Replacing the CM-5

[Zypkin]

Hi guys,
I had a look at the previous posts here and I noticed that when talking of getting rid of the CM-5 you always turn to Atmel... is there any special reason???
Since you have to start from zero, why not to choose some faster MCU? ...it could help a lot to have some more computational speed... I mean the Atmega is ok... but 16MHz are not that much...

...what do u think?


Zyp

[JonHylands]

The main reason people are choosing Atmel is because those chips are what Robotis uses, and they have a built-in hardware serial port (at 8 or 16 MHz) that can talk at 1.0 Mbps, which is the bus speed.

Many faster processors have chosen a clock that maxes out their serial speed at 921 kbps, which is not fast enough.

If you want a really fast processor, then do what I'm doing, and use a USB system to talk to the bus - then you can either use something like a gumstix verdex (600 MHz), or a laptop to talk to the bus directly.

- Jon

[cdraptor]

Has anyone checked out the DiosPro which you can get from CrustCrawler. I notice that is comes with library for AX-12+. Was curious to hear any thoughts from people about this. Considering getting one to play with.

http://www.crustcrawler.com/electronics/diospro/index.php?prod=21

[limor]

There were a few microcontrolers that were considered as replacement for controling the AX12 bus. search this Bioloid forum for microcontroler (with 1 or 2 l's).
the 1mbps bus speed is an issue for most but Atmel.
Personally i prefer the idea of a "multi-rate control" mechanism whereby low-level control is done at the servo level, with with a fast communications bus, high-level control including adaptive control done by a faster processor that can be on board like a gumstix or remote PC with slower communications link and higher latency.

[Merlin]

Has anyone checked out the DiosPro which you can get from CrustCrawler. I notice that is comes with library for AX-12+. Was curious to hear any thoughts from people about this. Considering getting one to play with.

http://www.crustcrawler.com/electronics/diospro/index.php?prod=21

Nor Limor or JonHylands has given an answer to cdraptor question.
I am also looking into DiosPro and I wanna know about previous experiences from other builders... any opinion?

merlin

[limor]

The current "state of the art" is to use an FTDI USB-to-Serial chip.

Because : Most embedded computers have USB host controller and a device driver for FTDI.

You could use Jon's board or you could use an Arduino or Sparkfun board or any one of hundreds of FTDI based boards or slap something together. (on the serial side of the FTDI you short the Rx and Tx and hook up to the Bioloid bus. this has been heavily documented here on the forum.)

Why limit yourself to 40mhz and a massive board of the DiosPro?

I'm in the middle of setting up a gumstix (verdex) together with Jon's board to control the Bioloid (using Simulink as to design the closed-loop control software).

[JonHylands]

I didn't reply, because I haven't ever looked at the Dios line. If I was looking for a micro-controller to connect to the bus, I would almost certainly use an ATmega128. The DiosPro looks like a pumped up Basic Stamp, which means I would have no interest in it. Why run at 130,000 commands per second when you can run at 16 million instructions per second...

- Jon

[kennethw]

Why run at 130,000 commands per second when you can run at 16 million instructions per second...

I have an incredibly newbie question: Does the gumstix live on-board or are you tethering? I ask because a 600mhz processor must draw much more power... How do you power it in this configuration?

[Robo1]

Hi there

The gumstix doesn't draw that much power compared to the rest of the system and yes you put the gumstix in side the robot.

I also have a question for JonHylands about your USB interface. What do you do about the direction selector or doesn't the AX's need one. This is inregard my DX/RS485 setup.

Thanks Bren

[JonHylands]

The gumstix probably uses less power on average than a single AX-12.

Another interesting option that I'm going to be trying out soon is the Hammer (http://www.tincantools.com). It has USB host, so it should work just fine with the FT232 board I have.

You don't need a direction selector, because the FT232 takes care of that for you. It basically always runs in receive mode (from the bus) except when it is transmitting data.

For talking to the DX/RX RS-485 bus, the FT232 handles switching the direction pins on the MAX485.

- Jon

[siempre.aprendiendo]

I'm in the middle of setting up a gumstix (verdex) together with Jon's board to control the Bioloid (using Simulink as to design the closed-loop control software).

Would you publish your work?

I'm very interested about replacing CM-5

[kennethw]

The gumstix probably uses less power on average than a single AX-12.

I'm just curious how this setup (with the accompanying boards) compares with using the original controller, in terms of power consumption. I'm sure that it's hard to determine if the actuators are the major draw.

[JonHylands]

Guaranteed the actuators are the major draw. An AX-12 peaks at 800 mA, so it probably averages a quarter of that, which is 200 mA at 10 volts, or 2 watts. If your robot is moving a lot, its probably closer to half that, or 4 watts.

The gumstix (without wifi) draws somewhere around 300 mA at 5 volts, or 1.5 watts.

The CM-5 probably draws well under a watt, probably close to 0.5 watts.

- Jon

[kennethw]

Guaranteed the actuators are the major draw...

Thanks, Jon. Sorry to hijack the thread with my newbieness.

[limor]

I'm in the middle of setting up a gumstix (verdex) together with Jon's board to control the Bioloid (using Simulink as to design the closed-loop control software).

Would you publish your work?

I'm very interested about replacing CM-5

sure. thats the plan