Schaublin 125-CNC retrofit.

Grrrrr I lost my post..
Darn, that just happened to me too. Had a whole essay for you
I appreciate being able to bounce my thoughts off you and the forum. Really helps to firm up my ideas.
I found your video about the variator. I had watched it once before (on page 32) What is actually the back gear ratio? 1:7 or 1:6.5?
Thanks for picking that Up. I said 1:7 from memory (failing). It is 1:6.5

At your high speed, what is the variator RPM range? eg something like 2200-3000 RPM = 800 RPM range?
The variator has much more authority. More like 1:4 ratio change range.

What is the variator actuator travel distance? Is it linear?
The linear screw, acts through a bellcrank on the spring loaded pulley sheath. There would be some non-linearity in the rotational motion of the bellcrank, but that is irrelvant given the non-linearity of the belt riding in the pulley, slip, belt wear, thermal expansion etc. The speed verses actuator relationship is probably ideally quadratic or exponential though. It is not linear.

I am wondering if you couldn't use one of these to sense variator position?
thesensorconnection.com/product/shock-ab...potentiometer?v=3003
We just need to work out how to sense the resistance value. A mesa THCAD could sense the voltage change with resistance very accurately if you have a spare encoder input.
I also started off with ideas about servo control of the variator. This could be done, but is unnecessary. The variator ratio changing through a 3Ph motor means that the rate of ratio change is constant. I could only alter and control that rate by replacing with servo or stepper, or approximate it with bang bang control. Given the wide range of speed control by the VFD, I dont need to finely control variator ratio. It take 5 sec to drive the variator from one stop to the other. Using 5 one second pulses, counting how many have been given gives me a 5 speed gearbox. Or with  half second pulses, it is a 10 speed, closer range gearbox.
The encoder closed loop VFD control  can deal with all the speed change between ratios.

Since it is a lathe, on CSS cuts, the spindle rpm needs to be constantly changing.

Plan A
Home the variator to an end stop on start up. Afterwards keep track of variator gear ratio by counting gear changes.
At a spindle on command, decide on high or low gear. Set it. It is fixed till next spindle stop.
Start spindle, set the variator to the lowest ratio closest to the commanded speed. Once the spindle is on speed, continue with the program.

As the facing cut progresses, linuxCNC will be increasing commanded speed, and the VFD will be increasing frequency until it approaches its limit.
At that point command the variator to speed up for 1 sec. As the VFD has much more precise and responsive control, during that sec, as the variator changes ratio at it's fixed rate (most likely too quickly), the VFD will probably reduce frequency to hold spindle RPM.
After  the variator stops adjusting,  the VFD frequency will again increase, towards its limit.
Rinse and repeat up to 5 times.

I am thinking if you can sense the actuator position, you select the back gear, then assuming say a 50 Hz VFD frequency, get as close as you can get with the varistor, then trim it with the VFD. Let the encoder feedback act on the vfd.
Adding a varistor or LVDT is still not a measure of gear ratio. You could define a function to map that value to ratio, but it would be subject to the inaccuracies of bellcrank motion, belt wear, thermal expansion, slippage due belt contamination etc. Plus, there is really no need to such accuracy in variator control, as the VFD will  compensate automatically.

The back gear is easy. Before the spindle starts, either high or low. Since that can only be based on the first speed commanded, I need to consider in any CAM PP, to stop the spindle between different types of cuts, otherwise the gearbox can be stuck in the worng range. For a really large facing path, I need to either accept 800 rpm top speed in the middle, or break the cut into two parts and stop and change gear between cuts.

Makes my head hurt again!
You and me both, but a problem shared is a problem doubled

Speaking of head hurting, thinking about it, the  current variator ratio could be constantly calculated. Motor RPM / VFD frequency is proportional to input voltage, so the variator input rpm is knowable in real time.
The back gear condition is known and one of two fixed ratios, and the spindle RPM is measured by the encoder, so variator output RPM is also known, and thus current ratio can be calculated.

This could be helpful. The arrays as look up tables are used in your Comp to inform Linux CNC what overall ratio to is in use, so it can scale the mapping of VFD input voltage to RPM. If I can constantly calculate that value, no look up table would be necessary.

I suspect someone is going to pop in now to inform me that I will also need to vary the PID values, as those which work with an overal drive ratio of 1:1 are completely different to those that work with 1:90, roughly the Schaublins overall rpm control range.

a closed loop stepper with a home switch, easiest way to have it "shift"

also easy to create a comp for

That whole motor, variator, gearbox module is really heavy, and PITA to pull out and reinstall and reconnect. The brackets, stepper motor mount, wiring, driver, and PSU would all need designing, manufacturing and installing (I have no space left for the driver and PSU in the cabinet). My Mesa motion hard ware has no Step/gen (although that could probably be flashed).
The stepper motion would have to be mapped to a non-linear ratio change, and still needs an integrated heirachy of control with the VFD.
So if that is all done, how does it improve things?

How is that easier than using what is already there, beautifully made to last forever, already connected, already electrically and mechanically working*  and just waiting for me to work out the comp? .

A comp which can keep track of five "gears" based on counting how often it has sent "change up" or "change down pulses" must be doable, although I am starting to doubt classic ladder for it, and see the advantage of a comp.  There is no need for accurate control of the variator, as closed loop VFD speed control will cover up it's inaccuracy automatically.

Rod's comp is a big help to me, as was doing that Aduino garden watering project. I am no programmer, but am getting the confidence to have a go at doing a LinuxCNC comp for the schaublin. I am sure I will struggle with the programming, and would appreciate your programming support once I start doing that.
Cheers,
Mark

*once my EMI is addressed

Thinking out loud here
Might be possible to make the variator motor as part of the PID loop, have VFD set ranges like 500, 1000, 2000 and use the pid to keep track using variator.
?

exactly what kind of motor is on the variator and what kind of control does it offer ? does it have a brake to hold it ? I am completely unfamiliar with the mechanical and electrical properties of the device so it's hard to give any kind of engineering input. Buuuuuttttt, a simply gilmer belt connection to a closed loop stepper with a integrated drive, and you could pwm a output on the card and write a simple hal comp.

That said, I'd be glad to help with whatever control scheme, but what exactly is the hardware in the variator box ? do you have any documentation you can point me to ? I'll gladly read it.

edit found your video about the variator. do you have a spare vfd ? if so, can you put it in TQ mode and see if you can control the variator with a 0-10v type signal for position ? I'd bet the spring force may be high enough to make this work. if not, a cheap servo drive and a encoder would be the next best bet. any 3 phase motor can be used as a servo basically. 

Hope that didn't come across as argumentative. I'd rather it be an expression of my laziness
I whipped up a quick video to explain it


Isn't that hardware gorgeous!