Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Design of an inductive spike supression on a relay

Status
Not open for further replies.

ti_chris

Newbie
Newbie level 3
Joined
Aug 4, 2020
Messages
4
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
74
Hi,

I've been working on this project to control my pool pump remotely using an ESP32 processor combined with a set of AZ21501–1A–120AE relays (one for the pump, one for the light). I expect it to generally go on/off once per day (so relatively low usage). I was hoping to gather some feedback around dealing with the spikes that would be generated by my 8.8A@115VAC 1/2HP WFE-2 Pentair pump. My choices as I see them:

No surge suppression:
My relay's life will obviously be diminished. The relay is rated at 40A so there's a fair margin here. It can easily still handle this after de-rating it to 40%. Would this really be that bad? I'd be happy if it lived for 5-10 years before requiring replacement.

MOV supression:
My current design is utilizing MOVs to absorb some of the spikes. The advantage is that this will tolerate some amount of beating at a relatively cheap cost. I'm not to certain how long they will last however. The failure case of MOVs in this particular case is also a little scary.

RC Snubber:
Alternatively, I could implement a RC snubber to absorb some of the shock. What I don't like as much about this is that it would leak current all the time and it may also pose a problem for the light if the current that leaks is high enough to turn it on. I'm also struggling a little bit to compute reasonable values that could be used for various demands. I've seen many various suggestions generally speaking.

Method 1:
Using Ohm's law it seems like the smallest resistor I could use for my pump (assuming we round the current to 10A) is 120V/10A = 12Ohm.
I don't have my motor's inductance or an easy means of measuring it with the tools that I have.

Method 2:
Some guide book that I read suggested a basic design guideline of 1Ohm per Volt of power supply rail R = 120Ohm (rated @ 2W)
Following that same guide for the capacitor, it's suggesting 0.1uF per Amp. 0.1*10 =10uF

Neither of these methods make me feel very good. In general, I'd much rather trust someone who knows a little more about these things. Any thought about what a combination of reasonable values would be. ie: Ideally, I wouldn't want to optimize the snub 100% towards this pump. If I change it later to one that's slightly different, I'd like to have some tolerance.

SSR:
Using an SSR would solve the arcing problem, but I'm not a fan of negatives that it brings with it. Notably, they waste a ton of energy @10A and require a thoughtful heat sink. The failure case is also problematic. I don't want my pump to burn out because the SSR failed short and can no longer shut down. I also dread the bill that this would generate before I could realize what's going on.

Contactor:
These tend to be quite robust and built for this sort of thing. The caveat is that they're rather expensive and don't really fit nicely on a PCB like a relay does.

I'm including all of my work on this project so far as well for context. I welcome & appreciate any feedback regarding anything else as well.

Schematic:
1596535597199.png


PCB:
1596535621706.png


Parts list:
 


As someone who has worked extensively with electrical motors and controls...... you require to use a proper contactor, rated for motor use.

Electrical motors take very significant amounts of current during start up, the first half cycle current might be 20X the rated full load current. As the motor gains speed, the current drops to a value of 4X to 8X the full load current. This diminished current however may last from 1/2 seconds to a pair of seconds, depending on the load's inertia.

Because of the motor's inductive nature, there will be significant arcing when the contacts open. That is the reason, among others, that contactors have a pair of contacts in series for each pole, to ensure that current is interrupted in two places.

I know that what I've told you won't dissuade you from using a relay. If so, then:

1) Use a 2_Form_A unit, and wire the two contacts in series.
2) Use a socket to make the relay replacement a simple task.

IMPORTANT EDIT; Optocoupler datasheets show how to implement a proper snubber circuit. Use it!
Second IMPORTANT EDIT: Optocouplers are not designed to drive a load directly. It is clearly spelled on every single datasheet. They are ONLY designed to drive thyristor gates. Use a small 1 amp Triac to actually drive the relay coil.
 
Last edited:
  • Like
Reactions: ti_chris

    d123

    Points: 2
    Helpful Answer Positive Rating

    ti_chris

    Points: 2
    Helpful Answer Positive Rating
As someone who has worked extensively with electrical motors and controls...... you require to use a proper contactor, rated for motor use.

Electrical motors take very significant amounts of current during start up, the first half cycle current might be 20X the rated full load current. As the motor gains speed, the current drops to a value of 4X to 8X the full load current. This diminished current however may last from 1/2 seconds to a pair of seconds, depending on the load's inertia.

Because of the motor's inductive nature, there will be significant arcing when the contacts open. That is the reason, among others, that contactors have a pair of contacts in series for each pole, to ensure that current is interrupted in two places.

Thanks for sharing your experience. I was thinking; what if I drove a contactor with my relay? It seems like that would generally do the trick without killing the relay quickly. Toughts about skipping the snubber in that configuration? The RLY540-2-120 seems like a pretty decent option for that.

Second IMPORTANT EDIT: Optocouplers are not designed to drive a load directly. It is clearly spelled on every single datasheet. They are ONLY designed to drive thyristor gates. Use a small 1 amp Triac to actually drive the relay coil.

Hmm you're right; it does say that on the datasheets. For my own education, what issues would you expect from doing this? It works in practice with this relay since it requires < 2W to trigger.
 

Semiconductor manufacturers, when they write a device Datasheet, expect that the device will be designed into a product which hopefully will be sold in the many thousands units. An operate correctly for long periods of time.

As such, the advice they provide has to be conservative considering all the statistical variations.
You may be lucky and build a circuit which doesn’t follow the datasheet, yet it operates correctly.

If you ask me what would I do? I would use the optocoupler to drive a small Triac, which in turn drives a proper contactor coil.
In the end, the decision on which path to follow is a compromise between performance and cost. If your original proposal works for you, then go ahead with it.

But I would still insist on a socketed relay, for easy replacement.
 

    ti_chris

    Points: 2
    Helpful Answer Positive Rating
Hi,

Safety:
Since your circuit deals with dangerous high voltage you need to take care about safety. Especially clearance and creepage distances.

EMI/EMC:
I also miss a GND plane, fast decoupling capacitors and filter (capacitors).
You switch load current, maybe with fast current or voltage spikes.
Bouncing GND, supply and/or signal lines may cause malfunction.

Klaus
 

Hi,

Safety:
Since your circuit deals with dangerous high voltage you need to take care about safety. Especially clearance and creepage distances.

I did pay attention to that. If you see areas of concerns, I'd love to hear about it to learn from it. I used the tables from IPC-2221B. The traces are not expected to be external. Where possible I've tried to keep them as far apart as possible.

EMI/EMC:
I also miss a GND plane, fast decoupling capacitors and filter (capacitors).
You switch load current, maybe with fast current or voltage spikes.
Bouncing GND, supply and/or signal lines may cause malfunction.

Klaus

I have 2 ground planes on the digital side as marked by the polygon in Eagle. Additional filtering may be necessary here. I'd have to check everything under a scope to see how it perform in reality
 

Hi,
Where possible I've tried to keep them as far apart as possible.
It's not a question of "possible" or "try". There are regulations, so it is a "must" and if it's not possible, then you must not sell your circuit.
You need to know what distances you need .... and you need to avoid hard edges.
I don't know what creepage distance you needs..it depends on many parameters.
I assume 5mm is a good value to start. And the minimum value of your PCB counts.
But at your PCB it's maybe just 1mm.

Use RATSNEST to show the GND plane. And mind: a copper pour cut into pieces is no GND plane.

Klaus
 

I used the tables from IPC-2221B. The traces are not expected to be external.
I hear a misunderstanding. All traces on a two-layer PCB are considered external (B2 or B4 according to IPC). Also solder mask isn't considered as safe coating, hence B2. But you seem to design for 120 Vac (< 150V according to IEC 601010). In this case, the clearance and creepage is fine, at least for pollution degree 2 (only non-conductive POLLUTION occurs except that occasionally a temporary conductivity caused by condensation is expected). It might even work for 230V.
 
Hi,

It's not a question of "possible" or "try". There are regulations, so it is a "must" and if it's not possible, then you must not sell your circuit.
You need to know what distances you need .... and you need to avoid hard edges.

Yes I did my research although as FvM pointed out, I miss-understood the definition of external. What I meant is that I entered the minimum in Eagle and where possible, I added more distance between them. I'll need to review them properly for B2/B4 and may wnt to consider a pollution of degree 3 given that this is placed outdoor and condensation is a concern.

Use RATSNEST to show the GND plane. And mind: a copper pour cut into pieces is no GND plane.

Thanks. I'm aware of that. I just tend to think that it's a little more readable that way. You're right about it not being a true ground plane. I may have some ground loops in there. That said, I placed my entire digital side on one layer so it wouldn't be all that hard to change every component to SMD and add a proper ground plane underneath it.

Has anyone mentioned using a solid-state-relay of the proper capacity to control the pump?

I mentioned SSR at the top and reason I'm, not a huge fan of it in this case. Worries are primarilly around the failure cases that it presents & heavy power dissipation.

I hear a misunderstanding. All traces on a two-layer PCB are considered external (B2 or B4 according to IPC). Also solder mask isn't considered as safe coating, hence B2. But you seem to design for 120 Vac (< 150V according to IEC 601010). In this case, the clearance and creepage is fine, at least for pollution degree 2 (only non-conductive POLLUTION occurs except that occasionally a temporary conductivity caused by condensation is expected). It might even work for 230V.

Yes I think you're absolutely right. I miss-interpreted that definition. I'll review it with that in mind. I'm also considering pollution degree 3 since this is going to be outside and condensation is going to be quite real. Either way, it wouldn't exactly hurt if I added a couple of cutouts to be extra safe.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top