Version 2 Crystalyte Controller information

[fechter]

Fechter,
I would concur with you that it is only 2-sided since you have examined the board carefully .

From the photos of the board, It seems that the S-legs of No 3 and No 7 are connected to the -ve bus on the component side via through-holes and tracks which are not beef-uped by "solder and wires".

Here's a pic of #3 with some of the solder mask scraped off. The trace goes to 3 through plated holes to the -ve bus. You could add copper/solder to the top side. If you want to be really anal, you could drill through and solder a wire directly to the bus.

 

[The7]

Fechter, thanks for down to earth discovery.

It doesn't seem that the trace from the S of No 3 to the plated hole could carry very high current (say over 25A).

 

[rkosiorek]

the 4 layer info was included as part as an email from kenny at Crystalyte. i just passed it on as gospel. sorry for the inaccurate info about the board. the schematics i posted are accurate though. those were generated by tracing out pads on the board by testing with a low ohms meter.

rick

 

[The7]

The controller may have a cut-off frequency similar to Shenzhen (ecrazyman) Controller.
See http://endless-sphere.com/forums/viewtopic.php?f=2&t=4282&start=15

If It has such cut-off frequency (say 325 Hz), it wlll not cause any problem for the direct drive motors because the cut-off speed is well above 100 km/h. It should have no problem since it is designed specially for Crystalyte motors in mind.

However, using this controller for geared hub motors might result in a very low cut-off speed. The Puma motor might give a cut-off speed of about 20-25km/h which is too low for any practical usage.

Wonder any of you use this controller on direct dirve hub motor like Crystalyte 5xxx/4xxx AND who use it on geared hub motor like Puma?

 

[solarbbq2003]

thanks fetcher for caps link at mouser, will see if i can find some down here.
Really I'm looking for what mods can be done to make controller more reliable, I'm assuming from whats been posted that the caps might be one place to start ( my knowledge on controllers very limited but learning!) would this be a possible modification scenario to reduce likelihood of failure:
1. double up red caps? ( can parallel two together?)
2. use low esr on big caps ( would that also apply to red caps to use low esr if available?)
3. increase max voltage on caps from 100v to higher level?
Also controllers I have, seem to use 470uF or 1000uF on big caps ( pic attached), would there be an a preferred uF value to use?
( I assume the 3 caps in with arrows in pic below are the ones that would be changed??? )
I've only used this controller with x5( from memory) with 12 lots of 4110 caps on 48v no probs. Did use the earlier 6 fets versions but they all failed ( had 4310 fets)
6 version fets were early ones crystayte supplied, as far as I know they dont do the 6 fet version now, so anyone still selling those type, my guess is the stock is a bit old.

 

[solarbbq2003]

to the7:
could you identify which cap to change on a pic , the one that might make running puma/bmc with this controller possible?

 

[solarbbq2003]

just reading ebike testing section, seems v2 does work on bmc/puma motor......

 

[fechter]

just reading ebike testing section, seems v2 does work on bmc/puma motor......

At least one person seems to have that combination working... Not sure what the problem is though.

 

[The7]

to the7:
could you identify which cap to change on a pic , the one that might make running puma/bmc with this controller possible?

I don't have the board in hand.

Perhaps rkosiorek could help to identify these 103 cap (0.01uF).
Rkosiorek said - [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.]

Seems that these C30, C31 and C32 could be the 103 CAP ???

 

[Jozzer]

I've had the oppurtunity to test some 20 of these controllers on 4 different models of BMC/Puma hubmotors, and not had a single one work right.
Also, these controllers and a test motor were sent back to Kenny who spent 2 weeks trying to "recalibrate" them to work with a Puma. By the end, he had killed the puma, and sent back the controllers still unwilling to operate the Puma. (though he claimed he had fixed it)
Fechter, the boards I sent to you have been "doctored" by Kenny, and I recently found out that they wont run properly on an X5! One of the boards that wasn't sent back to him DOES work on the X5. I assume he messed with the programming. Hope this doesn't "cramp your style"!

Steve

 

[rkosiorek]

good call, yes those are the 103 or 0.01mfd capacitors. and just to the right of those are the pull up and series resistors. they connect directly to the hall input connection pads.

rick

 

[fechter]

Fechter, the boards I sent to you have been "doctored" by Kenny, and I recently found out that they wont run properly on an X5! One of the boards that wasn't sent back to him DOES work on the X5. I assume he messed with the programming. Hope this doesn't "cramp your style"!

Steve

thanks for the heads up.

The main capacitors are obviously an issue. Also, if you're running 12 FETs, the gates of pairs are tied directly together. This is a no-no, as at high currents you're asking for parasitic oscillations which can cause destruction. Jeez.. how much does it cost to add 6 little resistors. Gates should each have their own resistor.

If I get time (eventually?) I'll populate one of the boards you sent me to see how it does on my test setup.

Steve: with the Puma motor, did any of these controllers allow the motor to get up to full speed with no load?

 

[Jozzer]

Afraid not Richard. Some almost made it in reverse I seem to recal, but even then they wern't running right, and no combination of halls/phase or reverse could make it run that way forwards.

 

[Knuckles]

Is it fair to say that the x-lyte v2 problem with high rpm PMGR is exactly what fechter describes here?
http://endless-sphere.com/forums/viewtopic.php?f=2&t=4282&

If so then this is clearly a design issue that controller fabricators haven’t taken the necessary time to really GAF about!!!
Don Grubee is selling PMGR motors and controllers (but he never exceeds 48V). Maybe he has a clue about the 333hz jitter-shift mystery!

I guess I'll ask that root'n'toot'n'cowboy! But don't hold your breath ... Don's a good engineer but a bit of an @&!%-hole.

'shift happens'

cheers

 

[The7]

In the other post, the power frequency for Puma motor is about 480Hz at 360rpm on 26" wheel at an speed of 45km/h.
Obviously most e-bikers would like their e-bikes using Puma to reach speed > 45 km/h.

The V2 controller seems to have problem for an output frequency of 480 Hz (or higher).
This may be caused by
a) the excessive delay due to the 2.9 kHz LP filter for the Hall signals,
b) the inbuild program in the MCU for a certain upper frequency limit (no sure if it has such program), and
c) the possible delay in the digital process.

V2 is designed for Y-connected Crystalyte hub motor.

Is Puma motor delta-connected?
If so, there will be an phase difference of 30 deg between the delta-connection and the Y-connection.
In such case, there will no combination of motor and hall signal that can correct this "30 deg" difference.
Such 30 deg difference will lead to a poor power factor in the motor and the motor will run hotter.

Let us use some fundamental EE formula to illustrate the effect of poor power factor.

For 1-phase ac cirucit
P (power) = V (voltage) x I (current) x pf (power factor)
where pf varies from 0 to 1.

Suppose V = 10 volt and P = 100 watt

If there is no phase difference, pf = cos 0 = 1
Then 100 = 10 x Current X 1
Hence Current = 10 amp

If there is an phase difference of 30 deg, pf= cos 30 = 0.866
Then for the same power 100 watt
100 = 10 x Current x 0.866
Hence Current = 11.5 amp

The current is increased by 15%.
And the motor is at least running 33% hotter due to resistive loss.

 

[fechter]

The Puma and BMC are delta connected, but the hall sensors are located for proper phasing.

I've never tried switching a motor from delta to wye or vice versa.

With the analog controller, the BMC runs great and the efficiency seems to be quite good.

 

[The7]

The best way for verifying the correctness of the phase relationship is to compare the V2 with the Puma dedicated controller using oscilloscope.

 

[Jozzer]

the Puma dedicated controller? U mean the one made by thier factory? Doesn't work very well tbh Second to last one I had blew yesterday.
Controllers that DO work include the Kelly BLDC controller and the analogue x'lyte, also one of the Ananda 48v controllers seems to be holding up OK.

We've been comparing them all on scope, time to start taking some pictures to see anyone else can see anything usefull perhaps.

 

[Tiberius]

I've never tried switching a motor from delta to wye or vice versa.

I've done it with an AC motor, but then no Hall sensors to worry about.

The delta/Y change alters the operating voltage by a factor of root 3. Here in the UK there are two voltages of 3 phase supply commonly found, 415 and 240 V (or 400 and 230 in practice) and the delta/Y change can switch a motor from one to the other.

Nick

 

[The7]

We've been comparing them all on scope, time to start taking some pictures to see anyone else can see anything usefull perhaps.

I will be very interested in seeing these pictures when availabe.

Some suggestions:

a)For a certain controller, try to show the motor phase voltage (say the motor yellow wire of Puma to the battery -ve) and the Hall signal of the same colour (say also Hall yellow signal of Puma) at the same picture for partial throttle (say about 50% speed).

b)Repeat (a) for full throttle.

c) Repeat (a) and (b) for other controllers.

 

[solarbbq2003]

novice quetion!:
what is low pass filtering and how are you working out 2.9kHz and 4.8kHz as limits?
below novice quetion?: if lowering uF of caps in pic above might resolve it would it matter if the caps were surface mount or other?

 

[fechter]

The Shenzhen controller has the exact same hall input filter setup.
When I get a chance, I will scope the hall signal before and after the filter.

 

[The7]

novice quetion!:
what is low pass filtering and how are you working out 2.9kHz and 4.8kHz as limits?
below novice quetion?: if lowering uF of caps in pic above might resolve it would it matter if the caps were surface mount or other?

Rkosiorek said - [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.]

The LP filter is formed by R-C circuit at the Hall signal input using the LP filter formula:

Freq = 1/ (2 x 3.1416 x C x R ) Hz

.For C = 0.01 uF = 0.01 / 1000000 F ;
and R = 3.3 k ohm = 3300 ohm
Then Freq = 4822 Hz = 4.8 kHz.

If R - (2.2k + 3.3 k) = 5500 ohm, then freq = 2.9 kHz.
Using the rule of thumb (say 10 times), the delay due to 2.9 kHz Lp filter will be negligible if the power frequency is 290 Hz or lower.

If you change 0.01 uF to 0.005 uF, the LP filter will be 5.8 kHz. Then it will not cause any problem if the power frequency is 580 Hz or lower.

It doesn't matter if they are surface-mount type or ordinary wiring type.

However, if it is pre-programmed in the MCU to limit the max power frequency, the change of these caps may not of any help.

 

[The7]

The Shenzhen controller has the exact same hall input filter setup.
When I get a chance, I will scope the hall signal before and after the filter.

Does Shenzhen have the same value of R and C as V2 ?

 

[Jozzer]

http://uk.youtube.com/watch?v=Icslk6jMyJY
Let me know if this is what you need to see, if so I'll do the rest of the controllers!