Brushless DC with toshiba TB6537 ?

[Cannibal]

Myself and two of my friends at the office (grad. students) are planning to play around with a small UAV this summer, since this seems the most fun way to waste our time and money. Our research group is a navigation group, so each of us has experience with building inertial navigation systems and often integrating them with sat nav. With this in mind we are confident of navigation systems, but none of us has very much experience with motor control.

We're looking into using brushless DC motors on a quad rotor platform, but from what we have read most commercial controllers do not have the update rate desirable for smooth control. Our inertial systems function at 100Hz+ so we were hoping to have the control loop work at this speed as well, and for this reason I'm looking into building a BLDC controller capable of supporting this update rate (I have a bachelors degree in EE, but I've not touched motor control for a long time, and never outside of a tightly controlled lab).

The best option we have come up with so far involves using a toshiba TB6537

http://www.toshiba-components.com/motorcontrol/pdfs/TB6537F_N.pdf

to drive a set of BTS7960 half bridges (integrated drivers)

http://www.infineon.com/dgdl/BTS7960_Datasheet.pdf?folderId=db3a304412b40795011 2b408e8c90004&fileId=db3a304412b407950112b43945006d5d


My main concern is that the TB6537 sheet 'reference design' does not show us any information on what parts would work for the suggested BJT layout, so I'm not immediately comfortable with switching to a pre-driver+MOSFET based layout. Judging from the voltage waveforms shown for high and low, it should work, but experience can be expensive when you're wrong. Does anyone have any experience with the output levels from the TB6537 when driving a High -Z load?

My secondary concern is that the switching times for the output drive section does not seem to be defined anywhere in the document - for example, what is the minimum on and off time that the output drivers will be forced to switch - I'm asking since the Half Bridge I am interested in has a turn on/ turn off time of a micro-second, and the *maximum* cycle time of the controller is 1us with the slowest possible oscillator. Should the TB6537 switch at fosc, I would be in trouble here ;)

The current feedback circuit is easily modified in my eyes, so I won't dwell on that too much, I'll just have to keep it in mind after I've selected a motor.

Thanks in advance for your input.

-Cannibal

[David T]

The new Spektrum DX7 'SE' claims an 11ms frame length. Not far off your 100Hz. I've not seen them warn people that normal servos and controllers won't work. So I'd buy a cheap ESC first and test. A lot easier than going the DIY route.

Search this forum for the conversion of commercial ESC's to DIY software. Also easier and cheaper than starting from scratch.
David.

[Cannibal]

Hi DavidT,

Thanks for the input - I had read up on the procedures for modifying one of the commercially available ESC's to take commands at a high rate over an i2c bus (I think it was a 'towerhobbies' model). I have only dealt with PIC 16,18,30,33 series, ARM7, ARM11, and 68HC11, never with atmel, so I consider that route as a potential excuse to learn a bit more about atmel products.

In the mean time though, I've discovered an application note for the TB6537 which joy of joys shows the parts and parameters used! If you're interested, it's located here:
http://www.marktechopto.com/pdfs/Toshiba/ACFAAB.pdf

Now that I've been able to look up the type and specification of transistor they're using, I'm considering going this route more seriously, but I'm still somewhat on the fence. My remaining questions include wondering if the TB6537 will callibrate its duty cycle to stay within the overcurrent limit that is set via the feedback loop, or if it will be up to me to determine this and modify my input throttle accordingly - On the one hand This is a hassle, on the other hand it looks like commercial controllers are also a hassle to program.

-Cannibal

[tune by tito]

cannibal you not need the toshiba TB6537 to be able to drive the BTS7960 half bridge. the BTS7960 have every thing you need,

check datasheet page 4:
The BTS 7960 is part of the Novalith IC family containing three separate chips in one
package: One p-channel highside MOSFET and one n-channel lowside MOSFET
together with a driver IC, forming a fully integrated high current half-bridge. All three
chips are mounted on one common leadframe, using the chip on chip and chip by chip
technology. The power switches utilize vertical MOS technologies to ensure optimum on
state resistance. Due to the p-channel highside switch the need for a charge pump is
eliminated thus minimizing EMI. Interfacing to a microcontroller is made easy by the
integrated driver IC which features logic level inputs, diagnosis with current sense, slew
rate adjustment, dead time generation and protection against overtemperature,
overvoltage, undervoltage, overcurrent and short circuit. The BTS 7960 can be
combined with other BTS 7960 to form H-bridge and 3-phase drive configurations.

for a brushless ESC you will need 3 BTS 7960, 1 microcontroller and few external components, however the recomendation of David T is easy and very cheap, when you do a DIY project you have to take many variables into consideration: pcb layout, micro software, components, plus the price of put every thing in one functioning device.
best regards tito

[Cannibal]

Hi tito,

Thanks for the input, but do note that when I said I was considering using the TB6537 to drive the half bridges, I should have said to 'Control' the half bridges - Essentially to serve the purpose of the micro.

At the same time though, I've been looking over the microchip application note that uses the DSPIC30f2010 as a brushless motor controller - it provides all of the code for a very nice serial interface where one can specify the number of poles etc for the motor at hand. I could play around with it in terminal, and then when ready, modify it to work over SPI or I2C.

Finally someone over at the sparkfun forums pointed out a "Dynam 25A Electronic Speed Controller ESC for Brushless Motors" that they say can function up to 200Hz. Maybe it would be best for me to pursue this first and skip the PCB design/code fiddling.

Which route would you follow?

Thanks again for the input.

[tune by tito]

Cannibal if you check the data sheet for TB6537 the IC need a Crystal of 4Mhz and also will need a microcontroller for driving TB6537 PWM lines and direction, for that reason i do not recomend using the device, because you can use a micro to do the same function.
Best regards

[Cannibal]

Hi Tito,

I am aware of the need for the pwm and direction control to come from outside - I'm very comfortable with embedded system design, but trying to get away from doing the development from the ground up for motor control - I have apprehension about the level of work involved in doing this properly from the ground up.

In terms of using a micro (my favorites are dsPIC30 and 33's), the PWMH and PWML pairs on the motor control pics are only 3 - so I think I would need a micro for each channel plus the 'main' controller to interface with other equipment and wireless.

I'm leaning more and more towards off the shelf ESC's that are known to work at 100Hz - in the last few days some helpful people over at sparkfun have pointed out a few that are worth consideration.

[tune by tito]

Go a Head, Be all you can Be.
Best regards tito