Another PUMA robot: NOKIA RM-1 Sfera-36

[ulihuber]

Hi all,

I am right in reviving an old NOKIA PUMA robot. First I started to post in another PUMA thread, but making up a new one might make it easier for others to follow. I'll try to unfold my best English, which probably will not work always for someone located in Bavaria/Germany.. ;-)

Let me skip the first and essential cleaning of the arm an shelf, as it is of no real use for others.

The robot arm is a license built of the Unimate PUMA 560. The controller is a NOKIA make in terms of hardware, as far I can see now, software is also not identical to the MKII controller. It is, however, based on an English dialogue.
Manuals and descriptions are solely in Russian language, referring the English prompts and commands. Hard for someone who can't even read Cyrillic letters....

But I made some progress.
I had a hard time locating several completely different problems. They might have been the result of a planless try of the preowner to get the robot to life by interchanging all boards of the two controllers. There was even one controller without ROM board which stuck in a wrong place in the second controller.

What I have fixed so far is:

- Replaced CPU board by one that seems to function.

- Found out that both, upper and higher RAM board did not work on both machines. There were different effects, two boards causes 'double bus errors' which sounds like an adressing problem, two just did not keep information. I saw that upper and lower boards have different names but are in fact the same PCB. Just different solder bridges. So I thought, that interchanging the address selection on both boards that caused bus errors could bring a small chance of avoiding defective adress lines. And voila - the processor came up.

- There was still an 'Error Prom Bank 8 and 9'. This looked really bad, because ROM sets on both cards are not identical. I decided to readout the Rom content and program 2716 chips which should be indentical to the russian EPROM. I was lucky enough to fix it this way. I have no explanation for the effect but I've done it before for an old CNC controller. The slow readout by the EPROM burner seems to get the bits while the CPU cycle is too fast for defective chips.

- Next obstacle was an error of different joint encoders. Looks like two of my arms have problems here. Solved just by selecting a working one.

- Last problem was tougher: One pin of the arm connector was loose and did not make contact but was pushed back in the housing

Ok, enough text, let' look at a few pictures:

My NOKIA arms


The controller shelf


The power section


The logic section


The Terminal section


More to follow in next post...

Best regards
Uli

 

[ulihuber]

OK, let's coninue with pics.

This is the CPU-card


Rom card


Ram card (2x)


Axis network


Axis controller (6x)
...

..more to follow

 

[ulihuber]

3rd. part:

Servo controller cards (6x)


Serial card


Parallel card


Something undefinable card


Having worked through all these cards, having cleaned them, having separated the working from the non-working ones (good to have some spare parts...) the situation looked much better.


No Duble Bus Errors any more. No CPU fatals, no memory errors, no joint encoder errors. It boots straight through into the operating console/user interface.

I can use all the nice commands from the russian manual, never knowing what their meaning could be as description is not readable for me :-(((

The mobile interface communicates with the controller.

But now the problem: It shows 'Arm power off'. No way to move the arm.
And - in contrast to the Unimation or Stäubli controller - there is no 'Power On Arm'-button.

I have no idea where to look at or what to do next. There could be an external interface, signal line or a command. Now I need someone who is able and willing to translate.

More to come later...

Best regards
Uli

 

[bluethunder]

Thats too cool. I would love to have an old industrial robot at my house..

Problem would be convincing the wife, and getting 480/600v 3 phase service put in.. Not sure which would be harder..

 

[tomrobot]

Hi Uli,

not sure if I can be of help for you in the future. I just bought a Stäubli PUMA MKIII some days ago. It is still on my trailer and I have to put it to my workshop first (what is a project in itself). I'm located in Germany (Baden Würtemberg). I got the documenation that comes along with the PUMA (not much technical stuff but the arm and the VALII handbook).
But since you wrote that the Nokia types are not the same as the Stäubli ones and since my controller is looking completely different I doubt that I can be of much help for you. Anyhow I need some time to get it up and running (if I did not damaged the PUMA carrying on the trailer it should work - I saw it working before).
It is always good to hear that there are people in the area dealing with these vintage robots. I'm deeply impressed how far you got with you stuff that looks pretty used. So it seems that you have a profound understanding of these things.

Best regards,
Thomas

 

[ulihuber]

Hi Thomas,

yes, I think I collected some knowledge over last 50 years...

But this robot gets me to my limits. Whereever I think I found a way to go, something shows up that is impossible to overcome.

I wonder how long it will take to get it moving...

Uli

PS: Are you active in Peters CNC Ecke ? How about the 'Schwabentreffen' next week ?

 

[tomrobot]

Hi Uli,

never heard of Peters CNC Ecke - I will register to see if it is of interest. The different subforums are looking of interest but maybe not for my Puma project - let's see. And what about 'Schwabentreffen' - what is it about. Sounds like a community of people dealing around with CNC or robotics. Is it more about "how to use CNC best" means more oriented on the goals building a work piece out of a piece of metal/plastics/wood or is it more about robotics?
Maybe you have a hint for me for a specific problem I have with my new Puma after I put it in my workshop, wired it together and got it working.
The controller and the arm looks fine. I can boot it, move and and program it but when the arm is a position where the joint 2 (arm itself) is under load (straight position) and I turn off the arm power the brake system is activated but it falls off 1-2cm before it stops. I guess this is not itended. I supposed to stay in the same position as it was when the power is on. Do you think this is a problem with the mechanical part (the brake system) or more with the controller part?
I'm concerned that this issue could damage tools when it handling it (e.g. I'm plan to add a plasma cutter in the near future).
Any help much appreciated.

Many thanks, Thomas

 

[ulihuber]

Thomas,

I just suspected you were one specific person who posted about a newly bought PUMA arm last weeks.
Peters CNC Ecke is German largest forum for CNC professionals and hobbyiststs. And robotics is a small section there too.
The 'Schwabentreffen' is an in-person-meeting over 2 days where the 'virtual people' have the chance to get real. Just a lot of talk, dicussions, a little bit of swap exchange, showing pieces of work etc.
Might not be the number one event for someone not beeing active in the forum however.

About your break problem: No, I don't have any idea here. First, my robot is - while using a software very close to VAL - based on completely different hardware. Second, up to now I did not get to the point where an arm gets power from the controller. Stiil a lot of hard work to do.

@others: I made some progress last days. A friendly guy in Kiew sent me some binary files with test routines for the controller. He enclosed a software which should be able to write floppies for the Sfrea-36 controller but this little nasty program does not like any of my mainboards, floppy-controllers and/or floppy drives in any combination. So I decided to sniffer the serial protocol between the controller and the disk drive. It took me some hours to make an adapter cable, to copy all coming and going data while doing disk operations and to reverse engineer the serial protocol. The worst problem was an unknown algorithm for an 8-byte checksum on 256 byte data blocks. After some unsuccessfull hours with trying to brake the encryption by brute force (variing start value and polynome for an 8-bit crc) I got help from someone who had the knowledge of the algorithm. Two hours later I had my little DOS-Turbo-Pascal programm formating a disk and copying the necessary files from the PC to this Sfera-36 system disk.))

Back to the workshop at midnight and - wow !!!- the diagnosis starts on the screen ! Hard enough to stop here for being able to make the next day in work....

I'll keep you informed about further progress....

Uli