mouse-to-5200 trak-ball mod

[Swami]

Hi,

I was going to attempt to make a spinner and handheld trackball from one of my atari ST mouse emulators, which convert USB mouse ouput to pairs of 90deg out of phase quadrature trains. The emulators, being digital synthesizers, output fairly perfect square waves, like good Schmitt trigger processed quad square wave pairs. I have five of these types of TOM-like devices (although different brands) which output the same type of Xpair/Ypair 4-lines of quad data, so at least one of them should be useful as a starting point (another option is a less perfect modern optical encoder - although all you have to do is stick a knob on it for the spinner's human interface).

Now, my first question is, what would be the last best point to add these Atari ST mouse outputs into the circuit diagram (see attached)? My notion is that it would have to be at the split-off points right before the 4013B wave synthesizer IC, since the SW trains are split and differently processed after this point and then recombined at the 4011B NAND gate and are most of the way to being processed into frequency proportional pot-like voltages for left/right and up/down.

Second, I'm wondering if one could simplify the electronic processing of the digitized square wave pairs (X pair and Y pair) after the above entry point with an available modern IC or would one need to, or simply be practical to, duplicate the slew of IC's, gates, resistors, etc., after the op-amp assembly? Apparently, something of the sort exists, but you only need 1kHz or so, not 400 kHz, so it should be a lot cheaper and this one isn't made anymore.

 

https://www.usdigital.com/products/interfaces/encoder/converters/EDAC2

Thanks

 

[BigO]

I assume you are going to interrupt the circuitry prior to your signal injection point.

 

Myself, I'd probably inject at the inputs of the 339 just to be sure that there's no issue with signal levels, floating inputs, etc.

 

Note, however, that the 339 inputs that accept the encoder signals are pulled up. That might not be needed or may be in conflict with the signals from your device.

[Swami]

I assume you are going to interrupt the circuitry prior to your signal injection point.

 

Myself, I'd probably inject at the inputs of the 339 just to be sure that there's no issue with signal levels, floating inputs, etc.

 

Note, however, that the 339 inputs that accept the encoder signals are pulled up. That might not be needed or may be in conflict with the signals from your device.

 

I would prefer it would be from scratch, rather than tied to the Trak-ball, so the electronics before the quadrature input would be deleted. It would be nice to have a couple cheap 5200 trak-ball boards to test both placements on, but that would involve cutting traces before the quad input points. It is not a large circuit, though, so hopefully it could be fit on one of my small breadboards. With the USB to Atari ST mouse adapter I can use any ps/2 or usb mouse I want while leaving the firmware to the experts. I have read the op-amps are there to amplify and straighten the square waves, so the question of including them depends greatly on the adapter quad SW amplitude vs the Trak-ball's after the op-amps. Unfortunately, I am short one oscilloscope, but I may be able to work around it.

[BigO]

I see. I wasn't sure, but thought you wanted to adapt the CX53.

 

It's really just a few bucks worth of parts, not counting a custom circuit board if you go that way.

 

I wouldn't be concerned about amplitude too much if the signal is coming from digital circuitry. You likely would be able to skip the 339, but may still need pull-up or pull-down resistors.

 

As for circuit design from scratch, I'd be really tempted to try using a microcontroller to replace the active circuitry.

 

Not sure if interrupts support detection of both positive and negative edges, but that would be convenient for decoding the quadrature inputs. Probably doesn't work quite that way.

 

Direction can be determined by looking at the state of one bit when the other bit changes state.

[Swami]

I see. I wasn't sure, but thought you wanted to adapt the CX53.

 

It's really just a few bucks worth of parts, not counting a custom circuit board if you go that way.

 

I wouldn't be concerned about amplitude too much if the signal is coming from digital circuitry. You likely would be able to skip the 339, but may still need pull-up or pull-down resistors.

 

As for circuit design from scratch, I'd be really tempted to try using a microcontroller to replace the active circuitry.

 

Not sure if interrupts support detection of both positive and negative edges, but that would be convenient for decoding the quadrature inputs. Probably doesn't work quite that way.

 

Direction can be determined by looking at the state of one bit when the other bit changes state.

 

One would think 30 years later you could convert the cleaned up quadrature from the adapter into bipolar frequency proportional voltages with a couple ICs, a few components and a micro-controller, but I'm not sure what I'd be looking for or if they are something common enough to be mass produced. If it can be done for $20 or so, it is worth looking into, though.

[BigO]

 

One would think 30 years later you could convert the cleaned up quadrature from the adapter into bipolar frequency proportional voltages with a couple ICs, a few components and a micro-controller, but I'm not sure what I'd be looking for or if they are something common enough to be mass produced. If it can be done for $20 or so, it is worth looking into, though.

I haven't studied that final stage of the CX53 output enough to completely understand what it's doing, but it's just a handful of passive components on that end. Having thinkering with this a few times here and there myself, I figured I'd just leave that final stage of resistors and cap as-is, at least for a first pass.

 

Considering the RC network at the end of the X axis, essentially the digital circuitry is supplying pulses to one half of the RC network when the ball is spinning one direction and is supplying an inverted version of that same pulse train to the other half of the network when the ball is spinning the opposite direction. The pulse train itself is pulses of a minimum width (as controlled by the 4538B One Shot - retriggerable configuration). Eventually, at higher and higher ball spin rates, the pulses of minimum width merge into an eventual steady state voltage. It shouldn't be too difficult to output comparable signals from a microcontroller.