BOSS 8 modifications

[jonny]

I've done a few mods to my mill. After once running the wrong program (I didn't watch closely enough and neglected to delete one program before loading another), I altered the BP firmware to only hold one part program. Any new program you load automatically overwrites the existing program. I also added a feature that allows me to toggle between 0.0005 increments with the hand wheel and 0.0001 increments. That is particularily usefull when hand feeding with micro endmills (the 0.0005 steps tend to break the small endmills) and when tramming a part. I redefined the illuminated MDI button, which I never use anyway, to serve this purpose.

The BP OS does a crc check when it boots (if you've had a rom failure, then you would have seen an error message saying something about crc) and I alter the crc word so that the check is still successful. If my new roms become corrupted, the mill will not boot.

Yup, I have too much time on my hands. And it's kinda fun digging around in someone else's programming.

[machintek]

COOL!!

Retired or what do you do for a living (or used to do)?

George

[Jonp]

jonny, thats really great, I like to see improvements to these good ol mills. I'm just curious what the BOSS 8 uses for a processor and is all the code in assembler? keep up the work, Jon P.

[jonny]

The BOSS 8 has three processors.

Zilog 8002 on the EMC, which takes commands from the keypad, and stores and interprets the part programs. BTW, one way to make better use of the limited RAM is to strip all the blanks and comments (except the comment after M6Tx, tool change). Ever wonder what that push button is on the EMC? It opens up what BP call ODT (Onboard Debug and Test?), a window into the OS. I'm not sure we're supposed to know about this, but there was the switch in plain sight.

Intel 8748 on the EAF, which passes data through, controls flood and mist, collects error bits into error words.

Motorola 68000 on the FMDC, which gathers and stores the event log available with the FIST monitor, and controls the axis motors.

I've done some hardware mods, too. The mechanical handwheel "pot" failed, giving eratic motion, so I replaced with an optical unit. Apparently BP equipped a few machines with an optical quadrature switch.

My LCD failed, the center rows, which contain the - sign, were no longer visible. I contacted Epson, the maker of the LCD, and they had no record of ever making that particular LCD. But they gave me several current LCDs, and I made a little daughter board to interface between the card behind the keypad and that new LCD. The EMC sends data to be displayed to a small 2k RAM on that card, and then the old LCD milks that data from the RAM and displays it. My daughterboard performs that milking operation. Only a tiny portion of the RAM is used for this, and to find out where the display data was stored I replaced the RAM with one of those battery backed non-volatile RAM, ran the mill, and then put the NVRAM into my EPROM reader and searched for the last displayed message. I was not willing to pay the $600 that BP wanted for the original LCD. BP no longer supports BOSS 8, so that LCD is even more expensive. I see someone on ebay wanting ~$900.

I also found a replacement for those power-hungry PALs that do the quadrature decoding for the handwheel and axis motors.

Wrote my own 2-d CAM based on Generic CADD. It's kinda crude in that it requires lots of cutting and pasting. I have to hand write all the Z motion. And if I want arcs in the XZ or YZ planes, I have to write those portions by hand. But it's specific for the BOSS 8, eg, it knows to put in G99s according to the rules of the BOSS8, and other features.

I did a little lathe work in my mill by putting the workpiece in the spindle and clamping a cutting tool in the vise.

[jonny]

My guess is that all the code was written in assembly language, though I'm not a programmer and I don't know how to tell. Would a compiler have existed for a Zilog 8002 back then?

[Gail_in_NM]

Jonny,

I have a V2E3 BOSS 8 that I bought new shortly after they first came out. Still running fine with a few periods of down time for required maintance. Many thousands of hours of hours on it.

A few problems over the years.
Like you, I had to replace the handwheel encoder.
Couple of brush changes on the Powertron motors, but thats normal maintance.
One H-Bridge transistor set replacement.
All three of the BEI encoders failed by the incadencent bulb aging until it had too much black in the bulb to give sufficient light. I replaced each bulb with a pair of IR LED's and a 47 ohm resistor all in series and removing the lens. It takes 2 LED's to provide coverage for both the posiition and index pickups. One of the clues that the bulb is going bad is when position is lost during rapids causing emergency shutdown, but everything is normal at lower feedrates.
Most difficult to find problem was an intermittent fuse. Located in the servo amplifier enclosure it would put noise on one of the coumputer supply lines after about 4 hours of operation. Open the door and it would cool down and everything would work again. Close the door and it would heat up again and fail.
EMC board had a cold solder joint that showed up after about 10 years of operation.

Not too many modifications. I added a pair of remote switches near the spindle for spindle start and program start. Just paralled a some reed relays across the control panel switches. On the spindle start I close both the spindle enable and direction for high range. Seldom use low range, but start it from the control panel when I do need it.

For programing, I wrote a set of LISP routines that run inside AUTOCAD that generate G-code and write it to a file. It's all interactive in ACAD and I have run it on ACAD 12 through 2000. My son has run it on newer versions and will lprobably check it out on ACAD 2006 which he has just started using. He uses it,with modiifications, for programs for 4 machining centers and a large laser cutter. File transfer to the BOSS 8 is a Pascal program that strips out blanks and comments and provides transmission error checking.

I am interested in some of your firmware revisions. Although not a programmer, I used to write a lot of 8080/Z80 assembly code and still do some programing with 8051 and PIC's.

Gail

[jonny]

Hi Gail, my EMC firmware is from my friend's V2E3. I changed the max X travel from his value (22.0000 inches?) to my 18.0000 inches. My Z-axis would go dead after turning on and off the spindle, and his mill didn't have that "feature" so I asked if I could copy his EPROMS. It served him, too, because now he has backup copies of his firmware.

What mods are you interested in? I could send you snippets of my firmware, including the hex code and location, so you could find the code in your firmware. The LOAD mod was simple; change a load (or move or whatever) from a variable source to a fixed source. The source value was the address of the next available RAM address, so I simply said that the next available RAM address was the beginning of the RAM portion that is used for part program storage. I wrote a BASIC program to generate the new CRC word.

Yet another mod was to install a switch on the front panal so that I could put a pause in a part program with M51. My friend did that to his V2E3 also.

Your fix for the quadrature encoders on the axis motors is a good tip for all to remember.

Jon

[arsenix]

Jonny did you use a decompiler to look at the rom dumps? I poked around with decompiling mine but I the few decompilers I tried didn't help much. If you've got a recommendation for one I'd love to hear it.

I was trying to find where the M51 command is processed in the code, but so far I havn't been able to find it.

EDIT: I forgot you are dealing with the EMC controller, which I know nothing about... i have a boss 9 which has another 68000 running it's processing board (the bsp).

james

[Lasershop]

All very impressive. Way over my head. My handwheel pot failed also on my V2XT. Where would I get the repalcement optical unit?
RonO

[jonny]

James, I found in the EMC code where the M51 command turns on the WAIT LED and sets two other bits, but that does you no good. BUT . . . after the M51 is executed the machine stops and waits for K9.1 (ground) and K9.6 (index out) to be shorted together, allowing part program execution to continue. Both these are on EAF, which I think you have. Consult your manuals. K9.1 may be called K9A.1, and K9.6 may be called K9B.2.

As for disassembling the dumps, a friend with an Amiga did the 68000 for me. I think the Amiga has something like debug. I wrote a disassembler for the 8002 based on an assembler I bought from a Canadian company. The 8748 is so short that I disassembled it by hand.

[jonny]

RonO, I first used an HP/Agilent Tech HEDS-5700 series, but the shaft bushings wore out quickly. And 50 cycles per rev was not available. Then I bought a http://www.usdigital.com/ p/n S1-50-F2, but the drag is so light that the weight of handle on the knob causes the knob to rotate. I removed the handle and put a shallow depression in the knob for my finger and now it's ok. Just tried to access their site and got the "The page cannot be displayed" message so I don't know if they still exist. There must be other optical shaft encoders out there. BTW, you will need for find a power source for the encoder. Both the encoders I used need five volts. Jon
The last phone number I have for usdigital is 360.260.2468 , 800.736.0194 sales in Washington state.

I misspoke yesterday. The device that wore out quickly was HRPG-A D32 #16R, a HP part. The HEDS5700 series is far more robust mechanically. The problem with the HEDS5700 is that the slowest resolution is 96 cycles per rev. The mechanical switch had 50, so the table would move nearly twice as fast. Maybe no problem. But, if you were to play with the LS7083 (also from USDIGITAL, which is online now, don't know where they were yesterday) you can replace the power hungry PAL and wire the LS7083 to yield one output pulse per cycle and then use the 256 cycle/rev version of the HEDS5700 and get very close to the original overall resolution. (50)cycles/rev X (4)output pulses/cycle X (0.0005) inch/output pulse = (0.1000) inch/rev.