RoboSavvy Forum

New Brainbot

2007-10-17T01:01:17+01:00

Well, of course the bus is limited to 1.0 mbps, so you aren't going to get better than that. What you might be able to do (and I'm going to be trying fairly shortly) is to run two FT232 chips off a single verdex, and each one will run at 1.0 mbps. The bus is the same speed, but you can talk to both at more or less the same time, so you get double the throughput.

Being able to split the bus is essential if you plan (like we do) to run 25-30 bus devices at 100 Hz.

- Jon

Statistics: Posted by JonHylands — Wed Oct 17, 2007 1:01 am

New Brainbot

2007-10-16T22:07:12+01:00

I will look in to usb and the FT232L in more detail tomorrow but would be very interested in it if we can push it above 2-3mbps then you could have a couple of atmel chips running of it.

What I meant by bottle neck is if you can get the usb bus faster then 1mbps then there will be a bottle neck of the 1mbps speed limit to the servos.

Bren

Statistics: Posted by Robo1 — Tue Oct 16, 2007 10:07 pm

New Brainbot

2007-10-16T17:59:16+01:00

No, mine would be:

Gumstix
|
FT232L
|
MAX485
|
RX/DX servo

The gumstix can talk over the USB port at 1.0 mbps.

Right now I'm using AX-12 servos, so my setup is this:

Gumstix
|
FT232L
|
Tristate/inverter
|
AX-12 servo/IMU/IO Board (any bus device)

Which end are you referring to for buffering, and what bottleneck?

- Jon

Statistics: Posted by JonHylands — Tue Oct 16, 2007 5:59 pm

New Brainbot

2007-10-16T16:42:38+01:00

Cheers for that Jon

I've just got my gumstix talking to my DX's . At the moment I'm just using the stuart and a max485. In your setup is it going

Gumstix
|
FT232L
|
FT232L
|
MAX485
|
Servo

How fast can you make the Gumstix talk to the USB. Do you use the same usb line to talk to the other atmels. After thinking about this afternoon it does seem like a good way to link up different parts, as I'm using SPI line to talk to the sensors (Arduino) and the stuart talking to the servos. But having them on one fast bus would be better. Is there anyway of buffering the signal at the other end to stop the bottle neck.

Bren

Statistics: Posted by Robo1 — Tue Oct 16, 2007 4:42 pm

New Brainbot

2007-10-16T13:38:22+01:00

Its just the linux driver for the FT232 - that's how it works. It creates a device called ttyUSB0, which you can access just like any of the other serial ports, except you can open it a 1.0 mbps.

To talk to the DX-117 (or the RX-series), you need the FT232 chip, plus an RS-485 chip (I just picked up a couple MAX1487 chips from Digikey for this). The FT232 data sheet has a nice schematic showing how to hook it up.

Once you get that, let me know, and I'll point you at the FT232 drivers for the gumstix. Are you set up to compile stuff like this on the gumstix? My brother takes care of that for me, so I just end up copying the binaries over, but depending on what buildroot you are using you need to recompile stuff.

- Jon

Statistics: Posted by JonHylands — Tue Oct 16, 2007 1:38 pm

New Brainbot

2007-10-16T11:54:38+01:00

Hi Jon

I know this is a bit cheeky but could I possible have a look at the code for the virtual serial port. I'm writing the code at the moment for my gumstix to talk to my DX-117's and want to run them at there fall speed of 1 mbps. Is it similar to your example SPI bit bang code.

Bren

Statistics: Posted by Robo1 — Tue Oct 16, 2007 11:54 am

New Brainbot

2007-10-16T01:52:26+01:00

The Lynxmotion track kit is by far the best track kit I've ever seen. For the price, I don't believe you can do better. If I was building a RoboMagellan robot, I would definitely use these tracks, although I would probably use bigger motors.

- Jon

Statistics: Posted by JonHylands — Tue Oct 16, 2007 1:52 am

New Brainbot

2007-10-16T00:37:48+01:00

Wow Jon, that is really cool!

So, how do the Lynx tracks hold up outside? Of course, I do realize you might not have enough time on them for an honest opinion but your post was several hours ago and I'm sure yhou been doing more experimenting (playing).

Carl

Statistics: Posted by quickster47 — Tue Oct 16, 2007 12:37 am

New Brainbot

2007-10-15T22:51:20+01:00

I'm not using the UART on the gumstix to talk to the bus. I'm using a virtual serial port running through one of the host USB ports. This is connected to an FT232 chip, which interfaces directly (using a tristate and inverter) with the bus.

I open the port from the gumstix at 1,000,000 baud, which the USB interface can do easily.

There is no USB network. The network is between the gumstix and my PC, and its wifi (802.11g) running somewhat faster than 1.0 Mbps.

The devices on the bus all have AVR chips inside, including the AX-12. My devices use ATmega168's, which can connect to the serial line in the bus directly, with no added hardware required.

- Jon

Statistics: Posted by JonHylands — Mon Oct 15, 2007 10:51 pm

New Brainbot

2007-10-15T22:30:23+01:00

Hi Jon

How did you manage to communicate with the verdex at 1mbps, I thought the uarts speed was 921kbps. Also could you spill the beans about the usb network. Are use using FT232L chips connected to all the different controllers.

Bren

Statistics: Posted by Robo1 — Mon Oct 15, 2007 10:30 pm

New Brainbot

2007-10-15T19:36:45+01:00

Congratulations, Jon

It is very interesting and amazing to me to follow your projects, specially your use of verdex.

Statistics: Posted by siempre.aprendiendo — Mon Oct 15, 2007 7:36 pm

New Brainbot

2007-10-15T19:14:34+01:00

So last night I finally got wifi running with the gumstix, and now BrainBot can tool around completely wirelessly. Today at lunch I made a short video of BrainBot driving around my driveway (I'm controlling it with a joystick attached to my PC).

phpBB [media]

The tracked base is being controlled with one of my bus I/O boards, using PWM through a Dimension Engineering motor driver board. The I/O board is controlled from the gumstix, which forwards commands from my PC over wifi to the bus device using an FT232.

Finally, the dream is here: 1.0 Mbps wireless from a PC to the Bioloid bus...

- Jon

Statistics: Posted by JonHylands — Mon Oct 15, 2007 7:14 pm

New Brainbot

2007-10-11T04:18:39+01:00

So, after a lot of delay, I finally got the new printed chest, and have the third BrainBot together...

Here's #2 and #3 together:

Hopefully within the next week or so I'll have the tracked base fully operational with wifi...

- Jon

Statistics: Posted by JonHylands — Thu Oct 11, 2007 4:18 am

New Brainbot

2007-09-28T16:17:20+01:00

The main advantage to the larger chest is the ability to put a head yaw servo inside the chest, between the shoulder servos.

As far as the hip yaw servo, I don't see any problems with the way it is done - you have a lot more trouble with the weight on the hip roll servo than the hip yaw servo (the one you're talking about). The robot doesn't weigh enough

If you were building a larger, heavier bot, I would use either a thrust bearing or a tapered roller bearing, and have the servo drive the rotation, but not hold any of the weight.

- Jon

Statistics: Posted by JonHylands — Fri Sep 28, 2007 4:17 pm

New Brainbot

2007-09-28T16:04:41+01:00

Jon

Like the idea of a larger main Torso. The Robotis stock mouldings may be neat but there is no space for addins/addons.

Everyone, myself included seems to have the Torso to leg rotation servos connected such that all the load is a shear on the top servo bearing. It would seem better from a loading point of view to utilse a U bracket arrangement, as most other joints.

I admit I can't see how to do this with any acceptable geometry. What are your thoughts on the loading on these servos?

Dewey

Statistics: Posted by Dewey — Fri Sep 28, 2007 4:04 pm