Open Source Scanfail Unit (phase 1)

[drlava]

Minimum scan rate:
This is not too difficult. My opinion is that if the full-field scanning is slower than around 10 Hz, we consider that to be a scanning failure. 10 Hz is just a rough number, and the actual number would depend on things like "distance to the audience" and "width of the overall scan field". Nevertheless, experience has shown that 10 Hz is a reasonable number. If there are other opinions, lets hear them .

Response time:
I think somewhere between 1 millisecond and 10 milliseconds is attainable and realistic for this kind of project. 1 millisecond buys you a single-pulse MPE of 10 milliwatts per square centimeter. If we have a faster response time, we start "eating into show content" and the artistic element of the show starts to suffer. Plus, it would be difficult to get better than 1 millisecond out of a simple circuit.

William Benner

I have been thinking about these two suggested limits and think there will be some difficulty in reaching them both. While the scan rate measurement is somewhat separated from the response time, they are still intertwined. Consider the 10 Hz full scale scan rate minimum. 10 Hz has a period of 100ms, during which there are a trough and a peak of the sine wave where the scan speed is slow, slower than the maximum deflection speed in a 5 hz scan rate. These periods last at least 25 msec. So, on one hand a 10Hz cutoff could be reached with low pass filtering of the absolute value of the velocity, but the filtering response time would have to be slower than 10ms. On the other hand, we could keep the 1 or 10msec response time, but at a cost of blanking the edges of deflection of the 10 Hz scan. This is all only considering one channel, where if it were a line it's likely the edges of the 10 Hz scan WOULD be unsafe after all.

Now if we consider both axes, and if we are generating a 10 Hz circle then at any given time one of the axes is generating the velocity required to keep the scan safe. Thus, the summation of the absolute value of the scan velocity of the two axes (as in the above circuit) can achieve the 10 ms or less response time while not blanking in 'safe' scanning situations such as a 10 Hz circle, but it wil blank if one of the scanners fails. This is another reason why, if single axes checks are to be implemented separately as above (in addition to the dual axis check), their response time needs to be slower than the response time of the sum of the deflection velocities.

What does this mean for safety? It means that the system above will allow a lower safe minimum 2-axis scan velocity and/or have a faster response time than a system that only accounts for each axis individually.

[Pangolin]

Suffice it to say that the numbers that I set forth early in this discussion are not numbers that I just pulled out of my a**. These are real numbers that are rather easy to hit with a properly-designed circuit. They are also real numbers that have positive implications for audience scanning laser safety.

I recommend you read the ANSI Z-136.1 Standard for the Safe Use of Lasers, and also read the IEC 60825-1. Really these are indespensible documents for anyone trying to achieve safety and create a laser-safety-related product. Once you understand the implications of those documents for visible wavelengths and their application in audience scanning, you will understand all of the other things myself and a few others have been talking about.

If you don't want to start with such dry documents, you could start with John O'Hagan's Theses here:
http://www.pangolin.com/resguide09b.htm

And my own article on how to do (or at least verify the safety of) audience scanning here:
http://www.pangolin.com/resguide09a.htm

Both of the latter documents imply my 1-to-10 millisecond response time (with 10 milliseconds being perhaps more attainable with a cheap circuit) and 10Hz wide angle scanning (which has an implication on radial and linear scanning velocity).

William or Bill

[dave]

Time to drag this thread out and dust off the cobwebs?

[heroic]

Time to drag this thread out and dust off the cobwebs?

How about just using the analogue inputs of an Atmel ATtiny attached to the feedback on the scanners, a little software (in C, even, if you like) to watch for if it's stopped moving, and reset the Atmel's onboard countdown timer every time it's still moving. If it's stopped, the countdown timer reaches zero and the output pin associated with it goes low, switching on a FET and hey presto, shutter activates. Costs about a buck...

[buffo]

One of the goals was to do it using all analog components. Easier to test/predict the failure modes, thus greater intrinsic safety.

However, I agree that a digital solution would be cheap!

Adam

[drlava]

I don't think anyone has built the above circuit yet. I have simulated it, if anyone wants the spice it's available.

[dave]

I had a play around with something similar late last year - I should dig out the schematic and post that..

Given a bit of time, I would be happy to proto your circuit, however, I'm a bit suck for time these days

[dave]

In light of recent events, maybe its time to dig this thread out again.. I have a completely free weekend, so its time to get to work prototyping

[Lasermad]

This thread is acheiveing a very HIGH "View" or read count , its running at typically double the amount of "reads" or "views" per post ..... so people are reading , and im certainly learning , a lot of it is out outwith my ability but i am reading and researching ! .and by the look of it so are many others ... just wanted to thank all for this very useful and informative thread.


PAUL

[dave]

I think that might be partially due to the fact that it got mentioned on laserfreak..