Friday, September 22, 2017

uBeam Funded? Or on Fumes?

It's seems uBeam have been in fundraising mode these last few months - the newspaper articles, the demos, the apparent closure of the San Jose office, and that the last known round was in summer 2015, all point to an ongoing fundraising effort. It happens to all startups - you get profitable, you get another round, or you go out of business, no way past that. As far as I know there's no product, so uBeam need to fundraise. There's been no indication of how that's been going (publicly, I've spoken to plenty of people in the VC industry, it's interesting...) - until now.

It seems that the building that uBeam use as their headquarters is going to be available for rent from January 2018. Not just some of it, it seems to be the whole building. You can see the ad here on Loopnet, and there's a "For Lease" sign out front right now.


So what could this mean? A few things stand out as the strongest possibilities - uBeam has raised (or has nearly raised) their next round and are ready for their next expansion and are preparing to move to larger premises, or they're taking their time in negotiating a renewal and the landlord is being cautious to ensure no voids, or simply they don't have the money to commit to a long lease and the landlord is seeking the next tenant.

It could be any of these, or some other innocent explanation. If it's preparing to expand they must be about to hire a lot more people, however their usual job ads haven't changed. If you look here it's the same three key positions of Lead Acoustic Engineer, Lead Systems Engineer, and Lead Hardware/Software Engineer (seems kinda key to have those people...). Perhaps they're so busy signing the next round and counting the money they forgot to put the job ads up, and with my reminder they'll do so in the next few days. All production, sales, and marketing perhaps, since they have to be ramping up to consumer sales soon. Right?

Seriously, I do expect that they will get some funding, though the mix of cheque size, valuation, and % equity is something I've never been able to make work in my head with rational numbers. But I'm just a dumb engineer, what do I know? So it's most likely that they are in discussion trying to get a round signed, that can take some time if it's large. And that would be just as well for a few reasons. If there's no deal on the table from a VC yet, knowing that the company is on a timer to get funding can weaken the negotiating position, but even more importantly can you imagine coming to work as an employee and seeing that "For Lease" sign outside? You'd worry that your job was about to disappear, and begin to look for other work, and that can kill a company. So if they're working through that paperwork, management can tell the team:

Don't worry, we're just working out the details and the next round is imminent.

Of course, if you've been around startups long enough you know that even up to the day the money runs out the employees can be being told:

Don't worry, we're just working out the details and the next round is imminent.

I don't know which it is - Funded or Fumed - but it looks like we'll know by January 2018 at the latest.

Oh, and if you're looking for office space in the area, I'd suggest looking at the place - the office, the location, and the landlord are all awesome.

Update 28th Sept: There's a possibility they may be consolidating in the old uBeam offices in San Jose. These are offices that look to have been leased in Q1 2016, around 8500 sqft for 3 people, and then closed less than a year later (see here for details, last page). Loopnet lists the office as available since Feb this year, but still available. I had thought they had been subleased, but perhaps not. If that's the case, the company has been paying likely around $40k a month for an empty office, and if the lease in Santa Monica runs out, they may be in a position where there is no choice but to move to an office that they've had sitting empty for most of a year.

If true they'll have opened an office, closed it, and opened it again all in the space of less than 2 years, having paid near $500,000 for it to sit empty in the interim.

Monday, September 18, 2017

Wireless at-distance charging may be here - and not how you were expecting

There are a number of wireless power companies out there claiming all sorts of amazing specifications of their seemingly perpetually "soon to be released" systems - 15 feet range, faster than a wire charging, 10's of devices, cheap, safe, efficient - yet few can answer detailed questions as to the physics, or the release dates, or the safety, just vague future promises that all will be revealed. One exception was Disney who went into extensive detail about their "quasistatic cavity response" method which they demonstrated worked, but clearly was utterly impractical (hence the detailed reveal).

Well, things may have changed today, with the public announcement by Pi of their charging solution. If you want to watch the demo they gave today, you can watch it here, or read some of the other press release material here, and articles from Techcrunch or DigitalTrends on it.


Above is a picture from their website as to typical usage, and it's basically a cone shaped transmitter that looks a bit larger than an Amazon Echo or Google Home in height, and looks to charge devices within a short range of it, in any orientation. If you go to the "Tech" page there you can see a representative breakdown of what's in the box. It uses magnetic resonance, controlling magnetic fields to transfer power in the same manner that Qi does - you may have heard of that recently, as Apple announced last Tuesday that this form of wireless charging will be in the iPhone 8 and iPhone X (it's already in most Samsung high end phones).

The limitation to Qi has previously been that a charging pad is required, and the phone has to be within a centimeter or so of the pad, and well aligned. This makes the phone almost unusable during that charging time, and limits how many devices you can charge to the size of your pad. 


Pi's approach seems to get around most of those limitations by being able to 'beamform' the magnetic field, their claim is that the equations to do so have been simplified so that they can be done on the processors available in consumer goods. The device itself seems to have multiple coils, at least one of which is at 90 degrees to the others, and is likely needed to be able to direct the fields as required - it also is what likely sets the cone shape of the transmitter.

They claim about 20W total power output from this device, but that it can go higher with a larger transmitter. Qi calls for 5W in its early version and 15W in the latest. Pi state up to 4 devices can charge at full rate and then it starts to slow after, which would point to them using the 5W rate. Their website says 10W per device, so what rates are used are still a little unclear. Still, in theory 5W will charge your phone in an hour, although in practice it will be longer than that - but it's a reasonable amount of power.

The downside to this device? The working distance is about 30cm, so not the huge distances we've been promised by other companies, but in terms of making the base technology better, it's a 10x improvement so it's impressive work. It's unfortunate that they may receive criticism because expectations have been set so high by companies that I don't believe will ever deliver their initial claims.

It's a usually-safe, well known technology, and unlike uBeam or Energous, Pi have actually published some aspects of their approach, such as here, so you can actually begin to analyze what they've done. It's not complete work, absolute numbers are not included, and it's not peer-reviewed, but I can follow their approach and there are no huge red flags appearing.

Now I've been critical of uBeam and Energous for 'showing a charge light coming on' or claims of charge rates incompatible with FCC regulations, so why am I not so critical of Pi? Quite simply, it's because they're not claiming anything that isn't readily believable based on the physics of what they are doing, and they are explaining what they are doing. It builds on an already proven (though limited) technology, and uses some smart ideas to move it forward and make it more practical. 

Now, I do want to see more - I want to know the efficiency under various conditions, we may find it's hideously inefficient, though I expect it's well into double digit %. An efficiency at 10 to 20% is decent for at-distance wireless, but poor compared to wired and has societal implications at-scale. I want to test safety and what happens when you start to put metals in the path randomly. I'd love to have one and test it and take it apart, but at first glance I'm not seeing any claims that aren't realistic.

Release is slated for next year, and given the demo seen, they're closer to production equivalent devices than prototype, and unlike other companies who seem to think you can go from prototype to consumer sales in weeks, this is a more realistic timeline.

So let's summarize what we have:
  • Based on a proven power transfer technology
  • Legal and to my knowledge safe
  • Using a smart method to overcome existing limitations
  • Works with inbuilt receivers in modern phones (cases for older ones)
  • A transmitter priced for consumers (under $200)
  • Multi-device charging
  • Reasonable charge rates (5W and up)
  • Can use the device in any orientation while it charges
  • Works internationally
  • Modest distance, 30cm
  • Unknown efficiency ranges (estimated below)
  • Timing near perfect to match Apple's AirPower release

It's an easy, minimal effort for a user, they don't have to change their phone if it's new enough (no need to integrate with the phone manufacturer is a huge plus), cost isn't prohibitive, and they can piggy-back on Apple's AirPower marketing. It also leverages a massive prior investment in Qi by many other companies. If this is as good as it is presented, I'm going to say that in my opinion this is as good as at-distance wireless charging is going to get for some time. You can complain about it only being 30cm, I actually don't know if it's going to be enough, but I can see it working for some office desks, hotel, and home locations. They do mention larger ranges from bigger transmitters (don't expect 15 feet though).

Oh, and apparently they've done this on just $3.5 million of investment. Tip-of-the-hat to the Pi team, and I'm looking forward to seeing more detail on this. 

I think uBeam and Energous are going to have a harder time explaining their value proposition after today.

(Also, if you sign up early enough on their website, it's $50 off the transmitter when it comes out).

Update Sept 19th: Some people have been asking why there are no efficiency numbers given. First of all, they won't be great, as in "90%+", and second it's a complex thing to explain to a lay-audience when your efficiency and charge rate are dependent on distance and orientation as it's not a single number. The paper here gives some indication, as it shows charging time (not efficiency) vs distance and orientation. If input power remains constant, and the results here are correct, you have a 50% efficiency reduction from baseline at 30cm. The orientation data doesn't make it clear what the baseline they use is, but seems to be from about the same to up to a 66% loss (take this number with a pinch of salt, that graph isn't the best). This data shows Qi as around 60% efficient as the baseline, so that would indicate Pi goes anywhere from about 60% efficient down to 10%, I'm going to hand-wave estimate a 20 to 25% in most typical use cases. Great for at-distance wireless, not good compared to a wire. (Caution, these are very hand-wavy calculations)

To put into context, imagine they get 25% market penetration on 1 billion Qi enabled phones globally. At 250 million devices, if they all charge once a day and evenly spread and given a 120 minute charge time, that's ~21 million people charging at 5 Watts, but if 25% efficient using 20W, so it's around  0.5 GW of extra global generation capacity. At around $3,500 per kW to construct that's $1.5 billion in new power stations, and at 12c /kWh around $1 million a day burned as heat.

Update Sept 21st: Just a few thoughts. The term "beam forming" is a bit much I think for having two to three coils, and it's not really a beam, it's the magnetic field. Likely "field shaping" or "field biasing". I expect they've tried to balance the number of coils and the associated electronics with cost, efficiency etc and come out with this as what they think is the right mix (<$200 is pretty compelling as a price point).

Update 1/1/18: Quite a few people are saying that Pi charging is a fraud because they've seen 6.78MHz noted for the system while Qi standard Apple uses is nearer 140 kHz. It looks to me like the founder used 6.78 MHZ Rezance standard during his thesis work and then switched to 140 kHz for this product. They clearly state "Qi standard" in articles, but don't mention the frequency in any place I can find. I'm looking for that and will post a link as soon as I can.

Update 5/26/19: Turns out it was too good to be true. It seems to have been basic Qi charging, then a higher frequency for the modest distance that needed a special case. Some updates on that here and here, but overall a lesson in digging deeper and "trust, but verify".

Final note: As every time I post something on wireless charging I get accused of writing for financial gain in trying to drive a stock price down, or pump one up, I'll once again note that I have no financial stake here, either long or short, nor am I being paid by any wireless power company. So many people find it hard to understand why I write these posts if not for money, and unfortunately there's nothing I can do to persuade them that other motivations do exist.

Thursday, September 14, 2017

Scientists Prove You Can Charge Your Phone With Ultrasound - In Just 7 Weeks!

There's been very little out there regarding ultrasonic wireless power transfer that anyone interested in the numbers can really dig into, so anyone critical of the claims of some companies has very little to point to when discussing the topic. uBeam have been extremely tight lipped when criticized, often simply claiming that no-one understands the field, or their assumptions are wrong, but never actually correcting those assumptions or providing detailed alternative numbers. While I've shown some information, as has Dave Jones of EEV Blog (among others), as a "disgruntled former employee with an axe to grind" apparently I'm not to be trusted in my analysis. Things have just changed, however. A prestigious group of ultrasound researchers at Stanford has just published a paper on "Wireless Power Transfer to Millimeter-Sized Nodes Using Airborne Ultrasound" that goes into some depth on the topic. 

It's peer-reviewed, which means other scientists read it, critiqued, and have judged it meets a standard that it is a novel or major contribution to the field, and there are no significant mistakes they can find in the work. The publication is the IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control, which is the highest ranked journal related to ultrasound (by Impact Factor). This is going to be hard to dismiss as partisan, biased, incorrect, with false assumptions, or incorrect numbers.

Even better, the authors have made it "Open Access" which means that anyone can download it without being a journal subscriber or paying for it. Please, if you're at all inclined, download it from the above link or at the one below to make sure the authors get credit for their great work.


I'll highlight a few points here from the abstract as a taster for you:

We propose the use of airborne ultrasound for wireless power transfer to mm-sized nodes... We show through simulation that ultrasonic power transfer can deliver 50μW to a mm-sized node 0.88m away from a ~50 kHz, 25 cm2 transmitter array... We also argue that longrange wireless charging at the watt level is extremely challenging with existing technology and regulations... 

To translate for everyone that's 50 microWatts from a 25cm2 panel, out to around 1 meter. Scaled to my estimate of uBeam's 45 by 45 cm transmitter that's around 4 milliWatts. A typical phone battery is around 5Wh so you're looking at 1250 hours to charge the phone with this method, if it were switched off. 

So there you go, proof that uBeam can work and charge your (switched off) phone in no more than 7 weeks! (Are there ways to go faster, yes, but 1250 times faster? While I'm painting closer to the worst-case picture, if you look at some of my earlier posts you can find different numbers that change the outcome to a higher number, but is it safe, efficient, practical, legal? It's now really up to uBeam to show.)

It's a well written paper, please read it if you have any technical background. For those of you who don't, they make a pretty reasonable statement that for "Internet of Things" (small devices that need very low power charging very intermittently) it's a possible solution, but phone scale devices (or larger) are unlikely to be practical. Further, they point out this is contingent on the 145 dB safety levels that the USA used to allow, but appear to not allow any longer. At ~115dB, the level almost every other country has always given as a limit, things get 1000x worse than this.

This is not the first time recently that peer-reviewed papers have called into question the use of high power ultrasound. Last year, the Proceedings of the Royal Society published a study of the potential negative health effects of the increasing use of high power air coupled ultrasound in our environment.

Over to you uBeam, it will be very interesting to see your response to this...