Pis and serial ports...

In that case you will have problems with MAX3232 as well. The charge pump makes only about 5 volt from the 3.3v supply.

You aren't grasping what I am saying. The spec is defined to allow for a large noise margin. All single ended RS-232 receivers have a threshold at approximately 1.0 volts. This is sufficient to meet the spec. If you drive the receiver input high and low enough to cross the threshold it will work.

If you don't believe me, try it...

And I think you are not understanding what we are saying. The receiver threshold is relatively unimportant, provided it can handle large plus and minus voltages (in fact a good quality receiver will have about a volt of hysteresis to deal with transition problems). The issue is that some of these devices will also have only a low transmission voltage, such that in a typical high noise industrial environment false signals will be generated.

If you only want to use your Pi for clean-room projects this is not a problem - indeed I've commissioned quite a few successful AVR direct links working continuously at 115200 baud 24/7, but if you want to go out in the jungle then you *need* to adhere to the spec. and make sure your receiver can correctly handle the full range input.

Either your design is wrong or you have a loading problem.

Never seen that on any MAX232 or MAX3232 or similar device. Output voltages are speced in the 12 to 15V and I have measured that on many systems open circuit and under CORRECT loading.

BTW have been using those devices for around 15 years.

For MAX3232 type device with Vcc = 3V3 and temp = 25C the Low threshold is min of 0.6V TYPICAL is 1.1V and NO maximum stated so could be anything from 1.2V to maybe even 1.8V. Which means from device to device it has a wide spread, and this will be typical of all likely RS232 receiver inputs.

So it wont take much noise of dirty connection, or gnd shift between ends, to miss them.

In a clean room environment or very short cable you can GET AWAY with Pi TX to PC RX on a LOT of systems, but then you are relying on a lot of things that wont necessarily work on all PC or similar RX ports.

Even in Single Ended your gnd wire has impedance and can shift you input signal relative to the RECEIVER's gnd. Especially as quite a lot of Pis are floating relative to mains earth and your other end may be fixed to mains earth.

Whatever you do you NEED an INVERSION of logic levels for it to work anyway.

You must have been using negative impedance as the load...

A MAX232, when powered with 5 volt, has little in excess of 9 volt output of the charge pumps and delivers +/- 9v on the output. A MAX3232, powered with 3.3 volt, has less than 6 volt output of the charge pumps.

Of course it is within RS232 spec. But in the old days you could use output pins as a powersupply for a low-current device, and that has become a little more difficult.

or current-loop kit with converters.

OK, so what you are saying use a NOT ready made solution? LOL I'm getting confused here.

Nope cable to various PCs and other devices with their own receivers.

Have not used 1488/1489 for OVER 20 years as NO system in last 15 years plus had the suitable PSU as 1488/1489 requires +/- 15V.

Not used current loop for comms in over 30 years

Yeah yeah, would be nice though. Typically one SD TV channel is about that bitrate too on the average or maybe a little higher so shouldn't be a problem for the RPi to handle at all. Even considering the ethernet on the RPi is on the same USB bus too.

You are reading a spec sheet. Try measuring some devices. I am not saying this is something you would want to use for anything other than hobbyist designs or lab queens.

Yes... but when you say "all" you mean it won't work with more than some

You are really digging in the dirt on this. The floating ground is what will make it work. If both ends were grounded to the mains you could easily end up with a ground loop which *would* give you a difference in ground voltage or worse, a 50/60 Hz signal on the ground line.

Yes, I believe I have said that several times.

Yes, that is the point. I understand exactly what you are saying, but you seem reluctant to pay attention to what I have written. I am not advocating using this for any real design work other than hobbyist or lab queen work. The point is that the threshold of the typical RS-232 receiver is a fixed voltage around 1.0 volts. It will work just fine when driven by a 3.3 volt TTL compatible signal. Just don't try to use it in a design that has to deal with long runs or noisy environments.

AFAIK Maxim is only an integrated circuit manufacturer, so they would not have been around in the days before integrated circuits.

But there were RS232 transceivers built with discrete components way back then.

In comp.sys.raspberry-pi message , Thu, 12 Sep 2013 08:52:57, Paul posted:

Not according to

I remember the 1488 being most often powered by +-12 volts. That is one reason why early PCs had those voltages. Even if that is not the same as +- 15 volts, the point is valid, most systems don't want to provide the negative voltage to power this driver so devices like the MAX232 devices were born.

Yes, the receiver has always been a single low voltage because you don't need a high PSU voltage to detect the transition between a high and a low (or a mark and a space as the spec says). The receiver just needs a standard threshold of about 1 volt and tolerance of +-25 volt inputs, easy to do with 5 volt power.

Correct the last systems that I designed kit for using 1488 was PDP-11s, which had the + & - 15V rails, after that next to no systems had negative rails. I had forgotten that we genrally had 12V regulators around for the drivers. But hey this is some time ago so some of the details are hazy as that knowledge has been superceded.

For many years was doing a lot of RS422 comms which is different levels not needing negative rails.

The 1488 was commonly powered from +/-12v.