Getting appropriate voltage for relays

[JohnRoberts]

I believe i shared this already but when I troubleshot and repaired my sump pump float controller I determined that the transistor driving the relay to switch the pump mains power failed short circuit collector-emitter ( evidence of a high collector voltage punch through).

I have seen this same transistor fail twice, first in the original unit, then in the replacement unit. I repaired the second unit by replacing the failed transistor with a higher voltage transistor I had sitting around in my back lab, and so far it is still working. The controller already had a clamp/flyback diode in place (looks like 1n400x series). My research suggest that almost any diode can be used there, while they behave differently WRT turn-on/off times. Apparently that protection diode didn't protect squat.

JR

 

[warpie]

The disadvantage is that it might not work. There is a threshold voltage required to properly energize the coil. If that is 18V, then that should be fine. But if your supply dipped a little for some reason like because the supply is linear and there was a momentary "brown-out", you might get a bunch of relays that flip momentarily.

So, if I power the relays with 22V instead of 24V I can still be on the safe side and save a couple mA per relay, right?

If they are all switched together you can use one flyback diode. But you should probably use a beefier rectifier diode and not a little signal diode. Any relay coil that is switched separately needs it's own flyback diode.

OK in that case a 1N4007 for 7-8 relays would be more than sufficient I guess.

Regarding the "dirty GND" this is what I ws planning to do anyway. Having said that, how do you treat the "Dgnd" on the PCB. Are you using a seperate groundplane or a big fat trace that returns to the reservoir caps or thereabouts?

 

[Bo Deadly]

So, if I power the relays with 22V instead of 24V I can still be on the safe side and save a couple mA per relay, right?

Look at the switching current and coil resistance in the datasheet and calculate how how much current you're making with 22V instead of 24V. But I can say with confidence that yes, using 22V instead of 24V should be no problem at all.

OK in that case a 1N4007 for 7-8 relays would be more than sufficient I guess.

Definitely.

Note that there are new diodes that are designed for switching power that would be slightly better (lower capacitance = less switching noise). There are lots of new parts like diodes and MLCCs and ICs and such that should be used in new designs. The only reason for using the old parts would be if the newer parts didn't have great availability (which unfortunately is a common problem these days) or unnecessarily expensive or because you're just copying some circuit and don't know how to interpolate the design.

Regarding the "dirty GND" this is what I ws planning to do anyway. Having said that, how do you treat the "Dgnd" on the PCB. Are you using a seperate groundplane or a big fat trace that returns to the reservoir caps or thereabouts?

Generally you should separate your "dirty" stuff from the non-dirty stuff. Then there's some boundry (or more likely a peninsula or fingers) that separate the two types of circuits. Then yes, they would have separate ground planes. If the PCBs are modular (very common in DIY designs), then you would have a separate wire-pad for DGND and 0V. If it's one PCB, then just make separate big fat traces to the 0V of the last filter cap of the PS.

Note that another possibility for driving relays is to run them between +V and -V and not dump relay currents into ground at all. You can also put a bunch in series and then add mosfets like 2N7000 to just bypass the coil. That can greatly reduce power consumption as well because you're using the same current to power mutliple relays (depending on the state of switching). The mosfet also already has a builtin protection diode. And you can drive the mosfet with an RC to make it nice and smooth. A 2N7000 SMD part is $0.20 USD.

 

[abbey road d enfer]

Note that another possibility for driving relays is to run them between +V and -V and not dump relay currents into ground at all.

I wouldn't recommend that. Remember that the rails are somewhat connected to ground.Assuming taht the paths to ground are balanced is a risk I wouldn't take.

You can also put a bunch in series and then add mosfets like 2N7000 to just bypass the coil. That can greatly reduce power consumption as well because you're using the same current to power mutliple relays (depending on the state of switching).

How would you do that? Do you mean powering a chain of relays via a CCS, like it's done with LED bars?

 

[Bo Deadly]

How would you do that? Do you mean powering a chain of relays via a CCS, like it's done with LED bars?

Not sure what you mean by CCS but you could have a chain of relays with mosfets in parallel with each coil. When the mosfet is on, current is diverted through the mosfet instead of the coil. Turning the mosfet off causes current to divert through the coil activating the relay. Etc. You would need to play around with resistor values to make it all work smoothly but if you have +-15 volts you might power two 12V relays or maybe as many as four 5V relays in this way using only the power required by one. In theory anyway. I've seen schems that have done it. Can't recall where.

 

[JohnRoberts]

in case it isn't obvious relays are a mature technology so most tricks for driving them have been explored. There are lots of boilerplate examples around.

JR

 

[moamps]

As a rule, relays should be supplied with the nominal voltage for which they are designed. The force with which the contacts inside the relay collide is very important for achieving good electrical contact and a certain level of cleaning of the contacts. This force is lower if the supply voltage is lower than the nominal one and in these conditions a shorter relay service life may be expected. If the consumption is too high in a design, relays that are low power (prepolarized etc.) can be used.

....... you might power two 12V relays or maybe as many as four 5V relays in this way using only the power required by one...

Aren't some laws of physics being violated in this way?

 

[Bo Deadly]

Aren't some laws of physics being violated in this way?

Let me rephrase that. You might power four 5V relays using the same current required by one. But if you're using +-15, you're wasting power which is what I think you're taking exception to. But +-15 is all there is, using multiple relays in series might save power. Assuming the switching scenario allows for it. If you need all to be active or not active, as opposed to some always being active, then it would almost certainly not work because with only one or no relays active too much current would be applied.

 

[abbey road d enfer]

Not sure what you mean by CCS but you could have a chain of relays with mosfets in parallel with each coil. When the mosfet is on, current is diverted through the mosfet instead of the coil. Turning the mosfet off causes current to divert through the coil activating the relay. Etc. You would need to play around with resistor values to make it all work smoothly but if you have +-15 volts you might power two 12V relays or maybe as many as four 5V relays in this way using only the power required by one. In theory anyway. I've seen schems that have done it. Can't recall where.

CCS=> Constant Current Source.. If you don't do that or any current-limiting device, when all relays are de-activated, the current draw becomes excessive.

 

[moamps]

Regarding the use of flyback diodes, I would only add that I use (if possible) a series connected zener diode and universal diode (1n4001 type) if it is necessary for the relay to go to open state as soon as possible when the voltage is turned off. This delay is slightly higher if only a flyback diode connected directly to the coil relay is used.
Figure 3 in this doc.
https://www.te.com/commerce/Documen...v&DocNm=13C3264_AppNote&DocType=CS&DocLang=EN

 

[abbey road d enfer]

https://www.te.com/commerce/Documen...v&DocNm=13C3264_AppNote&DocType=CS&DocLang=EN

This paper is interesting, however it would be interesting to compare with a snubber.

 

[oran.outan]

Hi all

Great discussion !
Maybe of topics but curious about...

- I have a JLM 8 relays switching (which is great by the way) and his using a ulm2803 before the relays.
If you have several relays is there an advantage to use this switching transistor before the relays ?

- in the past, Link with a micro controller, for avoid residual noise I used relay reed (which if i’m Not wrong are kind of light emitter relay and not coil relay). But they were not in the signal path, they were used to turn on/off an electronic circuit closed to pickups. (ie turn on/off e-bow on electric guitar
So maybe a stupid question but why I never see them in an audio circuit !?

Thanks
F

 

[JohnRoberts]

Hi all

Great discussion !
Maybe of topics but curious about...

- I have a JLM 8 relays switching (which is great by the way) and his using a ulm2803 before the relays.
If you have several relays is there an advantage to use this switching transistor before the relays ?

darlington transistor arrays require fewer placements for lower assembly cost. Darlington devices have high current gain so can interface directly with logic.

- in the past, Link with a micro controller, for avoid residual noise I used relay reed (which if i’m Not wrong are kind of light emitter relay and not coil relay). But they were not in the signal path, they were used to turn on/off an electronic circuit closed to pickups. (ie turn on/off e-bow on electric guitar
So maybe a stupid question but why I never see them in an audio circuit !?

reed relays are small, generally low power but still mechanical relays.

I never used one in any kind of product design.

JR

Thanks
F

 

[abbey road d enfer]

- I have a JLM 8 relays switching (which is great by the way) and his using a ulm2803 before the relays.
If you have several relays is there an advantage to use this switching transistor before the relays ?

All that JR said, but also ULM2803 has built-in protection diodes.

relay reed (which if i’m Not wrong are kind of light emitter relay and not coil relay).

Reed relays are electromagnetic mechanical relays. You are confusing with solid-state relays, which indeed use an LED to trigger a photo-element.
They are appreciated for interfacing low-level circuitry with mains-connected circuits, where their inherent isolation is a bonus.

 

[oran.outan]

Hey!
big thanks for your responses and clarifications !

So, was definitely relay reed as the e-bow work on 9/12v dc... but no noise/click when they're activate...
Was mixing/confusing about that solid State...

Best
O-O

 

[abbey road d enfer]

Hey!
big thanks for your responses and clarifications !

So, was definitely relay reed as the e-bow work on 9/12v dc... but no noise/click when they're activate...
Was mixing/confusing about that solid State...

Well, there are low-level optocouplers that can be used in a similar manner to a relay. Check H11F1M.

 

[Newmarket]

All that JR said, but also ULM2803 has built-in protection diodes.


Reed relays are electromagnetic mechanical relays. You are confusing with solid-state relays, which indeed use an LED to trigger a photo-element.
They are appreciated for interfacing low-level circuitry with mains-connected circuits, where their inherent isolation is a bonus.

Yes. I've experienced / used reed relays (Pickering comes to mind) in scientific instrumentation where there are high voltages (talking kVs here). They sometimes 'stick' but this can be usually sorted by 'flicking' them with your finger !

And yes '2803' is a classic solution for switching relay arrays with minimal component count and hence pcb estate.