Compaq Deskpro 386s/20n: More Disk

29Sep/160

Compaq Deskpro 386s/20n: More Disk

So, the old beast is up and running once more. All is well, but I want a little more hard drive space. I've had a 256mb Transcend IDE "Disk on Module" lying around for a long time, so I've decided to install it into this machine.

The IDE connectors on these devices are reversed as they are actually meant to be installed directly into the motherboard. This Compaq does not have a standard IDE connector on the motherboard end, so I'll need a little bit of trickery to get this up and running. I'm going to call this device an SSD from here on; writing Disk-On-Module each time uses my shift key too much.

IDE Pinouts

What I'll try and describe below is where pin 1 is on either end of the cables and how the Transcend module doesn't mate correctly with the 'hard drive' end. For all intents and purposes, we're going to pretend that this Compaq Deskpro actually does have a standard IDE port on the motherboard end. We know it doesn't, but for demonstration purposes, let's stick to the IDE standard.

IDE cables are all female-to-female. Most of the time they have more than one female plug down the cable to allow both a master and a slave to be connected on one channel. The IDE cables in this Compaq are the original 40-pin type. These 40 wires are all connected straight through, down the cable and into the plugs. The drives then determine, via jumper settings, which is to be master and which is to be slave. In this scenario, we can take it for granted that this will occur and that we don't have to modify cables or plugs to allow selection of master/slave automatically.

The best way to explain the pin mapping is sideways. The diagram below shows pin one and two from the motherboard to the HDDs. The important part here is that mirroring occurs and that 1 and 2 seem to 'leapfrog' over each other on the way.

As you can see from above, if we plugged the SSD straight into the IDE cable IDC header then we'd have a mismatched alignment of pins. The top row of pins needs to be swapped to the bottom row. To do this, we either need to put an IDC header on the 'other side' (upside-down?) of the IDE cable, or create a new cable that'll swap the pins around for us.

Due to the staggering of the pins on the IDC header, it doesn't matter which way you orient the plug. The side of the cable is what matters. Then again, this isn't the total truth. If you were hooking up another hard disk, then the rotation definitely matters as the 'notch' must be facing upwards so that pin 1 is to the left (when looking at the 'hole' side.)

Confusing yet? The issue here is that everything is mirrored. Once you then add the rotation, it's a real mind-f##k moment. Either way, for this to work, we'll want to have a plug on the wrong side of the cable. Actually... is that what I really want to do?

You mean, ruin the Compaq IDE Cable?

The IDE cable in this Compaq is proprietary. The motherboard-end is an edge-connector onto the backplane and I can't imagine finding another cable like this, ever. Therefore, we're going to do as little damage as possible to add a slave drive. I could just slap the IDC connector onto the reverse-side of the cable cable, and this would actually minimise any work I had to do, but the issue is that I'd then have to leave a note in the machine to tell the next unsuspecting victim what I'd actually done. Unless you know what you're looking at, you won't have any idea that it's backwards and a normal disk wouldn't work.

Instead, I'll add an IDC connector the correct way, a little way down the cable, as per a standard slave connection. From here I'll then build an adaptor cable that will allow me to female-to-female plug in the SSD. Check the next diagram to understand this.

Note that the 'adaptor cable' seems to be miswired? Pin 1 is pointing to Pin 2!? This is perfectly fine, as that's the cables purpose in life. You couldn't do this with an actual device, but you can with a cable. In the end, if you follow the wires, you'll realise that we now have the pins lined up perfectly. We also have a magical space for the existing HDD!

Building the adaptor cable

After thinking about this for a while, I thought I'd google to see if anyone had done it before. Most people put IDE-CF cards in their ancient laptops, I've even tried to do this before. I then came across an article where Michael B. Brutman installs an SSD into an old IBM L40SX. It was great to see that my concerns of just plugging the 'SSD' into the IDE cable as-is was wrong and that the technique to build a cable was proven.

I went to the shops and bought the required components: ribbon cable and 3 40-pin IDC headers. I already had some rows of header pins available in the cache-o'-crap.

Constructing it was easy enough, just make sure that both your headers are facing up and, just for fun, align the notches the same way. We won't be caring about the notches... but do it anyway!

Now that we're done, we're going to hack the original Compaq cable slightly... let's put a header on the cable below the current header, allowing for a slave to be connected. Except that we're going to put the master here, using the old header at the end for our new drive. You can see from the shots below that the original cable had enough slack to allow us to fit the header in.

At this point, prior to plugging in the SSD, test the modification we've just done to the existing IDE cable! After this, the final trick is to just slap it all together. Make sure to set the SSD as the slave! I used the header pin rows, sliding the plastic down mid-way. I then found a power cable Y-Splitter and jammed it all into the case.

Configuring your new disk

Back to the BIOS, you'll need to select a size that is 'just-below' your SSD size. I found a 212mb that I was happy with.

This all worked like a charm. I was then at the usual DOS Prompt. FDISK indicated that I had a new fixed disk and allowed me to to switch to it by using menu option number 5. Doing so showed the old Linux partitions!