How to design redundancy for relays?

It's become well obvious that this is something that will be beyond my capabilities or understanding for the foreseeable future. Maybe I'll just shelf redundancy, and focus on robustness. :slightly_smiling_face:

I beg to differ.

You ask for information, but complain if it is not served with sugar coating,
a layman explanation and a thorough justification?

In a forum where private members voluntarily try to help each other?

How dare you!

Let me put it another way. It was a STUPID response, that made no real attempt to be helpful. I'm much better at coding than electronics. Perhaps to newbie coders, I could give the equivalent reply, "write less crappy code".

In a private forum, where private members voluntarily try to help one another, there are no good excuses for being a smart arse. If one believes there is, then one's time is better served doing other things. Or, if you think I'm wrong, perhaps we can find out if you are interested (at a beginner level) in something that I excel at. I'm sure that you'll fit rather nicely under my shoe.

I'm genuinely grateful for the help that I receive from helpful members - and there are loads of them. One hit and run post from someone who isn't actually invested in the topic, doesn't really detract from the fact that somebody else might learn from my questions and experiences. So how about we just all get ourselves square again, and move the topic forward? (great idea, mate)

No, it was not.

It is what you really use as an answer.

I beg to differ. See reply #4.

I'm all for being told straight - but if I'm buying crappy relays, perhaps it's because I don't know what NOT crappy relays are, right? What's your motivation in answering, if not to steer me right?

People don't know what they don't know.

Again, see reply #4.

Good luck with your project and attitude. :face_vomiting:

Cheers.

@failsafe7
I work in telecoms. Telephone exchanges are all but gone in the traditional sense because telephony is migrating to the internet, but their design might be of interest. Telephone exchanges were designed with a lot of redundancy so single failures* do not affect the whole thing. Redundancy was provided at multiple levels. Some kind of processor was needed to receive the dialled digits and route the call through the switching network. However the processor worked there were more than one of them, possible 5 or 10, possibly a lot more than that, depending on design considerations and exchange capacity. If one of them failed or had to be taken out of service for maintenance then the others could cope. The switching matrix, however it worked, had multiple possible paths for any one possible call to take, again, lose part of it and the rest could cope.

Telephone exchanges have thousands of lines and need to be able to connect any to any, there needs to be very many instances of the same kind of equipment, meaning there are many duplications and thus opportunities for redundancy without having to have (much) more equipment than is required to just do the task. I would go so far as to say it is impossible to build a 10000 line telephone exchange without any redundancy, the nature of the task means redundancy is inevitable.

That is as long way from trying to provide redundancy for one or a small number of relays, as others have pointed out that is difficult as best and quickly increases the likelihood of failure, so is self defeating.

At the risk of being crass, just do as suggested and buy decent relays. Cheap relays for hobby use are less reliable than expensive, industrial quality relays. If you can find relays certified for aircraft or military or spacecraft use they will be super reliable. Whether you want to pay for them is another question. In between hobby and military quality there are industrial quality relays, which are probably what you need.

* Except one instance I know of where a single transistor failed and caused a 10000 line exchange to stop processing any calls.

If you want a project to work for more than a few minutes, avoid crappy parts and buy documented, quality controlled parts from a reputable manufacturer, via distributors like Digikey or Mouser.

Relays must be purchased with specs that exceed the demands of the load and matching the power supply characteristics, which you failed to mention.

The quality of the advice that you get on this forum depends on the quality of the information that you supply.

That is correct. As I said "These are for AC usage only.."

But looking back at the OP's original post(s) it seems the goal is 120VAC at 15A. The boards I was thinking of can only handle 2A. So they would not be a solution.

I agree with some others that your language is not appropriate. If you like any explanation on a very obvious advice, you can ASK for it. There is no reason to be so brutal.

So I still just have that one last little question about how I can be sure that a relay works for my application. What is the key variable to look for when trying to deduce if my Arduino can trigger the relay?

I have no problem buying quality components for the post-breadboard execution. I can do simple stuff, like basic perf board soldering (to hold), and wire wrap to make connections. 3D printing, for packaging, and we're off.

All in all, even with good quality components, my projects aren't going to break the bank. I invest more in "R&D" to learn, than I'll ever get back in any individual project. So I do appreciate the advice on sourcing and spec-ing.

@PerryBebbington - telephony, at the basic level that I understand it, was one of the ideas that I associated with redundancy. I have to say that I'd probably persist, if I were building something along the lines of an autonomous roving vehicle, that was meant to stay in the wilderness for extended periods of time. But I'm currently just exploring some IoT applications, related to leaving unattended processes while I travel for weeks or months, on end. I just naively wondered if it were already easily achievable, and documented by some member much more knowledgeable than myself. Certainly, I won't try to invent what's already being done.

@jremington - thanks for the sources. I clarified in reply #6, but I don't mind stating it again: 120VAC, 15Amp, using Arduino 5V to trigger.

I found Digikey to be a good source to identify mainstream components.

Their sorting capability works well for me. I usually select the characteristics I need then have the table sorted by number of parts in stock. I focus on those parts with the most parts in stock assuming they have a "following".

Look here for a sort with a 5Vdc coil and 15 - 20 A capability

You need to expand your ideas. Think of you project as a "black box" that you cannot see into nor measure anything inside the box.
You can have an external computer system monitoring the black box inputs and the black box outputs and verify that the black box it operating correctly.
If the black box is not performing correctly, then the monitoring systems needs to be able to switch to a backup system and report the problem to you.
Paul

But at that point, it almost sounds like standalone Arduino might not be the best tool... Like maybe using Raspberry Pi over multiple Arduinos... I'm not exactly sure what a backup system looks like in that scenario, short of a complete duplication of the system. (all hardware and peripherals)

What's the system you're building?

Yes, very much so. However, I hope I gave an indication that in telephony the nature of the task means building in redundancy is relatively easy.

I think it's a fair question. I've seen at least 2 people on here involved in satellite design, they might have something helpful to say, but that's up to them.

Redundancy in telephone, and ever the internet, does not mean the communication continues with a new path when a failure occurs. A failure requires the connection, or process must begin with a new initiation of the connection so a new path can be followed.
Paul

1 Like

Going to skip all the vitriol and just get to the heart of what you need. The Operate (aka pull-in) voltage is the minimum voltage needed for the relay armature to make reliable contact and the Release (aka dropout) voltage is the maximum voltage at which the armature will spring back to neutral position.

As a general rule, an Arduino pin output cannot directly drive a relay. There are exceptions to this for relays with very low current coils such as reed relays used for switching low level signals. Even then you will need to protect the output pin from the inductive spike that results when the relay coil is deactivated. I'd avoid doing that if I were you.

Most power relays need a fair bit of current to overcome the spring that holds the relay armature in the resting position so you will need an external transistor to drive the relay coil. If you want a more reliable approach than the Songle relays (I should state that I've never had one fail, BTW, but I wouldn't trust them for anything critical), the easiest method is probably to get a standard arduino relay board and desolder the songle relays and replace them with something along the lines of the one I linked to above: the pinout is standard. The rest of the electronics are likely just fine.

As far as reliability: I have an arduino-based product that is used to control large industrial machines and my customer would come screaming to me if he started seeing them fail. I have also built or consulted on dozens (if not over 100 by now) of other arduino-based projects and I have yet to see any failures over a period of years.

You do need to be careful: these products are primarily built for hobbyist-level usage. They don't have the protection that industrial products normally would, so you need to make sure you stay well within whatever specs the manufacturer provides, or design in that protection yourself. Do that and I think you'll be surprised at how reliable the arduino clones & peripherals can be.

1 Like

I will keep my response as simple as possible. Choosing a relay is much like choosing any component. You choose based on your intended application. You consider not only the load and current the relay contacts must handle but also environment the relay will function in. Quality relays like any other quality component include a specification sheet. That specification or data sheet will tell you everything you want to know about the product. Relays can be a science unto themselves. Here is an example of what I am getting at. The .pdf file list dozens of relay interfaces and the dats for each. Learning to understand and comprehend the data sheets has a learning curve. All the factors need to be considered.

When looking at for example an Arduino board or a Raspberry Pi board to drive a relay you need to consider the relay coil current and the available Digital Out current of your board. Some common inexpensive relay boards include a transistor driver and some don't. You need to consider the needed current and allow some head room. This is also true of relay contacts. If you want to switch a 10 Amp DC load you want a relay with contacts rated above that, like about 15 Amps. This helps assure long life. Also, I believe there is a section in these forums about "Flyback Diodes" and why you need them. I suggest you read it.

Next as mentioned, there are cheap off the boat relays intended for basic hobby applications and there are industrial relays intended for applications where failure is not an option. When an aircraft takes off those on board would like to think it will arrive at its destination safely. When a submarine submerges those onboard would like to have confidence it will surface. There is a considerable trade off in price. I suggest you continue to read up on relays as there are pages and pages right down to things like contact materials. I can't add anything to what cedarlakeinstruments covered above. There is good and there is cheap but there is no good cheap. :slight_smile: Also a full and complete description of a project goes a long way in getting good help and suggestions.

Ron