EEVblog #945 - Thermal Powered Smartwatches Are GIMMICKS

[EEVblog]

Thermal powered smartwates are nothing more than gimmicks.
Dave runs the numbers on the Matrix Powerwatch from Indieigogo.
Also some exercise testing shows up an unusual and undesired effect when using thermoelectric generators on your wrist.

Bluetooth Power Consumption Application Note: http://www.ti.com/lit/an/swra347a/swra347a.pdf
The Peltier effect TEG used in the experiment: http://www.cui.com/product/resource/cp85.pdf
The Wurth Energy Harvesting Kit:
http://www.we-online.com/web/en/electronic_components/produkte_pb/demoboards/energy_harvesting/energy_harvesting.php




Thunderf00t's video:

[rs20]

I wonder if using a 100 ohm load was a mile away from the correct load to use to achieve maximum power (I assume a TEG is much like a solar panel, in that choosing a random resistor risks wasting a lot of potential power). However, your point was clearly to show relative comparisons between different situations, and your points were clearly proven nevertheless.

Still might have been interesting to pass it through the demo board, which presumably performs MPPT for you, and would therefore presumably produce a similar amount of output power over a relatively wide range of load resistances?

[EEVblog]

Still might have been interesting to pass it through the demo board, which presumably performs MPPT for you, and would therefore presumably produce a similar amount of output power over a relatively wide range of load resistances?

Yeah, probably should have done that, as it would have included typical converter efficiency as well. But I don't think there is much in it IIRC last time I experimented with this.

[EEVblog]

I wonder if using a 100 ohm load was a mile away from the correct load to use to achieve maximum power (I assume a TEG is much like a solar panel, in that choosing a random resistor risks wasting a lot of potential power).

The datasheets for these do not show any power generation data, let alone at the lower end. So you'd have to characterise them fully to get any data in this respect.
Given that Matrix say it "generates more than it needs", and they have basically confirmed around 20uW, consumption, I think the 50uW figure I picked is going to be not far off the mark.
Also given that 500uW is the maximum exercise figure they quote is best case condition, that's further evidence for an order of magnitude lower resting output in average conditions.

[Fungus]

 

My CASIO runs on sunshine. That's also free and natural. I expect it will last longer than me.

[Jay_Diddy_B]

Hi group,

In the video Dave talks about the time to charge a flat battery as 2.6 years. The opposite is also true, if they ship the watch to backers with a charged battery, it will run for a while.

740mWh in the battery / 20uW = 37000 hours

4.2 years, too late for a refund.

Kind of like the LED flash lights that you shake with a lithium cell in them.

Regards,

Jay_Diddy_B

[Fungus]

Hi group,

In the video Dave talks about the time to charge a flat battery as 2.6 years. The opposite is also true, if they ship the watch to backers with a charged battery, it will run for a while.

You're all completely missing the point of this.

The people who buy this are obsessed with burning calories. This watch forces them to burn a minimum of 5000 calories per day, just to keep the damn thing running. Who'd want the shame of admitting they didn't do enough exercise to keep a watch running?

[EEVblog]

The people who buy this are obsessed with burning calories. This watch forces them to burn a minimum of 5000 calories per day, just to keep the damn thing running. Who'd want the shame of admitting they didn't do enough exercise to keep a watch running?

 

[timb]

I wonder if using a 100 ohm load was a mile away from the correct load to use to achieve maximum power (I assume a TEG is much like a solar panel, in that choosing a random resistor risks wasting a lot of potential power).

The datasheets for these do not show any power generation data, let alone at the lower end. So you'd have to characterise them fully to get any data in this respect.
Given that Matrix say it "generates more than it needs", and they have basically confirmed around 20uW, consumption, I think the 50uW figure I picked is going to be not far off the mark.
Also given that 500uW is the maximum exercise figure they quote is best case condition, that's further evidence for an order of magnitude lower resting output in average conditions.

There are actually a few TEG companies that have products designed specifically for power generation. They provide characterization in the datasheets.

I can't remember the companies off the top of my head, but I can dig them up if you're interested. They're mentioned in some LT app notes/datasheets on energy harvesting.

[nuclearcat]

I believe calculating mwh for their battery as 200mAh * 3.7V is not very correct, because lithium battery voltage drops to 2.8V at end of discharge, and at beginning it is ~4.2V. But it is hard to find, what are curve of lipo discharge at such low discharge rates, looking to 0.2C capacity and approximating, capacity might be even higher than expected! BLE will draw almost 0.5C, which is reducing capacity, but i believe it wont be significant difference, because LiPo is quite good with high discharge currents. So approx it might be correct, but interesting to know what are more precise values.
Also for LiPo self-discharge is 5%, so we need to compensate approx 37mWh monthly(1.23mWh/daily), but if as stated it generates only 1.25mWh/day, we just getting very close to be out of juice, because of self-discharge!
Another thing, for example CR2032  with 3.0V fully charged to 2.0V end of discharge, with their 230mah, is just 645mW(dave took 600mW, which is very real). If we calculate as with LiPo, with 3.0V it should be 690mW, but seems real "average" voltage on discharge curve is 2.83V. It can be seen on discharge curve:

P.S. I had really serious issues with CR2032, because at 1beacon/sec rate with BLE, with such battery intolerance to high currents, it lasts much less than expected, because battery vendors specify 230mah for very low discharge current.

[rollatorwieltje]

I wonder if using a 100 ohm load was a mile away from the correct load to use to achieve maximum power (I assume a TEG is much like a solar panel, in that choosing a random resistor risks wasting a lot of potential power).

The datasheets for these do not show any power generation data, let alone at the lower end. So you'd have to characterise them fully to get any data in this respect.
Given that Matrix say it "generates more than it needs", and they have basically confirmed around 20uW, consumption, I think the 50uW figure I picked is going to be not far off the mark.
Also given that 500uW is the maximum exercise figure they quote is best case condition, that's further evidence for an order of magnitude lower resting output in average conditions.

There are actually a few TEG companies that have products designed specifically for power generation. They provide characterization in the datasheets.

I can't remember the companies off the top of my head, but I can dig them up if you're interested. They're mentioned in some LT app notes/datasheets on energy harvesting.

Micropelt produces TEG modules for applications like this. Wouldn't be surprised if this is what Matrix is using as well. http://micropelt.com/thermogenerator.php

[EEVblog]

Micropelt produces TEG modules for applications like this. Wouldn't be surprised if this is what Matrix is using as well. http://micropelt.com/thermogenerator.php

They are TEG researchers, and IIRC they claim to have developed their own?

[rollatorwieltje]

Micropelt produces TEG modules for applications like this. Wouldn't be surprised if this is what Matrix is using as well. http://micropelt.com/thermogenerator.php

They are TEG researchers, and IIRC they claim to have developed their own?

I was under the impression they only made lab-scale prototypes, but looking at this presentation they indeed might actually be able to produce something themselves.

[derGoldstein]

Timeline in reverse order:

3.4 years: a CR2032 would run out
3 years: warranty runs out (a generous estimate)
2.5 years: mechanical failure due to rough use conditions (if you're in the gym this thing will bump into things)
2 years: they stop updating the iOS/android apps for this version of the product
1.5 years: they come out with the new version of this product (nobody leaves money on the table -- if this is profitable they're going to make  another one)
6 months: the hipters that bought this watch will have found some other gimmick they want to put on their wrist.

It will probably work as advertised, but why not be honest with your sales pitch? This is an initial version that will mostly work as a proof of concept. Having a large userbase will give them the feedback (as well as the money) they need to keep working on future products. This particular product might not be the *best* use for this technology, but it is a functioning one that will do the job. You get a cool gadget to wear and show off, and you've helped them with their research.

Maybe I'm naive, but I think that if they did this they'd still make their monetary goal, as well as giving people confidence that they're being 100% honest with their goals.

[ChunkyPastaSauce]

Micropelt produces TEG modules for applications like this. Wouldn't be surprised if this is what Matrix is using as well. http://micropelt.com/thermogenerator.php

They are TEG researchers, and IIRC they claim to have developed their own?

I was under the impression they only made lab-scale prototypes, but looking at this presentation they indeed might actually be able to produce something themselves.

Seems to based on this work
http://www.wag.caltech.edu/home/jamil/Nature-SiNWs-TE.pdf
http://www.wag.caltech.edu/home/jamil/Nature-SiNWs-TE-supplement.pdf
Looks to be pretty major based on citations (cited ~2000 times), and where it was published (nature)
Primary researcher was Boukai (who is also CEO for matrix / siliciumenergy)

Youtube video of him talking, basically said:
Current low temperature decent performing TEG devices are fundamentally expensive due to materials
~2008yr his research team was able to make a decent performing TEG device using cheap si material (papers above).... but process was expensive
So since then has been working on making the process cheaper / scalable.....completed, forming silicium energy
Say extremely low power / high function device availability growing, tipping point where energy recovery/harvesting makes sense...so currently targeting

[mux]

Some background on how much better their TEG could theoretically be:

TEG efficiency is basically completely dependent on the temperature difference and the 'figure of merit' which is material dependent. More detailed info: https://en.wikipedia.org/wiki/Thermoelectric_materials

Theoretically, an infinite FOM TEG would have the Carnot efficiency limit: eta = (Th-Tc)/Tc, where Th = hot side absolute temperature and Tc = cold side temperature. The figure of merit reduces this efficiency by (sqrt(1+FOM)-1)/((sqrt(1+FOM)+Tc/Th). The larger the FOM, the more this quotient tends to 1.

Bismuth chalcogenides, the material probably in Dave's Peltier module, has a FOM of 1, so its relative efficiency is 0.17. The best commercially available materials, PbTe, has a FOM of 2.2, which has a relative efficiency of 0.28, so about 1.6x better. Theoretically, graphene and other hype materials can get FOMs of over 20, which would have a relative efficiency of 0.65, or about 4x better than a jelly bean peltier. And that is pretty much the theoretical limit as well.

Even if these people manage to build something better than anyone else has built, it will only provide about 4x the power output of a run-of-the-mill peltier device. And let's be honest: that watch almost certainly has only 1/4 of the surface area, i.e. all those gains are immediately lost.

Edit: the quoted papers above only have a ZT (figure of merit, FOM) of about 1, which is identical to bismuth chalcogenides/Dave's experiment.

[2N3055]

That is pretty much what  Mr. Boukai is saying.. He claims efficiency close to bismuth telluride, but using cheap silicon...  That would make is cheap to use.. Not magically efficient..
And very good for energy harvesting in some scenarios.....

[ChunkyPastaSauce]

Edit: the quoted papers above only have a ZT (figure of merit, FOM) of about 1, which is identical to bismuth chalcogenides/Dave's experiment.

Primary target (I think) looks to be reducing the cost of low temp TEG units by switching materials to Si.
Bi and Te are both expensive.... if they can get similar or better performance in Si, seems possibly significant. 

I don't think their main business is the watch, it's the TEG. The watch gives them some funding, demo/exposure for their TEG, and some sales (vc money likes).

[EEVblog]

It will probably work as advertised, but why not be honest with your sales pitch? This is an initial version that will mostly work as a proof of concept. Having a large userbase will give them the feedback (as well as the money) they need to keep working on future products. This particular product might not be the *best* use for this technology, but it is a functioning one that will do the job. You get a cool gadget to wear and show off, and you've helped them with their research.

Maybe I'm naive, but I think that if they did this they'd still make their monetary goal, as well as giving people confidence that they're being 100% honest with their goals.

I think they could have made this more honest approach work.
People like cool stuff, and they like contributing to research.

[EEVblog]

Even if these people manage to build something better than anyone else has built, it will only provide about 4x the power output of a run-of-the-mill peltier device. And let's be honest: that watch almost certainly has only 1/4 of the surface area, i.e. all those gains are immediately lost.

Yes, the area and radiator is going to be small and inefficient.

[iromero]

One thing that killed any credibility Arrow's "certification" program could have was this newsletter about one of their first certified projects:

If anything it could be used as evidence that projects with Arrow certification are certified muddy business....

[rs20]

One thing that killed any credibility Arrow's "certification" program could have was this newsletter about one of their first certified projects:

Oh geez my head is hurting. This is bullshit-ception. Part A: A "certification" program that only ostensibly checks manufacturability rather than the claims being made. Part B: Solar Roadways, need I say any more. And then this newsletter, which makes the completely provable false claim that B was successful because of A. Part A was announced on May 20, 2016, nearly two years after the Solar Roadways main campaign finished. And why did half of the "Arrow-certified" SR panels fail?

So what we have here is a bullshit program being advertised using a bullshit logical fallacy about a bullshit idea. I really don't want to go deeper, someone please make it stop!

[rs20]

Something just crossed my mind. This watch seems to fill a really interesting niche, or at least it would if they replaced the TEG with a bigger battery. I mean, a phone-connected, bluetooth, sunlight-readable always-on watch with a 1 year battery life? Sounds brilliant for certain usecases.

Even a Pebble only has a ~1 week battery life, right?

I'm not trying to defend the TEG watch btw, just thinking I might want everything about this watch except the TEG. (Setting aside the fact that the current prototype doesn't actually display notifications).

So does such a thing exist on the market right now? A no, please don't reply with "my bluetoothless Casio has been working for 10 years". I mean a bluetooth-enabled, phone-notification-displaying watch with > 1 year battery life.

[Mr.TalkingMachine]

Like it was said, this thing could work but it's pointless. By using this power source, they had to cut many features of a regular smartwatch.

If the deal is that you power it by wearing it, then you'd rather stuff this tech in a simple watch, and then you would not, since there are plenty of better ways to power one. Like those that exploit the waving motion of the hand while walking to either wind a mechanical movement  or charge a battery like Seiko's kinetic watches.

It's an overly complicated solution to a problem that is probably not that much of a big deal. They mentioned in that video that tons of batteries are discarded, and modern society consumes enormous amounts of energy, but it kinda goes into a fallacy. What fraction of this impact is due to the sodding smartwatches? no idea. For what I've seen, at least where I live the fraction of people who use smartwatches is small enough already.

And I don't even get the point of their existence, anyway. Most of their functions become redundant if you happen to have a cellphone that isn't more than a decade old, like most people. And if you don't have one, I think it would be priority to get a proper phone than a smartwatch.

I have a timepiece that I power myself, it's a mechanical watch (about as old as my grandparents). And I don't even bother if I forget to wind it, because I have a smartphone anyway.

Below: Eco-Friendly timepiece, 1941.

[rs20]

And I don't even get the point of their existence, anyway. Most of their functions become redundant if you happen to have a cellphone that isn't more than a decade old, like most people. And if you don't have one, I think it would be priority to get a proper phone than a smartwatch.

You've missed the point. Even if you have an infinitely capable phone, the fact that you need to get the phone out of your pocket to use it adds a lot of friction. I mean, your phone can display the time, right? So why do you need a watch at all? And if you say that you want to see the time more conveniently, why does that argument not apply to other things, like phone call/message notifications and driving directions?

I was fortunate enough to be given a smartwatch, as I was sort of skeptical about the concept too. But I can leave my phone on silent now and never miss a phone call; whereas before there was a constant struggle between the phone going off during meetings or me missing calls when I was on my bike. Being able to see the next turn on your wrist while using driving directions, all while saving the phone's battery by not using its screen and while not needing to lug around a phone-holder to every car you hire, is another tiny but slightly-life-changing advantage.

It's a luxury for sure, but it's not at all self-evidently pointless as you seem to make it out to be.