Converting a PS/2 TrackMan Marble to USB

This is my favorite pointing device ever - positioned to exhibit its tragic flaw, an obsolete PS/2 interface:

I don’t think the ergonomics of the original TrackMan Marble T-CH11 have ever been improved upon since it first shipped around 1996. The hand-fitting top curve, the thumb-operated trackball, the real middle button without obtrusive wheel. Just to make it clear what we’re talking about:

Yes, that’s the T-CH11 Trackman’s modern descendant, the Logitech M570, hiding its head in shame behind the TrackMan. The M570’s ergonomics aren’t bad, except that a scrollwheel that tends to scrollwheel when you want to middle-button is a deal-killer - and I’m not a big fan of having to rely on batteries, either.

Enough other people feel similarly that there’s a vigorous market for used TrackMen on eBay and Amazon - pretty impressive considering the age of the product! Even the later, inferior 2-button-and-wheel versions are apparently in heavy demand.

This is a Sanoxy adapter. Passive PS/2-to-USB adapters don’t do it for the Trackman - you need actual protocol conversion, which is what the wart in the middle of the cable does.

This is what a Sanoxy that has met with a tragic accident on the slow boat from China looks like:

Note the picturesquely shattered wart casing. I never did find the third piece. However, the adapter actually still works with a TrackMan; the only detached wire is a keyboard lead.

Here is the adapter, fetchingly draped across the Trackman’s exposed interior. What this should make clear is that there’s plenty of room inside the Trackman case for the adapter microboard.

(Note: the three microswitches you can see near the bottom are a cheap, standard part called the Omron D2F-01F. If your keys start to stutter or go dead, it’s a few minutes' light work with a soldering iron to replace them. I have actually done this to resurrect my wife’s favorite trackball.)

And so, the plan.

If you want to simplify the recipe, you can just leave the USB cable attached to the microboard in place and mate it to a female-to-male USB cable extender.

To make this work, we need to know what the leads on both sides are doing. Here’s a low-angle shot of the PS/2 on-board connector in the TrackMan:

We don’t know the part number for this on-board connector, so we’ll just call it the OBC.

The leads from the TrackMan PS/2 cable are, in sequence: Red, Brown, Yellow, Green, Black, Orange, White. Six of these correspond in some unknown way to the pins in the male PS/2 connector; the seventh is unoubtedly a ground to the connector shell.

Here’s a closeup of the Sanoxy microboard:

This is intended to document the fact that the green (mouse/trackball) connector goes to the inner row of solder pads. Next, a closer view:

The mouse (green) connector is the lower one. It supplies four leads: Black, Green, White, and Red, corresponding in some way to pins in the female PS/2 connector.

Now it’s time to look at the PS/2 DIN port. The Wikipedia page on the PS/2 port helpfully informs us that the four active lines in a female 6-pin DIN jack like the one on the Sanoxy are:

Lines 2 and 6 are active on some laptop and notebook computers, where they can be used with a Y-shaped port-splitter. While that is noted for completeness, it’s not relevant to this build.

On the TrackMan they seem to have a different purpose. There is direct evidence, in the form of T-CH11 pictures like this one from eBay, that a passive adapter sufficed to mate the TrackMan’s PS/2 cable to a serial DB/9 connector. Thus, it is likely that Pins 2 and 6 are an RS-232 TX/RX pair.

Here’s a pinout:

Unfortunately, it gives no standard color correspondence. I am informed that there was none - colors on these leads varied randoimly, sometimes between different revisions of the same device.

USB is standardized. There are two color schemes, these wires fit the more common one

The pin numbers are from the right as a Type A male connector is held with the empty half side below, electrodes facing up.

We can see that these colors match the wires attached to the upper (USB) connection in the picture

As for the PS/2 connector, time to break out the trusty voltmeter… By doing simple continuity checking on the PS/2 cable we get the second column of the following table. Looking at the DIN pinout lets us deduce the signals on each wire.

At this point, my friend Phil Salkie, the expert hardware troubleshooter I recruited to do the hand-work on this build because my skills aren’t so great, made an inspired guess.

Here’s an ASCII-art diagram of the Sanoxy microboard, with lead colors R = Red, W = White, G= Green, and K = Black (B would be Blue, and L for Brown) attached to the rectangular solder pads:

(I have taken apart another similar Y-shaped active adapter not branded Sanoxy and found an almost identical-looking microboard inside. Best guess is that it will be found in most, if not all, active adapters.

Phil’s guess was that the match of the standard USB lead colors on the PS/2 side was not a coincidence, but a tip-off by a smart and conscientious designer that the PS/2 assignments were as close to a signal match with their USB counterparts as possible. In other words, this:

We could have verified this by continuity-checking the leads in the Sanoxy mouse cable, which is how I’d expected to do it when I first planned the build in my head. But I agreed that Phil’s theory seemed extremely likely. So we forged ahead, knowing we could desolder a mistaken guess pretty easily.

We weren’t wrong.

My original plan had been to identify the OBC type, get a male one, and crimp leads from the PS/2 end of the microboard into it. That way the original PS/2 cable would still be usable if our Frankensteining attempt failed.

This plan went in the dumper when Phil informed me that the machine we’d need to do the crimping costs $1700. Oh well…build plans, like battle plans, often don’t survive contact with the enemy. One adjusts.

Instead, Phil cut the PS/2 cable a few inches from the OBC, producing this:

He then tinned the ends of the leads and soldered them to the microboard’s mouse pads, after first desoldering and removing the keyboard leads. The result looked like this:

We tested it. It worked. We than replaced the short gray USB cable from Sanoxy with a 6-foot USB cable, actually a salvaged USB printer cable with its Type B connector cut off. The result looked like this:

To prevent the microard from making unwanted contact with the TrackMan board, we wrapped it in electrical tape.

The result: A wondrous creature not found in nature - all the simplicity and ergonomic win of the original, but USB and hot-swappable.

Alas, the first ending was not entirely happy. Turns out the Frankenball with six-foot USB cable is picky about what USB transcievers it will talk to. Success on a Thinkpad X60 and an HP Pavilion D7. Failure on the Xeon-ES1650-based Great Beast and an older Dell Latitude laptop. The Great Beast’s USB hub can see and enumerate the device (as "13ba:0017 PCPlay PS/2 Keyboard+Mouse Adapter"), but won’t register events from it.

After some investigation with a scope and and cable-swapping we were able to determine that the problem seems to be signal attenuation downstream of the adapter. Reattaching the shorter original cable restores correct behavior on all machines we tested.

This might be solvable with a higher-quality PS/2-to-USB adapter, or a lower-loss cable. Here is a simple though inelegant solution cannibalizing a second Sanoxy with an intact case:

Don’t open the Trackman case at all. Instead, cut the the DIN connector off the end of the Trackman’s PS/2 cable, pull the wart open, and desolder the Sanoxy’s PS/2 cable. Mate the Trackman PS/2 leads to the Sanoxy PS/2 pads as in the previous recipe, then press the wart back together.

As a result of publishing this HOWTO, we may now know why the original ergonomic shape was degraded in later Logitech models. There was a patent dispute over it! See Gart vs. Logitech. Though Logitech prevailed on appeal, they may have altered the shape to be less Gartian at some point in the proceeding. Adjudication of the parent claims was complicated by a dispute about whether Gart’s C&D letters met the requirements of the law.

I’ve documented and photographed this build carefully in hopes that new makers might learn not merely the recipe itself but something about the kind of thinking that goes into composing one.

We apologize for the blurry pictures. You can blame my not-very-good phone camera, or my not-very-good technique, or both.

This is how I banished the last remaining PS/2 requirement from my house. It was also a learning experience for me in basic makering. I’ll do the handwork for the next one on my own.