Magnetic DRO Scale question (not for squeamish!)

[graham-xrf]

Thanks for the links. Magnetic scales are far more complicated than I had envisioned. I was expecting a technology similar to that used for computer hard drives.

The error analysis has given me pause for though. Particularly the difference between high end and the low cost scales. What would concern me is that there is no practical way of knowing what the performance of a magnetic scale is prior to installation and trusting the word of a vendor who most likely has no technical knowledge of the actual function of the product seems rather foolhardy.

As pointed out in the second article, there is a difference between accuracy and precision. Just because we can read to six decimal places doesn't mean that the reading is an accurate display of fact. Unfortunately, we have a tendency to accept a digital display as fact without really confirming that for the most part. There is no value in having a DRO that can resolve to 1 micron of the displayed value can't be trusted to better than a thou or two.

Aside from a manufacturing or installation defect, I would be concerned with issuws arising from use. The possibility of ferromagnetic dust altering the pattern of the scale in an environment where such dust is commonplace is disturbing. Another disturbing thought is magnetic domains shifting. In the early days of PC's, hard drives had to be reformatted on a regular basis to avoid CRC errors and video tapes would fail after a time due to magnetic domain migration. Magnetic media is generally not considered a good long term storage solution.

Verifying a DRO scale along its entire travel would be a tedious task, especially if looking at the resolution down to a few tenths of microns. Were I to attempt it, I would probably use a set of trusted gage blocks in concert with a .0001"/div. dial test indicator.

Considering all that, I think I will stay with my capacitive scales on the lathe for a while yet. They may not have the precision but at least I am aware of their limitations

You point out what we often see on multi-meters, and some calipers. The number of digits precision of the display does not represent the fundamental accuracy of the measurement. The extra digits just change as "noise".

In the case of DRO(s), the common resolution of the grating I find is 5 microns, though available usually to 1 micron for very few dollars more. The digital display electronics, and displayed numbers is usually good to 1 micron regardless, and can suit both kinds. In metric units, if it sees a 5 micron type, the last digit is always a 0 or 5. In conversion to display inches, the minimum shift results in a number increment on the last digit that is meaningless, but of no consequence, because it represents about 0.00019".

I deliberately bought 1 micron gratings (for mill), and I use metric units, so the last digit does increment true. Even though the DRO may be truthful to 1 micron, I still can ignore the last digit anyway, because neither I, nor the machine with me driving its cutters, can actually make stuff to that precision. Two tenths may be standard fare for Stefan Gotteswinter, but I can only do that stuff in dreams!

I am thinking the glass scales are pretty much uncompromising, compared to how well magnetized regions on tape strips can be. Also, the glass ones have such economies of production numbers, they are affordable. My passing thought is "how do magnetic strips survive in the presence of magnetic chucks with leak fields orders of magnitude above what it takes to 'write to' magnetic strips"?

One aspect in the trade-offs you did not mention. It is the speed of the response. Some digital sensors deliver with a processing delay of some microseconds. It does not matter for a human using a DRO, but when it is the feedback for a CNC control loop, the instability can be devastating!

[ Edit: In this context, one has to avoid the phrase "economies of scale" ! ]

 

[ChazzC]

When I fitted my G0602 lathe with a DRO, magnetic scales weren't available and glass scales were too large for the x axis so I went with the iGaging capacitive scales (similar to those used on the ubiquitous digital calipers) and used the TouchDRO. The capacitive scales have a resolution of 10 microns or .004" which, when doubled to indicate diameter, is somewhat limiting but if I need to turn to sub thou precision, I use a micrometer anyway so it's not that big an issue. Were I to install a DRO on the lathe today, I would use 1 micron magnetic scales..

My x axis mount was very similar to Clough42's mount. I mounted the z axis scale on the back side of the lathe rather than the front as it was less obtrusive there. I mounted a DRO on the tailstock for the third axis. It has been in service for over seven years. Here are the whole gory details: https://www.hobby-machinist.com/threads/another-lathe-dro-install.34106/ (disclaimer; no real blood or gore).

10 microns = 0.0004”

Resolution of the SS scales is stated as 0.01mm (10 microns)/0.0005”; accuracy ranges from 0.001” for a 6” scale to 0.002” for a 24” scale.

As with anything, you get what you pay for. I find that my iGaging Absolute SS scales paired with TouchDRO is well matched to my mini mill: having a DRO system that is capable of working to 0.0001” on a mill that cannot deliver that degree of precision doesn’t make sense.

 

[RJSakowski]

10 microns = 0.0004”

Resolution of the SS scales is stated as 0.01mm (10 microns)/0.0005”; accuracy ranges from 0.001” for a 6” scale to 0.002” for a 24” scale.

As with anything, you get what you pay for. I find that my iGaging Absolute SS scales paired with TouchDRO is well matched to my mini mill: having a DRO system that is capable of working to 0.0001” on a mill that cannot deliver that degree of precision doesn’t make sense.

Thanks for catching the dropped digit. I will correct it.

As near as I can figure, the spec for accuracy on the iGaging and similar scales refers to full length measurements, Since most measurements requiring precision made in the hobby shop are fairly small, the .0004" resolution spec is most likely the most important. Additionally, if there is a scale factor, the Touch DRO has a calibration routine. With careful work, one should be able to accurately correct for any scaling errors. To be fair, I haven't checked out scale linearity but it would be fairly easy to do.

 

[ChazzC]

Thanks for catching the dropped digit. I will correct it.

As near as I can figure, the spec for accuracy on the iGaging and similar scales refers to full length measurements, Since most measurements requiring precision made in the hobby shop are fairly small, the .0004" resolution spec is most likely the most important. Additionally, if there is a scale factor, the Touch DRO has a calibration routine. With careful work, one should be able to accurately correct for any scaling errors. To be fair, I haven't checked out scale linearity but it would be fairly easy to do.

Da nada.

Mine match the hand wheel dials over full table travel and also a 2” indicator, so I’m happy.

 

[homebrewed]

I'm not sure why magnetic DROs have to be restricted to using composite magnets -- ferromagnetic powder in a rubber matrix. The first "tape" recorders didn't use tape at all, they used wire. A hard ferrous alloy (in the magnetic sense) would be much more uniform in composition than something that is essentially a kind of refrigerator magnet. I've actually seen an old wire recorder but didn't get the chance to see if it still worked.

Issues regarding local magnetic fields affecting the readings can be addressed using magnet shielding materials like permalloy. But perhaps using these approaches renders the approach less cost effective compared to other technologies.

Other issues like drift in the overall magnetization might be addressed by tracking variations based on the ratio of peak vs current mag-field information instead of measurements of drift-prone absolute magnetic-field intensity.

These alternatives seem obvious enough to suggest that some vendors are overstaing the issues regarding magnetic DROs. Maybe they really want to sell something else.

 

[Dabbler]

I'm not sure why magnetic DROs have to be restricted to using composite magnets -- ferromagnetic powder in a rubber matrix. The first "tape" recorders didn't use tape at all, they used wire

Long ago and far away, I was working with a guy fixing a wire recorder. Learned a teensy bit. The problem with continuous materials is twofold: The first is that it is nearly impossible to get crisp magnetic domains with a uniform material. This is why Mylar tape with particulates superseded wire recording as soon as they made tape feasible. Wire recorders had to run 2-3X faster than a Mylar tape machine to get similar fidelity.

The second reason is that the magnetic domains in a continuous material are much fainter: the particulates act like thousands of very short magnets, and the domain in continuous ferrite acts like a single dipole. So rubberized tape is vastly superior to a ferrous strip.

 

[JimDawson]

I have been using magnetic scales on machines since the 90's, I have installed them on routers, industrial saws, milling machines, and lathes. I normally buy the 1 micron resolution read heads. As far as I know the magnetic tape ''line'' spacing is 2 mm in all cases, the resolution is controlled by the read head electronics. The read heads come in 2 flavors, a sine/cos output and a pulse output. For our use, the pulse output is more useful. I'm pretty the pulse output read heads use the raw sine/cos data and convert it with electronic magic to the pulse output.

Thus far I have not seen a failure in the magnetic system. Accuracy has been consistently good for years. My mag tapes are continuously soaked in way oil and coolant. The only thing that affected them is acetone, when I spilled some on my mill table. The tape backing came loose and I had to replace the strip, about a $30 and 10 minute fix. The magnetic tape can be cut to length as needed.

I have used both Renishaw and Ditron units. Overall I highly recommend magnetic scales for all applications. I don't buy the ''kits'' I buy only the read heads, 10 mm wide tape, and if needed, the DRO readout as separate parts. There is always a flat surface on a machine to stick the tape onto, and if not, you can always build a flat surface.

 

[graham-xrf]

I have been using magnetic scales on machines since the 90's, I have installed them on routers, industrial saws, milling machines, and lathes. I normally buy the 1 micron resolution read heads. As far as I know the magnetic tape ''line'' spacing is 2 mm in all cases, the resolution is controlled by the read head electronics. The read heads come in 2 flavors, a sine/cos output and a pulse output. For our use, the pulse output is more useful. I'm pretty the pulse output read heads use the raw sine/cos data and convert it with electronic magic to the pulse output.

Thus far I have not seen a failure in the magnetic system. Accuracy has been consistently good for years. My mag tapes are continuously soaked in way oil and coolant. The only thing that affected them is acetone, when I spilled some on my mill table. The tape backing came loose and I had to replace the strip, about a $30 and 10 minute fix. The magnetic tape can be cut to length as needed.

I have used both Renishaw and Ditron units. Overall I highly recommend magnetic scales for all applications. I don't buy the ''kits'' I buy only the read heads, 10 mm wide tape, and if needed, the DRO readout as separate parts. There is always a flat surface on a machine to stick the tape onto, and if not, you can always build a flat surface.

This is great stuff!
Who knew that sticking a strip of tape to a machine was all it took to provide a measurement reference? The hope is the tape is stable enough that the accuracy is maintained over at least the length of the piece you are using.

One can find loads of videos of folk getting up to fabricating all kinds of bracketry, chip guards, etc. on how to mount DROs. Some describe modifications to the ends to avoid limiting the travel (e.g. in the Y-Axis. I have seen one video on what it takes to cut and re-mount a glass grating.
I am thinking that, especially for lathe use, a stick-on strip may be way more convenient.

I don't know of any common brands, or sources for this kit, as separate parts. Is it seen on eBay, or Amazon?

Also, it begs the question as to whether the commonly available displays, mostly offered with glass gratings, can be compatible with the magnetic pulse detection head. They seem to be compatible across brands, using a sort of de facto standard of D connectors.

Regarding the kinds of outputs. The pulse output derived from detecting thresholds, as I mentioned in #11, is perfectly fine for humans, but the several microseconds phase delay in delivering the count is a problem when it is the positional feedback in a CNC system. I understand that the sin/cos outputs may be more suited to CNC control systems, but I have not yet learned enough about magnetic strip sensors.

 

[Dabbler]

My Sony DRO was purchased in 1981. If you set up a long string of AA grade gauge blocks on the mill, and measure them with the DRO, it is still accurate to 2 tenths, after all these years. The tape used in those days was a little thicker, but essentially the same. Mag particles in a rubber-like matrix.

This mill has made all kinds of chips over the years, seen oil and soluable oil. Still working fine. (the electronics needs a capacitpr or resistor changed every 8 years or so, but that is easy)

 

[JimDawson]

This is great stuff!
Who knew that sticking a strip of tape to a machine was all it took to provide a measurement reference? The hope is the tape is stable enough that the accuracy is maintained over at least the length of the piece you are using.

That's pretty much all it takes. The overall accuracy is rated at +/- 20 um/M Long term accuracy seems to be OK, I'm still making good parts after all these years.

One can find loads of videos of folk getting up to fabricating all kinds of bracketry, chip guards, etc. on how to mount DROs. Some describe modifications to the ends to avoid limiting the travel (e.g. in the Y-Axis. I have seen one video on what it takes to cut and re-mount a glass grating.
I am thinking that, especially for lathe use, a stick-on strip may be way more convenient.

Yes, way more convenient and faster to install.

Here is the X axis scale on my mill, just hanging out there exposed to the world. Just a simple angle bracket. Been on here just like this since 2013.


and the Z axis for my lathe DRO

I don't know of any common brands, or sources for this kit, as separate parts. Is it seen on eBay, or Amazon?

Renishaw or Ditron. There may be others. I prefer the Renishaw LM-10 units, or the Ditron equivalent. I have seen a few used Renishaw units for sale on eBay, they used to be about $80, but that batch seems to be gone. I have also seen complete Ditron DRO kits on both eBay and Amazon, but no individual parts. None of those kits seem to use the 10mm wide tape.

Here is the data sheet for the Renishaw LM-10 https://www.rls.si/fileuploader/dow...ncremental-linear-magnetic-encoder-system.pdf

Renishaw sells through industrial controls distributors world wide.

You can buy factory direct from Ditron. sales@dcoee.com The tape (MS-200) is about $25/M, the DMR200 read heads were $90 each, and the 2 axis DRO I bought for my lathe, D100-2V, was $70. These are prices from about 3 years ago.

I have done a side by side comparison of the Ditron and Renishaw units. 2.5M rapid, back and forth for one hour, disagreement at the end of the test was 0.0001''. Close enough for me.

Also, it begs the question as to whether the commonly available displays, mostly offered with glass gratings, can be compatible with the magnetic pulse detection head. They seem to be compatible across brands, using a sort of de facto standard of D connectors.

I think for the most part 4 channel differential quadrature (RS422) is pretty much the standard today.

Regarding the kinds of outputs. The pulse output derived from detecting thresholds, as I mentioned in #11, is perfectly fine for humans, but the several microseconds phase delay in delivering the count is a problem when it is the positional feedback in a CNC system. I understand that the sin/cos outputs may be more suited to CNC control systems, but I have not yet learned enough about magnetic strip sensors.

I don't think there is much delay. The 1um scales are rated at 4M/sec, much faster than most CNC machines. I have not noted any problems and most of my installations are connected to CNC equipment, only one is connected to a DRO readout. The controllers I use are rated at 22MHz encoder input, and I'm sure that most modern CNC controllers are in that range.