Manufacturing the Bioloid I/O module

[JonHylands]

I have been including the buffer but could just use code like you. However, I suspect that Robotis use the buffer to improve line driving and possibly even processor protection. Have you found any problems running without the buffer with lots of servos on the same line? Robotis must use them for a very good reason.

I haven't found any issues with it, and I've been driving 19-20 servos. Its probably not as well protected as using the buffer, but its worked pretty well so far.

I personally have no problem with surface mount either - but I know there are a lot of people out there who do have a problem with it.

- Jon

[limor]

OK so at least two people said they would buy 2 BioSensor-boards at US$30.
4 down 496 to go!

I think this board should be based on pepper's version-1 concept board.

Very simple tiny board: Atmega8 with the RX,TX switched by PD6,PD7 (isnt just one PD enough to switch between RX state and DX state?..) to drive the 1-wire interface. The driver circuitry may be implemented by other than that particular driver IC used by Robotis AX12. (but knowing Robotis are good at what they do, I suspect switching RX/TX in software like JonHylands succesfully does, may be prone to problems)

As for adding a gyro or accelerometer to the BioSensor-board:
I wouldn't exclude adding one onto the board if an IC can be sourced for US$2. Thats because everyone is going to want at least one accelerometer or gyro on their robot. However, small boards like this and this (that will be available at RoboSavvy store shortly) can be glued onto the BioSensor-board.

One option that Clarence suggested (and I think pepper implemented something similar in his Orangutan daughter board), is to create through-holes connection that would allow for Clarence's tiny accelerometer board to be pushed onto the sensor board.

In addition:

3.3V regulator - to driver most accellerometers and gyros ICs
5.5V regulator - to drive other sensors
1 x female Bioloid socket (on the edge, parallel to the board)
10 through holes - connected to different pins on the Atmega8
12mm x 12mm - can be glued to any surface of the bioloid

[pepperm]

OK so at least two people said they would buy 2 BioSensor-boards at US$30.
4 down 496 to go!

Limor, two were down to me though

Atmega8 with the RX,TX switched by PD6,PD7 (isnt just one PD enough to switch between RX state and DX state?..) to drive the 1-wire interface

We could use just one pin to change direction but we would then need an inverting package somewhere. We would need one invertor for the switching then 2 invertors for each data line, that's 5 in total, two of which would have to have tristate outputs. I don't know of a chip that does this that is small I am afraid. We could use could use something like the 74241 but that is very big at 20 pins. I guess it's why Robotis went for the non inverting, tristate 74126. We could do it with 2 chips but that means extra board space being used up. I'm open to suggestions though and still haven't ruled out the shorting of the pins approach. Pev and me will be testing that shortly I hope.

I think the daughter board approach is the easiest and simplest for the short term. I could provide regulators, line drivers and connection points for various gyros/accelerometers on this board. Next time though I would mount the daughter board on the "back" of the Orangutan board. You could then access the pot, leds and programming connector from the front.

The problem as I see it for a board with the processor and everything on it is that the processor would have to be of the smallest package variety which would preclude home assembly and thus require more costly manufacturing. A double sided daughter board would be easily manufactured and sold cheaply through Robosavvy for easy home construction (just a couple of easy smd chips, resistors and capacitors then a few sockets). We could even hand assemble boards for those who don't want to brave it and sell kits of parts as needed.

One question for everyone. Do we need 1 or 2 Robotis bus connectors? I have 2 at the moment for pass through but 1 occupies less space on the board.

12mmx12mm is very very small!

Mark

[JonHylands]

Mark,

The ATmega168 package I'm planning on using is 4mm x 4mm for a 32-pin chip... That's small...

I'm planning on putting one Bioloid bus header on the "extremeties" boards (the foot pressure sensor boards and the IMU). I may, if I decide to build a more generalized board, make it a touch larger and put 2 of the connectors on it.

I think, in general, these boards should have two connectors. Each connector is 10mm wide roughly, so you could squeeze a couple on the end of a 20mm board. However, keep in mind for any board that has any kind of gyro or accelerometer, the board has to be securely fastened to the robot chassis in order to be useful, which means 4 mounting holes.

- Jon

[pepperm]

Mark,
The ATmega168 package I'm planning on using is 4mm x 4mm for a 32-pin chip... That's small...
- Jon

The 4mm square package (specs say 5mm by the way) is the MLF package which I think is impossible to solder manually. It doesn't have pins as such, it has pads. It's an oven job. The TQFP package is 9mm square and is just solderable with care and the use of solder mop.

[limor]

i just thought that if the board has one Bioloid bus connector it makes it easier to manufacture and also to place on some external part of the Bioloid.

it's easy to provide a split cable connector

[pepperm]

it's easy to provide a split cable connector

Yes or just run a cable from another connection point.

So, is the daughter board idea the best option for us at this stage. Obviously there is some discussion going on on the other thread as to what processor board the daughter board should fit to.

Mark

[cat007]

i just thought that if the board has one Bioloid bus connector it makes it easier to manufacture and also to place on some external part of the Bioloid.

it's easy to provide a split cable connector

Not a difficult job. My friend does cable harnessing for me. we made a custom bioloid extension cable before, 15meters long for him to control AX-12 servos in another room !!

[JonHylands]

The 4mm square package (specs say 5mm by the way) is the MLF package which I think is impossible to solder manually. It doesn't have pins as such, it has pads. It's an oven job. The TQFP package is 9mm square and is just solderable with care and the use of solder mop.

Yeah, its going to be an oven job. Fortunately, my friend and I put together an oven for doing surface mount boards, so that isn't a problem. Having an oven makes putting together these surface mount boards much simpler in general.

- Jon

[Pev]

it's easy to provide a split cable connector

Yes or just run a cable from another connection point.

So, is the daughter board idea the best option for us at this stage. Obviously there is some discussion going on on the other thread as to what processor board the daughter board should fit to.

Mark

My opinion would be go with the daughter board to start with, if this is found lacking in someway then address it. I actually think the Baby Orangutan looks like a good starting point and will provide all we need with the ability to stack the daughter board and if needs be sensors (am currently looking at this as a compact sonar we could include - http://www.robot-electronics.co.uk/htm/srf02tech.htm using it in I2C mode

Pev

[limor]

Atmega8 with the RX,TX switched by PD6,PD7 (isnt just one PD enough to switch between RX state and DX state?..) to drive the 1-wire interface

We could use just one pin to change direction but we would then need an inverting package somewhere. We would need one invertor for the switching then 2 invertors for each data line, that's 5 in total, two of which would have to have tristate outputs. I don't know of a chip that does this that is small I am afraid. We could use could use something like the 74241 but that is very big at 20 pins. I guess it's why Robotis went for the non inverting, tristate 74126. We could do it with 2 chips but that means extra board space being used up. I'm open to suggestions though and still haven't ruled out the shorting of the pins approach. Pev and me will be testing that shortly I hope.

Can't it be done with a simple analog circuit?