1

Edit/tl;dr: What are some alternatives to reed switches?

I'm looking for a switch that can be closed one-time that doesn't require any case penetration. The design is inside a water-proof enclosure. The switch goes between an 18V, 25Ah battery and some electronics. I'm obviously aware of reed switches, which seem like the perfect solution, but I'm concerned about reliability.

The application is a scientific instrument which is deployed in the ocean and is not serviceable. The switch is necessary because the device is in-transit/storage for sometimes more than 1 year. The device operates for approx. 2 years on this battery.

I've done a lot of searching on this topic. I'm convinced I just do not know the right terminology to get me to my goal. This post most widely breaches this subject with lots of great suggestions but no great solution.

What I really want is something like a fuse, only instead of opening when it 'blows' it would close. This can be a complex circuit or a simple device, the only requirements are low power consumption(10's of uA or less) and no case penetration.

SSVT
  • 23
  • 7
  • A latching relay triggered by a magnetic reed switch? – PlasmaHH Oct 19 '17 at 14:26
  • Hi @PlasmaHH, thanks for the quick reply! I feel like in that scenario the primary actuation is still a reed switch. It does make for a reliable connection once closed but I'm concerned about things like early activation and failures from the reed switch. – SSVT Oct 19 '17 at 14:32
  • Why exactly are you so concerned about reed switches? With their contacts hermetically sealed in glass, they are some of the most reliable switches ever made. – Dave Tweed Oct 19 '17 at 14:37
  • @SSVT: I don't see any problem with early activations, and also not really for their reliability. They only have to switch once in their lifetime. If you want to reduce the risk put several in series and parallel so that you have the possibility of one failing and one early activating not doing anything (or scale up if you are scared). Or add logic for early activation that you need to hold the magnet for a minute. – PlasmaHH Oct 19 '17 at 14:40
  • You can just cover a regular switch with a water proof rubber cover. – user3528438 Oct 19 '17 at 14:46
  • I was a little worried that this question might just turn into everyone telling me why reed switches fit my application. The fact is you are probably right, and that may very well be what I have to use. I've posted an edit to the question that might get people back on track and a little more productive. :) – SSVT Oct 19 '17 at 14:52
  • If you don't want to use reed switches, your tl/dr should probably have been your question title. – Wesley Lee Oct 19 '17 at 14:57
  • @WesleyLee I totally get that. I was perhaps naively thinking that people would offer suggestions beyond 'your concerns are invalid'. – SSVT Oct 19 '17 at 15:00
  • I've seen lots of things ship with pull-out plastic tabs between the battery and its terminals. Would something like that work? – pjc50 Oct 19 '17 at 15:06
  • A pull tab wouldn't work unfortunately. I would definitely prefer that solution, but it is not waterproof, and sealing it would be challenging. – SSVT Oct 19 '17 at 15:15
  • But you could use a spring loaded rod held back or released by a pull-out plastic tab... – Solar Mike Oct 19 '17 at 15:42
  • @SolarMike Interesting...Could you sketch that for me? I'm not sure how that can be done without the same opening in the case. – SSVT Oct 19 '17 at 15:49
  • Can you have a small window in the outer case and put a photocell behind it? Arrange the sensitivity such that you need a super bright light to activate it. – Steve G Oct 19 '17 at 15:53
  • That's actually a pretty damn decent solution @SteveG. Thank you for the comment! First non-reed switch proposal. – SSVT Oct 19 '17 at 15:56
  • Just thinking again that the sun is super bright, so perhaps you need an opaque label (a metal foil label) to cover the window until you're ready to activate it. – Steve G Oct 19 '17 at 15:58
  • How about a coil tuned to 125kHz, with additional detection circuitry. Present an external 125kHz field to it to activate the product. A lot of older RFID tags worked at 125kHz so you may be able to use some of that technology. – Steve G Oct 19 '17 at 16:07
  • You could activate a 5V latching relay (or a solid state circuit) using a 5V out wireless charging receiver. Maybe use a sequence to turn it off again. You presumably want something that draws uA or less when quiescent. – Spehro Pefhany Oct 19 '17 at 16:14
  • 1
    Without valid reasons why a reed switch would be "unreliable", this question devolves to one about a religious issue, and should be closed. We can only assume your reasons for avoiding a reed switch are silly, because they certainly seem so, unless shown otherwise. The burden of proof is on you. – Olin Lathrop Oct 19 '17 at 16:17
  • 2
    It now seems this question is about how to avoid a particular technology for arbitrary religious reasons. Such discussions are a waste of volunteer resources. – Olin Lathrop Oct 19 '17 at 16:24
  • You say "and no case penetration" .. How do you plan on activating this whatever the solution is....? – Trevor_G Oct 19 '17 at 19:15
  • I'm very sorry you feel this way @OlinLathrop, I can't really do anything for you. It doesn't really matter what my concerns are, at the bare minimum, I've already provided enough of a response to this discussion, which is that I'm attempting to do my due diligence in investigating alternatives. No religious convictions; for which I might add, your unfailing support of only one possible solution absolutely reeks of. – SSVT Oct 19 '17 at 19:48
  • @SteveG thanks again! The device in question remains in shipping crate until just before deployment, so sunlight isn't too much of a concern for me. Not to mention the specific ocean in question, is the Arctic, and these are often deployed during extended sunless periods. – SSVT Oct 19 '17 at 19:49
  • @Trevor, other solutions in the past have used reed switches, but over the approximately 30 years we have been building and deploying devices with these requirements, those have occasionally been a point of failure, a rather embarrassing one to be sure. Other solutions have used external batteries that had waterproof cable connection(very costly). I don't have a great solution, which is why I'm reaching out. I'm certainly not asking for magic, just hoping to leverage a little brain power. – SSVT Oct 19 '17 at 19:53
  • 2
    You really need to lose the attitude and cooperate with the people who are trying to help you. This is an engineering site, so your concerns with reed switches are entirely relevant. And have those "embarrassing failures" ever been analyzed for root cause? We're trying to help you solve a specific problem, not speculate wildly, which would be off-topic here. – Dave Tweed Oct 19 '17 at 20:58
  • 2
    @SSVT The fact that you've been seeing occasional failures, along with the nature of those failures, should have been in the question, because they are the actual problem you're having. It places your question in context, helps people understand why you want to avoid reed switches (and whether or not that want is justified), and solve your actual problem. The matter of fact is, reed switches are the go-to solution for a scenario like this. If they're not working for you, we need to understand that, and why. If not, people are going to suggest the go-to reeds, and you get what happened here. – marcelm Oct 19 '17 at 21:42

2 Answers2

1

You could use a reed switch directly, or have it activate a latching electronic circuit.

If you can permanently arrange to keep a magnet in the right place, then just a reed switch could work. The magnet placed in a special holder or something then becomes the on/off control. This should be fine as long as your current requirement is low enough for a reed switch to handle. Considering you are talking about running for 2 years from batteries, the current should be well within the capability of a reed switch. In fact, you may have the reverse problem of not having enough wetting current.

Having a latching mechanism has a downside. If it ever gets accidentally activated, then the device is irreversibly on. A single reed switch closure seems rather risky for this. That's why I was thinking of keeping the magnet attached for the duration the device is supposed to be on.

Another option is to have a reed switch activate a electronic latching circuit. The pass element would be a FET with low enough RDSON for the voltage drop to be inconsequential. Check the leakage current and make sure that wouldn't cause too much battery discharge during the 1 year storage time.

If using a read switch as a momentary input to latch the device on, then you should require a sequence that is very unlikely to happen by accident. At least require the switch to be closed for some minimum time, like a few seconds. That way vibration and shocks, like accidentally dropping the unit, won't latch it on.

In my KnurdLight project, I required two button presses to turn the light on. Each had to be about ½ seconds with ½ second gap between them. To do something like that, the reed switch powers up the processor, which then turns on the electronic switch. If the right reed switch sequence isn't detected, then the processor turns itself off again.

Yet another totally different option is a pressure switch. The unit doesn't power up until it is some minimum depth below water.

Olin Lathrop
  • 310,974
  • 36
  • 428
  • 915
  • Thanks Olin, I do like the idea of including some logic to help with early activation concerns(Truth be told, I'm already doing that). If you were going to attempt your design without a reed switch, how might you do it? – SSVT Oct 19 '17 at 14:58
  • @SSVT: I don't understand what you've got against reed switches. They are a simple way to get a single bit of information across a barrier, without taking any power at all until activated. I'd use one only to initially power up the micro, then as a signal input. Once properly activated, the electronic switch takes over and the reed switch becomes irrelevant. – Olin Lathrop Oct 19 '17 at 15:03
  • Thanks again Olin, I definitely understand the sentiment. But there are some minor issues with reed switches.Just an example of a common issue would be switch magnetization. If the reed switch contacts became magnetized, [see this post](https://electronics.stackexchange.com/questions/80886/reed-switch-sticks-until-tapped), there would be no recourse for 'fixing' this issue in the field. The cost associated with this type of failure is extremely high. I'm definitely not against reed switches, but I need to do my DD and search for alternatives. – SSVT Oct 19 '17 at 15:22
  • 2
    @SSVT You really should include such information in the question itself. Please detail what failure mode you're worried about (early activation, deactivation in the field, or both?), how large the impact of failure is, why a reed switch would suffer those failure modes, etc. As your question is currently phrased, you're gonna get a lot more people saying "use a reed switch". – marcelm Oct 19 '17 at 15:33
  • Well I've edited the question now to reflect the preference for non-reed switch solutions. I'm fully aware how reed switches work and their benefits. I don't see much benefit to posting my concerns with reed switches, I do not need or want to told why a reed switch will work, similarly I do need or want to convince you why it wont. Lets just assume it will not work, and move on. If no solution is discovered, come back around. Maybe that's the scientist in me. – SSVT Oct 19 '17 at 15:53
  • 2
    *"I don't see much benefit to posting my concerns with reed switches"*. The benefit is that we might take you seriously. Your aversion of reed switches smells strongly of a religious conviction rather than a solid engineering tradeoff. It is therefore a waste of time, and off topic here, to delve into. – Olin Lathrop Oct 19 '17 at 16:22
0

You could use a P-MOSFET and a fuse as shown in the circuit below.

schematic

simulate this circuit – Schematic created using CircuitLab

However, that is a one shot deal and open to accidental enable and you would not be able to power it up to test it and shut it down again. The circuit also has a small trickle current when the power is off.

Instead you could use a P-MOSFET in a latching circuit as shown below.

schematic

simulate this circuit

When the on switch is pressed it will pull the output up to the battery voltage which starts charging C1. When this charges up to the minimum Vgs of M2 the latter will turn on pulling the gate of M1 low latching the circuit on. I have shown values that require pressing the button for 5s to latch it on.

Pressing the off switch immediately turns off M2 and M1, then C1 begins to discharge through the load. Once it discharges enough, again around 5s with a 10K load, you can release the off button and the power will stay off indefinitely.

You should play with the charge and discharge paths to get it the way it needs to be for your load.

OR.. you could just bring the power to two insulated bolts that go through some bulkhead and slap a jumper across them when you want to power it up and then slap silicone over it...

Trevor_G
  • 46,364
  • 8
  • 68
  • 151
  • Wow, thanks Trevor! I really appreciate your input on the "one-shot" design aspect. There are a number of reasons that this one-shot option is preferable to something that can be reset. I also appreciate your final alternative, might be the best option yet. ;) – SSVT Oct 19 '17 at 19:41
  • 1
    Why are you showing normally-closed pushbuttons in your schematics? If I understand what you're doing, they should be normally-open pushbuttons. Which, by the way, still leaves open the question of how to activate them without penetrating a case. – Dave Tweed Oct 19 '17 at 20:55
  • @DaveTweed they were normally open, no idea why it drew it like that though. Having issues with the editor today... :( Ya I just showed switches.. no idea how he is going to activate them really.. maybe magnets... external wires... – Trevor_G Oct 19 '17 at 21:09
  • 1
    @SSVT and yet you didn't give it a like... LOL but your welcome :) – Trevor_G Oct 19 '17 at 21:39
  • 1
    Sorry @Trevor, it seems I don't have enough cred to give you one. As I assume this thread is dead, I've accepted your answer. – SSVT Oct 20 '17 at 16:59