Has anybody played with a MAX232 at the wrong voltage?
It’s designed to use 5V supply and 5V TTL levels and generates +/-12V for RS232. I’ve used many over the years but all at 5V. Well 5V is very 20th century, everything is 3.3V or 1.8V now. You are meant to use a MAX3232 for 3.3V but I don’t have one to hand.
Has anyone abused one and driven it at 3.3V? I have one here running from 3.3V levels and it’s generating about +/- 8V. It’s wired to a real PC serial port and rattling away 115k2 baud perfectly. It’s driving about 50cm of ribbon cable then a 1m 9pin D genuine host 1980s vintage serial cable. Waveforms look nice and clean and everything in the garden is rosy.
It works but if anyone has any real experience of abusing them please speak up.
Do not have engineering answer to this, but found some nice reading on the MAX232 (3.3 V is not TTL signal from 5 V supply but I guess that is obvious).
My old desktop, which is due for retirement soon, has a genuine RS232 port. 9pin D and RS232 voltages. I have a good handful of USB to TTL serial level shifters. Most are older Prolific units that Windows vomits over. They work perfectly on the Linux desktop. Debug output is being put out to that real port currently. It has been truely excellent to have a real RS232 port at times. The software is being developed on the Linux machine but that runs headless, so it’s easier to use the Windows machine to display everything.
My old EPROM programmer has a 25pin RS232 port but he lives in The Sunnyland Home for Retired Embedded Tools most of the time as just everything has onboard FLASH memory now.
However, the true target is my KX2. That has a +12V RS232 output and real RS232 levels compliant. Whilst I’m developing the software the ability to flip between the radio and a terminal is useful. Once the software is past the early stages, terminal connection wont be needed and I’ll play about making some simple transistor level shifting probably using some TUP or TUN devices for those old enough to remember Elektor!
I was telling Barry @GM4TOE that I had to get the oscilloscope out to do some debug. I’ve spent so long writing code that either is a simulation of a semiconductor or writing software to run on the simulation, writing bare metal code for hardware again is different. In my day job not only can I use a software debugger (GDB or Trace32 etc.) on the embedded software but I can run the simulation of the hardware (CPU, memory, peripherals) under GDB or similar. So you can set breakpoints on the hardware and software. It’s like having a logic analyser on the hardware and having it trigger when a memory location is accessed and then you can step your software simulation of a chip and step through the software running on the simulated hardware together. And you can script combined simulated hardware and its software debugging so you can run up something and run the script to exercise for issues.
But it’s back to earth with a bump for this stuff at home! I should get an ARM SWD port so I can debug but for now nothing beats driving patterns on GPIOs and watching with a scope. Of course my scope is a 50MHz analog beast and whilst I’ve not used or needed a scope in the simulation world, the last ones used were all very highend Tektronix multi gigasample LCD based ones. So back to an analog scope and playing with the trigger and sweep delays and such to find the signal you are interested in takes me back to when I had hair! Oh and the brightness control, you don’t appreciate the bright trace you have with shiny fast LCD scopes till you go back to a real CRT
MAX232 of yore is quite slow, with pretty soggy output drive. Low voltage will exacerbate this significantly.
Designed them out of products many years ago due to poor drive performance at high speed.
But this matters if you are trying to run 115kbd down 300’ of cable. Down one cubit, not so much.
Later suffix versions or from other mnfrs are likely very different performance to the originals. They are also one of those chips where you will find a manufacturer has several “common” chips with different numbers and ostensibly different specs, and they actually just have the same die inside them.
The datasheet seems to think it will work with 3v3 inputs since the ‘high’ threshold is 2v but they are assuming that the chip has 5v power which might not be your usage case.
I expected there to be a lot more about this misuse on the web but there isn’t. I have a genuine MAX232 wired up to 3.3V signals and fed it off 5V. As the setup is on a breadboard I merely unplugged the 5V to the MAX and picked up the 3.3V rail. It’s working FB.
Well Simon it’s so long since I used RS232 for more than 1m-2m of cable I couldn’t remember what the limits were, I had to look them up. It brought back memories of long cables through the roof space of where I worked in 1989 which connected our Televideo 925s to the VAX. The best place to sit was near the few windows that opened but that was down the longest cables. You could only get 9k6 reliably there. You could get 19k2 if you sat nearer and used a shorter cable but that area was hot in summer.
I don’t have it to hand but my memory seems think that RS232 receivers will operate with ±3volt levels.
It also brings to mind the scene from Spitting image with (I think) spoof Status Quo puppets singing a song about an RS 232 interface lead…
Andy
MM7MOX
Back around 1990 I worked on a telecoms system where the code was in 2716 family EPROMs. It was a fun job loading up a new version of software - swapping 16 chips on the two main boards plus another one or two on all the peripheral boards and then cooking them at gas mark 6 for 20 mins in the UV eraser.
A colleague took a software update to Singapore. It was a tube of EPROMs in his carry on luggage. He freaked out when security insisted on putting them through the scanner. I assume they survived.
I left the company (and moved up north to sunny Yorkshire) in 1994 just when the EPROMs were being replaced by flash.
I used it since I moved to GM in 2000. I burned some EPROMS for Brian G4ZRP for a 4m radio in 2003/4 vintage.
Things changed markedly in the late 90s. I spent a lot of my work life downloading Intel HEX files and burning the image into 27C256 and 27C512 EPROMS. Then from 1995 we started using FLASH on the embedded stuff I worked on. Intel 28F001 (128k x8 boot block) and then Macronix MX28F4000 (512k x8). Then another 4Mbit device from AMD as the MX28F4000 did nasty things on brownouts. The only EPROMS blown were on new machines during early development to get bootloaders and downloaders working. Then the EPROM code was put in FLASH and you could update in-situ. Sweet.
I had a look in the box of “programmable stuff”. There’s real relics in there. HD637V01B, a 6301 (cmos 6800 plus peripherals) that’s UV eraseable and programable. 27C128/256/512 devices. A large number of GAL 16V8 chips. I cannot recall seeing anything using GALS for 25 years. And a bunch of 2716 chips. The highlight being a pair of unused Mostek 2716 UV EPROMS in ceramic not plastic with gold plated pins, date code 8013. That makes them 44.5 years old.
In the old digital devices trays there are 68008 68681s and 68230s, wow they are just such huge chips! 63C09, 6303, 6809, 6502, Z80… all your favourite 1st gen 8bit CPUS. One of the Z80s has a 1978 date code
I really should get that lot up on eBay as there will, no doubt, be someone looking for some of this old tat so they can keep some 70/80s electro-antique working. I can’t see anything being used by me. I did think of making a 68k based system as I have an 8bit CPU, DUART and PIT and plenty of 32k x8 SRAMs. But why other than to waste time. It would be big, slow, heavy and thirsty for a 4MIPS “thing” with 128k FLASH and 32k RAM. This Pi Pico here was £3.80 with 2x 125MIPS CPUS, UARTS, GPIO, USB, 2MB FLASH and 256k SRAM. It’s the size of an old 40pin DIP device.
For reference, the MAX232 has date code 9043 on it.
