DRO on the Cheap (Or Not)

[RustyArc]

Having more-or-less decided to get a cheap Ebay milling machine, my thoughts turned to kitting it out with a DRO, in particular, the TouchDRO setup that's designed to use a cheap Android tablet:
https://www.yuriystoys.com/p/android-dro.html

I like the idea of the extra features it offers and the scope for more features with updates to the app. And having a tablet at hand could be very useful (I use a Google spreadsheet on my phone for lathe feeds and speeds).

Also, the cheapness seemed attractive - I planned to buy a cheap mill that wasn't going to be the most accurate thing in the world, so dropping a bunch of cash on a DRO for it doesn't make much sense.

So cost of tablet aside, all that was needed was an MSP430 Launchpad - $20 a Bluetooth module - $6 and some scales. Hunting about, I came across this read head:
https://www.machineryhouse.com.au/D696 - $88

And this scale:
https://www.machineryhouse.com.au/D694 - $55

The scale comes as a 1.5m strip that can be cut to whatever length and simply stuck to the appropriate surface on the machine, making for, potentially, a very simple installation.

The 1.5m would be enough to do all 3 axes, and then it'd be 3 of the read heads.

That all adds up to $345 for an easy-to-install 3-axis DRO setup (or $268 with the H&F sale).

So I ordered all the bits, although just one of the read heads. The MSP430 and Bluetooth module arrived first - installing the firmware was pretty straightforward and that part was up and going pretty quickly. The scale and reader from Hare & Forbes took a bit longer, and when it arrived, I encountered my first challenge - the read head had a cable and mini-DIN plug, but no pinout showing power and signals. I lopped off the plug, and with a lot of Googling, found the colour code for the 8 wires in the cable. I had to guess the supply voltage was 5V (there's a range of possibilities) and hooking up the scope to the output of the quadrature encoder, saw a signal when I moved the read head along the magnetic strip. Except it was a *really* low signal, i.e. 75mV amplitude, not the 5V TTL signal I was expecting. I tried winding up the supply voltage to 7V, and the amplitude rose to 100mV, but I wasn't willing to try any higher voltage for fear of frying the read head.

The MSP430 expected to see a 0 to 3.3V signal. 75mV wasn't going to cut it. The head puts out a pair of complementary outputs for each of the A and B channel, which is good for common-mode noise rejection, and for the purposes of boosting the signal up to something usable, it's ideal for putting into the inputs of an op-amp, or even better, an instrumentation amplifier. Fortunately I happened to have a few at hand, and with the right gain selected, had a 0-3.3V single-ended signal.

So hooked up the A and B connections for the X axis, and it worked - the axis counted up and down as I moved the read head either direction on the magnetic strip. Only problem was, it didn't seem to increment as quickly as I thought it should. After sorting out the scaling factor, and running the read head next to a ruler, it proved to be accurate and repeatable over 300mm, but the resolution was only 0.5mm, which is kind of rubbish. The H&F page has "5µm" so 500µm is a few orders out.

Not really sure where to go from here. I kind of need the read head to be putting out 100 more pulses than it's doing at the moment...

[Poloris]

If there's no photos it didn't happen.
The cost of doing it on the cheap really add up.

There is a good article in Model Engineers Workshop 196 P36 which details the authors method of retaining the table stops when fitting a scale to the front of his mill table.
This is worth doing.

Mark

[RustyArc]

Not much to show really, this is the setup on the bench:
379802
The MSP430 Launchpad is upper left, connected to the little Bluetooth module. Breadboard with op amps in the middle, Android tablet on right and read head and magnetic strip at bottom.

This is the amplified waveform from the A and B outputs of the head as it's moved across the strip:
379801

Here the lower B signal is leading the A signal, moving the head in the opposite direction has A leading B.

I think I'm starting to realise there's a lot going on in a proper (i.e. considerably more expensive) head that puts out a processed quadrature signal. The magnetic poles on the strip are typically 1 or 2mm apart, so the head is using the change in the angle of the 2 signals to interpolate the change in position between the poles.

My current conclusion is that I either need to cough up for a proper read head, or go with cheaper scales.

[Vernonv]

I used the TouchDRO setup for a few years on both my milling machine and lathe, but now run "commercial" DRO units on both .... in hindsight I wish I went commercial from the outset.

Sure I had some fun playing around with it, but it lacks robustness.

Also when you look at the cost of commercial DRO units they can be had for the sort of money you have already spent - for example in the last couple of months I bought a 2 axis unit for my lathe for $250 delivered (head + 2 x 5um glass scales).

[RustyArc]

If everything had gone to plan, the hardware for 3 axes would have come to $300 (plus tablet), the attraction being the simplicity and compact installation of the magnetic strip and read head, and the TouchDRO app seems pretty neat, and I like the idea of extensible hardware and software. I have a Sino DRO setup on my lathe, and it certainly works, but it will never do more than the basic functions that it came with.

That said, if I can't get this to work, and it's kind of looking that way, the next option is to use cheap Shahe scales and buy a custom TouchDRO board from the guy in the US, as he doesn't offer his newer firmware that supports those scales for free download, the total for that option also comes to $300 for 3 axes, but I'm left with the bits I've already bought, although some of them I'm sure I'll find a use for...

Then again, I might come to my senses and just get a commercial unit :)

[Com_VC]

Can't you use the Sino Scales with the touchdro setup and that way you will get some of the advanced functions that the touchdro offers?

Won't there be issues using coolant with the magnetic scales?

[RustyArc]

Can't you use the Sino Scales with the touchdro setup and that way you will get some of the advanced functions that the touchdro offers?

Won't there be issues using coolant with the magnetic scales?

I'm not too fussed about fiddling with the existing DRO setup on the lathe. It was more for if/when I get a cheap Ebay mill where I'd need to kit it out with a DRO. The extra TouchDRO features also seem more oriented towards milling.

I don't see why coolant would bother magnetic scales?

[Com_VC]

I was thinking you were using the shale or igaging scales which are reported to have issues with coolant unless they are well covered. I have no idea about the H & F ones so this might not be an issue.

[RustyArc]

I was thinking you were using the shale or igaging scales which are reported to have issues with coolant unless they are well covered.

Ahh, OK, you're right, I am tossing up using the Shahe scales if I can't get this magnetic one to work, but I hadn't heard of the coolant issue - I'll have to find out what the story is.

That said, I've been discussing the issue on the TouchDRO forum, and someone has pointed out an application-specific IC that does sin-cosine interpolation - i.e. you feed it the raw signals from the head, tell it how many points you want it to generate between poles (up to 10,000) and it generates the intermediate pulses. The catch is that it's not cheap (around $18 per chip) and it's a 48 pin QFN package that's 7x7mm, so not exactly something you can throw straight on to a breadboard...