Roboteq HBL-2360 + Invacare Brushless motors

[shirley_hkg]

I am starting to look into the feasibility of deploying the RoboteQ HBL-2360 controller on the newer version Gearless Brushless Heavy Duty motors by Invacare.

Experiments have shown that these motors can spin over 450rpm at 50V (not loaded). That converts a speed of 18 mph with 14” conventional wheelchair tyres.

I’ve been using these motors for 10 yrs . Reasons that I choose to stick to GB-HD motors are ,

1,They are capable of very precise control and movement , essential when approaching table on carpet floor, etc. They follow the joystick exactly, and precision is only limited by one's hand movement.

2,They are highly efficient that they can run prolong hours loaded w/o overheating. Range is really good too.

3,They run deadly quiet; passing pedestrians un-noticed.

My goal is to let these motors run at 15 MPH , on a pack of 48V LiFeSO4 lithium battery.

All inputs are welcome.

Regards

Shirley.

[shirley_hkg]

Summing up network information on this subject, the first obstacle is get over the one of its kind HALL SENSOR SIGNAL from the motor, converting it to signal compatible with RoboteQ HBL-2360.

People are puzzled with the 2 signals : EEPROM data & EEPROM CLOCK.
What do they do?

OR, they do nothing in respect of daily running at all , don’t they?

Below shows the pin assignments of the connectors of the old & new versions of GB motors.

Both motors can be operated by new generation controllers, but not vise versa.

Only the new motors have these 2 connections.

Would these 2 signals be only useful in calibration or kind of ?

[Irving]

They are factory calibration. As you know (or maybe not) the rotating bit of a brushless motor (rotor) is a ring of high-strength magnets rotating inside (inrunner) or around (outrunner) the stator windings. In an ideal world they, and the stator windings, are all identical and therefore the torque generated by a specific current in a winding is known. In reality it varies. by as much as 5% depending on quality. If you were building one off motors you can select magnets for consistency and adjust # of turns to take into account stator material variations and winding differences. In mass production this is much harder to do, so instead, during post-manufacture testing its becoming common practice to measure the torque for each rotor magnet/stator winding position and record it in an eeprom. The controller can read this data out as a serial stream and adjust winding current dynamically to give a smoother torque response.

[shirley_hkg]

OK!
That is a set of data referring to ultimate magnetic flux per standard current and is stored in the EEPROM for the use of controller. That enhances the correctness of timing and smoothness.

In reality, we did put older version motor, which has no EEPROM connection , in use with the newer controller. They run as they are supposed to be. As far as I remember, there were occasions that lack of torque were noticed. However, the situation seems becoming less significant when we re-calibrate the setup again, using the pre-installed CALIBRATION command in the controller.

Therefore, my questions are :

“ Will the hall sensor circuit board inside the motor output a weighted signal with incorporation of the EEPROM data ?

Is it possible to convert the hall sensor signal (with or without the utilization of EEPROM data) to one that’s compatible to HBL-2360 ? “

[shirley_hkg]

That is the pair of hall sensors and circuit board :

[Irving]

OK!
That is a set of data referring to ultimate magnetic flux per standard current and is stored in the EEPROM for the use of controller. That enhances the correctness of timing and smoothness.

In reality, we did put older version motor, which has no EEPROM connection , in use with the newer controller. They run as they are supposed to be. As far as I remember, there were occasions that lack of torque were noticed. However, the situation seems becoming less significant when we re-calibrate the setup again, using the pre-installed CALIBRATION command in the controller.

Therefore, my questions are :

“ Will the hall sensor circuit board inside the motor output a weighted signal with incorporation of the EEPROM data ?

Is it possible to convert the hall sensor signal (with or without the utilization of EEPROM data) to one that’s compatible to HBL-2360 ?

Answer to Q1: No idea, but I doubt it. Hall sensor is about positioning/indexing only, nothing to do with winding current. Its there to assist controller to work out relative position of rotor and stator at low rpm/stationary to enhance torque at starting/low speed when insufficient back EMF to work it out that way. Hall sensors not needed on motors that always run at a min speed of several 100+ rpm

Answer to Q2: Often on 3-phase brushless motors you have 3 hall sensors at nominally 120deg apart electrically (may not be physically so). Here you have quadrature signals, sin and cos (which may not be from a hall sensor but from an encoder disc). Both can be used for position sensing/phase commutation but the approaches are different and requires a different controller setup. They also have different characteristics regarding speed ranges and low speed torque control. Arguable quadrature is better at the latter around zero velocity and directional changes. Having said that, many controller chips can take both types of input. I doubt the Roboteq can be adapted if it doesn't already have quadrature inputs.

[Irving]

Addendum to my last post. According to the Roboteq 2360 data sheet it does have encoder inputs but mentions them in the context of its use as a servo driver for positioning. Whether they are of use here is hard to say without more research. I suggest you ask Roboteq...

[shirley_hkg]

Irving , thank you for your in depth reply.

[Burgerman]

http://www.amazon.co.uk/A3144E-Effect-S ... all+sensor

At worst case, you will need to remove the stock sensors and fit 3 hall effect devices. Each has power, neg, and signal. Not sure if you need linear or switching.

Added. Now I do! Those above should be OK?? But you probably already have them.
That's all you need.

http://www.mouser.com/pdfdocs/Selecting ... Motors.pdf

[LROBBINS]

I've never worked with any brushless motors, but one thing to bear in mind about the Invacare motors is that they are (comparatively) very low RPM. If quadrature encoding is better at low RPMs, it may not work out too well to substitute that with 3 magnetic sensors. Sounds like Irving's advice to contact Roboteq about quadrature encoding is on the mark.
Ciao,
Lenny

[shirley_hkg]

Thanks for all your input. That's beyond my knowledge honestly. I now seek help from a collage student major in electronics and robotics.

Will post here the outcome and obstacles.

Lastest finding is the EEPROM DATA & CLOCK are not connected to the wheelchair controller. They are for manufacture.

[Irving]

:P Thanks for all your input. That's beyond my knowledge honestly. I now seek help from a collage student major in electronics and robotics.

Will post here the outcome and obstacles.

You're welcome, and would be interested in the outcome

Lastest finding is the EEPROM DATA & CLOCK are not connected to the wheelchair controller. They are for manufacture.

Doubt that else what's the point in having it? IMHO the EEPROM is programmed with data recovered during testing through a programming 'pin' on the PCB. More likely these motors are not specific to Invacare and someone else does want the data output... which therefore suggests the motors are available through another route....

[shirley_hkg]

It is good news , itsn't it ?

[Irving]

It is good news , itsn't it ?[/color][/size]

It could be, but finding who might not be so easy, got to find the OEM first.... there are ways...

[shirley_hkg]

Log the waveform of the sensor of motor.

[shirley_hkg]

This IC seems to be able to convert the analog signal to the u/v/w 3-phase commuter signal for RoboteQ itsn't it ?

http://www.ichaus.de/product/iC-TW2

[Burgerman]

I am surprised the roboteq wouldn't just see that as a square wave and work normally anyway. But maybe not.

[Irving]

This IC seems to be able to convert the analog signal to the u/v/w 3-phase commuter signal for RoboteQ itsn't it ?

http://www.ichaus.de/product/iC-TW2

Yes, that chip would do it. There are some limits to its stator tooth/magnet ratio emulation but for common motor combos it should be fine. Good find, might look at that myself as have been looking at a motor (not Invacare one) that has encoder option but not a sensored option. Tho adding an encoder is an expensive option its physically easier, and UVW for torque control is marginally easier software wise than A/B

I've not come across this company before, they have some interesting devices..

[shirley_hkg]

Will need an adapter to connect the IC to computer for programming, but it costs $250.00.

Is there any other economical alternatives to do it ?

http://www.ichaus.de/product/MB3U

[shirley_hkg]

I am surprised the roboteq wouldn't just see that as a square wave and work normally anyway. But maybe not.

They told us there will be new model which will accept analog sin/cos signals , in the coming year.

However, they said the handling current will be 50A per channel.

[Burgerman]

iS THAT:

50 total divided into 3 wires?
50 per wire?
50 equivalent to a brushed system?
Or 50 equivalent to one wire on a 2 wire system?

Or something else? How does it compare is WATTS compared to a brushed controller?

[Irving]

On a brushless motor two out of three windings are energised at any one time, and the current is PWM'd for torque control, while frequency = speed. you need low resistance and low inductance as current = torque and inductance is a killer when it comes to speed as you need the current in the windings to rise fast. Obviously the more stator/pole pairs you have the shorter the time to energise so high volts, low current is preferable. A 50A controller might be perfectly viable with 72 or more volts...

Those with a serious interest in BLDC motors should read this, especially the derivation of their torque characteristics from pages 37 to 46.

[Burgerman]

That beyond me.

Can we compare two 50v controllers.
Brushed 150A
Brushless 80A

Are they equal?

[Irving]

That beyond me.

Can we compare two 50v controllers.
Brushed 150A
Brushless 80A

Are they equal?

You can't easily make that comparison as it depends on the motor.

The motor equations for brushed and brushless are similar. Torque is proportional to motor geometry (length, radius, magnetic flux, number of turns of wire) and current. A brushed 2-pole motor only has one set of windings energised at any time interacting with 2 magnetic poles, whereas a brushless has two windings energised, but more importantly brushless motors have multiple stator teeth/pole pairs and the windings cover multiple stator teeth and so interact with more magnets. As a result, for the same physical geometry, the brushless motor theoretically generates more torque for the same current, or the same torque with less current, assuming a higher drive voltage. Unlike brushed motors, the speed of a brushless motor is not a function of voltage, but controlled by the commutation rate, therefore high voltage/low current/high torque operation is feasible. So comparing controller current output capabilities is not simple/relevant.

[Burgerman]

Mmm. So...

Forgetting motors, when they say "80 Amp" what does it refer to?

What I am really asking it which controller delivers the most total watts, or which takes the largest battery amps maxed out?

So forgetting how motors work, just controller capability in max watts?

EG the roboteq brushed cont. can do 150A x2 at 100 percent pulsewidth meaning it can deliver 150A x 2M x 50V - 15K Watts or 300 Batt Amps @ 50V maxed out.

How would the Roboteq 80A brushless controller compare? I don't know what the 80 is referring to, or how to do the math to convert that to 100 percent pulsewidth.

[shirley_hkg]

Will need an adapter to connect the IC to computer for programming, but it costs $250.00.

Is there any other economical alternatives to do it ?

http://www.ichaus.de/product/MB3U

ANY THOUGHT OF THE ABOVE ?

[DMC-Man]

It should be possible to build your own USB to I2C adaptor. Checkout page 3 of: http://www.ichaus.de/upload/pdf/MB3U_MB3U-I2C_B5en.pdf to see the circuit diagram. The iC-TW2 operate on I2C no BiSS/SSI is needed. The GUI software for the PC is availabe for free and could work just with I2C.The iC-TW2 inputs are sine/cosine from sensors. Your might see also: http://www.ichaus.de/BLDC_Integration_Trends if the motor has an integrated single chip encoder.

[Irving]

Will need an adapter to connect the IC to computer for programming, but it costs $250.00.

Is there any other economical alternatives to do it ?

http://www.ichaus.de/product/MB3U

ANY THOUGHT OF THE ABOVE ?

As far as I can make out the MB3U is a USB to I2C adaptor from ichaus and uses the opensource BiSS protocol. The GUI and comms stack is freely downloadable and works with any 3rd party product that uses the BiSS protocol, see http://www.biss-interface.com/. The adaptor is $188 + shipping

However, the datasheet for the TW2 chip gives you everything you need to know to program the chip via a I2C interface (SDA/SCK 2-wire), the BiSS software is just a convenience and 90% is irrelevant to this need, which is just tuning the 3-phase interpolation. All you need to be able to do is read/write the relevant registers on the TW2 chip. In which case there are several USB-I2C products on eBay or from robotics hobby outlets that will do the job at around $20, or roll your own. You could even 'bit-bang' a couple of parallel port output pins to do low-speed SDA/SCK directly which would be adequate for this and near-zero cost.

[Irving]

Mmm. So...

Forgetting motors, when they say "80 Amp" what does it refer to?

What I am really asking it which controller delivers the most total watts, or which takes the largest battery amps maxed out?

So forgetting how motors work, just controller capability in max watts?

EG the roboteq brushed cont. can do 150A x2 at 100 percent pulsewidth meaning it can deliver 150A x 2M x 50V - 15K Watts or 300 Batt Amps @ 50V maxed out.

How would the Roboteq 80A brushless controller compare? I don't know what the 80 is referring to, or how to do the math to convert that to 100 percent pulsewidth.

Assuming they are spec'ing it the same way as the HBL2350/60 it means 80A motor current through the 2 series windings as per this diagram in the manual. So 80 x 2M x 60v (HBL2360 equiv) = 9.6kW but you get double the torque as 2 windings active rather than 1. With a 3-phase controller you need 6 sets of MOSFETS per channel rather than 4 sets in a brushed controller so less space since box is same dimensions hence (maybe) reduced output capacity.

brushless.GIF (21.08 KiB) Viewed 8963 times

[shirley_hkg]

Thank you DMC-man & Irving for your replies.

Another 2 questions.




Q1, the log above is captured when motor was spinning at around 170 rpm (6.5mph @ 13.5" tyre)

If chair goes 15mph , motor will be spinning at 400 rpm. The data rate from the hall sensor will also increase.

Can you tell from the graph that HBL-2360 has enough bandwidth / processing power to do it ?


Q2, In CLOSED LOOP SPEED MODE with brushless motors, the controller will keep motor speed to the command with variance in load. That seems to have the same effect of the CLOSED LOOP TORQUE MODE , itsn't it ?

Then, which mode should be best to drive a wheelchair ?

Thanks.