QL serial ports and MAX3232 overheating

Nagging hardware related question? Post here!
Post Reply
Silvester
Gold Card
Posts: 436
Joined: Thu Dec 12, 2013 10:14 am
Location: UK

QL serial ports and MAX3232 overheating

Post by Silvester »

On another thread I mentioned that I had trouble getting USB RS232 devices working at anything like 9600 baud to the QL. The problem was they only output 0/5V TTL levels, so I tried using the same chips as used in the QL (MC1488/1489) to shift levels to +/-12V. That didn't work very well either.

The solution was to buy some of those tiny PCBs containing MAX3232 converters off Ebay, and power them off the same USB port. I think the problem was the PL2303 USB chip I used didn't like driving the MC1488/9 chips (old logic fabrication), and that they were powered externally.
tiny3232.jpg
tiny3232.jpg (11.61 KiB) Viewed 2341 times
Everything worked fine until one occasion, shortly after plugging serial connector into a QL, I could smell something hot (like soldering iron hot!). The MAX3232 chip was overheating alarmingly, and the Tx output from MAX3232 was only around 1V (normally +/-9V). After unplugging everything and letting it cool down, I tried it again. Rx worked, but Tx was erratic (though the voltage levels appeared OK).

Since I had 5 of them I tried another one. This time I didn't plug/unplug anything whilst everything was powered up. Again it worked fine for a while, then suddenly without any apparent reason, the MAX3232 went into melt down mode!

It didn't take long on a web search to find this is not uncommon with MAX232/MAX3232. The problem is the chips are prone to electrical noise/ESD on the RS232 side and it can cause latch up of the charge-pump circuit, causing excessive overheating (Tx then becomes impaired or broken).

After reading all the suggested fixes I decided to reduce noise from QL using 10nF capacitors to ground on Rx/Tx/RTS/CTS lines followed by 100 ohm resistors in series to MAX3232 pins. Also the 5V power from USB port to MAX3232 is fed through a 20 ohm resistor, this prevents any possible latch holding.

So far everything is working reliably to QL at 9600 baud (using 2 stop bits). Sometime later I'll try it with Q40 and see if it works up to 115200 baud.

Although the main culprit appears to be the Chinese fabricated MAX chips (which can work OK), it appears the original TI/MAXIM chips are not completely invulnerable to the same problem.

Someone recently did an interesting blog on decapsulation of MAX chips in acid and compared the smaller Chinese dies with original MAXIM:

https://blog.heypete.com/2016/09/11/inv ... 232-chips/


David
Post Reply