Shenzhen (ecrazyman) Controller Information

[The7]

Hi The7,

Your comment about 10 pole and 4.3 ratio is a bit spooky if I'm counting 86.

Nick

The motor in this picture (Knuckles) is quite similar to yours:
It has 10 pole-pairs (or 20 magnet poles) and the gear ratio is around 4:1 to 5:1 when estimated from the picture.
No of counts in 1 hub turn = 40 to 50 for this motor.


Knuckles said :"OK. Keywin says ...
... There are 10 pole pairs
... The gear ratio is 1:4.3
... and 1720 rpm at 72v is bafang motor"

The gear ratio of your Bafang motor as given by Knuckles is 4.3:1.
So the no of counts of your should be 43 for 1 turn of the hub.

Would like to know the exact circuit of your counting?

 

[The7]

Excellent test setup!

btw. I just use two 36V SLA chargers is series as an 88V PSU for no-load testing. That works perfect for my controllers and DD brushless motors.

Noted that you are using an DD brushness motor.
So don't think that the frequency would reach the critical frequency in this test.

 

[Knuckles]

Of course. I was just illustrating the chargers as PSU for no load testing.

 

[Stevil_Knevil]

While my Knuckle-troller PCB was out of it's housing, I figured that this would be a good time to 'ruggedize' the caps, implement a couple user-interface augmentations and generally improve upon an already good controller.

First observation- the PCB is sprayed with a varnish/conformal coating of some kind (?) to resist moisture. Solder wants nothing to do with this stuff, so I swabbed the areas where I needed to work with a Q-Tip soaked in Acetone, waited for it to dry (duh) then soldered the wires for a future CycleAnalyst connection.

Next, I decided to fatten the FET traces with solder wick and copious amounts of solder to keep the resistance and heat loss to a minimum.

If something goes left, this controller will throw fault codes. A carefully measured & step-drilled hole in the housing, then covered with a water-proof, clear tire patch will let me know if/what kind of 'event' we've had without the need for disassembly.

Finally, a couple blobs of silicone should keep the caps from getting perished when the going gets rough.

 

[dirtdad]

Very nice work. Knuckles should take note. On the other hand, maybe this should be left as an exercise for the student, that way we can say we did a controller mod!

 

[fechter]

Nice work Stevil!

I like the LED window. I was also thinking of replacing the LED with a wire and having it run to an LED in the throttle body or one mounted in the end plate. Hopefully you won't need to see the codes too often though.

 

[Tiberius]

Ok, guys,

This is embarassing. I rushed into print too soon yesterday.

I messed up counting the pulses from the Hall sensor. In my defence may I point out that I'm usually working with signals 10^7 faster. I had a nice 5V square wave from the Hall sensors and I fed it to the scope and the counter. But I left the counter threshold set to DC coupled and zero volts and there was enough bounce on the signal to accurately count both edges. With better settings the count is 43 Hall pulses per turn, and I've checked that with a stored trace of a quarter turn on the scope. So that ties up nicely with a 4.3:1 gearing and 10 pole motor

It also turned out that while 2 of the PSUs had 2 A cutouts, one was 1 A. That got changed to a different one (you don't want to know how many PSUs I have) and now I can go to 84 V full throttle.

The change in noise was down to my test jig coupling very well with the benchtop and hitting a resonance somewhere.

Here are the scope traces at various voltages, all full throttle. Bottom trace is a Hall sensor, top trace one of the winding connections to ground.

At low (sic) voltages there is an artefact at the beginning and a trapezoid ramp.

Getting higher, the ramp reduces. At 72 V its is almost gone.

After that its just a square wave. This is as far as I could go. You should be able to see the frequency readout.

Or maybe not; the photos aren't good enough. From 72 to 84 V the frequency hardly moved - 326 Hz.
There is some kind of effect or limit being reached at 72 V or 326 Hz. Mind you, with 43 count in a 26" wheel that would be 35 mph.

Nick

 

[Knuckles]

Is this good or bad? I think this is good? YES? Hell I feel like I don't know ANYTHING anymore!
So the Bafang caps out at 72V (326 frequency limit) but doesn't jitter to bad between 72V and 84V?
(Man! And I thought I was a geek! All you ES guys are like WAY OUT THERE!)
Under loaded conditions (on a 26" wheel) the 326 Hz frequency will not be reached until 35 MPH?
Party on Wayne! Party on Garth! This is all good? YES?

btw This IS THE BAFANG MOTOR

Knuckles said :"OK. Keywin says ...
... There are 10 pole pairs
... The gear ratio is 1:4.3
... and 1720 rpm at 72v is bafang motor"

The gear ratio of your Bafang motor as given by Knuckles is 4.3:1.
So the no of counts of your should be 43 for 1 turn of the hub.

 

[fechter]

...

Or maybe not; the photos aren't good enough. From 72 to 84 V the frequency hardly moved - 326 Hz.
There is some kind of effect or limit being reached at 72 V or 326 Hz. Mind you, with 43 count in a 26" wheel that would be 35 mph.

Nick

Good measurements. You reached the magic 325 Hz limit. When you get close, I noticed the waveform gets squared off as the timing goes out of whack. If you had a few more volts, you could push it into the range where it just jitters.

 

[Tiberius]

Good measurements. You reached the magic 325 Hz limit. When you get close, I noticed the waveform gets squared off as the timing goes out of whack. If you had a few more volts, you could push it into the range where it just jitters.

Yes, I am definitely running out of electricity, or "leccy wallop" as we refer to it in these parts.

The PSU's are struggling and the current limit on them is setting the voltage. I would like to explore that top end of the range a bit more to find out whether its the current, the voltage or the frequency that is setting the limit. My feeling is that its the frequency and that is causing the current to rise, which then limits the volts. To find out I really need more wallop, and then I'd be likely to break something.

But as Knuckles says, its way over the top anyway for this particular motor.

Nick

 

[Knuckles]

How is this for voltage ...
http://cgi.ebay.com/Battery-Charger-Electric-Scooter-Bike-48V-2-5A_W0QQitemZ300200309945QQihZ020QQcategoryZ11332QQcmdZViewItemQQ_trksidZp1742.m153.l1262
or this ...
http://cgi.ebay.com/Battery-Charger-Electric-Scooter-Bike-60V-2-5A_W0QQitemZ300222261516QQihZ020QQcategoryZ11332QQcmdZViewItemQQ_trksidZp1742.m153.l1262
and then there is this bad boy here ...
http://cgi.ebay.com/350W-24V-14-5A-switch-power-supply-12V-15V-36V-48V_W0QQitemZ300200543029QQihZ020QQcategoryZ296QQcmdZViewItemQQ_trksidZp1742.m153.l1262

cheers

 

[The7]

With better settings the count is 43 Hall pulses per turn, and I've checked that with a stored trace of a quarter turn on the scope. So that ties up nicely with a 4.3:1 gearing and 10 pole motor

It should read 10 pole-pairs (NOT 10 pole) motor. So my expection is justified.

Excellent traces from scope.

See the comments in this trace (60V FULL Throttle)
The spikes in the phase voltage are BAD.
The phase voltage leads the Hall signal by 30 deg.
The flat top spaces 120 deg which is a typical conduction period of a 6-STEP controller.
The phase voltage is symmetrical.

Could you run this again at partial throttle for 60V supply because I would like to see if the phase voltage is still symmetrical like mine.
If it is still symmetrical, then the appiled phase voltage is in-phase with the BEMF of the motor

 

[The7]

...

Or maybe not; the photos aren't good enough. From 72 to 84 V the frequency hardly moved - 326 Hz.
There is some kind of effect or limit being reached at 72 V or 326 Hz. Mind you, with 43 count in a 26" wheel that would be 35 mph.

Nick

Good measurements. You reached the magic 325 Hz limit. When you get close, I noticed the waveform gets squared off as the timing goes out of whack. If you had a few more volts, you could push it into the range where it just jitters.

Both fecther and Tiberius hit this magic figure of 325 or 326 on individual controllers.
Both have BAD spikes in the phase voltages.
So it shows that this is produced "digitally". most likely due to codes or program in the MCU.
For example, 16 bit register would have 65536 levels (0 - 65535) and it could not represent any level higher than the max level.

At 60V FULL throttle, it is 271 Hz.
Phase voltage becomes more square above 60V.
The frequency increases to 326 Hz until 72V.
The frequency remains almost constant at 326Hz until 84V and the phase voltage is "square".

It seems that the PSUs cut off the power if you try to exceed 84V ???
If you are using a real battery above 72V or 84V, the supply current would be very high and damage the controller.

Did you note the supply current at 60V, 72V and 84V from your PSUs?

The squaring off of phase voltage at supply voltage above 60V is definitely NOT the proper design of 6-STEP controller.

Perhaps there will be no damaging effect if this controller is operating below 271Hz where the conduction period of the phase voltage remains 120 deg.

 

[fechter]

The squaring off of phase voltage at supply voltage above 60V is definitely NOT the proper design of 6-STEP controller.

Perhaps there will be no damaging effect if this controller is operating below 271Hz where the conduction period of the phase voltage remains 120 deg.

I think so. Everything behaves nicely below the critical frequency. Once it starts to jitter, the current goes high enough to trip the current limit on the PSU. With a stiff supply, the current spikes will eventually trip the overcurrent protect feature in the controller and shut down.

This controller has a two stage current limiter. There is one slow analog stage that does the normal current limiting, and a second, higher peak detector that will trigger a fault and stop the controller. This seems to do a good job of protecting the FETs.

 

[Tiberius]

This controller has a two stage current limiter. There is one slow analog stage that does the normal current limiting, and a second, higher peak detector that will trigger a fault and stop the controller. This seems to do a good job of protecting the FETs.

That sounds right. When I was cranking it up yesterday I had a couple of trips that were not down to the PSUs.

Nick

 

[Tiberius]

Hi,

Here are pics of the inside of my controller.


It should be the same as the others, apart from one mod. It is nominally 72V controller but I might want to use it at 48 or 36, so I asked Keywin to mod the low voltage cutoff for me. The blue wires hanging off are part of this.

Nick

 

[Tiberius]

Could you run this again at partial throttle for 60V supply because I would like to see if the phase voltage is still symmetrical like mine.
If it is still symmetrical, then the appiled phase voltage is in-phase with the BEMF of the motor

Hi The7, sorry if I haven't answered all your questions. The 8 hour time difference is going to cause a delay sometimes. But right now I am v busy getting ready for a meeting in France next week. After that I might stay on and go to the Le Mans 24 hour race. Nasty ICE's I know, but...

I haven't taken photos of partial throttle yet, but from what I remember the trapezoid shape remains the same.

Nick

 

[steveo]

Hi to all on the Es!

I just would like to do a quick post on this controller

I just recieved 2 controllers from knuckles this week; I have nothing but to say great things about this controller so far!!
The controller stock does run a bit warm & even after shunt has been modded... compared the my 4110 controller; but is does work GREAT with my x5304 in a 20" so far!!

-battery connections & motor phase wires to anderson powerpoles 45amp
-5 pin molex for the hall sensors
-changed the throttle connector to the same male/female 3 pin connections as a x-lite throttle;
-Placed a 35amp fuse on the red battery lead
-1 amp fuse on the orange wire ( power switch wire )
-LVC 100k adjustable timmer pot at about 55v


Also added a Modification to beef the power up!!

I added another shunt wire from a spare crystalite controller i had laying around; Max amps is now 45amps!!! The best part about this mod is the controller already has the pcb holes for the extra shunt wire

next...
- 4110 mosfet upgrade & instalation of 160v 470uf caps!!
- heat sinks to the voltage regulators


-steveo

 

[Knuckles]

I am so very pleased to hear this! Excellent work Steve.

btw, Keywin has another engineer friend who works for Chinese company that makes bicycles.

ARE YOU THINKING WHAT I AM THINKING!

Custom designed e bike with EVERY structual tube filled with LiFePo4 bats! 2x2 motors! disc brakes!

ha ha ... We so CRAZY!

 

[The7]

I just recieved 2 controllers from knuckles this week; I have nothing but to say great things about this controller so far!!
The controller stock does run a bit warm & even after shunt has been modded... compared the my 4110 controller; but is does work GREAT with my x5304 in a 20" so far!!

Frequency at hub speed of 300 rpm:
A) Gearless Hub Motors
1)Crystalyte 4000: 300 x 8 /60 = 40 Hz
2)Crystalytle 5000: 300 x 12 /60 = 60 Hz
3)GL2 : 300 x 23 /60 = 115 Hz

B) Geared Hub Motors
1) Puma : 300 x 5 x 16 / 60 = 400 Hz
2) P2 : 300 x 4 x 10 /60 = 200 Hz
3) Schwinn AL1020 : about 400 Hz

Running C5303 at 600 rpm will have an motor frequency of 120Hz
Even at 900 rpm, it is only 180 Hz which is very much lower than the critical frequency (325 Hz).
This shows that C5xxx will hardly cause any problem to the S-controller.

 

[Stevil_Knevil]

ARE YOU THINKING WHAT I AM THINKING!

That it might be time for a new thread :wink:

 

[steveo]

I just recieved 2 controllers from knuckles this week; I have nothing but to say great things about this controller so far!!
The controller stock does run a bit warm & even after shunt has been modded... compared the my 4110 controller; but is does work GREAT with my x5304 in a 20" so far!!

Frequency at hub speed of 300 rpm:
A) Gearless Hub Motors
1)Crystalyte 4000: 300 x 8 /60 = 40 Hz
2)Crystalytle 5000: 300 x 12 /60 = 60 Hz
3)GL2 : 300 x 23 /60 = 115 Hz

B) Geared Hub Motors
1) Puma : 300 x 5 x 16 / 60 = 400 Hz
2) P2 : 300 x 4 x 10 /60 = 200 Hz
3) Schwinn AL1020 : about 400 Hz

Running C5303 at 600 rpm will have an motor frequency of 120Hz
Even at 900 rpm, it is only 180 Hz which is very much lower than the critical frequency (325 Hz).
This shows that C5xxx will hardly cause any problem to the S-controller.

4110 mosfets & 160v 470uf caps to come .. stay tuned!!


p.s. .. my s-controller with 45shunt still pulling strong on stock fets @ 66v

 

[fechter]

I did an interesting test. The MCU is clocked by a 16mhz ceramic resonator. A PIC16F72 is rated for 20mhz. Digging around my junk pile, I ran across a 24mhz crystal, so I tried that. With the clock at 24mhz, the status LED indicator blinked noticably faster!

I hooked up my test motor and was able to run it up to 413 hz, where I ran out of voltage on my power supply. It ran fine at 413 hz. After running for about 5 minutes, the MCU was still cold to the touch. Everything ran smooth as silk with no abnormal current spikes.

In theory, the critical frequency should now be around 487hz with the clock running at 24mhz.
The crystal unit I used was a FPX24. Since it's surface mounted, I had to install it on the bottom of the board, which works fine.


That one is $1.91 from Mouser: http://www.mouser.com/Search/ProductDetail.aspx?qs=mzRxyRlhVdtKKC3YNoGMbA== There might be something in a 3 SIP that will fit exacly where the stock unit goes.

I dug around the junk box more and came up with a 32mhz one. I installed that, but no joy. Oscillator does not run. That does not mean the PIC can't run at 32mhz, but it won't with this particular part.

Anybody ever overclock a PIC before?

 

[tostino]

Would you say that this is a viable solution to the problem? or is it just some quick fix?

 

[The7]

I did an interesting test. The MCU is clocked by a 16mhz ceramic resonator. A PIC16F72 is rated for 20mhz. Digging around my junk pile, I ran across a 24mhz crystal, so I tried that. With the clock at 24mhz, the status LED indicator blinked noticably faster!

I hooked up my test motor and was able to run it up to 413 hz, where I ran out of voltage on my power supply. It ran fine at 413 hz. After running for about 5 minutes, the MCU was still cold to the touch. Everything ran smooth as silk with no abnormal current spikes.

This test shows that the followings are NOT necessary the cause of the critical frequency:
1) the LP filter of the Hall signal, and
2) the data representation in the MCU.

The shows that the critical frequency depends on the clock frequency.
The rate of executing the programming instructions depends on the clock frequecny??

Could you view the voltage waveform again at partial throttle to see if there are any BAD spikes?