Controling a 10 turn potentiometer with 51200 microstepping stepper motor?

[sourcecharge]

Has anyone done this before?

If it was a 10 turn precision pot, that was linear throughout it's range, I'm wondering if that would make a variable electronically controlled potentiometer with 512000 positions using the microstepper CWD860?

So, because the poteniometer is electronically controlled, the position would be known and could be converted to logic inputs of 7 segment readout. 

A quick check on digikey shows their "best" digital pot have only about 1000 positions, and you would have to deal with the coding with a uC.

I'm guessing that the number of digits would be 5 1/2 ? (512000/10/10/10/10/10/5) 

Edit:  So, if a 10V reference was used, it would be about 51200 steps per volt, the output would need to be calculated for the correct voltage.

If the micro-stepper controller had more positions than the digit readout could be increased if this is correct.

So a 5 1/2 digit read out of a compared voltage to a known reference would give a pretty accurate DC voltage measurement which is the goal here...

This would involve 6x 74192 up/down counters, 2 OR gates and an inverter, so that the up/down can be switched, and a clock for the counters (any VCO or 555 type will do) would also be be the stepping clock.

If the ref was 10V and the input was lets say 50V, then the input voltage would of course have to be divided down before the "measurement" but this could be done with relatively cheap precision fixed resistors.

If the ref was 10V and the measured signal was divided down to 5V, the stepper would start the "measurement" by turning the pot connected to 10Vref and 0V while the wiper output is used with a comparitor against the 5V signal.  So the microstepping would give 512000 position, and can be precisely controlled.  When the comparitor goes high, then the stepper changes directions.  This starts the pot to go in the opposite direction, and when the comparitor goes low, the stepper changes direction, and this happens over and over again while centering on the position that has equaled the input voltage.

6 linked up down counters would be counting the whole time and now bouncing between a range of voltage points that the motor is controlling.   The BCD counters could directly output to my 7 segment output displays (I've got alot of TIL311 which have a 4 bit latch, decoder and driver built in them). 

Do you think the 10 turn precision pot would be linear?

Is it really 5 1/2 digits?

If the pot was calibrated, with regression test points to verify the resistance points the output could be adjusted by having stop points on the pot.

Well, what do you guys think?

 

EDITED: due to fast typing and poor grammar.

[Brumby]

Has anyone done this before?

I don't know - but there's a lot of things I don't know.

Do you think the 10 turn precision pot would be linear?

How precisely do you want to define "linear"?

Is it really 5 1/2 digits?

Depends on what you define as linear - but if I take your question as asking whether the practical measurements would follow the mathematical inference of "5 1/2 digits" then I would be somewhat surprised if it was.

[PartialDischarge]

Its been done but its a thing of the past. I’ve seen this in electromechanical laser ranging systems (designed in the 70s 80s) for example. Your calculations forget that you can use gears also, so that 5 turns in the motor translate into 1 turn at the pot.

That said, nowadays there are cheap and good DACs, so that digital to analog conversion usign a motor and a pot is mostly useless complexity

[sourcecharge]

Its been done but its a thing of the past. I’ve seen this in electromechanical laser ranging systems (designed in the 70s 80s) for example. Your calculations forget that you can use gears also, so that 5 turns in the motor translate into 1 turn at the pot.

That said, nowadays there are cheap and good DACs, so that digital to analog conversion usign a motor and a pot is mostly useless complexity

I'm quite a noob regarding DACs or ADCs, what would be an excellent ADC that can put out BCD for digital output without using a uC or memory that can support a variable number of 7 segment displays?

I was guessing that any ADC or DAC would require supporting ICs that would require programming, but I was hoping for an easy solution with digital logic.

The CWD860 has a 10000 step function and this would allow direct calculation of the voltage without using complex programming.

The motors might be a thing of the past, but that actually says nothing about it's accuracy.

I was thinking that the digital potentiometer was how the ADC was going to measure, but those had at best 1024 positions possible.

Was anyone able to get good accuracy with such a setup?

The gearing of such setup would also be interesting but the "slop" between the gearing would have to be zero if such a stepping factor was involved.

I think that might be somewhat unrealistic.

That's why I was thinking that a direct coupling was necessary.

Using the 10000 step function would give a 100uV resolution of 1V, and it would give direct conversion using up/down counters for a 5 1/2 digit output.

Also, if different reference voltages were used, even lower voltages could be measured.

I would suspect that in the past, these cheap and accurate micro-stepper drivers were unavailable.

Wouldn't that require a reassessment of this idea?

I'm glad though, that I'm not the only one that thought of this, as my ideas are somewhat uncommon, or possibly even outdated.

[JS]

For the motor, microstepping is more about smoothness than movement precission, don't trust the micro steps positions to be precise or have any decent repeatability. Also, while micro-stepping you need to keep the motor powered with a PWM signal, so noise to the system might be a problem. There are really cheap steppers with reduction on them but the speed is sloooow to wait for 10 turns.
https://www.aliexpress.com/item/Frees-hipping-5V-4-Phase-Stepper-Step-Motor-Driver-Board-ULN2003-with-drive-Test-Module-Machinery/32706559510.html?spm=2114.search0104.3.1.6ff54c6bsrqFvB&ws_ab_test=searchweb0_0,searchweb201602_3_10152_10151_10065_10344_10068_10547_10342_10343_10340_10548_10341_10696_10084_10083_10618_10307_5013511_5013611_10059_5013715_100031_10103_10624_10623_10622_10621_10620_5013815,searchweb201603_45,ppcSwitch_5&algo_expid=5640d650-5daf-4922-8082-9b96ce80ad45-0&algo_pvid=5640d650-5daf-4922-8082-9b96ce80ad45&transAbTest=ae803_1&priceBeautifyAB=0

  Mechanical movement won't be as precise either, even if you do closed loop axis movement the backlash between the wiper and the axis is out of your control. The linearity of the pot won't be that precise either, check some datasheets of 10 turn pots and see what you get. Then there is stability to worry about, temp and long term, I don't know the specs of this kind of pots but I don't think they will be to 5 1/2 digits level.

  I don't think you are going to get 5 1/2 digits out of this system, but it could be used for the last step of the system using other way to control the higher decades. I think it might be useful to have a really fine adjustment for the source but the trusty digits should be switched selection. A binary divider (R2R or binary KVD) with 19 relays could do 5 1/2 and the most limiting factor will be the resistors of the higher bits, so you can use cheaper resistors as you go down. The other approach is the power design approach, which uses a variable resistor instead of a divider, which makes switching easier but resistor values are all over the place, not just a few different values, but for 5 1/2 digits might be fine, they made 4 1/2 digits half a century ago...

JS

[Mechatrommer]

dont forget backlash, your design will have backlash, the trimpot itself has backlash. you need to study how much reverse turn to compensate backlash and put your "mechanical DAC" system back into linearity. but as others said, why bother if you can make it in full silicone die. i've thought of this mechanical stuff in situations where silicone ic will not be feasible, but i have not yet the need for your 51K position accuracy.

[sokoloff]

I also don’t think that 9+ discrete digital logic chips is easier than a micro. Pair up with somebody who can do the micro coding if you must, but this seems tailor-made for a micro.

You’ll be able to do all kinds of backlash compensation and other chicanery in code that you can’t realistically do in fixed digital logic.

[PartialDischarge]

I'm quite a noob regarding DACs or ADCs, what would be an excellent ADC that can put out BCD for digital output without using a uC or memory that can support a variable number of 7 segment displays?

I was guessing that any ADC or DAC would require supporting ICs that would require programming, but I was hoping for an easy solution with digital logic.

Unless you really are trying to do something different for the fun of it (we've all done it), don't waste your time and learn something that will give you x100 more results.

Many DACs can be controlled with just many parallel jumpers to their inputs, MX7538 for example. While you're at it get a Psoc 3, 4 or 5 evaluation board from Cypress, that will open many doors for you.

[Richard Crowley]

For the motor, microstepping is more about smoothness than movement precission, don't trust the micro steps positions to be precise or have any decent repeatability. Also, while micro-stepping you need to keep the motor powered with a PWM signal, so noise to the system might be a problem.

In one of my previous jobs I ran an E-test lab where we performed electrical tests on 200 and 300mm silicon wafers. We used automated wafer probers (Electroglas) which use a clever 2-dimensional (X/Y) flat-bed linear stepper motor scheme. By using microstepping and calibration tables stored in ROM, they could achieve positional accuracy and repeatability over the better part of 1m, good enough to accurately probe quite small bond-pads on the dice.  They reputatedly used laser interferometer position measurements to create the calibration tables unique to each machine.

We routinely made quite sensitive measurements and never encountered problems from the continuous PWM stepper motor driving current.  They may have been smoothing out PWM current to achieve this.

Which is not to say that a mechanical lash-up of stepper motors and pots seems in any way reasonable here in the 21st century as a replacement for a DAC.

[sourcecharge]

Well, I gave it a try with a 5k 10 turn pot.

Although it was not a precision pot, it should have been good enough.

I found that with a 10000 step, the resistance would skip, as I would step it one step at a time, it would increase either by about .5 ohms every step, or it would skip a step and then increase by about 1 ohm.  The actual resistance per step varied and was not exactly the same resistance even if it didn't skip.

The forward back linearity resistance was aweful too between 1 step and back.  So the backlash was definitely a problem.

The 40000 step didn't help either, the linearity of the pot just was not good enough for a microstep turn of 360/10000 degrees.

So, ya it was a bad idea.

Thanks all

[ogden]

Using the 10000 step function would give a 100uV resolution of 1V, and it would give direct conversion using up/down counters for a 5 1/2 digit output.

I am sorry to disappoint you, but most multiturn pots are far from 5 1/2 digit resolution. Before you continue with your project, you shall get familiar with what's inside multiturn pots, at least do some testing w/o stepper. [edit] I see you did testing and it was conclusive