Ossia and FCC Approval

Earlier today (Thu 27th) the FCC published that Ossia, the RF at-distance wireless power company, had been granted Part 18 approval for their charging system. This has many similarities to the Energous system that got similar approval around 18 months ago, but also a number of differences. The first similarity is that it seems a lot of cost and effort to deliver very little power across an annoyingly short distance though in a more elegant manner than Energous. If you want to look up the data for yourself, you can go the the FCC OET Authorization Search page and enter "Ossia" under Applicant Name. Ossia are clear in its intended application:

The Cota power system is intended to power sensors, actuators, small displays and other devices in a commercial or industrial environment for which changing batteries and/or connecting wires is impractical.

So no charging your phone, or any other significant electronics, and also it seems to be for fixed receivers, not mobile.

This is a large area transmitter, around 0.36m^2 (imagine a ceiling tile), and outputting 5 Watts, not that much more than a cellphone or a wireless router. For comparison, at the claimed 145 dB SPL uBeam would be emitting 100 Watts acoustic from the same area panel.

Having had a little time to read the 100+ pages of data, I'll summarize the results:

• Uses 2.45GHz band (12.25cm wavelength), compared to ~900 MHz used by Energous (33cm)
• Tries to be tightly +/-10MHz, to limit interference with WiFi?
• A 16 by 16 antenna array (256 total), compared to the linear 12 antenna used by Energous
• A 60 by 60cm ceiling tile size transmitter, compared to a flat bar for Energous
• Means a 0.3 wavelength spacing, so good steering possible
• A maximum of 5 Watts emitted, compared to 10 Watts max for Energous
• Claiming 1 Watt delivered, unclear if to the receiver or battery, possibly 500 mW to 1 Watt actual, max. Energous in the 30mW to 150 mW range.
• Limited by a SAR of ~1.42 W/kg, compared to ~0.97 W/kg for Energous (FCC limit is 1.6)
• No increasing power from here
• Max range 1 meter, compared to 90 cm for Energous
• Creates a more confined beam, better steering, but no "hotspot"
• They claim to be able to steer around objects, some viability to that claim with a large phased array and test results 
• A 15 by 15cm receiver with unspecified output
• Ossia website says "meaningful power at 1 meter" but no numbers given
• No indication of a "keep out zone" that Energous 
• Ossia website claims the phased array steering means it's not needed
• Fans on the transmitter, possibly for compute/ADC heat phase-shift electronics
• If 30+ Watts for electronics, efficiency is likely ~1.5 to 3.0%
• ~10 mW CW output from the receiver, for location
• No consumer use allowed, professional installation only and a 20cm offset from people from transmitter and receiver
• I cannot see a significant business case for this system, even at OK power the range and stationary nature limits it
• No personalized message of congratulations from Ajit Pai, FCC Chair

So the summary seems to be that Ossia have a larger 2D phased array at a higher frequency and so can control the beam better than Energous, transmit half the power, to a fixed receiver the size of two phones, needs professional installation and never closer than 20cm to people, barely enough power to charge a phone, but potentially a very low efficiency. They also cannot increase the power beyond where they are without breaching the SAR safety limit (I'd guess they scaled down to the 20mW per antenna to get under the SAR limit, and some indication they started at 40 mW per antenna). The worst case seems to be immediately behind the receiver, on page 94 of the RF Exposure report Part 2.

Front of the Ossia Cota Transmitter Panel

Without impedance information it's hard to calculate what the power is even if we assume a half-wave dipole and multiple antenna, and know the field strengths. Obviously it's less than 5W received, they claim "meaningful power at 1 meter" but no numbers given, and no use cases such as "charge your phone" so I would guess it's starting to confirm the low power number in the high 100s of mW. (Update: This article claims around 1 Watt delivered so I'll guess it's somewhere from 500 mW to 1 Watt at the battery, max under ideal conditions) If it were much greater than a Watt I think they'd be saying that very loudly, so ~10 to 20% efficient on receive (emitted to battery and *not* overall system efficiency, which is wall socket to battery). (These are estimates, I will reserve the right to update later! Which I did...) I'd be interested to see how WiFi and Bluetooth work around this, even if they have restricted to one or two of the channels.

Back of the Ossia Cota Transmitter Panel

The receiver outputs a signal at a constant 9dBm, which is <10 mW. This is the signal that's used to determine location. Now that signal I expect gets received at up to 256 channels on the transmitter, sampled (assuming the same number as transmit), so Analog to Digital Conversion (ADC). Sampling will likely be in the high GHz to get the phase accuracy needed. You can look at ADCs in the 10GHz+ range, such as this AD913 part, and they are going to consume (say) 5 Watts per channel. If sensing/location duty cycle is 100% then that's 1.25 kW of heat just from the ADCs, and so fans will be needed. Wording in the report indicates they do this 50 times a second, so I expect a lower duty cycle, let's guess 10%? Even at 100 Watts for this with simplifications and good engineering, the system is using 105 Watts to charge ~1 Watt, so overall end-to-end efficiency is <1% which is not good. (This is not certain it's what they do, but given what I'd guess the phase delay precision requirements are to make this work, it's the most obvious solution).

Update 29th June 19. I'm glad I left the caveat on different ways of doing things, as a conversation with a colleague who has a lot of narrowband phased array radar experience covered alternative ways of doing this. My expertise is most heavily in broadband ultrasound for medical, and applications of using time reversal have been really precise in sending back the exact same signal, not just phase shifting a CW train. It's a reminder that general engineering expertise gets you so far, but specialized knowledge is invaluable. If anyone from Ossia was reading it they were probably rolling their eyes :). Very quickly, they would expect that they use phase shifters rather than ADCs, and that they'd expect the electronics to be around 30+ Watts to do this. With distributed electronics, it's just easier to put a fan on the back rather than heatsink everything. They also estimated you'd manage to contain about 80% of the power into the beam, would have around 75% loss of power in the 1 meter propagation path, and then around 50% receive efficiency, so you'd see about 1 Watt RF at the receiver, and about 500 mW converted to usable power. Assume they do a good job, that's in line with the 500 mW to 1 Watt estimate. That would put it at around 35 Watts in to get between 0.5 and 1 Watt out, so 1.5 to 3% overall end-to-end efficiency.

The compute on this to work out how to send the signal back should be pretty minimal. As an alternative to this system (barring, of course, a length of wire) I note the receiver is about a 15 x 15 x 3cm block so about 775 cm^3. At 500 Wh per L, that means around 390 Wh in a Li-ion battery that size, so anywhere from 2 to 4 weeks runtime at 0.5 to 1 Watt requirement.

Why limit to 1 meter? Well with a panel 60cm across, and 12.25cm wavelength, the near/far field boundary is at around 75cm, so it gets hard to have a bounded energy region beyond about 1 meter. That may be a consideration here on the regulatory side.

The report also makes it seem like the phased array nature of the array is used not to track a moving receiver, but to act as the safety system and try to route the signal around obstructions. Both the transmitter and receiver are likely fixed. How this is then useful in most circumstances is very questionable.

I'd be interested to compare this to a modern multi-antenna wifi router with MIMO like the ASUS AC5300 with the firmware hacked to up the power transmitted and see what gets received. Might be a fun 1 to 1 comparison.

What Ossia Want the Public to Think Their Tech Is

Most of the press articles on this are pretty light reading, simply regurgitating the talking points handed to them by Ossia. In the 18 months since Energous manipulated them, the press still fall for this approach, and it remains as frustrating as ever. Interestingly, what Ossia wants them to talk about is not this system, but mythical future ones, that what they have here is "a critical first step". It's like they are embarrassed by it, even the pictures they've given out are for non-existent desktop products. This is a shame as I think the engineering team has done as good a job as could be expected, this data shows they are ultimately limited by laws of physics and regulatory safety rules - ultimately I think this is as good as it gets. It's a successful approach though, they get the press talking about phone charging even in the headlines - something that's hard to do when you can't be within 20cm of your phone... Read for yourself the similarity in articles from IEEE Spectrum, Tom's Hardware, Tom's Guide, and VentureBeat.

What's the importance of this approval? The nominal use case of charging sensors in an industrial environment IMO doesn't hold up when everything is fixed and <1m distant, a wire or large battery does the job cheaper and better. It seems to me about as important as Energous' equivalent approval from 18 months ago, good for marketing/smoke and mirrors, but meaningless if you want to charge anything.

As far as I can see, no-one has beaten Powercast yet for just getting the job done without hype.

Note: I posted a version of this article earlier, but wrote it late at night and realized when I got up I had made some mistakes, and didn't have time to correct, so took the article down for a few hours until I had a chance to fix. So if you think you saw it and then disappear, you are correct.