Can't decide on relay coil voltage

[HwAoRrDk]

In one of my projects, I have decided to use some relays, but I can't decide on which voltage (and therefore, which variety of relay) to use on the coil side.

My intended setup is a micro giving inputs to a ULN2003 darlington driver, which operates the relays. The switched voltage will be a nominal 12V, the micro (and all other logic) on 5V. But, given that with the ULN2003 has a wide operating voltage, I could be using either 12V or 5V to drive the relay coils. I can't really decide what to do.

So far, I've only thought of a few pros and cons for each choice:

12V
Pros:
- Lower coil current, and therefore lower power dissipation in the ULN2003.
- Does not add to load on 5V regulator.

Cons:
- Uses 'dirty' unregulated 12V supply.
- 12V level is only nominal, and may realistically be anywhere between 9-14V.

5V
Pros:
- Uses 'clean' regulated 5V supply.

Cons:
- Higher coil current, greater power dissipation in ULN2003.
- Added load on 5V regulator.

Regarding the greater current when using 5V, it's not too much of a negative factor, as I'll only be driving a maximum of 3 relays at any given time, so the largest additional current draw to the 5V supply will be (with the particular relays I'm initially looking at) around 90mA - which can be accommodated with a beefier regulator. I think this should also be acceptable in terms of power dissipation on the ULN2003. However, I have a feeling I may need some caps to buffer the 5V supply to avoid side effects when activating 3 relays simultaneously - is this a concern?

My concern with the varying level of the nominal 12V supply is that if it happens to drop to a low enough voltage (e.g. 9-10V), it may not be enough to actually operate the relay properly. However, I suppose this depends on the exact specification of the chosen relay.

Are there any other factors I haven't thought of? Which would you choose?

[ebclr]

Relays can work perfect with dirty unregulated power supplies, I bet on the 12V ones, don't forget the protection diode.

[HwAoRrDk]

In that respect, it's not the relays I'm concerned with, it's the ULN2003 darlington driver IC (which has integrated protection diodes, by the way).

[Ian.M]

Automotive application?
If so, reducing the load on the 5V rail makes it much much easier to design a regulator with input clamping to survive load dump transients.  Also the ULN2003 uses Darlingtons so has a fairly high Vce_sat even at low currents.  You'll at best get 4V across the relays if you use the 5V rail which, depending on your relays may be marginal for pull-in.

I'd use the 12V rail to feed the relays. If pull-in at low voltage is a concern,  use 9V relays and two ULN2003 outputs per relay. One connects direct, the other via a resistor of 1/3 the relay coil resistance.  Activate both to pull in then 0.2 seconds later deactivate the direct connected one to run the relay through the resistor at a reduced holding voltage.  An alternative approach is to read the 12V rail voltage with the MCU's ADC then PWM the relays to maintain an average of 9V across them.  If your MCU doesn't have enough PWM outputs, one AND gate per channel between the MCU and the ULN2003 would let you 'regulate' all of them from a sigle PWM output.

[jitter]

In one of my projects, I have decided to use some relays, but I can't decide on which voltage (and therefore, which variety of relay) to use on the coil side.

My intended setup is a micro giving inputs to a ULN2003 darlington driver, which operates the relays. The switched voltage will be a nominal 12V, the micro (and all other logic) on 5V. But, given that with the ULN2003 has a wide operating voltage, I could be using either 12V or 5V to drive the relay coils. I can't really decide what to do.

So far, I've only thought of a few pros and cons for each choice:

12V
Pros:
- Lower coil current, and therefore lower power dissipation in the ULN2003.
- Does not add to load on 5V regulator.

Cons:
- Uses 'dirty' unregulated 12V supply.
- 12V level is only nominal, and may realistically be anywhere between 9-14V.

Not really a problem as the coil voltage of a relay is also a nominal value.

If you look at the datasheet of e.g. this series of Axicom relay, you'll see that the "must operate" voltage is quite a bit lower and the max. voltage can be quite a bit higher.
The (e.g.) 12 V / 200 mW coiled relay is guaranteed to work from 8.4 V and has a max. permissable coil voltage of as high as 24.3 V. And once a relay has latched, it will not release until far below that must operate voltage. In this case the "must release" voltage is 0.6 V.

In reality the guaranteed specs are conservative values. I work a lot on products that use the 5V / 500 mW version of this relay and though it's only guaranteed to operate from 3.5 V, I know from experience that it will actually reliably operate from 2.5 V (as the result of a design flaw that caused these relays to be driven at that low voltage in a certain product). It went unoticed for many years until it was discovered by accident, while investigating another problem.

5V
Pros:
- Uses 'clean' regulated 5V supply.

Cons:
- Higher coil current, greater power dissipation in ULN2003.
- Added load on 5V regulator.

Regarding the greater current when using 5V, it's not too much of a negative factor, as I'll only be driving a maximum of 3 relays at any given time, so the largest additional current draw to the 5V supply will be (with the particular relays I'm initially looking at) around 90mA - which can be accommodated with a beefier regulator. I think this should also be acceptable in terms of power dissipation on the ULN2003. However, I have a feeling I may need some caps to buffer the 5V supply to avoid side effects when activating 3 relays simultaneously - is this a concern?

My concern with the varying level of the nominal 12V supply is that if it happens to drop to a low enough voltage (e.g. 9-10V), it may not be enough to actually operate the relay properly. However, I suppose this depends on the exact specification of the chosen relay.

Are there any other factors I haven't thought of? Which would you choose?

It does indeed depend. If the voltage drops to 9 V and the ULN2003 scrapes off another 1 V, then perhaps a 9 V relay would be a better choice?

[HwAoRrDk]

Yes, automotive.

The 5V version of the relays I have been initially looking at using are specified at 3.75V pull-in voltage, so I think there's some margin there. But again, I suppose I'll have to be careful here with my final choice. I'll make a note to remember that!

That's a nice idea with the dual ULN2003 outputs, but I'd be loath to consider it, as it would double the number of components in that area. As I have a total of 6 relays, I'd need two ULN2003s (unless anyone knows of a 12+ channel darlington driver DIP IC ). And I'd need to use an IO expander or two, as I wouldn't have enough MCU pins. Also, not sure they do a 9V relay in the series I was looking at.

I wasn't aware you could PWM relay coils. Are there any good resources you would recommend to learn more? Is it a good thing to do anyway? I guess it would save power (not that that'd be a concern for me, though).

[jitter]

Yes, automotive.

I wasn't aware you could PWM relay coils. Are there any good resources you would recommend to learn more? Is it a good thing to do anyway? I guess it would save power (not that that'd be a concern for me, though).

I wasn't either, but apparently it is. Here's an automotive aplication note on it.

[Ian.M]

PWMing relays isn't much different to PWMing any other inductive load.  Assuming a single switching device, you *MUST* have an anti-parallel diode across the coil rated to handle the full coil current.   You'll want the PWM frequency up in the low ultrasound range to get it well above any mechanical resonances of the armature, avoiding audible noise and contact fatigue, but keeping it as low as possible to minimise switching losses.

Another option would be 5V relays with a 6V regulator to power them - that should run OK down to 8V and will keep all the hash from switching the relays off your MCU power rail.

What are you doing about load dump protection? You cant simply expect a 7805 or LM317 (or any other low voltage regulator) to survive if fed directly from an automotive '12V' supply. 

[HwAoRrDk]

Hmm, solutions for either choice to mitigate problems with potentially not enough voltage for proper pull-in are all involving a bunch of extra components. I'm starting to think maybe go with the 12V choice and damn any problems with too-low voltage. At the end of the day, if the vehicle's 12V supply drops low enough to not be able to operate the relays, it probably means the operator has bigger things to worry about than the in-operability of my little device.

For 5V supply I plan to use an LM2931 (which I've used before). It has a bunch of protection features, including 60V load dump protection. I believe they're designed primarily for automotive use.

If I go the 12V route, are there going to be any problems with a ULN2003 operating directly on 'dirty' 12V? I don't want to leave that particular component open to chance of failure due to bad power if I can help it.

[Ian.M]

ULN2003 is only good for 50V abs max.   You might be better off using 100V low threshold MOSFETs and separate antiparallel diodes across the relay coils, which also gets rid of that nasty 1V saturation voltage loss.  Otherwise, unless the vehicle has central load dump protection, you'll need to limit the voltage to the ULN2003, with a series resistor and polyfuse, shunt TVS diode (approx 30V) and some bulk decoupling, and you'll probably also need that for the 5V regulator as well as you need to be able to handle a 101V load dump pulse.

[Delta]

12V. No questions. Stop worrying and crack on!

[HwAoRrDk]

What would cause a 100V+ load dump? I've not seen reference to anything of that level in the (admittedly brief) research I've done regarding handling automotive power transients. But I'm not at all well-versed in the details of that stuff, so I've probably missed something. I was under the impression load dumps/over-voltage of magnitudes around 24-60V are what commonly occur - from things like sudden battery terminal disconnection while the alternator is running, or jump starts.

Anyway, you're right - 50V maximum on a ULN2003 is asking for trouble. I think that tips the balance of decision back towards operating the relay coils at 5V.

But, if you think I need more protection on the power supply all round, then how about this for a plan:

- Have a 'safe' 12V supply protected by resistor, polyfuse, TVS, etc. so will see no more than 30-40V.
- Feed this to both the 5V LM2931 regulator and the ULN2003 to operate the relays.
- Relays directly switch the 'dirty' 12V straight to the loads.

...

Alternatively, I could use something like an SN75468? That's rated for 100V. Hmm...

[jitter]

But, if you think I need more protection on the power supply all round, then how about this for a plan:

- Have a 'safe' 12V supply protected by resistor, polyfuse, TVS, etc. so will see no more than 30-40V.
- Feed this to both the 5V LM2931 regulator and the ULN2003 to operate the relays.
- Relays directly switch the 'dirty' 12V straight to the loads.

Yes, sounds good.

[Ian.M]

http://www.sft-lab.com.cn/uploads/ISO_16750-2-2010.pdf
More than you ever wanted to know about the abuse your automotive gadget  would go though in testing if the client had specified ISO 16750 compliance.

That's were I got the 101V max load  dump peak from.   If you can control the make and models of vehicle its fitted to, to make sure they already have central load dump protection as standard, you only have to withstand 35V as the load dump should be clamped at or below that at the alternator.

The objective is for your gadget to *just* survive a load dump or jump starting idiocy that is nasty enough to take out the vehicle ECU.  If it lets the holy smoke out too far below that level, its liability waiting to happen.

[HwAoRrDk]

Thanks for that link. Some light bedtime reading!

I have no idea if any of the target vehicles (which would be few - my proposed device isn't particularly universal) have centralised load dump suppression. I do know that my own vehicle (that is one of the few) is described as "regulator-less generator with power transistor controlled by PCM", if that has any relevance.

[jitter]

PCM as in Pulse Code Modulation?

[HwAoRrDk]

Powertrain Control Module. More commonly termed ECU.

[hamdi.tn]

i think this is a case of overthinking a design choice .. 12V is one to choose.
however i have to say that i don't really like that ULN2003. If they fail due to a load short-circuit they may take with them the micro-controller driving them (been there, i did intentionally killed a uC driving an ULN2003 by overloading the ULN). i suggest to use relays comply with EN 61810-1 that garantie that the relay coil will only fail as an open circuit (most relays on the market comply with it so it should not be a problem finding one) and do the usual power input protection.

[Jeroen3]

For class A the operating voltage is 6 to 16 Volts. That rules out 12 Volts relays without boost converter.