If the SSC-32 board were to be re-layed-out, I would suggest the following:
Add a jumper (or something) to allow the on-board regulator to be disabled, so that an external supply can be wired in without back-driving the regulator. This may also allow the board to be run at a lower voltage, to save power.
Add a jumper to allow disconnecting the power from the MAX232 when it’s not needed.
Finish the FW to support the 4-legged critters. (Although pretty soon I’ll be far enough along that I won’t care about this.)
Find better connectors for the servo power inputs. (No, I don’t have a good suggestion that wouldn’t be too expensive…) I wonder if “mini-bananas” would be good?
The current draw feature would be well appreciated… Just “smoked” three HS-5654MG’s on my Hex… That is an expensive mistake I would surely pay to avoid again in the future…
only the latest line of Hitec servos (the robonova ones) can send any data back, such as pot readings. A servo controller could be modified to read this data on only these servos, but to get this data out of any servo see openservo.com Bonus points for i2c!
Just thinking out loud:
A couple of approaches for the current-limit idea:
Put a 0.05 ohm resistor in the ground that is common to all the servos. Use an A/D input on the micro to periodically measure the V at the top of the resistor. The firmware would compare the reading to the # of servos that are currently expected to be moving, and decide if the current draw was too high, and stop the servos. The code could make the current measurement value available via the serial port.
Add current-limiting circuitry (either part of the SSC-32, or as a separate small board). A very simple way is to just put a resistor (say 0.5 ohms) in series with each servo’s power input. This would “tend” to limit the current, but it would not be ideal. A better way is to use a device that has natural current-limiting characteristics, such as an incandescent lamp bulb. Auto lamps (such as for dome lights and such) might be good. The specs for the lamp determines the current limit.
Build an actual current-limiting circuit. This is easily done with a 3-terminal regulator (like a 7805) and a resistor. With a 7805 the max current is 1 amp, but there are bigger versions of those regulators available (definitely 3 amps, maybe 5 amps).
Ahh, but many robots are not controlled by PCs. I know mine isn’t. From a microcontroller point of view, it is much easier to do standard serial communications than it is to do USB. If the SSC is redesigned, please don’t make USB the only interface option.
As for modifications I would like to see, the aforementioned jumper to turn on/off the RS-232 logic would be a nice touch.
I agree with Qwindelzorf. USB is only practical where a PC is in control. Hobby-robot microcontrollers typically don’t have and don’t need USB.
RS-232-type ports are widely used, and are easily supported on PCs for the foreseeable future (a USB-to-serial adapter is easy, if your PC doesn’t have a serial port).
If another interface were added, I’d suggest IIC. With many/most micros, it’s supported without any additional HW. It’s just two wires.
Yes, but it is (1) already becoming practical (the Gumstix is an early examle of this trend) to have a PC motherboard embedded on your robot with (a) the new nano-itx form factor and (b) the new low-wattage computer chips. Both of those trends will only continue into the foreseeable future. In a couple of years (which is what I understand this thread to be discussing) you will have one more step down in form factor, a couple more years of Moore’s law, improvements in battery technology, more complicated robot motor control programming requirements and an simple ecomonies of scale working against the current bot boards…
Simply put, specialized robot microprocessors are going to be almost totally eliminated sooner rather than later. The big pond of PC processing and motherboard development is overflowing into their kiddee pool… In 2-3 years, are you going to want a latest generation mobile PC chip running at 10+ times the cycle speed (while drawing the same current), with integrated graphics processing for sensor inputs or a bot board? Even if the transition isn’t as fast as I think it will be, more and more of the hobby boards are putting in USB interfaces, we’re still moving to the same place…
You are always going to need a servo controller though because the PC people have no reason to put the servo headers. Keep in mind, I’m talking abot recomended changes for say, 2-3 years from now.
