Topband: Detuning Relay Flyback Diode

[David Cutter]

Jim

I agree with your idea about the zener and a diode but I disagree about 
using signal diodes on their own.  In my early career I recall changing a 
few 1N4148/1N914s on small cradle relays.  Changing to 1N4000 series stopped 
the failures.  I also agree, that for small relays there is no need for 
1N5400 size.

David
G3UNA




----- Original Message ----- 
From: "Jim Garland" <4cx250b at miamioh.edu>
To: <topband at contesting.com>
Sent: Sunday, December 24, 2017 3:40 PM
Subject: Re: Topband: Detuning Relay Flyback Diode


> I'm always hesitant to challenge anything Jim VE7RF says about amps, 
> because he's almost always right, but here I have to respectfully disagree 
> with a couple (not all!) of his points. I, too, call protective diodes 
> "back EMF" diodes, because their purpose is to clamp the inductive voltage 
> spike that occurs when a relay coil is interrupted too quickly. As Jim 
> notes, that voltage spike can be much larger than the relay operating 
> voltage and can quickly fry, e.g., a switching transistor that operates 
> the relay.
>
> In the olden days, builders just put a diode directly across the relay 
> coil (cathode connected to the positive coil terminal), and this clamped 
> the inductive spike to about 1V, which is the forward, turn-voltage of the 
> diode. The breakdown voltage rating of the diode only needs to be greater 
> than the coil voltage rating, so fast signals diodes like the 1N914, or 
> 1N4148 diodes work fine.
>
> The problem with this simple approach is, as Jim noted, the diode alone 
> slows the relay's release time. The forward-connected diode acts like a 
> low value resistor R when the back EMF drives it into conduction. The 
> current through the coil decays with a time constant of L/R, where L is 
> the coil inductance. Depending on the diode and coil inductance, this time 
> constant, which determines how long it takes for the relay contacts to 
> open, can be many msec.
>
> Jim's fix is to put a resistor in series with the diode. This raises the R 
> in L/R and shortens the release time somewhat. Unfortunately, it only 
> partially clamps the reverse voltage spike, so breakdown of the switching 
> transistor can still happen unless R is chosen carefully. The trick to 
> pick a series resistor that is as large as possible without exceeding the 
> breakdown voltage of the transistor. I've tried this, but under best of 
> circumstances, I still end up with an undesirably long release time. It's 
> a tradeoff between zapping the transistor or excessively delaying the 
> release time.
>
> My solution is to put a small 24V zener diode in series with the clamping 
> diode. This automatically limits the back EMF to 25V (24V from the zener, 
> 1V from the forward-biased diode). I've done numerous workbench tests and 
> find this is the best compromise solution I could come up with. It still 
> delays the release time slightly, but much less than a resistor-diode 
> combo.
>
> Two final points: First, none of this discussion pertains to the closing 
> time of the relay contacts. When the relay coil is keyed up, an inductive 
> EMF fights the buildup of current through the coil, but doesn't generate 
> destructive voltages.(The exception is if you try to switch the relay on 
> with a high impedance current source, but nobody would do that.)
>
> And lastly, there's no need to use a big rectifier diode (1N5408, etc.) to 
> clamp a relay coil, and especially no need to series several diodes to 
> increase their breakdown voltage. The diode is always forward-biased by 
> the inductive spike, so its breakdown voltage is immaterial, so long as it 
> is at least 100V or so. The momentary peak current through the diode will 
> never exceed the steady-state current of the relay, which is 100 mA or so. 
> Thus a single small 1/2W zener (1N5252) and a small signal diode (1n4148) 
> will work just fine.
>
> 73,
>
> Jim W8ZR
>
>