[TowerTalk] Relay lightening protection

[jimlux]

Kipton Moravec wrote:
> I had never heard of that before. The ULN2003 is a seven darlington
> switch and has the diode built in. 

Neither had I, but several other relay manufacturers say the same thing, 
as do the reliability folks in aerospace. I'd venture that the ULN part 
was designed to match what folks were doing with discretes, and those 
folks often were engineers working in small shops who never gave much 
thought and just copied a circuit from somewhere else.

Or, they just decide that it's ok to have the relays fail after a 
shorter than normal but still long and happy life.  In consumer gear, 
particularly, the cost of an extra component is significant (part cost, 
hole in PCB cost, assy cost, test cost, etc.), and if the box makes it 
to the warranty period, everyone is happy.

> 
> Also MOSFETS usually have the diode built into the switch for the diode
> in parallel with the switch case they show in second part of Figure 1.

Yes, but the built in diode for FETs is not always sufficient. Sometimes 
it comes for free as a function of the semiconductor process. If you're 
building a 4 quadrant DC motor drive, for instance, you're well advised 
to look at the properties of the diode and see if you need a separate 
discrete (or just use a bigger tougher FET) AND.. that's the diode 
across the switch model, not the diode across the relay coil.  The 
latter is what Tyco recommends not doing.

> They recommend the silicon transient suppressor diode to optimize for
> normally open contacts.

> 
> 
> But the diode is effective for a normally closed switch. 
> 
> "Now that we have provided suggested suppression techniques based on
> normally-open contact performance, we must add a qualifying comment
> concerning the normally-closed contacts. When the primary load is on the
> normally-closed contacts (and a small load or none on the
> normally-open),
> it may be desirable to use a rectifier diode alone as the relay
> suppression
> (or perhaps a rectifier diode and a lower value of series resistor). The
> retarded armature motion that adversely impacts normally-open contact
> performance will typically improve normally-closed contact performance.
> The improvement results from less contact bounce during closure of the
> normally-closed contacts. This results from the lower impact velocity
> created by the retarded armature motion and has been utilized in the
> past
> improve normally-closed contact performance on certain relays."
> 
> 
> So of course that begs the question, what to do with double throw
> switches?

Those sorts of tradeoffs are why they pay engineers to design things..eh?