OK to use 1N4148 100V-max diode to suppress relay coil transients?

[JDW]

I've always used 1N4004 or 1N4007 diodes to suppress relay coil transients in my engineering designs without problems, despite their switching speed being slower than a 4ns 1N4148.  Faster switch speed is typically preferred for something like this, but as you all know, a 1N4148 can only handle 100V max (some SMD types are only 75V max) whereas a 1N4004 can handle up to 400V.  Why does that matter?  Well, for most automotive relays that I use, I've measured the transient peaks to be well over 150V in most cases.  So for that reason I feel it best to go with a slower yet safer 1N4004 or similar due to its max voltage spec.  And while there are high voltage Schottky diodes out there, they are more expensive and therefore less desirable for coil spike suppression.

The reason I ask this question about the 1N4148 is because I have been asked by a client to price-quote an existing design where a large number of relays are used and external 1N4148W diodes used to suppressing relay coil spikes.  I am thinking it is best for me to advise the client AGAINST use of those 1N4148 diodes, but before I do that I want to hear the thoughts of other EE's out there.  Specifically, I would appreciate hearing some sound engineering advice on this point that goes beyond the mere, "well, I've used 1N4148's to suppress coil spikes for years and haven't noticed them fail."  Datasheets aren't everything, but the specs really do matter.  The fact remains a 1N4148 is rated at 100V max and a 150V+ transient from a coil is over-voting that diode.  In my mind, that is a recipe for failure at some point down the line, especially if the relay switches quite often.  Therefore, is it really a good idea to use 1N4148W diodes to suppress relay coil transients, especially in a design that requires long term reliability? 

The relay coils in my client's application are rated for 12V, 24V and 48V. The application where this circuit board will be used is in a 48V autonomous vehicle used in factories. Power on the circuit board which is feed to the relays is supplied by the following 4 DC-DC converters, ensuring a stable power source regardless of vehicle battery voltage dips:

• CHS4004848
• CHS1204824
• CHS604812
• CHS3004824

I look forward to hearing your thoughts.

Thank you.

[jmelson]

I've always used 1N4004 or 1N4007 diodes to suppress relay coil transients in my engineering designs without problems, despite their switching speed being slower than a 4ns 1N4148.  Faster switch speed is typically preferred for something like this, but as you all know, a 1N4148 can only handle 100V max

The purpose of the diodes is to recirculate the current in the coil, thereby PREVENTING the voltage spike.  So, as long as the diode is working, there will be only a very slight voltage transient when the coil is turned off.  So, the 1N4148 should work just fine.

Jon

[JDW]

I've always used 1N4004 or 1N4007 diodes to suppress relay coil transients in my engineering designs without problems, despite their switching speed being slower than a 4ns 1N4148.  Faster switch speed is typically preferred for something like this, but as you all know, a 1N4148 can only handle 100V max

The purpose of the diodes is to recirculate the current in the coil, thereby PREVENTING the voltage spike.  So, as long as the diode is working, there will be only a very slight voltage transient when the coil is turned off.  So, the 1N4148 should work just fine.
Jon

Jon, I certainly appreciate your speedy reply and your advice; however, I still have two concerns:

1. The phrase "SHOULD work" which implies "hope" that it will work rather than guaranteed success.  I cannot leave the door open to chance.
2. Based on what you said, what then about a 50v-max Schottky diode across the coil?  Surely at some point, the voltage spec of the suppression diode really does matter, and in that case, what is the voltage cap we need to consider here?  So long as the max voltage spec of the suppression diode is higher than the steady state voltage of the relay coil it's safe to use?

Please also consider that some relays have some pretty massive coils.  That is why I specified that some relays will be 12V across the coil, others 24V and yet other 48V across the coil.  Some of the relays have 4 switches in them.  The transients I am talking about are therefore different from the average relay you may have dealt with, hence the concern expressed in my original post. 

You seem to basically be saying that "voltage spec of the diode doesn't matter because if the diode is working, even if the diode is a 50V-max part (a diode voltage rating higher than the steady-state coil voltage) it should be okay because it surpasses the transient."  I am just trying to confirm if that is really so, regardless of relay coil voltage and coil current.

Thank you!

[uer166]

Something is wrong with the assertion that a diode-clamped relay coil can generate 150V, that is not how the circuit works. The diode needs to be rated to the relay coil working voltage + some overhead. So in case of a 48V coil, you can't use a 20V-rated diode, it will simply break down when the relay turns on.

I've used a 75V diode, as well as the 1n4148 for a massive 48V coil relay with no problems, and verified that no spec of the diode is exceeded at all temperatures. It turned out that the 1n4148 had very large margins.

When you say "150V spikes", you might be talking about classic resistor-based snubbers, I've seen that in automotive. That doesn't apply here at all since there is no resistive snubber, just a recirculating/clamping/snubbing diode whatever you want to call it.

This is a huge topic, did you know that using a zener clamp instead of a recirculating diode can massively improve relay contact life? That is because the relay opens much quicker, since the coil energy is dissipated in the zener, and the contact arcing lasts for less time.

[Someone]

This is a huge topic, did you know that using a zener clamp instead of a recirculating diode can massively improve relay contact life? That is because the relay opens much quicker, since the coil energy is dissipated in the zener, and the contact arcing lasts for less time.

Zener for the win...  however when the OP is agonising over pricing of 1N4148 vs 1N4007 it is possibly outside their constraints. If cost is the driver I'd be flipping the problem on its head and going for an SMD load of as many parts as possible, but thats getting off topic.

[Sredni]

What's the coil current?

[uer166]

SMD load of as many parts as possible, but thats getting off topic.

When I said "1n4148", I really meant "1N4148W", the SMD version 

[JDW]

When you say "150V spikes", you might be talking about classic resistor-based snubbers...

No, I'm talking about the UNSUPPRESSED spike.  Measure voltage across the relay coil with NO DIODE, then make your relay switch multiple times.  Measure the maximum peak.  That is what I am talking about.  If you get a 190V-peak transient, that energy must go into whatever suppression diode you ultimately choose to use.  So I am trying to get sound EE advice on that point with regard to the diode used, and especially use of the 1N4148 or 1N4148W.

This is a huge topic, did you know that using a zener clamp instead of a recirculating diode can massively improve relay contact life? That is because the relay opens much quicker, since the coil energy is dissipated in the zener, and the contact arcing lasts for less time.

I've heard those arguments, but I've not seen Zeners as the go-to suppression part of choice.  Even in some relay makers' data sheets they use a standard diode, not a Zener.  And if anyone should be concerned about relay life and conveying that info to buyers of their parts, it should be the relay makers.  If you do an exhaustive search for info on that subject, which I have done some years back, you find that about the only place where people actually use Zeners across a relay coil is in discussions about the topic on online forums.  That's not to say it's a bad idea necessarily, but if the idea is sound, I would expect to see more Zeners used across relay coils, which I do not.

Here's a datasheet for one of the relays used on my client's board.  This is one relay I need not worry about since it has the diode internal to the relay, but note that it is a normal diode, not a Zener.

[eblc1388]

...I am just trying to confirm if that is really so, regardless of relay coil voltage and coil current.

Of course the relay coil current matters, also the relay operate frequency too. Every time the relay switches off, the full coil current is going into the diode and some will be dissipated as heat to heat up the diode. Will a small 4148 diode good for a massive 12V relay that takes 2A coil current?

[uer166]

I think there is a fundamental misunderstanding of how a recirculating diode works.

"No, I'm talking about the UNSUPPRESSED spike" That spike is supressed to 0.7V over the power rail by the diode operating in the forward conduction mode, not reverse blocking. This test is completely irrelevant, in fact, the "real" unsupressed spike you should be seeing is in the kilovolts range.

[JDW]

For those of you asking about the COIL CURRENT, it varies by relay.  Some are only 5mA, other coils 8.5mA.  The largest coil current is 40mA for the G7L-2A-P-CB DC48 relay because it is 48V across the coil and a 1220Ω coil resistance, with between 6H~15H of inductance, varying by armature ON/OFF state.

[uer166]

In my design it was about 200mA at 48V (big coil, ~9W), at about 200mH (although note armature inductance changes depending on contact position as they open).

It had plenty of margin, your 48V at 40mA will be perfectly fine with a 1n4148, even with infinite armature inductance. Before suggesting to use random diodes, I would draw the circuit out, and fully understand the potential problems, only then come up with a solution.

Edit: you already know about inductance change 

[uer166]

I would expect to see more Zeners used across relay coils, which I do not.

I would expect relay manufacturers to never use zeners, because they don't know your drive circuit voltage withstand capability. When selecting a zener clamp (which btw has more variables than a recirc diode), it's a tradeoff between how fast it collapses the coil current, and how much voltage the FET/BJT/whatever need to survive.

Zeners are used routinely, especially in bigger contactors with economizers.

[srb1954]

The peak unsuppressed voltage doesn't appear across the diode since the the diode goes into forward conduction when the relay driver is switched off clamping the voltage across the diode to about 0.7V.

The diode is in reverse mode when the relay is on so the total supply voltage is then applied across the diode. This supply voltage, plus some safety margin, should be what determines the diode voltage rating.

A slow diode like a 1N4007 has an issue with clamping voltage overshooting a little (maybe 5V)  when the relay driver is switched off as it has a transient higher forward conduction voltage during the forward recovery time before it settles to its normal forward conduction voltage of 0.7V.  This transient overvoltage puts a little more stress on the driver transistor or FET so it is an additional consideration when choosing your driver device. A faster diode like the 1N4148 diode has an advantage that it has a much quicker forward recovery time and therefore clamps the peak voltage more effectively.

[JDW]

A faster diode like the 1N4148 diode has an advantage that it has a much quicker forward recovery time and therefore clamps the peak voltage more effectively.

A Schottky would be faster still, yet we don't often see them used as relay coil suppression diodes.  Would you say that is due primarily to reasons of cost and leakage, or something else?

[srb1954]

Cost and leakage would be a factor but also the more limited availability of higher reverse voltage ratings for Schottky diodes.

There is really no need for the much greater speed of a Schottky diode in this application as the rise time of voltage across the relay coil is slowed by the turn-off time of the driver device and stray capacitance associated with the relay coil and a reasonably fast Si diode can cope perfectly OK.

[Messtechniker]

To be on the safe side I'd do two things here:
1.) See if you can find any specs for the diode integrated with relays where such a diode is present.
2.) Exercise say 10 relays per operating voltage with your envisaged diodes over a few days and review the results.

[tggzzz]

The simple requirements are that:

• a diode should not conduct when reverse biassed, i.e. the relay coil is not energised. That means a 100V PIV spec should be compared with the PSU voltage powering the relay coil
• a diode should be able to conduct the same current flowing through the relay coil. Thus if the coil current is 100mA then the diode's forward current rating should be >=100mA
• when the coil is de-energised, the coil current I_coil will continue to circulate through the diode until energy has been dissipated in the diode and coil resistance, P[diode]=V_f*I_coil, where V_f is the diode forward voltage. That can be related to the diode's pulse current rating how often the coil is de-energised

If you want to shorten the time the diode conducts, then have a series snubber resistor to dissipate some of the energy.

Depending on the application, you may need to take account of spikes on the PSU rails.

You can confirm your understanding by using a spice simulation.

[trobbins]

• a diode should be able to conduct the same current flowing through the relay coil. Thus if the coil current is 100mA then the diode's forward current rating should be >=100mA
• when the coil is de-energised, the coil current I_coil will continue to circulate through the diode until energy has been dissipated in the diode and coil resistance, P[diode]=V_f*I_coil, where V_f is the diode forward voltage. That can be related to the diode's pulse current rating how often the coil is de-energised

In effect, the list point for continuous rating is not a requirement - it is the pulse rating that is a requirement.  How long a duration the diode will conduct the decreasing inductor current as it dissipates its energy through its own winding resistance will depend entirely on the relay, and that may need some testing to determine if the diode's continuous current rating is exceeded.  The 1N4148 has a 200mA continuous rating, but has a higher peak rating, which can be assessed as to whether the application is imposing a repetitive action or non-repetitive action.  The application appears to fit easily under both the continuous and peak rating limits, so there is no need to assess whether the peak current rating is exceeded.

[tggzzz]

• a diode should be able to conduct the same current flowing through the relay coil. Thus if the coil current is 100mA then the diode's forward current rating should be >=100mA
• when the coil is de-energised, the coil current I_coil will continue to circulate through the diode until energy has been dissipated in the diode and coil resistance, P[diode]=V_f*I_coil, where V_f is the diode forward voltage. That can be related to the diode's pulse current rating how often the coil is de-energised

In effect, the list point for continuous rating is not a requirement - it is the pulse rating that is a requirement.  How long a duration the diode will conduct the decreasing inductor current as it dissipates its energy through its own winding resistance will depend entirely on the relay, and that may need some testing to determine if the diode's continuous current rating is exceeded.  The 1N4148 has a 200mA continuous rating, but has a higher peak rating, which can be assessed as to whether the application is imposing a repetitive action or non-repetitive action.  The application appears to fit easily under both the continuous and peak rating limits, so there is no need to assess whether the peak current rating is exceeded.

Whether or not the pulse rating is relevant depends on the time constants and energy.

[JDW]

If you look at some 1N4148 or 1N4148W data sheets, you will see that the 1N4148 can handle more continuous forward current then it’s SMD “W“ sibling. However, both variants of the diode can handle up to 2A (non- repetitive) for durations of up to 1.0 µs. and many data sheets say the diode can handle up to a 500 mA surge for a duration of 1 second. But that data varies wildly by datasheet. for example, I’ve seen some datasheets that say it can handle a surge of 1.0 amp for 1.0 seconds. Some say the surge is non-repetitive while others do not say that.

[magic]

Depends on vendor.

If you want to be really pedantic about it, find some diode with transient thermal impedance ratings.