> From owner-piclist@MITVMA.MIT.EDU Mon Jun 2 16:58:10 1997 > X-Msmail-Priority: Normal > X-Priority: 3 > From: Didier JEANJEAN > [cut] > Optoisolators are really poor for induced voltage because of the internal > capacitor between the input and the output. They are only good for > isolation of low and middle frequence or low voltage but induced voltage > are some times 4000V with a rise time of 50nS. How about Nuclear EMP: 10KV/1ns? Lucky the cold war's over. OI's can be made as safe as you like. True, the standard type in a DIP cannot withstand much common mode dv/dt. The solution to this is to increase the distance between the LED and the photodiode, or add a grounded shield between the same. Increasing the distance is taken to extremes with optic fibres, but you can get good results with an ordinary IR LED (narrow angle output) and a large area photodiode, separated by a few cm. For screening, use brass mesh since it doesn't cut down on transmission too badly. E.g., I have been using the following design quite successfully: Take one IR LED for remote control usage. Take one BPW50 photodiode and position it so it butts up against the lens tip of the LED. Reverse bias the photodiode with >= 5V. You will find that the current transfer ratio is about 2.5%. I have also used 5mW IR lasers (e.g. Sharp LT021). In this case the average CTR is only 1-2%, but the differential CTR (near the laser threshold) is 4-5%. One advantage of the laser is that you can separate the circuits by many cm. Another advantage is an incredible speed increase. An IR LED only has about a 1MHz bandwidth, and thus limits the data transfer speed since a photodiode has a risetime limited only by capacitance. The laser, on the other hand, can be modulated at >= 100MHz. Contrary to popular belief, near IR SS lasers are quite linear between their threshold and their maximum output -- unfortunately the range is rather small (2:1) and if you exceed the maximum for only a ns... goodbye magic smoke! Oh, and yet another plus for IR lasers: some models have an independent photodiode at the 'non-business' end. This can be used in a feedback circuit to linearise the output. I can't wait to try this out, but unfortunately Sharp (in their questionable wisdom) connected the PD in a completely awkward arrangement, with the WRONG PIN commoned on the 3-pin case. But I have digressed. Regards, SJH Canberra, Australia