Using power stealing RS232 devices with a microcontroller?

[ledtester]

I have a couple of RS232 devices which steal power from other signal lines (e.g. DTR) and I'm wondering what the best way is to use them in a 5V microcontroller environment.

I'm not worried of the level shifting of the TX and RX lines - there's lots of ways to do that and I have MAX232s.

I suppose I could try:

- see if the MAX232 charge pump can generate enough current to run the device
- just connect DTR to 5V and see if it works
- create a mini boost converter with a small inductor and a PWM line of the uC
- create a charge pump circuit ala https://sites.google.com/site/wayneholder/12-volt-charge-pump

Any other ideas? Thanks!

[Kleinstein]

Usually the RS232 lines provide something like 8 -12 V. This is too much to directly power a 5 V µC. So one would need a voltage regulator, at least a linear one. Depending on the software running, there might not be a positive signal on the handshake lines. In this case one would need some kind of charge pump to convert a negative voltage to a useful negative.

Usually one could use the voltages provided by the incoming RS232 lines direct to drive the RS232 outputs. So one would use a max232 without the 5 to 10 V charge pump (leave out one of the capacitors). Not sure the max232 is supposed to be used this way, but I see no reason not to do it.

In theory the max232 might work the other way around (e.g. coupling +-9-11 V and making some about 4-5 V from that. However there can be a problem with startup. This might be an option if the µC can run from slightly less than 5 V.

[alm]

Minimum RS-232 signaling levels are below 5 V, so a linear regulator would not be a reliable way to power a 5 V device from RS-232. Fluke used the power from handshaking lines trick in one of their serial-to-IR converters for their 18x DMMs, and they had to publish a list of tested USB-RS232 adapters because not all adapters would work properly (probably because either insufficient voltage or current).

[Nusa]

The first question is how much power do you need for your application? If this number is too big, we can forget this discussion entirely.

The second question is how much power can the hosts rs232 drivers provide without overheating (their cooling solution was designed for signaling, not power delivery).

Are designing for a known host, or need it to work with a generic host? +/- 12V may be a typical voltage, but the actual standard is much less restrictive. You may run into ports that only transmit +/- 8V or 15V or even 5V. And on reception, your design needs to tolerate anything up to 25V.

[ledtester]

Just to make sure we're all on the same page...

My microcontroller has a 5V supply. I want to interface a power-stealing RS232 device with it, which means that the uC has to generate some power on the DTR line to power the RS232 device. I figure that the RS232 device can't require much since RS232 interfaces haven't typically provided that much power on signals such as DTR.

I'm just curious what kind of solutions ppl would attempt.

EDIT: For concreteness, you can assume the RS232 device is a serial mouse - i.e. a commercial product that was designed to work with a "standard" PC serial port.

[Nusa]

Ah, I thought you were trying to power a microcontroller off an rs232 port.

So forget everything I said except the 25V tolerant part. Just in case someone plugs something other than your theoretical mouse into the port.