Hi @BrettMargoSupplies,

Thanks for the link! I actually started this project with @Kasim21 so am aware of his work. I’ve been searching for solutions that don’t use nichrome wire or are easier to assemble. Kas also used a 20c battery for his setup and I’m working with a much smaller battery; it might still work but need to characterize. 



Per the electRelease datasheet, they claim it is activated by a minimum of 10V bias but recommend 50V. However, not much current though unless the surface area is large which could be intriguing for a collar application. If I can get my hands on some, I’ll recharacterize it in my lab and post the results. 



Besides eventually using the system on African Wild Dogs with @Kasim21, I’ve been talking with potential collaborators in the states to pilot the system with domestic dogs and dogs in training facilities. The sensors aren’t particularly constrained to be used on a dog but am designing it with the African Wild Dogs as the target species. 



I’m also working on distributed microphone systems, quantized ML models, and real-time low power inference focused on pollinators in Argentina: 

Read "Overview ‹ BuzzCam — MIT Media Lab" on MIT Media Lab

Overview ‹ BuzzCam — MIT Media Lab

The South American Bombus dahlbomii is one of the largest bee species in the world. It can grow up to 1.6 inches long and, because of its size and furry appear…

MIT Media Lab

Goodness me! That bee could almost carry a sensordrop! 

If you do manage to get your hands on some electRelease I'd be very interested to hear how it goes. Happy to help with a preloading design. Those are fairly straightforward.

I finally got a hold of the lab behind electRelease and will get a sample sometime in the next couple of weeks for testing. I'll put it through a few tests and will post the results in a follow-up wildlabs post. 

Hi @Rob_Appleby,

Hmmm. Yeah I read that there was another company that licensed the product to use it in packaging but it seems they discontinued it. But given the website is still up, I’d imagine they still sell. They’re 30 minutes from me so maybe they’d let me stop by. If I make any progress, I’ll update this thread. 

We’re exploring to use this system on African Wild Dogs. Currently, the payload alone is about 50g and 74x10x34 mm^3. I’m looking to interface an additional detachment mechanism to this payload where it can be controlled either via timer from my MCU or the onboard LoRa radio. How much power and applied for how long does your 10g prototype take to detach? 

I’ll wait to see if the the electRelease folks get back to me over the next few days but I’d love to reach back out and sync you up on a few things I’ve been working!

I've had successful, repeated detachment with as little as 3V and a few hundred mA of current. I also managed to get a coin cell to power a release a few years ago. It really depends on two main things I think: the gauge of nichrome (where thinner gauges, such as 34AWG generally require less current) and coiling the nichrome - I usually wrap the nichrome wire around a thick needle (e.g. knitting needle) and try and get at least two full turns (see image below):

Sometimes more turns is actually better in terms of reducing current. Our timer drop-off has an approximate 'sleep' draw of 50uA (although in practice, it's usually less) so keep that in mind if calculating deployment times. Another idea is to try a 'hybrid' set up using a low current battery and a lithium super (or 'ultra') capacitor - the latter should be able to handle the spike in current for a few seconds, but, it will take a while to recharge again. One additional option here is, if you think the collar and particularly the sensor package are likely to regularly get sun, you could use a small solar panel to keep the lithium capacitor charged. I use this approach for small BLE sensors and it works quite well.

I was just discussing with someone else about integrating the drop function into an existing system with an MCU. Either of our drop-offs are essentially just switches that turn current on to the nichrome via a MOSFET. It's not very complicated really, although we do have some built-in protections in the timer version to repeat drop attempts automatically and monitor voltage, so I guess it depends on how sophisticated you need it all to be. Essentially, if you can program a timer in your MCU to activate a free pin at a certain time and date, you can easily create 'switch' using an N-channel MOSFET. Happy to do up a simple schematic if you'd like, so just let me know. 

We did have a lot of trouble getting the disbonding to occur in the epoxy, but perhaps you'll have more luck. From memory, it took several minutes of relatively high voltage to disbond and, you need a physical force like a spring to actually separate the two side. We found it all a bit too complicated and decided to use nichrome instead. But, as I say, you might have more luck and please do keep us all updated if you have any success. 

All the best,

Rob

Can I ask what battery you are intending to use? That might help determine what nichrome gauge to try...You can get very narrow gauges, which I suspect could be heated with pretty low current...

Cheers,

Rob

Interesting. Is the idea that more turns requires more nichrome so higher resistance means less power but requires longer for the detachment to trigger? Or does the conductive heat transfer across loops result in similar detachment times?

50uA is pretty good. Taking into consideration battery self-discharge rates, you're kinda in that ballpark where lower doesn't get you a whole lot. In my case, I think I benefit from having it controlled by the MCU while in one of the low ARM power states since I already have it in my circuitry. Unless I want to create a discrete system that will trigger regardless of any failure modes of everything else. I guess that's an ideal application for something like your discrete timer solution! 

I like the idea of having a discrete timer system that can be triggered via an MCU GPIO but otherwise, has a failsafe that always releases when X condition is reached (e.g., dangerous levels of discharge of a lithium battery or X amount of time has passed). 

Hmm, requiring an added force could be problematic. Unless I include an insulated spring mechanism in the bonding layer that will push up against the two metallic pieces and over comes the adhesive forces once a bias is applied. I'll still try to get my hands on some and test it out in various conditions but as you mentioned, nichrome might still be easier. 

Do you have a mechanical mechanism in mind when thinking about detaching an entire collar? As I understand it, SensorDrop is intriguing but focused on dropping a subsystem off of a collar while leaving everything else intact. I was thinking of doing something in line with a typical domestic dog collar where the nichrome would burn and trigger a spring-loaded pin that separates two plastic pieces. An alternative would be to just sow in fabric that would burn off while there is a barrier between the system and the dog to ensure the subject isn't hurt. For obvious reasons, I don't necessarily like something that can start a flame as mentioned in the SensorDrop paper but fully understand that that is the easier and more feasible method than most. The epoxy, although requiring no heat, is also an ingestion hazard as per their MSDS so that isn't a clear winner either. 

Right now, I'm operating with a 3.7V 300mAh battery. 

Ok, so it looks like that cell you linked to can handle a maximum current discharge of 600mA, which should be enough to heat coiled 34AWG or 36AWG nichrome. 

I think the number of turns is possibly more to do with the conductive heat process, but the additional resistance is probably also a part of it. Multiple lengths of line distributed across the entire coil can sometimes take longer to melt, or thicker line, and the most central coil may get the hottest. Usually, it all happens pretty fast though. At any rate, I'd experiment with number, diameter and distance between coils, at least a little bit. I remember Kas almost setting things on fire several times when he first started exploring! So look for the sweet spot where the line melts, but the coils don't get red hot. Kapton tape can help keep nearby things safe (at least to a degree), but in our collar drop-off design (see below), we had the nichrome a good distance away from anything sensitive anyway, and well tucked away from the animal too (using 3mm acrylic and silicone). 

You could use something like low voltage as a back up trigger (i.e. if your battery gets to a certain level, drop-off)...we have that in our timer model (but not the remote). If you essentially use your main MCU and battery then really you are just talking about the nichrome element (and a MOSFET and some resitors and possibly caps). You could run a short pair of wires out to a nichrome coil perhaps...I've done that on collars that couldn't have a separate standalone drop-off...

Here's a link to a paper Kas led on the original timer drop-off and our Github page has all the necessary resources...This is the traditional style of whole collar drop-off. We have a few different versions, some of which aren't in the paper (mostly different housings), but it should give you an idea of the standalone version we often use. It's often just a block of polyurethane or epoxy with a small section to rewire the nylon and access to recharge and reprogram. Here's an example of a drop-off (circled in red) on a collar:

And here's a little Sketchup model of another (essentially the same) design, showing a completely separated cavity at the front for the nichrome (drill through the separating wall to attach the wires to the PCB in the main cavity) just as another example:

I remember the guy from the epoxy company was like, ah der, of course you have to use some kind of force to actually separate the parts after we just sat there, hoping for the parts to separate...We abandoned it pretty much after that...

More than happy to discuss further, so just let me know...

Cheers,

Rob

Your hybrid suggestion is right in line with my collar setups.

Works but is somewhat annoying waiting for it to charge/recharge.

Most cubesats have multiple lithium batteries so they don't need the extra current.

On dogs you should be able to get away with only using the nylon or dyneema as your mechanical connection @pchwalek . Provided it's sourced from a quality supplier and tested. I had "80lb" nylon fail at <10lb when actually tested. The method of tying/securing the line is also very important. I like a fisherman's knot with a tensioning method and I epoxy my loose ends for good measure. Still trying others methods though. For animals larger than dogs my testing has shown nylon/dyneema to be insufficient on their own. Working on mechanisms to assist.

A 3-5ohm resistor can also be used in place of nichrome for cutting the line. You do have to control your temperature more precisely though or you risk melting the solder joints.

Having a discrete system triggered by a gpio pin is pretty standard in aerospace. Although the way I've seen it done is as a deadmans switch. If the discrete system doesn't receive it's usual heartbeat from the pin then it is triggered.

I usually use a modified version of the 'surgeon's' knot - I do three overhand knots in a row, and the same in the opposite direction. If you wet the line slightly, you can synch it really tight and smoothly and it won't slip at all. I've thought a lot about alternatives - I've always liked the idea of a mini guitar peg-style tightening system, maybe using a small machine screw with a hole drilled through it, so the line goes through the hole and the line tightens as you turn it - then you could maybe Loctite it in place...But I always just end up tying it - because I am lazy, and it's probably the lightest solution...

Re: breaking strain, we found in some early tests that the surgeon's knot usually breaks at about 25-50% of the rated strength of the line. So if a line is rated at 10kg, it will break at 2.5-5kg when tied...so we usually use two independent strands to get back to around 10kg. We also often use the line as a passive 'breakaway', such that if an animal gets snagged, it can break the line, and usually aim for about 75% of the animal's bodyweight...

I tried surgeon knots at one point. Was having trouble with them but maybe i'll have to give it another try. 

I quite like having a tightening system. Makes it easier to tie the line and I have a preload system as well.

If you get a tightening system working, let me know please Brett!

@BrettMargoSupplies thanks for the dyneema reference. Have to get some to try out. 

@BrettMargoSupplies @Rob_Appleby Instead of knots, have you tried crimping methods? Not sure if that can potentially cut the nylon line instead of securing it. I've also been thinking of designing a similar system to a guitar string tightener using a small worm gear. 

As an aside, I got a hold of the company behind ElectRelease and am waiting for a callback but if I get my hands on some, I'll post updates here or in a dedicated post. 

Haven't got around to trying crimps yet but think they'd be a decent option.

I've avoided crimps, just in case it damages the line. I did think about a kind of 'cleat' design, where you wrap the line back and forth around two 'projections' and then maybe glue it in place...some kind of friction approach I think is better than crimping, unless you can be really careful...

Re: the conversation about alternatives to knots/tying nylon line, I wonder if something like these PCB mount screw terminals would be strong enough to hold line in place for a long time, but also not sever it? They could be mounted either side of the nichrome coil, with holes for the nylon to come through from underneath, pulled over the nichrome and then through the screw terminal, where it can be secured in place. The metal screws could potentially be replaced with nylon to make it a little 'softer' on the line maybe? Below is an absolutely amazing diagram of the idea! Could even make a PCB for it all (or mod the current drop-off PCB to have this at the front)...the only thing is they aren't the smallest things in the world...so good for larger drop-offs maybe (certainly ok for AWDs I reckon) but not smaller critters...

I've been trying to find something off the shelf along those lines @Rob_Appleby

This is a cubesat example that's neat but not really meant to withstand the load an animal would put on it.

Seems like most tensioning examples out there are custom made but I'm convinced it can be done with off the shelf parts.

Regarding worries of damaging the line you can use ptfe tube or polyester/fep heatshrink for added abrasion resistance.

The PCB mount terminals are actually quite robust and paired with a lock washer and something that would better secure the nylon (e.g., a piece of flat grooved plastic to add compressive loads across the string and not pinch it), I think it would work nicely. 

As a separate idea, I just designed a circuit with SMD threaded standoffs, which I am going to secure in plastic and screw onto aluminum pieces that I'll be testing the ElectRelease with once it delivers. The standoffs are mounted on a circuit board, giving a 50V differential with a single-cell LiPo, hopefully releasing the epoxied aluminum plates. 

One of my peers referenced this paper and designed his own. Its literally as simple as getting a larger through-hole resistor and wrapping fishing line around it. You then subject it to more power than it is rated for, causing the resistor to get accessively hot and burning off anything wrapped around it. However, depending on the duration of heating, the resistor can turn black and release volatiles.