Open Source Scanfail Unit (phase 1)

[mixedgas]

Looks great!, wil this circuit also detect if the operator draw only one beam, or a very small circle?

I do have doubs by using ADC, as they are probbably hard to find at a normal radio shack.

radio shack keeps dropping parts weekly, soon no more opamps.....
try:

www.digikey.com

ships worldwide, 25$ minimum order.

www.mouser.com

ditto, no minimum order, but more expensive shipping and lousy website search engine
however they will send you a catalog the size of the Bejing phone book 3 times ayear.

Steve

[mixedgas]

I will make a trial version.


are your board layouts drawn in Eagle ?
Steve

[Pangolin]

Hi all,

WOW, Audrey, that circuit you have is REALLY complicated. There are a lot of ICs.

What you should do is what's called a "single-point failure analysis". This means you analyze what happens if any single point in the circuit fails. (In your case, it means every single pin of every single IC). And in most cases you need to examine what happens if the point fails in two different ways (i.e. open, and shorted). If the result is that an unsafe exposure is allowed, then you will know you are on the wrong track...

As an example, while designing PASS, our analysis went well beyond single-point failures and PASS will prevent an unsafe exposure even in the face of five simultaneous failures. So this demonstrates that it is possible to design a product that has this kind of quality and capability.

But back to my main point. My point is -- more components = more possible points of failure. Why didn't MediaLas design a board with fewer components? I don't know. Why don't you ask them?

In any event, what we are talking about here is a little bit different. We aren't talking about (or at least, are trying not to talk about) a product for true laser professionals. I have said many times, and will continue to say that, if you are a true laser professional (which I personally define as being somone who actually receives money or other compensation for doing a laser show), then you need to be using a professional safety system from a professional company. If the MediaLas system meets your criteria, then sure.

But what we are talking about here is something very simple, that all hobbyists can make from Radio Shack components, and will give good performance. What I have in mind, frankly, would give better performance than most of the professional systems out there, and has only two analog ICs and a handful of passive components. This is along the lines of what Steve presented a few (dozen) posts ago.

You are certainly free to build Audrey's circuit and see how well it works, as well as building others presented here. But I encourage you to keep pushing in the direction presented by Steve. This job can be done very well with only two analog ICs.

And why don't we use shutters? Well, most hobbyists (and an awful lot of professional) laser projectors don't even have a shutter, and as was written earlier, shutters generally can't react in 10 milliseconds or less. And lastly, again, we always have to ask the question "what if something goes wrong". What if the shutter sticks? If that's your only line of defense, then you are screwed...

Best regards,

William Benner

[Miss A.]

Well William or Bill,

Please do not keep us waiting! Please reveal the SafeScan with the two IC's. Not only for the hobby laserist, but also for the audience who are in the danger zone right now.

[stanwax]

Audrey

See steves post on page 2 of this thread - with pdf attachements

Rob

[drlava]

Here's my take on an analog design around the basic ideas here (steve roberts front-end).
it's a basic differentiator circuit followed by an absolute value circuit and a summation amp. This is followed by a low-pass filter to provide a dwell time minimum and filter noise. The time constant for the summed speeds can be faster than the time constant for a single galvo, so that single scaner failure can be detected, but at a lower priority (sensitivity) to the complete scanner failure mode. Comparators determine the failure of the summed scanner speed, and each axis independently with a settable threshold for each. Finally, a latching or gate makes sure that if a failure occurrs, the lasers are blanked at least 100msec to make sure the problem has cleared. This can be changed to a manual reset requirement if desired. Also, with the addition of a transistor, amp, filter, comparator (all of which are available except for the transistor on this layout with current chips), a safe vertical zone guard could be implemented without much fuss.


what do you think? this schematic is close to being simulatable, I just need a LM2900 model.


EDIT: there's an error that has to be fixed: the comparators need to be switched around. it's late I'll get to that later.

[Pangolin]

This is *something*. You should build it and see how well it works.

If you test this thoroughly, the testing will reveal problems, and the problems will steer the direction toward solutions. You should eventually converge on what I have had in my mind for a while.

An old university professor used to tell me "Student tech thyself".

William or Bill

[Miss A.]

Audrey

See steves post on page 2 of this thread - with pdf attachements

Rob

Rob, ofcouse I know Steves pdf's.

But I think the boys from Pangolin already know the right, complete circuit with just 2 IC's, no doubt about it. I thought a forum is designed to share information, not hold it for yourself or even make it look like a quiz or puzzle: we receive a small hint and we have to think harder for the right answer.
When it's about (others) safety, you have to help each other. At least, that's the way we do it in Europe.

My contribution for the thread is over and will wait for the quizmaster with his rabbit out of his hat...

[drlava]

It's O.K, There's no good reason to assume our completed circuit will match William's concept he has in mind, in fact, there's little reason to try to make it match. As long as it works, that's what we need. If it's 4 chips instead of 2, and the cost is a few bucks more, it doesn't really matter. When he talks about multi-point failure analysis and keeping it working even after 5 point failures.. there's a simple workable solution to get there: redundancy. Simply have a nice tight design on a board and then duplicate it and feed the results go/nogo into a system that triggers if ANY of them show a scanfail, and design the final go/nogo check system so that if it fails, it won't fail to a false negative.
Dont' get frustrated because he's dangling a mystical solution out there, just keep with what we've got. I have found a few issues such as op amp selection on the above design, and will get a simulation running for it eventually for troubleshooting and tweaking.

EDIT:
Here is the latest Rev. It has been simulated and works, although the sensitivity and scan frequency cutoff haven't been checked. The differentiator gain may need to be adjusted and the set-points for the comparators need pots for prototyping and experimentation of those set-points, but the basic idea seems to work. The transistors and FETs aren't fixed, they can be changed depending on what you have on hand. Who's a wiz at prototyping?

-Andrew

[Pangolin]

I thought a forum is designed to share information, not hold it for yourself or even make it look like a quiz or puzzle: we receive a small hint and we have to think harder for the right answer.

Yes, well, if you are referring to my own actions, then it's like I wrote earlier -- this is supposed to be "Open Source" not "Benner Source". The problem is, if I just come forward with a working circuit, then everyone will just use it willy-nilly without any thought being put forth about it, or an understanding of how it works or why it is the way to go. More importantly, nobody would understand why other approaches are not as good. And finally, it is possible that someone comes up with something better than I have in mind.

This open forum-style allows everyone to contribute. This way, NEW ideas can be found. For example, I never thought of your own approach to a scan fail system. (I don't like your approach very much myself, but heck, who cares what I think. Build it and see how well it works.) There will be tremendous learning involved for all parties.

When it's about (others) safety, you have to help each other.

Yes, well, lets not forget exactly whose idea it was to have such an open-source project in the first place!!

Also, this really shouldn't be about (others) safety to begin with for two big reasons.
1. If you are exposing the public (especially for money), then you should be using a professional system from a professional company.

2. The scan fail interlock is supposed to be the LAST line of defense, not the first line of defence.

There is still a fallacy that, if you have such a thing, you don't need to do anything else. Divergence plays the single largest role in making shows safe. Show programming techniques play a secondary role (far secondary actually). A scan-fail interlock does not make a show safe. It could POSSIBLY keep a show safe, provided that it is already safe to begin with.

At least, that's the way we do it in Europe.

Huh? Do it in Europe? Could you please point out all of the European companies or people who have started an open-source scan-fail project, and their contributions to the project? I think this project wouldn't be underway if there was already someone "doing it in Europe"...


William or Bill