Em 29/08/2013 05:39, RussellMc escreveu: > What they all said, on average, and more, and less. Murphy says so. > > I would tend to use 1 x 1N400X per coil mounted at the relay - across coi= l > contacts if practical. > > It's quite likely that 1 only diode almost anywhere on the circuit would > suffice and it's possible that you MAY get a useful improvement with > esoteric diode types. > If I was building these in production volumes I'd look at measuring what > actually happens with various solutions. Those who like to do high wire > without a safety net could probably get a good enough result with eg SPIC= E. > SPICE is usually an excellent way of modelling real world events. > If "usually" is good enough for you then SPICE may be too. [FWIW about as > good a SPICE as you can get is available for free. LTSPICE is free and th= ey > have recentyly upgraded it.] > > You said the relays were close together and separated from the driver by = up > to 20' of wire, and you specified the diuode to be near the relays rather > than the driver. > In the following "reasons" I also consider the more general case of relay= s > further apart on a 20' circuit and of a diode at the switch end. This is = a > superset of your problem but allows the solution to be more universal. > > My above 1 x 1N400X suggestion is based on: > > - Seems likely to work. > > - Has usually done OK in the past. > > - Cheap, available, robustish. > > - 1N400x will allow some voltage rise above what you'd see with eg a > Schotkybut I'd expect it to be under a Volt extra and would be > surprised if it was > vastly more than a volt. Voltage rise is caused by the pre switch-off > current flowing absolutely unimpeded into the new load. Given infinite > impedance it will sigh and deliver infinite voltage. If only a perfect > diode but with finite switching speed was present you'd get infinite > voltage as it turned on. However, circuit capacitance and wire resistance > alone will act as a reservoir with a time constant that the current will > transfer into. The 20' of line will look (ATM) mainly like a small > capacitor but to some extent like a transmission line with finite > bandwidth. All told I'd expect real world situations to deliver and > acceptably small 'blip' with a 1N400X. > > - When on the 1N400X sees coil voltage. A 1N4001 is 50 V (or so long ago > memory says) so anything in 4001-4007 range will do at 12V. I usually buy > 1N4007s as the cost difference is usually minimal and you then usually > don't have to ask. > > - While 1 1N400x is rated at 1A continuous it has a surge current of > amazing (perhaps about 20A? - see datasheet). Foor Irealy <=3D 1A it just > works. Fir I_relay > 1A (and that would be an unusual relay) you can deci= de > if the dissipation is going to be OK and that will depend on switching du= ty > cycle but in 99.99% of cases a 1N400x will have adequate current and > dissipation ratings. > > - As others have noted, if the diode is at the other end of a length of > wire you will get radiation and also you will get increased voltage rise > initially. Best (or worst) case Murphy will be able to use the radiation > from your 20' slot antenna to crash your system or if used in a more majo= r > installation, take the whole plant down. Doing this is difficult. Murphy > thrives on challenges. > > - Schottky would be better notionally in the forward direction but this > does not seem of major advantage. In the reverse direction Schottky leaka= ge > rises magnificently with temperature - as long as you stick under 50 C th= is > is unlikely to matter. If you get to say 100C then SOME Schottkys have > stunningly large reverse leakage. Many are poor but not so much to really > notice in a relay driver. In some other applications Shottky reverse bias > leakage can predominate. > > - If your 20 foot circuit is never broken and no relay ever gets unplugge= d > or disconnected then a far end diode will usually protect the electronic > switch. However, if you have one only diode and it is at the far end then > if the cct gets OC for ANY reason and your system turns a relay off with > the diode missing you may need a new driver. > > 1N4007 is 1000 V rated and the lowest voltage versions usually encountere= d > are rate at 100V+. Schottky diodes start to get expensive above about 30V > and may be cheaper again at 20V or 10V. A Schottky diode is extremely > ESDsensitive and not overly bulletproof while 1N400x's can be used as > bullets. > Even a modest voltage pulse applied to a circuit with a say 30V Schottky = in > it could cause Schottky destruction. [[FWIW: That said, I have used many > 1N5817s & 1N5819's (through hole) and do not recall ever having > ESDdamaged one. ]] > > - Aren't those enough reasons? Why do you want more ? :-) > > > Russell McMahon > Applied Technology ltd > Auckland, New Zealand. If I'm not mistaken, apart from the reverse stand-off voltage, the only difference between the 1N400x parts is junction capacitance. The parts from 1N4001 up to 1N4004 have the same higher capacitance and the other devices have a smaller one. Even the price is the same, so I think it is better just to stock the 1N4007. That's what I do. Isaac --=20 http://www.piclist.com/techref/piclist PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .