at this time.

The V2 units supply the hall sensors with 5v, compared to 7-10v with the analog controllers.

In testing I did with a Puma motor, I noticed that the hall signals had a suspiciously slow rise time with a 5v supply, so I suspect this may be the problem here. Once the speed reaches a certain point, the hall timing will go retarded and start blowing things up.

The SS40 series hall sensors used in the BMC and Puma motors have an internal pull-up, but it is pretty weak.

By adding 2.2k or 1k pull-up resistors between the hall supply and the 3 hall signal wires, you can dramatically speed up the rise time. The hall sensors are rated for around 20ma output, so 1k should be OK.

Does anyone know if the V2 controller has any pull-up resistors on the hall signal lines? (I don't have one to look at). It would be easy enough for Crystalyte to change (or add) the pull-up resistors. One could also add them externally on the hall connector.

 

[rkosiorek]

there is a 2.2K pull up at each of the hall sensors. there is also a 3.3k resistor in series from the hall sensor input to the Cypress chip and a 103 cap from the cypress input pin to ground.

looks like the 3.3K and the .01mf cap form a low pass filter with a 4.8Khz rolloff.

Fechter since you have a scope what is the frequency of the output of the hall sensors at full rpm on the BMC? could this low pass filter be a problem?

rick

ps please check my math on that frequency.

 

[Jozzer]

Fechter, do you want a V2 board to play with? I've got a number of them stripped of leads and FET's, but otherwise intact. Send me your addy and i'll post it on, their doner card said they wanted to leave thier bodies to electrical science...

 

[fechter]

Fechter, do you want a V2 board to play with? I've got a number of them stripped of leads and FET's, but otherwise intact. Send me your addy and i'll post it on, their doner card said they wanted to leave thier bodies to electrical science...

PM sent

I'll check on the frequency. I doubt it's high enough for the filter to get it, but I haven't measured it.
2.2k pull-ups should be good. It could just be exceeding the speed of the processor at some point.
Overclock?

 

[rkosiorek]

it is an 8bit RISC based core running at 12Mhz . it also uses a hardware based PWM. i think it should be able to handle just about any frequency we throw at it.

the software on the chip is read protected. so it cannot be reverse engineered. but the design of the board is very simple. almost all of the connections go directly to the controller chip. the only exceptions are the FETS and drivers for the high power outputs and and the comparators tied to the current limiter.

it would be an interesting project for someone with enough time to use the V2 controller as a platform and write new software. perhaps delete some of the routines that give the error codes and concentrate more on the control software.

if it was me i would probably write the routines so that the hall sensors were used only for starting and a sensorless back-emf technique used for actual running. oh well hat would be a dream. i have too many other things on the go right now.

rick

 

[lazarus2405]

if it was me i would probably write the routines so that the hall sensors were used only for starting and a sensorless back-emf technique used for actual running. oh well hat would be a dream. i have too many other things on the go right now.

What would be the advantage of this? I assume there'd be good reason to go through all that trouble.

 

[solarbbq2003]

I notice eabs gets a mention on the led codes, eabs is chinese term for regen braking ( electronic assist braking system:eabs),
also notice a pin named abs on board, anyone know if v2 has regen braking?

 

[fechter]

there is a 2.2K pull up at each of the hall sensors. there is also a 3.3k resistor in series from the hall sensor input to the Cypress chip and a 103 cap from the cypress input pin to ground.

looks like the 3.3K and the .01mf cap form a low pass filter with a 4.8Khz rolloff.

Fechter since you have a scope what is the frequency of the output of the hall sensors at full rpm on the BMC? could this low pass filter be a problem?

The Puma motor I have runs a ~5:1 gear reduction. Nominal rated hub rpm= 320.
There are 16 magnet poles (32 magnet segments) on the rotor, so that works out to ~427hz at 320 rpm. If you were running seriously overvolted, you might get up to ~800hz.
At 60v, the BMC motor on my Vego tops out around 800hz, using an analog Xlyte controller.
This is not anywhere near the rolloff frequency of the filter (I trust your math more than mine).

I'll see if I can scope the hall signals if I get one of those v2 controllers. I might not hurt to lower the pull-up resistors to 1k.

Yes, a sensorless setup would be sweet, so would dynamically adjustable timing. A 'rounded' trapezoidal output could be done too, which would greatly reduce motor noise.
A glitch or loose connection in a hall sensor wire can cause the controller to blow, this can be sidestepped with sensorless control.

I like the error codes. Beats the hell out of trying to test hall sensors with a meter. If properly done, the software should protect the hardware from self-destructing and give you a clue of where to look for the problem.

 

[rkosiorek]

the chip only has 4kbytes of flash.

if you use either a vectored or variable timing approach a lookup table or map would seriously cut down on the computationl requirements but might significantly eat up that space. personally if given a choice between a customized timing map and a couple of blinking lights, i'll vote for the map.

but there may be enough memory space for both. wouldn't know until some code is written and experimented with. and i'd have to have a spare V2 that i would be willing to sacrifice as once i start screwing with the software it would erase the original code.

since i don't have a V2 to sacrifice and unfortunately at this time i have too many other things on the go i would not have the time to do this project justice. i just got a new mill and lathe and it looks like i got about a weeks worth of de-greasing before i can use either. i'm also thinking of converting the mill and rotary table to a 4-axis CNC. i already have all of the steppers and drive electronics.

on to the BMC/PUMA by physical measurement the Hall sensors are advanced 5 degrees from neutral. i think i'm going to try to set them to neutral and see if that makes any difference to the performance. the timing difference is so small that it would only make a slight difference in performance. they probably included it to prevent the motor from starting in reverse. at high rpm the timing difference could affect how well it runs if the PWM on time is long enough so the wrong coils are still energised when the halls report a new position.

rick

 

[solarbbq2003]

other controllers with regen braking seem to have just one wire connect to ground to engage regen, guessing a bit here but abs pin to ground "might' engage regen

 

[kbarrett]

Does anyone have a clue what the value of the "C16" cap ( the one just to the left of the last FET ) is?

The gentleman in this thread lost his.

 

[marty]

Are these Version 2 Controllers any more waterproof then old version controllers?
Got this one now [not installed yet]
It don't seem waterproof at all.

 

[Ypedal]

That's an easy fix..

Remove the 4 screws on the end caps.. add a thin layer of silicone or adhesive of your choice ( Personal recommendation = " Goop " )

Wire connections, electrical tape, a few good layers and pull the tape tight and it will keep the water out.

 

[kbarrett]

The only weak point I see in mine for water is the red on-off rocker switch. Everywhere else was easy ... I just used a bead of silicone sealer before re-assembling.

 

[solarbbq2003]

to kbarrett
cant see any markings on that cap (c16), but probably same value as the other caps in similar posistions, is there any way I can measure the value?? if so will give it a go

 

[kbarrett]

to kbarrett
cant see any markings on that cap (c16), but probably same value as the other caps in similar posistions, is there any way I can measure the value?? if so will give it a go

Not while it is on the board, unfortunately.