usb-serial adapters

Re: usb-serial adapters By: bob prohaska to All on Sun Dec 24 2023 01:55:43

In my case, serial terminal, on headless computers. And sometimes interfacing with retro stuff, for example I recently dug up my old HP48 calculators which talk 9600 bps and ymodem at best.

In my line of work (telephone call data recording), RS-232 is still in heavy use (although TCP/IP links are starting to replace them). We use IP-to-serial converters a lot.

In My Experience - MOST USB/SERIAL adapters work OK in PI/OS. I think, despite brand names, there are only a few actual chip-sets.

While Amazon reviews are a bit suspect, still DO read them, with particular attention to Linux remarks

IF you get stuck having to load virtual Winders drivers ... then you haven't chosen wisely

Pi's can also do RS-232 directly. A bit less civil, but it works.

If you can find the chip number, that's the main thing. If there is a link to Windows drivers, downloading them and taking a look at the .inf file is a good clue.

There is no support for running virtual Windows serial drivers in Linux.

Pis can do *serial* directly, but not RS232. They only have 3.3V signal levels, so voltage conversion needed for RS232 or another voltage (1.8V, 5V, etc).

Theo

Yep ... level-shifting chips are generally needed. The MAX3232 will do it.

As for converters ... I'd swear I got a USB<->232 unit working on a Pi3 ... but don't quote me on it. I'm kinda a hardware guy, so I would have made something with the MAX chips almost automatically.

Anyhow, peruse the usual spots, AdaFruit, SparkFun, Seeed. They may have a canned solution.

It does more than "level shift" it also inverts the signals. It will accept 3v or 5v logic, even though supplied from a 3.3v supply.

Maybe. But not as you say. You would need a simple 3v to 5v level shifter (so as NOT to invert the signal) then the MAX232.

sound advice.

Properly wired, the MAX232 can deliver "real" RS-232 ... including the negative voltages. There IS still some equipment that expects those kinds of signals ... albeit mostly early-80s. Some industrial controllers of that vintage are still in use however.

Flow-control ... sticky. For RECEIVING no modern cpu/controller is so slow that it cannot absorb such comms at even the highest usual RS232 speeds. However those ancient devices you may be communicating with may NOT be able to absorb the info so quickly. Ergo you have to respect their flow-control lines. Serial comms can be a pain in the ass. There are subtle issues.

Again, depends on what you are trying to converse with. Very old devices may expect/depend on those rather high negative signal levels.

These days there's often a canned fix for just a few dollars. Kinda takes the fun out of it though :-)

It was not too hard to build one with the MAX3232CPE. I even added leds indicating the data transmission on the RX/TX pins. It works with the RPI2 and RPI4 very well. It is not isolated and protected so proper wiring is important

It was not too hard to build one with the MAX3232CPE. I even added leds indicating the data transmission on the RX/TX pins. It works with the RPI2 and RPI4 very well. It is not isolated and protected so proper wiring is important. Here are my notes on this

The 3232 does provide the negative and positive voltages but only at

Indeed, which the max232 provides, +/- 12v. But it needs driving with 5v logic levels. Hence the need for the 3v-to-5v level shifters BEFORE the MAX232 inputs.

Which means the older stuff is broken and not really conforming to RS232 - the signals from the '3232 & Co *are* fully RS232-compliant.

Yes, I know broken stuff exists and if you need to interface to such stuff, you need better drivers, but the blame is on the broken equipment and not on the MAX3232.

When driving a RS232 load, the MAX232E output is specified at +-9V typical,

The MAX3232 output is +-5V minimum, and +-5.5V typical.

If you really need more output swing, you will have to use something like a SN75188 (with a 3.3V -> 5V level converter, and +-13V supplies) - at VCC=+-13.2V, these will drive +-9V guaranteed and +-10.5V typical.

With Vih=1.9V, these might even work from a 3.3V output without additional level translation.

cu Michael

Indeed, and as you say the blame is on the equipment - but in the late

Ah I mis-remembered - thanks for that.

cheers

Probably true. I think the 3232 is 3.3 in/out so some kind of level-converter would be required between it and the 232. Cheap, or you can make yer own, but it's an extra level of complication.

Thing is, some of that old equipment (oft industrial) IS still out there doing its job so a Fake-to-Real RS232 (might add a couple other RS's) converter might be a worthwhile bit of kit to have around.

There's also the issue of the flow-control lines. While modern stuff is fast enough to not worry about it, the old equipment might need to signal a brief stop while you are sending to it. Buffers were often very small back then (and processing slow).

Async is versatile, but a pain in the rump sometimes. Synchronous serial is more civilized. However most equipment you'll ever come across is async.

Synchronous was more common in the mainframe environments of old. It can be a real pain to set up, especially if you're trying to get hardware from different manufacturers to talk to each other. Synchronous protocols are better suited to sending files across the country than characters across the room. The handshaking overhead can easily eat up the speed advantage of losing async's start and stop bits.

but +/- 5.5v out. For equipment that actually, meets the RS232 spec is fine