Mitutoyo 982-537-1 DRO Display with Series 572 Linear Scales

[jpanhalt]

See related threads here:
https://www.eevblog.com/forum/microcontrollers/stymied-why-are-my-data-sometimes-doubled/
https://www.eevblog.com/forum/mechanical-engineering/my-new-dro/

Mine is an older unit from the early 1990's.  There are 2, DB-9 connectors. The pinout I determined is: Pin 1, Mitutoyo SPC (Digimatic) data; Pin 2,  REQ line (data request, active low); Pin 3, clock; Pin 4, DCt voltage of about 1.6V; Pin 5, circuit common (previously identified as GND); Pin 6, 5 VDC; and Pin 7 & 8 , chassis GND.  There are only 6 pins on the connector.  REQ  comes from the display; data and clock come from the scales (open collector).  Data format is shown in Attachment 1.  In brief,  thirteen 4-bit bytes are transmitted, MSB (first) to LSB.  Each byte is BCD.  Bit order in the bytes is reversed.  That is, 0001 = d.8.

Attachment 2 shows a typical trace.  Yellow = data, Blue = REQ, and Magenta = clock.  I marked off the 4-bit bytes.  The decimal value on the display was -0.0200 inches.  The earliest peak (0001) = d.8, is the negative flag.  The positional data read (decimal), 0,5,1, and 2.  The "2" is the decimal.  So, the whole value is 0.51 decimal.  The display read 0.0200", and of course, the linear scales are only in mm.  0.51 mm = 0.02008". 

Hope that helps someone.

Regards, John

EDIT: See update here for opening case and a picture: https://www.eevblog.com/forum/mechanical-engineering/my-new-dro/

EDIT2 (12/10/2023):  Something didn't seem quite right.  I opened the case again and probed around.  Circuit Common and chassis ground appear to be isolated.  Mitutoyo supplies a very brief design recommendation with RC  LPF filters (22k & 220 pF).  The Large display uses copious LC LPF's (10 uH & 330 pF).  Tentatively, it appears the linear sensors are powered by about 1.5VDC from the display.  No 5VDC to them; although, it is brought out to the DB-9 socket.  The 5VDC regulator is for the display and pull-ups on the sensors open collector outputs.   The same series of devices for the quill is powered by a single lithium coin cell.  The longer sensors for the mill XY axes do not have batteries that I know of. Will update as I find out more. 

[jpanhalt]

Here's an update.  Sorry to take so long.  The biggest hurdle was convincing myself of the correct pinout for the DB-9 connectors on the long linear scales.  Also, unlike the spindle scale here: https://www.eevblog.com/forum/microcontrollers/stymied-why-are-my-data-sometimes-doubled/  , the scales do not have their own power.   I finally built a breadboard rugged and mobile enough to test in the shop.  It works.  Results are in mm and the code from the spindle device will need that modification.  I used 1.5V to power the scales.  I could not find anything from Mitutoyo as to how high it can go.

Attached are photos of the breadboard and the schematic.  Space was limited, and I resorted to a couple of 1206 SMD's  (seen on the bottom view).

The buffer is a non-inverting, TTL, Schmitt trigger.  Mitutoyo warns not to use CMOS, but the DRO that came with those scales has a CMOS inverter.  The buffer is probably not needed, but it does clean up the raw oscilloscope views of the clock and ser_in data.

EDIT: As for the readout, the current LCD is a bit slow, so I added a header for SPI parallel to the 3-pin display header.

[jpanhalt]

UPDATE
I pasted together the program parts and can now go from a Mitutoyo SPD encoder in mm to inches.  Used the 24-bit BIN2BCD as for smaller numbers, it was slightly faster than the band aid 20-bit version.

The display is quite stable and two values are shown in the attachments.  Range is at least a meter.  In simulation, the time for conversion from mm BCD to ready to print inches BCD using an input of 959.59 mm (37.7791") took 382 Tcy.  Based on my experience, that would be on the high side, since most values during use will be a few inches.  For today's post, I used the spindle scale, so my range is quite limited.  Since sampling rate is limited by the encoder to about once/40 ms, that delay will be trivial.

Now, I need to clean up a few aspects in the code and start using an SPI or I2C display.  I will also add leading zero suppression and will post the final code when that is done. 

On Wednesday, I came across a "programmable" display ($15 on Aliexpress) and bought it (3rd attachment). It has a built-in processor and uses RS232 for communication.  I have no idea yet how fast it will go.  If it doesn't work, at least it is a nice box.

John

[jpanhalt]

Here's the code in MPLab ASM I used yesterday.  Any "one-wire" RS232 display that accepts ASCII should work,  Baud is 19,200.  You may need to change the codes used to control and clear screen.  The code simply displays the reading.*  One could add a switch to go between mm and inch.  It does not include zeroing.  I suspect, that is done by toggling power to the scale.  For the Longer linear scales, that would correspond to the 1.5V supply shown in the schematic I posted earlier.  My supposition is based on the spindle control with which zeroing is done by pressing the "on" button.  I doubt Mitutoyo will answer that question, but I will try to ask it. (Done)  Otherwise, I will use my scope with the control unit on my mill.  It's too cold here now to do that.

The code can be shortened significantly by using indirect addressing (FSRx), but execution time is increased.  I have that code available, but i doubt it is of interest.  It's tradeoff.

*Eventually my plan is to add controls, such as presets and stop or engage signals.

EDIT:  To my surprise I got a response from Mitutoyo this morning:

The 98-537A (or -1) toggles between the 1.5V output.

  So, it appears my suspicion was correct.

[jpanhalt]

This beautiful box just came:

It's better than I expected, but not quite usable as is in any project I am currently doing.  It is a nice box with 20 membrane switches that have nice tactile feedback.  It opens easily and the front panel is connected to the main PCB with 2 headers: 20 pin for the 192x64 GLCD with what appears to be standard serial communication with 2 controllers, each 96x64 pixel and a 9-pin for the multiplexed 20 switches.

The main board has the following:
1) NXP P89V51RD2FA MCU.  That's 8-bit, 80C51 type, 64K flash, 1024 byte RAM.  The crystal is 11.0592 MHz (128x24x3600).
2) SIT232ESE -- 120 Kbps 232 communication
3) 2x BL3085 -- RS485 communication
4) XE1509 -- 150 kHz buck converter, 5V @2A

Someone who is familiar with the 8051 chip might be able to use it as is.  For me, I will probably make my own main PCB with a PIC MCU.   The display was a nice surprise.  I was afraid it would be character only.  The price including shipping is not bad considering the box is ready to use.  I didn't shop around, and it might even be available cheaper.