Tokyo PC Users Group
	  Home Page
Members Only
Become a Member
Meeting Info & Map
Corporate Members
Workshops & Training
Other Clubs
Job Hunting?

Assembling a PC: Lessons Learned -- Part 1

Masayo Arai

Assembling a PC:

Lessons learned

Part 1 of 2

By Masayo Arai

Hi AJ readers!

Started as a hard disk crash, finally got a new handmade PC. I had quite an exciting and productive month of trials and errors in my first experience. I want to stress that you do not have to have PC hardware experience or knowledge to make a PC. If you are motivated by my story, try to make your own. Good luck!

I also hope this story provides you some hints if you have similar problems. See the table below. If you find any similar device(s) or a similar combination of devices installed in your PC, you might be having similar problem(s).

September 13, 1997

It was my first and most serious problem since I owned my old PC named Beauty. I had been using Beauty for about three years at home. Up to this day Beauty had contributed to me very diligently. Suddenly the series of problems started.

While I was reinstalling Windows 95 to upgrade from the 1995a version (4.00.95) to the 1996b version (OSR2), suddenly drive C: (540 MB Western Digital Caviar 2540) stopped reading and writing.

I checked everything of the crashed hard disk, the CMOS settings, the jumper settings, and cable connection. I also tried to swap drive C: and D:, but one of the two hard disks was not recognized by the system at all.

Lesson 1: Back up all the data you want to keep including all the customized INI files and personal dictionary files. Fortunately, I was using an MO disk for my working data. All I lost was the August family expense spreadsheet and personal dictionary for MSIME. Later on, It took a day to recover these data.

Then, I gave up trying to fix the crashed drive. I changed the jumper of the ex-slave drive to make it a single master drive. As I am using both English and Japanese OSs and applications, 540MB is not enough, but anyway I had to make Beauty boot up.

I started MS-DOS from drive A: and tried to transfer the DOS system by executing SYS C:. But there was no space to copy the DOS system. I had to backup everything and reformat it.

Lesson 2: Before starting to use any additional hard disk, reserve space for MS-DOS bootup files. Later on, it might be used as a primary drive.

September 20, 1997

I tried three or four times to make the reformatted hard disk bootable, but I could not. I tried the SYS command from drive A:. This did not work. Then I tried a thorough MS-DOS installation, but the result was the same. The BIOS told me ".....Boot 2." and stopped proceeding with anything. The solution finally was System Commander to make the boot sector. This idea was successful, Beauty started the System Commander menu and I succeeded in starting MS-DOS.

Lesson 3: When you cannot make your hard disk MS-DOS bootable, try installing System Commander. (I'm not sure if this is a good solution. Let me know if someone knows any other way to make a hard disk bootable.)

540MB is not enough to install both English and Japanese Win95 and applications for those. So I asked Mike to use one of his hard disk drives. (He is usually sitting next to me and playing games while am using Beauty.) He lent me an IBM 2GB SCSI drive. Thanks, Mike.

My attempt was to install Win95J on the IDE 540MB drive and Win95E on the SCSI 2GB drive. This did not succeed. Win95J worked OK, but Win95E did not start because Win95E Setup deleted some lines in the config.sys file which are needed to install SCSI devices. When Win95E needed to load files from drive D:, there was no drive D: on the system. I remember that Mike had a similar problem when he tried to install Windows on a SCSI drive though he may not remember this.

Lesson 4: If it is needed to load SCSI drivers from config.sys, it is recommended not to install Windows on the SCSI drive.

So, I needed two IDE hard disk drives for two Windows. I decided to make two partitions (200 MB and 300 MB) on the 540MB drive. I used FDISK, FORMAT /S, and System Commander. This time, I used the above lessons 2 and 3.

To make the configuration as simple as possible, I deleted all the SCSI related lines from config.sys and autoexec.bat before installing Windows again. This solution is to use the SCSI drive only after Windows starts. After two Windows were installed, I got the following three hard disk drives.

  • 200MB drive C: for Win95J system
  • 300MB drive D: for Wino5E system
  • SCSI 2GM drives E: for application programs and data.

Under this configuration, I could not solve the following mysterious problems:

  • NJWIN and some other programs did not start. Windows said "Program starting error" or "The device is not working". Even when I copied the program files from E: to C: or other drives, the result was the same.

  • Iomega Zip drive could not be installed. Guest couldn't find the Zip drive.

I tried changing the SCSI driver files, terminator settings, SCSI ID settings, and SCSI hard disk from IBM to HP (another SCSI drive from Mike), but the problem was not solved. The SCSI configuration of Beauty was:

  • Mistubishi Kasei MKIF/PC SCSI adapter (or AIC-6X60 single chip)
    Tried sparrow.mdp (Win95a) SCSI driver and AHA150X/1510/152X/AIC-6x6 (Win95b) SCSI driver. Both did not work.

  • Mitsubishi Kagaku 230MB MO drive (Olympus)

  • HP SureStoreDisk 2000S SCSI HD or IBM DORS 32160 SCSI HD

  • Fuji Film ZIP drive (Iomega compatible)

I could have asked Mike to let me use one of his IDE drives, because he had 6GB and 2GB IDE drives. If I had done so, this story would have been shorter. But I did not do it because the 2GB IDE drive is his main drive and I would have more trouble making his second IDE HDD bootable.

Lesson 5: The older type of SCSI controller is not only slow, but causes unexpected problems. It would work for one SCSI device but may not work with two or more.

The problem had not been solved for a week.

September 29, 1997

Up to this point, I did not spend any money. It was a big decision, but finally I decided to buy a new IDE hard disk. I went to Comdoc Mizonokuchi in the Maruetsu Triangle on my way home from the office. At the store, I pointed at the Western Digital 3.2 GB drive and asked "Will this work on my IBM compatible PC? I have been using a Western Digital 540MB drive on my PC." The Comdoc guy said "Yes. All these hard disks are compatible with DOS/V machines."

After I came home, I found that Beauty could not recognize the whole 3.2 GB. But I could not return it due to Comdoc's policy.

Lesson 6: Do not believe the PC stores.

Lesson 7: Study about what you want to buy before you buy it.

The new 3.2GB hard disk needed Ultra DMA. Beauty did not have such advanced technology. Its old Award BIOS did not have parameters for a 3.2GB hard disk. I also found that Mitac is not providing BIOS upgrades for obsolete models.

I had already spent 30,000 yen for a hard disk, so I had to use it even if the BIOS did not recognize the drive type. I used FDISK on Beauty but only 540 MB of 3.2 GB was usable. Then I connected the new hard disk to Mike's new machine "The Brute," a 266MHz Pentium II. The Brute easily found the correct hard disk parameters and FDISK worked no problem. On The Brute, I made primary 2GB and extended 1 GB logical drives on the new hard disk.

The result of this experiment was quite bizarre, Beauty's Win95 assigned three drives, 2GB, 135MB, and 1GB on the new hard disk. Beauty's DOS assigned only 2GB and 135MB drives.

I moved the hard disk back to The Brute again, and repartitioned with a different setting (2GB and 1GB of extended drives). This was worse. 2GB and 1GB drives assigned (F: and G:) after the SCSI drive (E:) on Windows, DOS did not see any drive from the new drive.

After several times of repartitioning and reformatting the new drive, I compromised to use only 2GB of the 3.2 GB space. I made it like this by assigning a 2GB primary drive only (all this partitioning and formatting was done on The Brute; Beauty did not work for larger drives).

My drive configuration resulted in:

A: 3.5" FD

C: 540MB old IDE drive

D: 2GB of 3.2GB new IDE drive (1.2GB couldn't be used)

E: SCSI 2GB borrowed from Mike

I: SCSI MO drive

K: CD-ROM 2x

(Z:) Zip drive was not recognized

Now I had both an unstable SCSI and an unstable IDE situation. Very strange problems happened:

  • When I changed the SCSI drive volume label, Win95 hung up.
  • When I accessed the MO drive, Win95 hung up.
  • When I accessed My Computer, Win95 hung up.

The MKIF/PC SCSI adapter had been working no problem before with only an MO drive. I decided not to use the SCSI hard disk. I returned the SCSI drive to Mike. The system seemed to be fixed and working properly.

A: 3.5" FD

C: 540MB old IDE drive

D: 2GB of 3.2GB new IDE drive

I: SCSI MO drive

K: CD-ROM 2x

I installed only Win95J on drive C:. Since the system did not know what kind of drive D: is, I used drive D: only for not so important data.

October 11, 1997

I had not used drive D: so much, but the system was working so-so. Still some programs did not start, but essential programs were working.

I tried to use drive D: to back up some large amount of data from the MO disk. While copying data from the MO to the new hard disk, the system reported many bad sector errors. Scandisk stalled due to too many bad sectors. I think this was caused by the mismatch of the hard disk parameters. So I gave up using maximum space on the new hard disk and repartitioned again. This time I assigned the space that Beauty could deal with. I do not know how many times I repartitioned the new hard disk. The resulting disk configuration was:

A: 3.5" FD

C: 540MB old IDE drive

D: 540BM of 3.2GB new IDE drive (most portion could not be used)

I: SCSI MO drive

K: CD-ROM 2x

This day, I decided to replace the motherboard to get a newer BIOS that could handle the 3.2 GB hard disk.

Lesson 8: Do not use a hard disk if the parameters cannot be set with your BIOS.

Buying the 3.2GB hard disk resulted in replacing the motherboard. If Beauty could have updated its BIOS, it might not have been necessary to replace the motherboard, but I was beginning to know that fixing small problems one by one takes more than fixing the basic problem.

October 15, 1997

Before buying a motherboard, using Lesson 7, I studied about motherboards, chipsets, CPUs, busses, and power units. Then I set the following goals:

  • Studying a PC is the first priority

This was my first time to assemble a PC. No experienced person was available around me. There was no assurance that I would succeed or not. Using minimal cost in case it failed. As a last resort if I were unable to make it, I would buy a set from Gateway 2000.

  • Use already owned parts as much as possible

We have a lot of peripheral devices Mike had collected for the late Beast (His previous machine which suddenly died a few months ago. In fact only the motherboard or power unit was dead and all the peripheral devices I checked are still fine. ). If I could not make them useful, maybe nobody would use them.

  • Balance among the parts is important

As I experienced with the 486 motherboard and 3.2GB hard disk, it's important to have good balance between parts. All the parts are one-year old or older. The motherboard and CPU didn't have to be the current most advanced ones.

I did not want to do time-consuming Akihabara hunting, so I again went to Mizonokuchi Comdoc. Using lesson 6, I did not buy what they recommended. Their recommendation was to buy motherboard and case with power unit all together. They said that case and board should have good coordination. They said Mitac's old case does not fit the new board. I ignored this recommendation. After I bought a board, if I found that the old case did not fit, then I could buy a new case. Following the goal of "spending minimal money", I only bought a motherboard and a CPU.

See you soon next month for part two of this saga!

© Algorithmica Japonica Copyright Notice: Copyright of material rests with the individual author. Articles may be reprinted by other user groups if the author and original publication are credited. Any other reproduction or use of material herein is prohibited without prior written permission from TPC. The mention of names of products without indication of Trademark or Registered Trademark status in no way implies that these products are not so protected by law.

Algorithmica Japonica

November, 1997

The Newsletter of the Tokyo PC Users Group

Submissions : Editor

Tokyo PC Users Group, Post Office Box 103, Shibuya-Ku, Tokyo 150-8691, JAPAN