Computer Corner by Bill Kibler

Welcome to our special on the PC/XT machine. I covered most of the concepts you need in the special and centerfold. I do get plenty of calls and just yesterday, I helped a friend fix his machine over the phone. With that in mind, I thought it might help if I covered the how-to of trouble shooting a system.

POWER

When working on any system the first place you start after a visual inspection is checking the power system out. I say after the visual inspection, since sticking the power cable back in the wall socket is certainly easier than any of the other steps I will outline. Your visual inspection is mostly to find glaring problems (like not plugged in) and get a feel for the system. After that we want to make sure you have power and the power supply is working properly.

The way I checkout the power is to turn it on and listen. With the older MFM hard drives, you could hear them ramping up to speed, sort of like listening to a jet engine taking off. Often when the power supply has problems, the older hard drives will have trouble reaching their normal speed and may stop, slow back down, or start hunting (going faster then slower then faster). There can be other reasons for that speed problem, but most often it is a bad supply or one that is overloaded.

Another sound to listen for is the fan speed. Oh yes, we have had a couple of cases where the newer 486 CPU cooling fans have gone bad, sounding like power supply fan problems. Take the cover off and listen and look at where bad sounds are coming from. Often lots of dust will be binding the fan up and some cleaning with a dry brush or vacuum will return things to normal.

Now you can have a part of a supply go bad as well. I had several older supplies in which there was a separate regulator circuit for each pair of cables. To find the bad one required putting a load on the circuit (like a floppy unit) and checking the voltage. You should see 5 and 12 volts, but might see 8, not 12, if it is not regulating properly. I have rarely seen overvoltage, but it can happen and burn out all components on your computer.

Most modern power supplies however have what is called a crowbar circuit. This circuit is suppose to shut down the power supply if a short or other damaging problems happen. So you might take a new supply and stick it in as a replacement only to still have nothing happen. You proably have a short somewhere and the supply is just shutting down. You can check this out by removing loads (undoing conectors) and then retrying the power. If your fan starts again, put connectors back on one at a time till you find the shorted cable. I don't beleive any PC power supplies will shut down without a load, as the fan current should keep the crowbar circuit from kicking in (some crowbars shutdown if load goes to zip.)

What Next

After checking and being sure your power supply is working, I like to check for system activity. Often I simply see if the fans and drives run at proper speed and then see if it tries to boot. Most BIOS programs will go through a number of steps that are visible and make sounds. Stepping the drives home (their normal starting position) is one sign the BIOS was able to run and thus the power and motherboard are proably working ok. I say proably, only because you might have a bad power cable to the hard drive and yet the floppy works fines.

It is at this point that we can start running into configuration problems. As I stated in my $10 XT article, a common problem can be a misconfigured system. The PC/XT systems use dip switches for setting what type of options you may have. On AT and later units, a menu driven utility stores system parameters. The newer options can cause many problems if set wrong. One option I sometimes do is to not have the A: drive be checked for boot disk. Thus watching for the BIOS to step and try to load from A drive might not be valid on one so set up. On 386 and 486 systems, there are many options having to do with how the CPU chip handles memory, keyboards, video ram, and 20 or 30 other possible settings, anyone of which might be a reason for not working properly. Set these options with care, write them down in case they get trashed, especially your hard drive settings.

I think all setup programs have an default option which should give you a safe starting point if yours becomes too badly messed up. For configuration problems such as wrong drive type, you will get an error message. Most will allow you to boot up still, but you must hit the F1 key to continue on. This will also give a false problem indication, because you might have an error that stops things before it steps drives. In my earlier case, the switch setting were wrong for monitor type and thus I did not have anyway to see the message. I blindly hit the F1 key anyway, and then my drives got stepped. That told me the system was running but with some error and possibly a misconfigured video option.

The solution is first to get video and you do that by changing the switch setting and repowering up the system. Suppose the switch is correct, then a bad video board or monitor is possible. To check this, is to replace the video board with a new or know working unit. I usually have two or three other systems setting around, so using spares is always possible for me.

RECORD!

While on the topic, it is very important to record, ON PAPER, what your hard disk setup values are. Most newer IDE drives have the setup values on the drive, but the older MFM and even some IDE drives did not. Even the newer IDEs have some latitude for setting their parameters and should you use non-standard values, these need to be recorded. There are many free text files that contain a fairly long and accurate list of what values of size, heads, and cyclinders most of the manufactured drives are. However it can be a real big problem getting data back off a crashed drive even if you know the values.

Some of the early MFM drive controllers have their own hard disk initialization routines and thus the drive setup information is contained somewhere on the disk. The different programs can have this information stored in different places and in different ways. So if you have an MFM type drive, chances are a drive controller problem might mean you are unable to retrieve the data from the working hard drive, without an identical controller.

I remember having to reformat hard drives each time I tested out a controller card. Each one had their own protocol, and yet they all seemed to format the same. You do a low level format from DEBUG, by going G=C800:5 . That jumps you into the configuration and format portion of the on board (controller board) BIOS ROM. Now if you want to upgrade you PC/XT to a newer IDE drive, which most likely will be a big step up, you will need to get an IDE controller that has it's own BIOS and of course is also a 8 bit card. The 16 bit cards are all for AT class machines and will not work in an older PC/XT.

I have one of the BIOS optional cards and they can sense the hard drive parameters and allow you to boot them. I recently went to a 1.2Gig drive on my main system, only to discover that the BIOS can't really handle that large of a drive. Most brand new BIOS's will have updated code for the larger drives, but most older BIOS settings have limits of 16 heads, or 1024 cylinders, and 64 sectors as their maximums.

A Good buy!

A very good program to buy that comes with the bigger hard drives, is Ontracks's Disk Manager. I first used this program when I was installing a Novel network on a non-standard hard drive in a 286 system. The Disk Manager program can firgure out what your drive type is and then set up drivers and other needed programs so you can use the drive. My Gig drive's boot sectors now contain a special program that makes the BIOS able to read the hard drive. For repairing many of the older MFM drives, you might find a version of Disk Manger your ideal tool.

A last note on drives is watch out for block size. You might run into problems when moving to the larger drives. Both DOS and CP/M have only so many bits set aside for block allocation. That means as the drive size gets larger the block size must also get larger, or you will not be able to utilize the entire disk. Generally under DOS it goes like this, below a 128 megs it is 2K blocks, 129 to 256 is 4K blocks, 257 to 512 megs are 8K blocks, and 513 to 1 gig is 16K blocks. Above one gig you have to use 32K blocks, which means the smallest file will take up 32K of space. So even a 2 byte record uses 32K if you leave your hard drive in one partition.

I always recommend breaking drives into multiple partitions or false drives. You should keep you boot drive to under 256 megs if you can, thus all the very small startup files will only use 4K blocks.

Lastly

There is a whole passel of things that can still go wrong, you can have bad connectors on interfaces, wrong jumpers for the memory you have just added, overheating CPU (although it usually isn't the problem), a bad com port (but the other one is good), and all of these will drive you crazy finding them. The best approach is to obtain some good software and hardware tools. Practice a little snooping around while your system is running OK and see how to use the tools. Most newer MSDOS machines have MSD, which is Microsoft System Diagnostics. This tool will give you access to most parameters and minor tests of ports. There are better programs that will fully test memory, com ports, video screens, and about any option you might have.

People willing to get in their and hunt around, usually develop a series of personal tools and skills that speed up the trouble shooting process. I once came into a work location and found a note from the person who was on duty the night before. He had spent the entire night trying to figure out what had failed on one machine. He gave up and went home. I had plenty of repair time on the machine and he was sure I would fix it. It still blew him away when I told him how long it took me to find the problem, about 30 seconds. The machine has a reset switch and my first testing procedure is to reset the system. When I pushed the reset switch it didn't feel just right. A quick check with an Ohm meter showed the switch in fact was shorted and thus the system was in constant reset mode.

The basic skill I used was a strong personal knowledge of how the system worked and how it feels when you use the devices. You too need to develop that sense of understanding, and then trouble shooting will no longer be a black art, but a skill you have mastered.

That is all the advice I have time for, so send me those questions and letters, and keep hacking! Bill.


Kibler Electronics, PO Box 535, Lincoln, CA 95648-0535, USA. / bill@kiblerelectronics.com
Copyright © 1995, Kibler Electronics.