Brushless Servo commutation chips

[JiriK]

Hi!

I´v been working on my own CNC mill for a while now.. The mechanical parts are at the stage where I can start building control for my servos.

http://www.saunalahti.fi/~jikuuk/Images/Project/cnc/
(In finnish..)

Most of the parts in my mill, including servos (Dunkermotoren BG63S55) I have found from local junk yard.. So no complete drivers here.

I found these custom made high current drivers along with the servos that seem to work using PWM and direction signals.. But current limit is a bit of a problem. Also, these don´t spin as smooth on low rev as I´s like them to spin.

So, I´v started to do some research on building my own controller cards.
Basic plan is to use a NSC LM629 as the PID processor, and a PIC or an AVR micro to control it.. Or then I make an ISA- bus card and let the computer control the LM629´s directly.

But that high current drive of the brushless motors..
They seem to be two phase motors, two coils like on a bipolar stepping motor + 2 hall- sensors for position feedback. Those two Hall- sensors don´t give enough precision, so I´m wondering wether there might be a chip somewhere that could count the steps from the rotary encoders (Optical, 1000steps/rev) of my motors and use them as reference.

Have you seen such chips?
Or do I have to add another uC to the driver board for the commutation...

[ViperTX]

Search for quadature decoder chips....SGS Thomsen, Freescale, HP, etc.

So, are you going to write your CAM software (CNC stuff)?

[JiriK]

I try to avoid codewriting as much as possible.. So I plan to use some freeware- or commercial CAM- software. Step / Dir - control would be simple to connect to PC.. But I´ll leave door open for other interface methods.

Performance Motion Devices, Inc. (www.pmdcorp.com) seems to have pretty good processors for my application... We´ll see how expensive they are.

[Al_The_Man]

They have an interesting motion card. I was curious as to what kind of motors you have there as typically DC brushless resemble three phase, what kind of commutation are you going for? One downside of the typical DC brushless motor and controller is the 'cogging' at low rpm due to the coarse switching of 6 step trapezoidal.
Al

[JiriK]

Datasheet calls them four phase brushless DC-motors.
Spec:
Dunkermotoren BG63 x 55S
24VDC / In = 6.1A, Ifm = 93A
Nominal torque 28Ncm
Starting torque >170Ncm
Nominal speed 3200rpm
Motor weights 2,0kg + brake + encoder.

Connector has has 8 pins:
Phase A,B,C,D
Hall sensor 1,2
and supply for Hall- sensors.

There seem to be two phase coils, Between pins A-C and B-D.
So, this requires a bipolar drive (2 H-Bridges)

This package also contains a brake (more like a lock) and a 1000step / rev encoder.

For smoother drive I´ll propably need sinusoidial (sorry if I misspelled that)
drive wich takes it´s reference from the encoder (and possibly from the Hall- sensors)

I also have a couple of Maxon EC40 3-phase brushless motors (118896)
but no encoders for those...

[da21]

here's a link to a brushless controller with software for an avr
although it is primaraly for modelers , i dont see any problem in expanding the design and the current rating to suit our requirements , at least it's a good start

http://people.freenet.de/khruehlow/T..._Platinen.html

if people are interested in brushless controllers , i could be persuaded to look further and build a prototype

[JiriK]

da21, please build a prototype brushless motor driver
I´m not the only one who has this kind of motors..

Back to the chips, I have some NSC LMD18245´s here.
They contain some of the needed high power circuits, including a PWM- current control etc. But they just don´t have enough capacity for these motors I have (3A vs 6A)
I wonder.. Are there other similar devices that have same kind of logic & control circuits like in LMD18245´s but with driving circuits for external power devices?

[unterhaus]

Originally Posted by JiriK Most of the parts in my mill, including servos (Dunkermotoren BG63S55) I have found from local junk yard.. So no complete drivers here. I found these custom made high current drivers along with the servos that seem to work using PWM and direction signals.. But current limit is a bit of a problem. Also, these don´t spin as smooth on low rev as I´s like them to spin.

That's a very nice machine. Are you sure that you can't use the drivers you found? Maybe modify them? Better is the enemy of good enough.

Originally Posted by JiriK But that high current drive of the brushless motors.. They seem to be two phase motors, two coils like on a bipolar stepping motor + 2 hall- sensors for position feedback. Those two Hall- sensors don´t give enough precision, so I´m wondering wether there might be a chip somewhere that could count the steps from the rotary encoders (Optical, 1000steps/rev) of my motors and use them as reference. Have you seen such chips? Or do I have to add another uC to the driver board for the commutation...

These are 4 phase motors, not 2. Two phase would have two independent coils, these are not independent. I have a BG83S. I believe you only need 2 halls to handle the 4 states of the motor.

There are chips from LSI computer systems that will handle the quad encoder. The ls7266 counters will handle 2 encoders. THey also make chips that will output to a counter, which would be nicer if you have enough counters on your processor. http://www.usdigital.com/products/lsi/

Seems like motorola/On semi/Freescale semi made commutation chips that would handle 4 phase, but probably only using the halls. To get better commutation, you may need to go to one of the dsps that are made for this. TI, Microchip, Analog Devices, and Freescale all make devices that will suffice. All 4 companies make devices that have quadrature counters built in. Siemens does too, I believe.

Freescale has some very nice app notes. Wish I could cite them for you.

[JiriK]

I´m not 100% sure wether I can use the drivers or not.
At their current condition they are pretty much useless,
current limit shuts down entire circuit and locks the motor.
(Brake is controlled with the same board) Turning power off and back on again restores the drive. There doesn´t seem to be a way to reset the board with power on..

Aboat the motors, These BG63 x 55S´s have four pins dedicated to phase coils, A,B,C and D. Coils seem to be between A-C and B-D. Other combinations, like A-B, B-D aren´t connected. Also, the drivers I have seem to contain two H-bridges, like the datasheet says: "Electronic commutation is executed by means of bipolar selection of the 4-phase-motor winding by employing 8 transistors in the output stage."

What kind of output transistor (FET) arrangement your drivers have?
Maybe this is just us talking aboat the same thing using different terms..

I called to local PMD corporation reseller.. Good devices but a bit too expensive for me.. ~150€ / axis At that cost I think I can spend some time and make the drivers myself, even if that requires for me to learn a lot of new things.. That´s no problem, because projects like this tend to be also for educative purposes.

[CNCAddict]

hmm...I've never heard of a servo motor that isn't 3-phase. You may have stepper motors unless that 4th wire is a center tap to the delta or wye winding. Oh, and your machine looks amazing. I just wish I had acces to all those parts...and equipment

[unterhaus]

Originally Posted by CNCAddict hmm...I've never heard of a servo motor that isn't 3-phase. You may have stepper motors unless that 4th wire is a center tap to the delta or wye winding. Oh, and your machine looks amazing. I just wish I had acces to all those parts...and equipment

There's nothing stopping a motor from having more phases. I just saw a 3 phase stepper on Ebay, and these are definitely 4 phase brushless. However, it doesn't sound right that there isn't continuity between all 4 phases. I'll have to check my motor.

[CNCAddict]

There are no need for more than 3-phases in a motor. I think there are very slight efficiency gains going over 3-phases, but that is lost in the very complex controller needed to run such a motor. I personally would stay away from odd motor configurations because finding a controller would be near impossible.