I thought I might chime in on this interesting tangent discussion about shutter speed testing too. 
First of all, the geometric effect that some are talking about that results from the finite opening/closing time of the blades, has been well described previously in terms of shutter efficiency. I am fortunate to have a copy of Photography Its Materials and Processes by Neblette, which has a very good discussion of such things, and quotes ASA Standard PH.3.4-1952:
The efficiency of between-the-lens shutters at a particular diaphragm opening and a particular exposure time setting is the ratio of the quantity of light actually transmitted by the shutter at those settings to the quantity of light that would have been transmitted had the shutter been fully open at that diaphragm opening for that total open time.
So the descriptions above have hit upon a real, and important subtlety that should be thought about when measuring and interpreting shutter speed test results. The efficiency can account for up to at least a full stop difference in measured "speed" at the high end, depending on how you measure/interpret the data. It is also a function of the aperture setting. When using the photodiode-timing approach for testing shutter speeds, that is just one of a number of subtleties that need to be considered. Other factors I thought of when I was first looking into this approach include the geometry of the optical layout in terms of collimated vs. flood illumination and/or need for aperturing. An example of this is that you might need a different configuration for a focal plane shutter compared to a leaf shutter. Another thing to consider is that ideally the photo-detector should not have so much light on it as to be saturated, since it will tend to rail at that maximum value too quickly and yield inaccurately long exposure times. Also, do you know what the response time of your circuit is, and whether it is fast enough for your highest speeds? Having said those last two, maybe published circuits already have this all worked out? Anyway, don't get me wrong; I am not saying that this isn't a good method - certainly it can work just fine - but I wanted to point out that there are some subtleties to take into account that are sometimes overlooked.
Over on photonet I talked about at least four different ways I could think of to measure shutter speeds. But when it came time to finally putting a system together I could use, I decided I really liked the digital camera approach as described here or here the best. It is hard to find a really good description of this method, but in a nutshell it involves using a digital camera (which must have a manual mode and be capable of generating histograms), and comparing light through the shutter/camera under test when that shutter is used, versus when the shutter on the digital camera is used. The nice thing with this approach is that it measures light throughput in a photographically meaningful way without having to worry about interpreting timing intervals, shutter efficiency, etc. In fact it automatically gives you what you might call the effective shutter speed, regardless of efficiency concerns or anomalies of the opening/closing behavior of the shutter. Of course this method has its subtleties too! And, I can offer tips if anyone using it. But I wanted to at least point it out for anyone who might not have heard of it before.
First of all, the geometric effect that some are talking about that results from the finite opening/closing time of the blades, has been well described previously in terms of shutter efficiency. I am fortunate to have a copy of Photography Its Materials and Processes by Neblette, which has a very good discussion of such things, and quotes ASA Standard PH.3.4-1952:
The efficiency of between-the-lens shutters at a particular diaphragm opening and a particular exposure time setting is the ratio of the quantity of light actually transmitted by the shutter at those settings to the quantity of light that would have been transmitted had the shutter been fully open at that diaphragm opening for that total open time.
So the descriptions above have hit upon a real, and important subtlety that should be thought about when measuring and interpreting shutter speed test results. The efficiency can account for up to at least a full stop difference in measured "speed" at the high end, depending on how you measure/interpret the data. It is also a function of the aperture setting. When using the photodiode-timing approach for testing shutter speeds, that is just one of a number of subtleties that need to be considered. Other factors I thought of when I was first looking into this approach include the geometry of the optical layout in terms of collimated vs. flood illumination and/or need for aperturing. An example of this is that you might need a different configuration for a focal plane shutter compared to a leaf shutter. Another thing to consider is that ideally the photo-detector should not have so much light on it as to be saturated, since it will tend to rail at that maximum value too quickly and yield inaccurately long exposure times. Also, do you know what the response time of your circuit is, and whether it is fast enough for your highest speeds? Having said those last two, maybe published circuits already have this all worked out? Anyway, don't get me wrong; I am not saying that this isn't a good method - certainly it can work just fine - but I wanted to point out that there are some subtleties to take into account that are sometimes overlooked.
Over on photonet I talked about at least four different ways I could think of to measure shutter speeds. But when it came time to finally putting a system together I could use, I decided I really liked the digital camera approach as described here or here the best. It is hard to find a really good description of this method, but in a nutshell it involves using a digital camera (which must have a manual mode and be capable of generating histograms), and comparing light through the shutter/camera under test when that shutter is used, versus when the shutter on the digital camera is used. The nice thing with this approach is that it measures light throughput in a photographically meaningful way without having to worry about interpreting timing intervals, shutter efficiency, etc. In fact it automatically gives you what you might call the effective shutter speed, regardless of efficiency concerns or anomalies of the opening/closing behavior of the shutter. Of course this method has its subtleties too! And, I can offer tips if anyone using it. But I wanted to at least point it out for anyone who might not have heard of it before.
