Selecting Scales for a DRO

Friday, December 6, 2013

When building a DRO the selection of scales is the most critical decision. Touch DRO Android application can work with a number of different scale types, ranging from inexpensive calipers to 1 micron glass scales. In addition to the cost considerations there are technical parameters that you should take into account. In the previous post I explained which parameters really matter in a DRO setup. Now let’s take a look at some of the commonly available scales and see how they stack up.

 

The specifications for each scale are based on the information provided by the manufacturers/distributors. Unfortunately it won’t be a completely apples-to-apples comparison since different manufacturers list different parameters, though.

Capacitive Scales

Calipers, “Standard” Chinese scales, iGaging DigiMag, and AccuRemote scales fall under the category of “capacitive” scales. Even though they come in different forms and sizes, they all offer roughly the same resolution, repeatability and runout but differ on other important parameters.

Inexpensive Digital Calipers

Chinese digital calipers are the least expensive option, selling for as little as $10 on eBay. They use the same 4-pin connector as the “standard” Chinese scales, which requires a proprietary (and hard to find) cable. They are available in lengths from 4” to 12” and can be easily cut down to a slammer size if desired. These calipers have one major flaw that makes them ill suited for a DRO application: after a period of inactivity they shut off, resetting the position. Their practical resolution is around 0.0005” and most retailer claim accuracy of 0.001”. In practice the latter is a bit optimistic, as on some Harbor Freight units I was able to measure 0.003” of runout over 6” span.

Inexpensive Chinese calipers come in different shaped and sizes,
but most are ill suited for a reliable DRO
Specifications
Resolution
0.0005”
Accuracy
+/-0.001”
Counts per inch
2000
Refresh rate
~3Hz
Available lengths
4”, 6”, 8”, and 12”
Material
Stainless steel

Pros:low cost; broad availability

Cons:lose their position on power-down; slow refresh rate; data cable expensive and difficult to find

Shahe Remote DRO/iGaging DigiMag scales

Shahe/iGaging* DigiMag** “Remote DRO” scales are marginally more expensive than the calipers and work reasonably well. The scales come in a number of sizes between 6”and 35” with the price ranging from $20 to $80 (US Dollars) . The main disadvantage is that the frame is made of aluminum and the reading head is plastic, so over time they will wear out with heavy use.

Specifications
Resolution
0.0004”
Accuracy
+/-0.002” per 6” of travel
Repeatability
0.001”
Counts per inch
2560
Refresh rate
up to 50 Hz
Available lengths
6”, 12”, 24”, and 36”
Material
Aluminum frame, plastic head

Pros: inexpensive, easy to read, good refresh rate

Cons: inferior materials, low repeatability, difficult to find outside of the USA

*iGaging and AccuRemote appear to be brand names used in the US and Canada.The scales are made by a Chinese company called “Shahe” and are sold under that name in the rest of the world.

** Many retailers have the word “magnetic” in the product name or descriptions. This is a bit misleading, since the scales are not magnetic; the DRO display can be magnetically attached to the machine. These are capacitive scales and should not be confused with magnetic quadrature scales.

AccuRemote Scales

AccuRemote scales use the same electronics as the iGaging DigiMag version and appear to be made by Shahe as well. The scales are offered in similar lengths, but the frame and the reading head are made of ground stainless steel. For a more in-depth comarison The AccuRemote scales are a bit more repeatable, have smaller runout and will definitely wear much better. The tradeoff is the cost, which is about two times higher for the same lengths.

AccuRemote and iGaging DigiMag scales are made by the same company;
other than the frame/head material, the electronics are identical
Specifications
Resolution
0.0004”
Accuracy
+/-0.001” 0-6”, 0.0015” 0-12”, 0.002” 0-24”, 0.003 0-38”
Repeatability
0.001”
Counts per inch
2560
Refresh rate
up to 50 Hz
Available lengths
6”, 12”, 24”, and 38”
Material
Stainless Steel

Pros: easy to read, good refresh rate, good repeatability

Const: scarce availability outside of the USA

Standard Chinese Linear Scales

“Standard” Chinese scales (also called 24 * 2 and “Sylvac” scales) have been the mainstay of DIY DROs for years and are commonly available in most parts of the world. The price (excluding the data cable) is comparable to the AccuRemote scales. Their main downside is the proprietary data cable that is next to impossible to come by. Additionally, most of these scales use 1.5V power supply with the positive side connected to the frame, whereas most other scales and encoders have the frames grounded. This makes mixing them with other types of scales a bit problematic. Parameter-wise they are almost identical to the AccuRemote scales but as far as I know don’t come in lengths larger than 24”.

Specifications
Resolution
0.0005”
Accuracy
+/-0.0008” 0-4”, 0.0012” 0-6”, 0.0012” 0-8”, 0.0016 0-12”, 0.002 0-20”
Counts per inch
20,480
Refresh rate
up to 3 Hz (50Hz in “fast mode” if available)
Available lengths
4”, 6”, 8”, 12” and 20”
Material
Stainless steel

Pros: resolution and repeatability, good availability across the world

Cons: susceptible to noise, difficult to mix with other scales, requires hard-to-get data cable

Quadrature Encoders

This category encompasses a wide range of different linear and rotary encoders.

Glass Scales

Glass scales are used in the majority of commercial DRO units. Most common glass scales have resolutions of between 5 micron and 1 micron (0.0002” - 0.00004”) with superior repeatability and reliability. They offer good protection from dirt/coolant and are much less sensitive to electric noise in the garage. The main downside is the cost that can be up to five times more than the cost of capacitive scales of the same length.

Glass scales are encased in sturdy metal cases, use shielded cables
and offer superior resolution and accuracy
Specifications
Resolution
1μm-5μm (~0.00004”-0.0002”)
Repeatability
varies between manufacturers (excellent)
Hysteresis
varies between manufacturers (excellent)
Counts per inch
5000-20000
Refresh rate
real-time
Available lengths
100mm-1200mm in 100mm increments (~4” to 47”, 4” increments)
Material
glass scale, metal casing

Pros: very accurate, real-time refresh rate, good protection and reliability

Cons: more expensive than capacitive scales, can’t be cut without special tools

Magnetic Scales

Magnetic scales are less common than glass scales but are gaining some momentum recently. They are very similar when it comes to resolutions and repeatability. The main advantage of magnetic scales is the fact that they are virtually dirt-proof. Moreover, they appear to have smaller cross-section, making them easier to use when space is at a premium (for example on a lathe cross slide).

Specifications
Resolution
1μm-5μm/0.00004”-0.0002”
Accuracy
10 μm/0.0004” over 1m of travel (for 5μm version)
Repeatability
10 μm/0.0004” (for 5μm version)
Counts per inch
5000-20000
Refresh rate
real-time
Available lengths
10”-20’ (yes, 20 feet); can be cut to size
Material
metal

Pros:very accurate, real-time refresh rate, dirt-proof, smaller cross-section, can be cut to size

Cont:cost, lower availability, more sensitive to misalignment

Rotary Encoders

Rotary encoders are sometimes used in situations where a linear scale might be impractical. It can be either a rack-and-pinion setup, a shaft encoder on the lead screw or a similar application. The cost of rotary encoders varies widely, with low-end units retailing for $10 to hundreds of dollars for “industrial” grade encoders. Major drawback of rotary encoders is the fact that they don’t address the backlash, though.

Pros:can offer digital readout where linear scales are impractical, relatively inexpensive

Cons:don’t eliminate backlash

Conclusion

The choice between the scales will likely come down to cost. Glass scales are more expensive than the capacitive ones but offer excellent accuracy, repeatability and reliability. These scales are pretty much immune to garage noise and are very well protected from the elements. Capacitive scales are much cheaper but require some compromises.

High-end solution

Glass scales are the best choice for a higher-end setup on a good quality, rigid machine, as they will let you hold much closer tolerances, etc. These scales are very reliable by design, extremely precise, accurate, and very fast. With some careful shopping [on eBay] a set of 3 5μm scales for a mid-size machine will cost rougly $350-$450.

I don't have first-hand experience with magnetic scales so I can't make a meaningful recomendation, but on paper they are comparable to glass scales in most respects but have some advantages. Please feel free to comment if you have any experience with magnetic scales.

Middle-of-the road setup

Outfitting a mill or a lathe with glass scales is a costly proposition. Moreover, for a mini machine (Siex X2 mill or the Mini Lathe) glass scales will be a ridiculous overkill and a waste of money. These machines have so much flex that the extra resolution and accuracy offered by the glass scales will be completely negated. A much more practical setup will be based on AccuRemote scales. With realistic resolution of 0.001" they offer a good compromise between accurace and cost, and are more than adequate for most of the work a hobbyist will come across. Such setup for a mid-size mill will cost roughly $200, and if treated with care the scales will reliably server you for a long time.

In AccuRemote scales are not available in your country, the standard Chinese linear scales offer similar performance and price point but are much more finicky to work with and will be more succeptible to electric noise.

Budget Setup

If you're on a strict budget, I recommend the iGaging DigiMag scales. They are almost as accurate as the AccuRemote version but cost roughly half. The main concern is the inferrior construction and materials. In practice, though, if they are properly mounted to your machine, the play between the plactic head and the frame won't pose any issues. Such setup will cost around $100, based on today's Amazon prices.

I would avoid the “Harbor Freight” calipers as they are pretty slow (3 refreshes per second) and reset when the display times out.

Finally for situations where a linear scales is impractical, don’t discount a rotary shaft encoder. Even though it will not account for the leadscrew backlash, you will still get some of the conveniences of a DRO. For instance, an inexpensive rotary encoder with 1000 pulses-per-revolution, installed on a lathe's compound slide will offer resolution of 10,000 - 16,000 counts-per-inch. Even though you will still need to be mindful of the backalsh, the DRO will be able to do the trig calculations on the fly.

44 comments :

  1. Do you have any information or links to information about the power/output connections for the different types of scales. Pictures would be helpful or just a schematic type line drawing.

    Thanks

    ReplyDelete
  2. Hi Yuri

    Would it be possible to connect optical encoders salvaged from printers utilising the myler strip? Rigging a swarf proof enclosure would not be that hard and a plentiful supply could be had. By the way this is a great project.

    Thanks

    ReplyDelete
    Replies
    1. Swarfing,
      This is an interesting idea, actually. I suspect the mylar strip works similar to the glass strip in the glass scales. If that is the case, you should be able to use the voltage shifter circuit but replace the fixed resistors with 10K trim pots, so you can adjust the "virtual ground" to whatever the encoder's 1/2 point is for each line.
      I have a broken printer in the garage and will try to play with it to see if that is the case.

      Thank you
      Yuriy

      Delete
  3. Littlemachineshop dot com "had the cables for the chinese scales-calipers" if my memory serves me, catalog page 48. (in DRO pages). quite a while since I looked there thou.

    I am a beginning coder for the arduino. A program to read the chinese scales using a backlit LCD??
    My shop evidently is much nastier than anyones who can use a touch screen anything. My 1951 Leblond lathe throws oil like a antique harley.

    ReplyDelete
    Replies
    1. David,
      Yes, LMS has the cables, but they are $18.95 each, which pretty much defeats any cost savings of using the "cheap" scales. One foot iGaging scales sell for $15 more on Amazon...

      A program to read the Chinese scales for Arduino is a no-go regardless of the output, unfortunately. Although the ATMega chip should, in theory, be able to handle three or four scales, Arduino "layer" doesn't provide enough pins with external interrupts. Going to Arduino Mega or skipping the Arduino abstraction layer and coding directly for ATMega defeats the point. If you're looking at an LCD-based design, I'd start with an ARM dev kit, such as LPC Expresso, MBED, TI's ARM TIva Launchpad, etc. That way you will be dealing with a 32-bit MCU that can run at 80+ MHz, have 5V tolerant pins, bunch of timers and real-time debugging. All that will cost between 1/4 to 1/2 of what a name-bran Arduino Mega will cost.

      Hope this makes sense
      Thank you
      Yuriy

      Delete
  4. Yuriy

    I'm glad you said what you did as it now has convinced me to have a go. I'm a bit frustrated though as i trashed a couple of A0 plotters that had some nice long mylar strips a while ago. Will have to see if there is a source for long axis to my mill and lathe. Most HP printers i have looked at have duel channel encoders which i would assume are quadrature? Excuse the ignorance. Paul

    ReplyDelete
  5. I would like to recommend glass scales purchased from Ditron. They are the manufacturer and scales can be purchased directly from them. You can choose from two resolutions and they are made to custom length. There is a Yahoo group, Echo is the contact/sales person.
    I have no interest in the company, just a satisfied customer. (have 3 on my B'port mill)

    ReplyDelete
  6. I am interested in the Ditron scales. I asked for some specs and got a bewildering array of information.
    Elkodude & Yuri,

    Echo was not familiar with the term "Quadrature" and I don't understand what it means either. I just want a standard scale to use with Yuri's DRO. I suppose it would be nice to have them plug into the box with whatever plug comes on the included cable.

    What is RS422 output signal , 9 pin TTL , 9 pin EIA-422-A, 7 Pin TTL?

    Thanks

    John

    ReplyDelete
    Replies
    1. John,
      All glass scales output one of the two signals: quadrature or sine/cosine wave. The former is much more prevalent among the "Chinese" glass scales. I haven't seen Ditron scales that don't output quadrature signal.
      With that said, take a look at this post: http://www.yuriystoys.com/2014/01/connecting-glass-dro-scales-to.html.

      Regards
      Yuriy

      Delete
  7. Very useful information, thank you!
    What about the new iGaging Absolute Digital Readout DRO?
    (http://www.amazon.com/dp/B00LOZF4LA/ref=wl_it_dp_o_pC_S_ttl?_encoding=UTF8&colid=1KUEA54W79EW6&coliid=I1D91KJHK93FOY)
    Will they work with TouchDRO? Or is it better to use AccuRemote?

    ReplyDelete
    Replies
    1. Victor, I haven't had a chance to test them yet. I will try to get a set some time soon and see what they output, but at this point I would assume that they are incompatible...
      Thank you
      Yuriy

      Delete
    2. Any update on the Absolute scales? The Accuremote scales appear to be no longer available (well, I may have found one vendor that mis-markets them, so they still have stock)

      Delete
    3. Also interested in knowing if the Absolute scales are compatible with TouchDRO.

      Delete
    4. A lot of people will surely be interested. I bought a set, and they are not directly compatible. I am working on this issue now & will post details, photos etc. in another question.

      -- Craig

      Delete
    5. Craig,
      If you can work out the protocol I can add the support to the firmware.

      Thank you
      Yuriy

      Delete
    6. Has anyone made progress on the absolute scales. We have one we may want to use with this project in the near future.

      Delete
  8. Cool site! I am thinking of getting a mini-mill, and building your project looks like an excellent excuse to get one.

    I have a couple of the HF digital calipers, and I have noticed that the display never times out, and even if you turn off and on, the readout does NOT reset. In fact, I measured the current draw of the battery, and it does not change significantly when you turn the unit off!

    ReplyDelete
  9. Yuri, if I have correctly understood, the protocol is defined here: http://www.yuriystoys.com/2012/09/android-dro-communication-layer.html
    and the controller simply needs to send for example x 23454; y -12; z 13532; and so on. Is there any provision for reference marks from scales? Bruce

    ReplyDelete
    Replies
    1. Bruce,
      Yes, that is the protocol used by the application.
      No, there isn't any provision for the reference marks. The thinking was that the application doesn't care about the low-level details of position reading. All that "magic" happens in the controller.
      Thank you
      Yuriy

      Delete
    2. Yuriy,

      You might consider adding one further symbol (say "r") to your BNF syntax definition. So x 23454; r y -12; z 13532; would mean that the scale reference mark for y was traversed at -12 counts.

      I have another question. My Heidenhain scales are rated at up to 48 meters/min, and have a 40-micron period. Thus the maximum counting frequency is 48 * 1000 / (60*0.040) Hz = 20 kHz. In order to track the phase accurately these must be sampled at least twice as fast (Nyquist) but better would be four times as quickly, so at 80 kHz. Since there are three scales (X/Y/Z) and each scale has two outputs (0-degree and 90-degree phase) that means that in principle the controller needs to be able to acquire and analyze 480k samples/second. (Of course what is sent to the DRO app is just a tiny fraction of this.) Is it correct that neither the MSP430 Launchpad or Arduino can handle this sort of sample rate?

      My own mill has a rapids speed of 400mm/min. That's a factor of 120 slower, so a sample rate of 4000 samples/second would be enough for 3 axes, though three or four times that would make the coding easier. What's a realistic maximum sample rate for the Launchpad or Arduino? The basic calculation is counting phase rotations. In principle one would need to calculate atan2(y,x) but this can be done well enough by using a 2-d array as a lookup table.

      Bruce

      Delete
    3. Bruce,
      As I said in the reply to your other comment, reference marks are low-level implementation details that the app doesn't need to care about. The whole idea of having a controller separate from the UI is the abstraction of the scale protocol.I.e. "magic" happens and TouchDRO gets correct position, and if the controller needs to read the "R" channel to ensure accurate position reading, then it should do so, not the app.
      I don't know off the top of my head, but I'd imagine AVR and MSP430 have sampling rates in hundreds of thousands samples per second. Arduino (if you use AnalogRead will be much slower due to the overhead, obviously). The tricky part will be the processing. At 4000 samples per second you will have at most 4000 CPU cycles to process each reading (much less, actually, since you will need to spend cycles formatting and transmitting the data, etc.)
      Basically you will need to be REALLY careful with your code to waste as few cycle as possible, since some things that look innocent can swamp the little CPU... Take a look at this article for example: http://embeddedgurus.com/stack-overflow/2011/02/efficient-c-tip-13-use-the-modulus-operator-with-caution/ ...

      Regards
      Yuriy


      Delete
    4. Hi Yuriy,

      The reference marks are not for the controller, they are for the UI. But that's an issue we can sort out later.

      I've ordered some hardware to play with, and looked at the controller code. It's straightforward enough that I'm sure I can get a basic version up and running pretty quickly.

      Regarding the data rates, I made a mistake in one of my calculations. The rapids speed on my mill is 1.2meters/min, which corresponds to 12k samples/second. However according to this page http://www.microsmart.co.za/technical/2014/03/01/advanced-arduino-adc/ I can speed up the sample rate to 50k samples/second, while only losing some resolution. That's fine, even 4 or 5 bits of ADC resolution would be enough for me to determine atan2(y,x) accurately.

      I have written some real-time code before so yes, the 4k instructions per read is an issue, especially if it drops to 320 instructions per read. In the worst case I'll have to look for a cheap single-chip sin->digital quad encoder and add some off-chip counter hardware.

      Cheers, Bruce

      Delete
    5. Bruce,
      You can speed up Arduino to AVR native speed by skipping the Arduino library altogether, so that's not an issue. But if you really have 320 cycles per read you're in trouble. Don't forget about the fact that you have UART transmissions to deal with and that your reading will need to happen inside a timer interrupt (so you are reading at the same intervals). That introduces ISRs (context switching), locks (to avoid race conditions) and all sorts of other overhead.
      If you are really serious about creating a new controller, either move to Arduino Due, (even better) find a hardware solution (which probably won't be cheaper either). Of course, I would be even more excited if you could prove me wrong :) Seeing what people squeeze out of the ATMega chip is pretty fascinating.
      Regards
      Yuriy
      P.S. Feel free to email me at ycroosh at gmail.com. I'll be happy to collaborate/help...

      Delete
  10. Hey Yuri - first and foremost, thank you very much for all the blood, sweat and tears you've put into this project. The previous owner of my mill installed AccuRemote (the blue ones) DRO units on the X and the Y axis but not the Z -- so I bought an iGaging (the black ones) unit for that missing axis. I was testing the scales against the Arduino on the floor and the AccuRemote scales would not work (numbers were erratic and jumping) where as the iGaging scale was working percetly. After switching the cables around and so on, turns out that the AccuRemote scales need to have their frames grounded. After doing so, they worked perfectly. Anyways, just an FYI in case someone runs into the same issue during testing / development. Thanks again for all the effort!

    ReplyDelete
  11. Hi Yuri. I'm another user building this and have the Chinese scales. Like most with these scales I'm struggling with connections.
    Soldering wires to the connector may work for a while but eventually the pads will pull off.
    Then I thought about Mini or Micro SMT sockets. I can solder them to the connections ?
    Their ~7mm body fits perfic in the cut out as well. BUT - they are 5 pin at 1mm SMT pitch. No cookie there then....
    There are however 4 pin Minature USB connections, used on some Digital cameras. ~7mm socket body. Only issue is the 0.8mm pin pitch SMT Pin layout won't quite match the 1.5mm connection pitch.

    Any other ideas welcome......

    ReplyDelete
  12. Hello,

    I was curious - has anybody tried to modify the read head on the AccuRemote or iGaging scales to get access to a raw signal? It seems like this should have a lower latency. They must be able to read the signal faster than they update the display, or they couldn't track motion effectively. If it output something simple to decode like quadrature, that would be all the better.

    Thanks,

    Paul

    ReplyDelete
  13. how can I cut the AccuRemote stainless scales. Thin abrasive wheel ???

    ReplyDelete
  14. Has anyone tried the Renishaw LM10 magnetic scales? They are very compact and have a 1um resolution. I think that they have the same output as a normal glass or magnetic linear encoder.
    Any feedback?

    ReplyDelete
    Replies
    1. I have a similar scale on the Z axis of my lathe (with 5 micron resolution). No problems so far, but I've only had it for under a year.

      Delete
    2. Where did you purchase them from? Ebay?

      Delete
  15. This comment has been removed by the author.

    ReplyDelete
  16. Can I assume that all will be ok with any of the Renishaw magnetic scales set ups. They sell the sensors in several forms and I fancy complete diy for building the scales.
    John
    -

    ReplyDelete
    Replies
    1. Yes, they work just fine. I have similar scale (from different manufacturer) on my mill.
      Do they have a distributor in the States, do you know?
      Regards
      Yuriy

      Delete
    2. Thanks for the reply. I'm in the UK and have looked at Renishaw's site several times. Their parts can be bought direct by the look of it but I haven't actually bought, just looked at prices.

      John
      -

      Delete
  17. This comment has been removed by the author.

    ReplyDelete
    Replies
    1. From the names of the pins you mentioned it doesn't look like it. They would need to output square quadrature wave. You could try contacting them and asking what signal your scales emit.
      Thank you
      Yuriy

      Delete
    2. This comment has been removed by the author.

      Delete
  18. Ok. Got my igaging scales in today and cut the first wire to solder directly to the board...although I wasn't expecting 5 wires. The red and black are obviously the power leads, yet there's also a grey, white, and green. Thus far, haven't found which color goes into which hole (other than the red for the "+" symbol and the black for the ground symbol). So which of the three remaining wires is "D" and which is "C"? And where does the third one go? Have I missed something in the instructions?

    ReplyDelete
    Replies
    1. John,
      Red and black are very likely not Vcc and 0V. iGaging cables are all over the place. Your best bet is to open the reading head and check continuity between the test points and the wires.
      Regards
      Yuriy

      Delete
    2. Even when using a micro USB connector? That'll be a challenge. With that and getting enclosures that are too small, will have to bash a couple together to get something oil tight.

      Delete
    3. There are test points on the PCB inside the head. Those are much easier to hit than Micro-USB leads.

      Delete
    4. Ok, found them. they are much easier to use. See there's only 4 leads from the circuit card going to the micro usb connector. The one marked "VDD" goes to the red wire, the one marked SSY goes to the white, the one marked data is green, and the black goes to the ground. Hence, presume the lead to "data" is "D" on the kit, that "VDD" is the "+", the black is "0". My question is this: "is "SSY" the clock?

      Delete