Spehro Pefhany wrote: >At 11:19 PM 2/5/2006 -0700, you wrote: > > >>Spehro Pefhany wrote: >> >> >> >> >> >>>A diode across the coil has been suggested, and is the simplest way of >>>handling the coil inductance. A 1N4148 is fine for any relay you'd be >>>likely to use. >>> >>> >>> >>> >>Alas, sorry, it is NOT fine and is no longer considered good >>engineering practice. I see >>this idea all over the place, and this is bad information. Even relay >>manufacturers include >>a diode, but it is not good practice. >> >> > >It's fine from the pov of protecting the transistor. Hundreds of millions >are in daily use. You won't see datasheet performance on the relay, though. > >What authority suggests that it is "not good practice"? And why would it >change? BTW, some of my designs use relays by the sh..er... pallet-load. > > > I was more concerned with failure of the 1N4148 itself. >>The 1N4148 is a "faster" general purpose silicon diode, used at one time >>because nothing >>any better was available. The 1N4148 cannot switch fast enough to >>suppress the edges >>of an inductive spike, so was most successful when a cap in series with >>a resistor is in parallel with >>the coil as well as the 1N4148. >> >> > >Distributed capacitance of the coil will limit dV/dt, although in more >sophisticated designs it may be desirable to add some external capacitance- >but not for protection reasons. You don't necessarily need a series resistor, >current will be limited by beta. For protection, even a slow-as-molasses >1N400x diode is fine. There are no nanosecond edges here to be concerned >with. > > > Actually, yes, the 1N4005 would be a better choice. >> Moreover, it cannot handle the high >>current pulse of the inductive >>collapse. I've seen wholesale device failures with this setup, even with >>tiny 5V relays. >> >> > >"High current pulse"? Where does this idea come from anyway? I have heard >this misconception before. The peak current through the diodes will never >exceed the steady-state current through the relay coil. There is no magic >here, the coil is generally just acting as an inductor (with some small >variations due to the magnetic stuff moving around, which can be ignored >for this purpose). > >The "strange" thing that can cause transistor failures is violating >the SOA of the BJT during "off" switching because of the inductive >load. Small transistors often don't have the safe operating area >explicitly defined, so it's best to err on the side of using a hefty part, >vis-a-vis the coil current, as I suggested. > >See, for example, Figure 3 on this TIP31 data sheet: >http://pubpages.unh.edu/~aperkins/pdf/TIP-devices/TIP31.pdf > > > > >>Use a Transorber, Varistor, or TVS device, with a voltage rating just >>above the voltage being >>applied to the relay or motor. If need be, just trust me on this one, >>its the truth. Go to the >>technical documents page of any transient voltage suppressor maker and >>you will understand >>what people are doing. These were originally invented by GE. >> >> > >Expensive and totally unnecessary from the pov of the transistor. They are >actually *harder* on the transistor because the voltage goes higher. > > >>Slow oscilloscopes won't pickup the unsupressed spike, so this can't be >>easily proven, but if you >>own a small portable radio, you can pick up the unsuppressed spike on >>the AM spectrum. >> >> > > > >>With a >>1N4148 it will still be present, with little improvement. If you then >>use a transorber or TVS placed >>AT THE RELAY COIL, the spike will be so suppressed that you can't pick >>it up (but if the suppressor >>is not placed as close to the relay as possible, the RF will still be >>generated by radiating through the >>conductors going from the relay coil to the suppressor). Try it. >> >> > >A TVS is useful where high currents are involved- for power supply >protection from load dump on automotive systems, where lightning may be >involved (that's where fast clamping is good). For relay coils, the >best approach if you feel like using a TVS is to use a diode or >regular zener and send the price difference directly to me. ;-) > > > Then why are these TVS devices becoming smaller and smaller? They are available for $0.25 USD in 0603 sizes. There is a huge market for inductive suppression (although I think the small ones are to allow US products to pass EU ESD specs). >One exception- if the layout is poor or if a crummy relay is used, it's >possible to have coil-contact flashover, especially from lightning. In >that case, a TVS can protect the driver (eg. telecom situation). But a >properly rated relay with solid-state (Transil, TVS, MOV) or spark >arrestors at the line is going to be more reliable long-term. > > > I spent 6 weeks solving failures on financial instruments (CC readers) and identified the 1N4148 across a telecom relay as the primary problem... it was shorting after a few months in service, and taking out the internal fuse on the PCB. The fix was to use GE TVS in place of the 1N4148. Its OK, Spehro, Sorry I snapped...I am just reacting to my experience. We agree on almost everything else. An item here and there won't hurt. --Bob >Best regards, > >Spehro Pefhany --"it's the network..." "The Journey is the reward" >speff@interlog.com Info for manufacturers: http://www.trexon.com >Embedded software/hardware/analog Info for designers: http://www.speff.com >->> Inexpensive test equipment & parts http://search.ebay.com/_W0QQsassZspeff > > > > -- Note: To protect our network, attachments must be sent to attach@engineer.cotse.net . 1-520-850-1673 USA/Canada http://beam.to/azengineer -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist