PIC based scanner safety?

[Frixxion]

My uncle has just started playing with PIC's and I've been listening in on what he has been doing and have started doing some research on my own as well. And I don't know if it has been mentioned or tried before but wouldn't it be possible to make a scanner safety with a a PIC? You would need at least 4 analog inputs but those PIC's are widely available. Most PIC's work at at least 1 mhz and our scanners work at 40 khz max?

I want to see if it is possible to write a program that compares the scanners position (trough it's feedback) and compares it to where the scanner should be (DAC output)? This is the way they all work right?

So is there anyone out there who's tried this? A quick search on the forum "PIC scannersafety" returns nothing

[-bart-]

...PIC based ABS...
...PIC based traffic lights...
...PIC based automatic pilot...

[Frixxion]

Is that a bad thing? Really I have no idea, all I know is that these things are incredibly easy to program and that one could do alot with these things

[mixedgas]

There is a bit more to it then just position.

usual place to start is to compare the sum of the squares of the velocity signal from the gavlo feedback chain. This gives you net vector velocity which is pretty much the speed of the projected beam.

You also have to detect galvo failures, and software failures, ie static beam, and graphics card failing etc

plus a watchdog on the pic to make sure its working.

Steve

[Frixxion]

ok, so this is something a PIC could do right?

Detecting galvo failures could be done by comparing galvo position to DAC output. Software failure could be detected by sqrt(x^2+y^2)>0. Is a PIC capable of doing these basic math operations? And watchdog is a feature that the PIC has right?

So would this mean it would pay off if I'd give it a try? I know I would not be able to make a professional product, but it's always better than not having any safety right?

Edit:

Sorry just thought that sqrt(x^2+y^2)>0 doesn't give you anything right? Is just gives you the length of the vector between x and y?

Read the output of the DAC on let's say channel X , wait 1 msec or so (to let the galvo position itself) and compare that value to the feedback of the galvo. If those values are off too much, the safety should spring into action.

[Banthai]

Hi Frixxion

Hope you are OK buddy and things are good in NL

we did start a thread many moons ago regarding an open source Scanfail unit ... it had a lot of input from many people but unfortunately never matarialised ....

one of the things that was discussed prior to that thread was the element of responsibilty .... lets say... if "I" design a scan fail circuit and then sell that unit on this forum and someone uses it in a commercial projector ... and it FAILS .... and someone gets hurt ( god forbid ).... it is VERY possible that "I" would be brought into the litigation and court case etc ...

ther was a debate that CPU based scanfail units should not be used as there were too many weak links ( software bugs, crashes, program coruption etc ) that could cause a scan fail unit not to do its intended job and allow stationary beams ... into the Audience ...

My personal view is that anything ( cpu or analog ) that minimises the risk is a good thing .... but unfortunately in the world of Legal proceedings anything that is regarded as an "Electrical safety device" has to be totaly failsafe ....

please check out our original post here .... http://photonlexicon.com/forums/showthread.php?t=4320

hopefully you can take over from where we left the project ...

look forward to seeing an update

PS on this subject ... i am currently looking at buying one of Gento's USB scanfail units .... as it seems to be a quality device and has been tried and tested ... I dont know how Gento protects himself from any liability etc .... i guess he will have insurance ...

again some kind of scanfail device ... has to be better than nothing at all

All the best ... Karl

[drlava]

I beleive Gento does this sort of programming for a living, (plus the safe program itself is rather simple) so it should be reliable. However, PICs are usually not used in mission critical devices, especially not PICs run on internal clocks. I'm all for an open source scanfail, just not on a PIC. It would be great if we could start an open scanfail project that also has a beam power pickoff for measurement, plus an extra input for the beambrush2 position feedback.

[Frixxion]

Wow! I did not notice that thread yet! Thanks for pointing it out to me, it contains alot of info on when the safety should spring into action. I see your point in the liability issue though. But fortunately, the Netherlands don't have such strict legislations as far as I know, so anything that I can come up with will be an improvement to my projector. I'll just keep it to myself how it works for now and just post any results I have. So if anyone wants my sourcecode etc.. mail me.

To dr lava, I'm sorry but I don't have any experience with PIC's yet so I have absolutely no idea of how reliable they are. And since I'm extremely stubborn I want to try and see for myself how they will handle this application. I learn alot by trial and error so this will might be another case of me making a big mistake. But anyways, I just want to try it

[Frixxion]

Btw, I have an ABB PLC that I might try and see what results it will give me. I know this PLC ain't fast, but I'll see what it can do for me

[buffo]

I'm all for an open source scanfail, just not on a PIC.

Andrew;

I remember you talking about this in another thread several months ago, but the exact thread escapes me. Were you thinking about an all-analog solution? And if so, how far did you get? I know there has been some discussion about one based around an actual CPU, but I'm assuming that would present many of the same problems as using a PIC would - in terms of reliability anyway... Do you think a digital solution of any kind is viable?

Honestly, for the hobbyist, it might be nice to have something like this - so long as it was reasonably priced. It doesn't need to be bullet-proof (or lawyer-proof, if you prefer!), because no one is going to do a commercial audience scanning show with one here in the US anyway. (If you want to even attempt to get a variance for crowd scanning here, you'll probably need Pangolin's PASS unit, and they're a lot more expensive.) But I think a lot of hobbyists would jump at the chance to install one in their projector for a price point of $100 or less.

Adam