Researching BLDC w/ Hall sensors

Basil

Hi All,

I'm in sales, and a customer of mine found out about my interest in electronics and tasked me with designing a BLDC controller to drive 4 motors (Maxon or Anaheim automation).

This is for a piece of farming machinery and has extremely simple requirements. Power applied, run all 4 motors at a constant 500RPM (ramp up 1 second). Power of, stop. Every off the shelf product is too feature rich and thus too expensive.

As I have no experience designing motor power circuits and firmware I'm wondering if there is a thread anywhere on this board I could read to get some prop specific tips. I'm busy reading everything i can on the Web.

I'm using search function but there are allot of threads which mention BLDC to get through and not many mention Hall sensors.

I explained I am not a qualified engineer and hes happy for me to give it a go anyway quite excited really

T Chap

I have a system with bldc and Anaheim. The prop drives a motor drive that uses encoders for precise movement. All the prop needs to do is generate a voltage ( a counter works easily ). Anaheim has motor drivers that are pretty simple and easy to use, why not use theirs?

The solution is to use an mc33035 bldc driver into an hip4086 MOSFET driver. Irf540 mosfets.

Basil

We are currently using Anaheim controllers for the prototype and are very happy with them. However we are in New Zealand, and once you consider freight and that we will need 4 controllers per truck the cost adds up. The motors themselves are already more expensive than his existing solution (hydraulic drives) but he wants to go electric as it offers many benefits. So im looking at ways to reduce cost,.

This may not go ahead once he gets to production, but in improving his product he'd like to consider all options.

T Chap

So are you looking to design and make your own board?

Basil

Yeah, thats the plan. I first thought of the prop because im familiar with it but if there are non MCU options thats fune.

T Chap

Well you would have to define a little more clearly the ultimate objective. Do you already have a source of the control voltage? Does the new design require any communication to other devices that require i/o, serial, LCD displays, button inputs, safety controls, rpm/speed control, etc. In terms of the Prop being used to drive other mosfet drivers or dedicated BLDC controllers, that is easy. As far as the Prop reading the hall sensor and turning on the mosfets, that is doable but I haven't seen much talk about BLDC drivers, although there may be some o the obex. Personally I prefer a dedicated controller and mosfet driver with the Prop doing the encoder management and speed control via DAC(counters in a duty mode). It is easy to ramp a voltage up or down with a counter, but then again you can ramp a voltage with an RC network also. Bottom line is, the Prop is an easy solution to do what you need to do.

Basil

T Chap wrote: »

Well you would have to define a little more clearly the ultimate objective. Do you already have a source of the control voltage? Does the new design require any communication to other devices that require i/o, serial, LCD displays, button inputs, safety controls, rpm/speed control, etc. In terms of the Prop being used to drive other mosfet drivers or dedicated BLDC controllers, that is easy. As far as the Prop reading the hall sensor and turning on the mosfets, that is doable but I haven't seen much talk about BLDC drivers, although there may be some o the obex.

Nothing complicated at all really. 24VDC supply (truck batteries), single external toggle switch to turn on/off. Single Pot to adjust RPM (same for all motors), single pot to adjust ramp.
Motors are the only 90W each, peak draw is under 5A from memory.
I might think about a small LCD to display the 4 actual RPM readings and thats all.

Oh and an ESTOP.

T Chap wrote: »

Personally I prefer a dedicated controller and mosfet driver with the Prop doing the encoder management and speed control via DAC(counters in a duty mode). It is easy to ramp a voltage up or down with a counter, but then again you can ramp a voltage with an RC network also. Bottom line is, the Prop is an easy solution to do what you need to do.

That's along the lines of what I am thinking. I like the idea of using a prop to control the pwm voltage and closed loop as it allows me to easily fine tune in software.

Ive been reading a few whitepapers from TI, Atmel and freescale among others and it doea indeed sound straight forward.

StefanL38

Hi,

in the model aircraft section you can get motors up to 500W. 3-phase BLDC driver circuits for 10A start at $10 / piece.
The only thing I do not know exactly is if the already give rpm-feedback to get a constant rpm under changing loads.
But this can be done by the hall sensors. So all your propeller-chip has to to is create a model-servo-compliant signal
pulswith controlled be hall-sensor rpm-feedback. Or even use a magentic encoder-chip from austria microsystems
http://www.ams.com/eng/Products/Magnetic-Encoders
https://forum.sparkfun.com/viewtopic.php?f=5&t=20813

so most parts can be bought right of the shelf. What cost-limit do you want to stay below?

best regards
Stefan

Basil

Hi Stefan,

If I can halve his costs he would be happy. The current controller from Anaheim is $179USD and he would need 4 per truck, so I have a rather easy goal of coming under $360USD for my 4 motor controller. I'm aiming for under $300.

The key with this application is that it needs to be reliable. It will be operating for years on a piece of farming machinery inside a sealed enclosure (to protect from fertiliser). This means heat tolerant, vibration tolerant, etc. Industrial grade essentially.

I have mentioned non-industrial grade products but my customer is not happy with that suggestion for the above reasons, and to be honest i wouldn't be happy selling them to him. This is why brushless motors were chosen as opposed to cheaper brushed motors. And why Maxon/ Anaheim motors and controllers were chosen despite their relatively high price compared to hobby products. Yes, this is price sensitive, but within reason

The load on the motor can and will vary, hence the need to a closed loop system and using motors will hall sensors.

T Chap

How many systems estimated to be built? Can all 4 drivers be a single board or are they separate boards? What are the cable lengths from each driver to the motor? Are all controls located at a single control panel? Since you already have built in hall sensors, you don't need encoders(easily $40-50 each). You can bring one of the hall sensors into the Prop and count revolutions. Assuming no gearbox is used, this will give you the RPM at the shaft. The hall sensor voltage on the Anaheim motor/controller is 6.3VDC, so you have to buffer or divide it to get it into the Prop. Also, when you speak of load on the motor, are you putting a radial load on the shaft? Under 75$ is easy as far as the board. More like 40 - 50$

Mark_T

You mention 24V, 90W, so that 5A is the working current - but the stall current is likely to be much
higher, perhaps even 100A. You'll need to implement current sensing and over-current limiting/shutdown.

Finding the right 3-phase controller chip would be a smart move, there's probably one
that does all you need.

There are some good MOSFET drivers for 3-phase, I've used the FAN7388 before and it
seems to perform well, its a reasonable back end (add 6 n-channel MOSFETs, 3 caps and 3 diodes).

For reliability make sure everything runs nice and cool, pay attention to fault conditions (fast
response to overcurrent, TVS to suppress over voltage, generous voltage ratings on MOSFETs).

PS I'd be very surprised if the hall-switch outputs aren't open-collector - should be easy to
interface too - be sure the hall sensor signals come in one a separate cable from the motor
windings, and a shielded one if possible.

T Chap

http://www.digikey.com/product-search/en?x=22&y=7&lang=en&site=us&KeyWords=+480-1999-ND%09+


This is the hall on Anaheim motors.

Basil

Mark_T wrote: »

You mention 24V, 90W, so that 5A is the working current - but the stall current is likely to be much
higher, perhaps even 100A. You'll need to implement current sensing and over-current limiting/shutdown.

While we are testing an Anaheim motor, they are leaning more towards the Maxon motor due to the flat construction(EC FLAT 90W). It has a starting current of 69A (we are using the 36V motor sorry), 4.76A nominal.

Mark_T wrote: »

Finding the right 3-phase controller chip would be a smart move, there's probably one
that does all you need.

There are some good MOSFET drivers for 3-phase, I've used the FAN7388 before and it
seems to perform well, its a reasonable back end (add 6 n-channel MOSFETs, 3 caps and 3 diodes).

Ill start a search. That's looking like the way to go, with the prop telling it what to do.

Mark_T wrote: »

For reliability make sure everything runs nice and cool, pay attention to fault conditions (fast
response to overcurrent, TVS to suppress over voltage, generous voltage ratings on MOSFETs).

PS I'd be very surprised if the hall-switch outputs aren't open-collector - should be easy to
interface too - be sure the hall sensor signals come in one a separate cable from the motor
windings, and a shielded one if possible.

T Chap wrote:

http://www.digikey.com/product-searc...80-1999-ND%09+


This is the hall on Anaheim motors.

Thanks for the tips.

Basil

T Chap wrote: »

How many systems estimated to be built? Can all 4 drivers be a single board or are they separate boards? What are the cable lengths from each driver to the motor? Are all controls located at a single control panel? Since you already have built in hall sensors, you don't need encoders(easily $40-50 each). You can bring one of the hall sensors into the Prop and count revolutions. Assuming no gearbox is used, this will give you the RPM at the shaft. The hall sensor voltage on the Anaheim motor/controller is 6.3VDC, so you have to buffer or divide it to get it into the Prop. Also, when you speak of load on the motor, are you putting a radial load on the shaft? Under 75$ is easy as far as the board. More like 40 - 50$

Initially 10-15 systems will be built, after that maybe 10-20 per annum would be a conservative estimate. Our volumes over here aren't quite up to US standards
I intended to put all 4 drivers on a single board. I considered individual boards so if one driver fails just that board can be replaced. But ultimately I don't want anything to fail so...

Cable length is currently 3-4 meters, but this was not my idea. I will be recommending he move the controller much closer to the motors (say within 1m) and separate the hall from the motor cables.
Controls (on/off switch as it relates to this project) will be in the cab, motors at the back of the truck. So there will be a single cable running the length of the truck, but its only for a push button/toggle so no big deal.

From the point of view of the motor, the load is a flywheel with varying mass (+/- 10% while operating), I need to keep the RPM constant. I cant go into details as i'm under various confidentiality agreements sorry but I hope this is enough of a description.
There is no gearbox to remove a point of failure. Even sealed gearboxes dont like hanging out on the back of a fertilizer truck.

Basil

Ok this is such a dumb question, but what's the reason all these 3 phase drivers are being suggested? Is there a reason one cant just use standard transistors to drive the MOSFET?
And why do they have such high voltage and current ratings if all they are doing is turning on/off the MOSFET?

This questions is due to my lack of understanding of how FET's work...

Mark_T

Basil wrote: »

Ok this is such a dumb question, but what's the reason all these 3 phase drivers are being suggested? Is there a reason one cant just use standard transistors to drive the MOSFET?
And why do they have such high voltage and current ratings if all they are doing is turning on/off the MOSFET?

This questions is due to my lack of understanding of how FET's work...

They perform several functions in one package, saving a lot of complexity:

Drive the gate hard - charge up and discharge the gate capacitance with high
currents to get fast switching - with power electronics you seek to tame the
switching losses which go up as the load power and PWM frequency and switching
time.

Provide a bootstrapped supply to a floating high-side driver (some chips can handle 600V
or more). This is because you use all n-channel MOSFETs in a high performance bridge,
p-channels are about 3 times worse in performance due to the low mobility of holes compared
to electrons. An all-n-channel bridge requires bootstrapped floating supplies to the high-side
switches - typically you add a few diodes and capacitors.

Handle shoot-through prevention and or dead-time generation to prevent catastrophic
device failure due to shoot through current / power spikes. Chips differ a lot in what
they provide here.

Provide hysteresis on the inputs to help prevent false triggering due to induced
switching noise

Provide under voltage shutdown for the gate driver supplies. Without this the gates
might float into the linear region if the 12V supply was failing, causing the MOSFET to
explode as its dissipation jumps to kW levels.

Allow the bridge to be separately powered from the gate drivers without
any special interlock.

Some also have zener protection on the gate/source drive to guard against gate
oxide punch-through.

Basil

Good enough for me

Thanks Mark, ill try translate and digest that!

Sounds like it makes life easier picking a MOSFET. I won't have to worry about input voltage being too low for output voltage for example. Low Rds on not as critical etc etc?
So in otherwords, for my application with a 100A peak (5A nominal) and 24V supply, I can just choose a suitable MOSFET with higher voltage and current capabilities (and thermal considerations) and ill be ok?

I guess switching time would be another considerations which brings me to another question.
To control speed/torque, looks like ill need to provide a PWM signal to whichever phase is active. Is there a nominal frequency? Or should I just go as high as the driver/MOSFET will allow?

ErNa

you will need PWM, 20kHz will be fine, as this balances switching and conducting losses of the mosfets. The motor current will show some ripple and the higher the pwm frequency, the less ripple you will have, also minimizing losses. 20 kHz is not audible too, but in this application noise should not be a limiting factor ;L-)

Basil

ErNa wrote: »

you will need PWM, 20kHz will be fine, as this balances switching and conducting losses of the mosfets. The motor current will show some ripple and the higher the pwm frequency, the less ripple you will have, also minimizing losses. 20 kHz is not audible too, but in this application noise should not be a limiting factor ;L-)

Great thanks
I have a plan in mind which would use 4 pwms anded with various masks to give me the required 6 outputs per motor. Shouldn't be too hard on the prop.

And no, noise isnt a concern haha