Hello,
I am a young business owner, and I am hoping someone here can help me with a problem. Just the other day, an electrical storm destroyed some equipment at my machine shop. We lost a phone, fax, laser printer, and a PC. The PC was hooked to my CNC milling machine via an RS232 serial cable. The machine was disconnected from the mains, but was damaged. The machine is 1985' vintage, and the controller hardware is all in a card-cage. The CRT now shows a flashing "error asynch communication".
I have pulled the main CPU board, as this has the RS232 port on it and I figure that is where the surge got in. There are 2 serial ports, and each goes directly to a 75154 quad line reciever and 75188 quad line driver. (pic zpu5.jpg) I would like to think that maybe one or both of these chips fried and is now shorting the boards power bus to ground, effectively stopping the board from functioning. I had previously socketed the 75145 reciever chip to test for an unrelated serial problem, and measured a very low resistance between Vcc and Gnd. I swapped out that chip and it still does not work.

OK, so here is my question: There are several NEC devices on the board that I think may be voltage supressor diodes? (pic zpu3.jpg) There are 2 right next to the card-cage headers, and they are marked "10M" and "35V" they also are stamped "NEC" "+3KI" but that is just the brand and maybe a date code?

Does that maybe mean 35 volt, 1000w voltage supressor? They are banded with white and the board is marked "+", so they have a polarity. What is the failure mode for these, and how can I test them? (will they fail and short?) If I need to replace them, will the replacement be an axial lead like a regular diode? I have never seen them in this square packaging before....

In the picture zpu4.jpg, there is a small glass component with what looks like a ferite coil over it. There is a marking next to it that looks to say TH60 degrees C. Is that a thermal fuse?

Thanks all for the help!
Carl Crawford

 

The component you have guess as a thermal fuse is almost certainly just that. It should measure zero resistance if it's good.

The NEC devices appear to be surface mount capacitors. My guess would be 10 microfarad at 35 volts. That's taking the 10M to be 10,000,000 picofarads. If you want to get replacements, check dimensions pad-to-pad. That is much more critical than height. If no 10 uF SMD capacitors are small enough to fit, then choose a smaller capacity that is about that volume.

If you replace the drivers and receivers, put IC sockets on the board so the next repair can be made faster and more easily.

 

Wow, that's tough luck
In a way, you're fortunate that the computer is as old as it is. Newer motherboards are populated with a great deal of CMOS devices, which get destroyed very easily by static/lightning - besides, they're all SMT devices; really tough to work on unless you have pro tools.

Before unsoldering the caps, apply power and measure all of your supply voltages on the motherboard. Most of the board will likely only require +5v. About the only place that +12v and -12v are used is for RS232 communications, so make sure they're present. If any of the voltages are out of spec or missing, try running the power supply without the motherboard connected; the supply may be damaged (particularly the +12v/-12v portions.)

You will likely normally read a pretty low resistance between Vcc and ground. Your computer is all TTL.

The item marked "TH60°" is indeed a 60°C thermal fuse. It looks like it's cracked on the left side, but that may just be the way the light reflected on it. Measure across it with your meter set to the lowest resistance scale. If it's not right near 0 Ohms, it needs to be replaced. However, I suspect it's OK because you ARE getting some kind of output from the computer (the error message on the monitor)

10M, 35V = polarized tantalum capacitor, 10uF, 35V with radial leads. The silver stripe indicates the + side of the cap, which is opposite of how electrolytic caps are marked (they generally have a stripe marking the - side instead) and diodes, which have the line on the cathode end.

You really need to replace them with tantalum caps - that is, if you have to replace them. Electrolytic caps might work in a pinch, but tantalum caps have much better electrical characteristics, and will be much more reliable. You could try unsoldering them one at a time (being very careful to not overheat them), and testing them with a capacitance meter. If you don't have a cap meter, you can use a DMM set on 2M Ohms and a 1K Ohm resistor to perform a rudimentary test:
1) Short the leads of the cap using a piece of wire for several seconds.
2) Clip one lead of the resistor to the + cap lead using a test lead with alligator clips on both ends.
3) Using a test lead, connect the + (red) probe to the other end of the 1k resistor.
4) Place the - DMM probe on the other capacitor lead. It should start at a low reading, and take roughly 12-17 seconds (for a 10uF cap) for the meter to count up and finally read infinite. If it starts out reading infinite, the cap is blown open, or it wasn't discharged properly. If it doesn't get up to infinite, it's "leaking", or partially shorted.

But really, check the voltages on the board first. You don't want to be unsoldering things if it's unnecessary.

 

Beenthere,
I doubt the caps are SMTs, because there are no SMT pad visible. Besides, SMT wasn't even around back then that I remember - everything was still thru-hole.

 

Hello,
Thank you for your reply, the NEC devices are through-hole.

 

OK, so maybe I should socket and replace the other chip, (quad line transmiter) and check power supply outputs to make sure it is all there...

 

Check the voltages on the motherboard to ensure that they are present and within limits:
a) +5v needs to be 4.75v to 5.25v, or you risk severely damaging all of your TTL ICs.
b) +12v and -12v should be within a volt. These levels are not nearly as critical as the +5v.
c) You may also need to have -5v present. I can't remember offhand if that's used in RS232, or the IC's themselves.

The +/-12v and -5v are much lower power outputs than the +5v, as they aren't used many places. They won't take much of an overload. That is the area I would concentrate on first.

 

Do you have an operating system on this computer, like MS-DOS?

If so, which version?

Windows came with MSD.EXE, which is Microsoft Diagnostics. It has tests for the RS-232 ports, including detection, diagnostics, and I believe a loopback test.

 

The 75154 quad receiver has ground on pin 8, and should have +5V on pin 15. If it's powered by +12v, that will be on pin 16. If pin 15 is not 5v, and there isn't anything on pin 16, then your +12v supply is down.

The 75188 quad line driver has ground on pin 7, -12V on pin 1, and +12v on pin 14. Voltage levels on the output are clamped via diodes to the power supply rails.

 

Hello,
I re-read my post, and I think it may not have been clear enough.... I am not concerned with the broken desktop PC, but rather the control on the milling machine. It is the proccesor card in the milling machine that I am working on.....

thanks!
Carl Crawford

 

OK, that's fine.

In that case, you might be seeing anywhere from 7.5 to 15v and -7.5 to -15v on the 75188 pins 14 and 1, respectively.

 

Since the board is buried in the middle of a cardframe, it is almost impossible to probe Vcc pins while it is in opperation. I did have a chance today to measure voltages at the power supply, they are a solid +5, +15, -15, and +12. I would think that if a chip on the proccesor board is bad, it would pull the power supply down on that line. (or at least go into protection) is it possible I could still see no or low voltage at some Vcc pins even thought the power supply still reads normal? Is the PS just chugging away and heating up the bad chip?

I wish I had a thermal camera so I could see if one chip is burning hot, or could get my fingers in to feel them while it is on.... I may have to get some 96 pin DIN extenders so I can probe Vcc pins while it is on...

Cheers!
Carl Crawford

 

Hello again Carl,
It's difficult to say precisely WHAT might have been damaged by the lightning. However, I suggest that it is perhaps more likely that something has been burned "open" rather than "shorted" due to this mishap. It's very difficult to try to assess the damages remotely, with only some relatively low-detail photos and discriptions of the symptoms to go by.

Do you have an oscilloscope? Have a look at your supply voltages with it. They might measure OK with a meter, but an O-scope might show a blow rectifier bridge or the like.

If you don't have an O-scope, you may have to resort to unsoldering things. Start in the vicinity of the RS-232 port and fan out from there. If the RS-232 port is connected to a bus, check those items directly connected to it next.

No really easy way to troubleshoot this - storm damage can be a real bear.

 

If you suspect lightning damage, how do you think the lightning got into your system?

I can remember on televisions, nearby lightning strikes almost always took out the two diodes of the bridge connected on the return line, as the impulse entered via the ac return line.

Those that entered in the cable input did more extensive damage to the tuners and such.

It all depends on how the impulse entered the equipment. Just something to consider.