Qi /Qi2 overheating

[artag]

I'm developing a product which has body-worn wireless sensors. It's very convenient to use wireless charging for these as it's possible to encapsulate them in a sealing package with no need for connectors (the sensor design is such that it needs to be airtight, not just showerproof).

Previously, I was using Qi receivers intended as aftermarket upgrades for phones, and low cost wireless power transmitters. These worked pretty well - I only need a small battery so they charge quickly. The only irritating factor was that the rx/tx distance needed to be short and the devices well aligned, or they didn't charge.

Recently, Qi2 chargers have become very widely and cheaply available, and they tend to be the default. I have not yet moved away from the aftermarket receivers although I probably will do as they're quite big and smaller options are becoming readily
available. But my typical setup at the moment is a Qi2 transmitter and a Qi receiver having no magsafe-style alignment magnet
.

I am seeing problems with the sensors overheating whilst on charge, with the transmitter current consumption going up to 2A even though the battery current is only 100mA. The inefficiencies of the wireless coupling are such that I'd expect the transmitter to take no more than 250mA.

I believe the problem is that the Qi2 transmitters are capable of putting out more power to compensate for suboptimal coupling, but when they do this they get hot. This then heats up the sensor. I haven't yet proved that the sensor doesn't generate heat or overcharges the battery but  I don't think so. The mechanism seems to be that the transmitter grills the sensor.

It's also possible that the reason for the poor coupling is that some other element of the system is absorbing power from the magnetic coupling. Quite early wireless charging designs used smart power control to avoid just building an inductive heater but compensating for poor coupling that is really due to an adjacent 'shorted turn' may be an issue.

Trying to find out how to stop this lipo-frying effect, I find many people complaining of their phones overheating due to wireless charging. It's apparently a growing problem and may point to a screwup by Qi or common implementations that get it badly wrong. These seem to particularly involve built-in chargers from large automotive manufacturers. I can imagine it becoming a very visible problem when it causes a car fire.

Is anybody else struggling with this problem ? I'd like to find app notes, recommendations etc. Perhaps a thermal cutout that stops the receiver maintaining transmission if it gets too hot. Maybe modern receivers have this built in whilst older Qi receivers don't. If so, the backwards compatibility has a rather major flaw.

 

[moffy]

If you put just one of your Lipo batteries on top of the Qi2 transmitter with nothing else attached to the battery, does it heat up when the Qi2 charger is on i.e. does it get hotter than the charger?

[NiHaoMike]

Can you disassemble the device and then place it on the charger? Try to figure out where the heat is coming from.

These seem to particularly involve built-in chargers from large automotive manufacturers. I can imagine it becoming a very visible problem when it causes a car fire.

If wireless EV charging ever becomes a thing, there definitely would be sensors to detect abnormally high temperatures, that's exactly what I have seen on a proof of concept demo unit. But the relatively poor efficiency means it's unlikely to be used while parked, a robotic arm would be just as "wireless" as far as the user is concerned, have the efficiency of regular wired chargers, and be cheaper to implement. Using while driving would be a much better use case for wireless charging, but I don't see that becoming practical in the foreseeable future. Tracking the vehicles would be a bit of a challenge to implement, especially with the need to keep the cost down to a practical level. Energized lines or rails similar to what trains use would be far more likely to be practical.

[jbb]

I can’t get too specific but I have some domain experience.

On the general heating problem: yes, there are devices out there which cut back the power flow to limit temperatures

Some basic questions:
- do you have any permanent magnets near the PTx or PRx coils? They tend to make ferrites saturate, which can cause excessive current flows in the coils
- are you using a certified Qi transmitter? There are allegedly an off-off-brand units out there which don’t behave right, don’t implement safety features etc.
- how much control do you have over the design of your receiver?

Some more advanced questions:
- is your receiver getting up to and regulating at the desired bus voltage?
- does the problem disappear (or get better) if you carefully align the receiver coil for best coupling?
- does your product have any ferrous material (eg steel) in it that isn’t related to the power receiver?

[artag]

Thanks for the input. I have a bunch of tests I know I need to do but was looking for initial directions from people who may have investigated already.

I have complete control over the receiver design but am trying to use off-the-shelf coils+controllers at the moment as they're cheaper when in low volumes. I am, however, moving from the 'vinyl-sandwich with USB' type to 'coil+pcb', and these seem to use newer designs. At least, they have an awful lot less passives. Maybe I should add some pics to illustrate that better. It's the older ones I'm seeing the problem with - but I haven't tried the new ones yet.

I want to be able to use off-the-shelf transmitters, partly to avoid another custom manufactured accessory but also because I expect the end users to use any convenient charger even though I will supply a tested one.

I have a selection of transmitters. Most are generic unbranded but of designs comon on ebay/aliexpress. A couple are branded but from phone manufacturers (both Samsung, I think). I have seen the problem on at least two types. I also have a presumed genuine apple transmitter (packaging clearly marked apple and bought from a very large supermarket chain with only a small discount). Fakes are possible but the only more certain source would be an actual apple store.

There are no Rx magnets at present but I plan to use an apple-like magnet ring to help with alignment. The first problem I saw was indeed with a no-name magnet-ring transmitter but I'm pretty sure the next was with a non-magnet type (though it was dual-pad). This adds to my suspicion that the Tx is increasing power to compensate for poor coupling without limiting the temperature rise. It's possible they rely on the receiver to disable itself which is probable on phones but not with the generic adapters. There is some steel but not large : a bmp280 pressure sensor and the shielding can over an nrf52840 bluetooth module. They might be nickel plated or tinned copper, I haven't checked for that.

NiHaoMike : Although the reports are from people with Teslas and other, I should make clear that this is nothing to do with EVs. It's a Qi phone charging pad fitted to the car by the manufacturer. Presumably a bought-in item but still capable of doing reputational damage if they set fire to a phone in a car.

[jbb]

I think photos could help.

The alignment magnetic rings are a bit of a challenge right now. Technically, you’re not allowed alignment magnets near the Tx cool under the Baseline Power Profile (BPP) and Extended (or is it enhanced?) Power Profile (EPP). The new Magnetic Power Profile - which is derived from the Apple MagSafe gear - is the only one technically allowed to have magnets.

As the PRx designer, you are allowed to reduce consumption whenever you like - ie if you’re getting too hot.

How do your Control Error (CE) packet values look. Do you reach a steady state with CE ≈ 0?

[artag]

I don't have any visibility of the protocol, I'm just using off the shelf modules that take 5V in and produce 5v out.

I'd very much like to capture the control packets though. How is this done ?

[jbb]

Well, the ‘proper’ way is to drop >$100,000 on the official test gear from Nok9 or Granite River Labs. Good luck with that purchase order :-)

I think you can get a ‘sniffer’ for the Comms that you just place right next to the Tx / Rd coils. Not sure where you’d get one, sorry.

[NiHaoMike]

Sigrok apparently supports decoding Qi, maybe one of those cheap mixed signal USB logic analyzers plus a loop of wire will be able to sniff what's going on.

[Marco]

The demodulation doesn't seem trivial, so just interpreting the protocol doesn't get you much.

[NiHaoMike]

The demodulation doesn't seem trivial, so just interpreting the protocol doesn't get you much.

Isn't it FSK with the frequency shifting done by switching capacitors to change the tuning of the coils? Should be possible to demodulate by recording the waveform and processing it in GNUradio.

[Jeroen3]

You are using the transmitters with a suitable 9V USB-C power brick?

[Marco]

What's the chip on the receiver? Qi 2 transmitters solutions have to be certified thanks to Apple, so there's some minimum quality to expect. The older receivers less so.

[jbb]

The Qi Comms are different depending on the direction.

Comms from Power Receiver (PRx) to Power Transmitter (PTx) are (approximately) Amplitude Shift Keyed (ASK). This is often done by switching a small (some tens of nF) in and out of the AC tuned circuit. You can check out the Communications Physical Layer of the Qi specs for more info.

Comms from PRx to PTx are mandatory.

Comms from PTx to PRx are done using Frequency Shift Keying (FSK); the PTx deliberately wiggles the inverter frequency up and down. Note that some PTxs employ spread spectrum frequency dithering to help pass EMC emissions tests; this can complicate FSK receiver design.

Comms from PTx to PRx are not required for Qi Baseline Power Profile, which is limited to 5 W power transfer.

[artag]

You are using the transmitters with a suitable 9V USB-C power brick?

Yes. But the problem I'm seeing is too much power delivered (possibly not right to the battery), not too little. I haven't actually checked what PD is delivering, that's something to add. Is 9V required ? If the transmitter requires it but PD doesn't offer it, will it shut down, do the best it can at lower voltage, or overheat ?

I can control these variables for my testing but I can't control them in the field. I need to provide a receiver solution that won't overheat even if the transmitter isn't Qi2 compatible. If that means I can't use generic unbranded receivers that's fine for me but not really viable as a safety strategy for Qi.
 

[artag]

What's the chip on the receiver? Qi 2 transmitters solutions have to be certified thanks to Apple, so there's some minimum quality to expect. The older receivers less so.

I have several types that I will identify in testing. However being certified by apple only helps me supply a suitable part. It doesn't control what the customer uses. I'm trying to find a solution that is always safe.

The situation I have at the moment is that I have older receivers that were sold as add-ons for non-Qi phones and unbranded transmitters of various types. I've started to see problems only recently and I would say this might be linked with using specific transmitters or receivers. I need to know what I can do to prevent this.

But I'm also curious. Can an off-the-shelf collection of parts assembled using consumer interfaces such as USB charging plugs result in fiery death ?

[jbb]

What's the chip on the receiver? Qi 2 transmitters solutions have to be certified thanks to Apple, so there's some minimum quality to expect. The older receivers less so.

At risk of being pedantic... Qi 2.0 MPP transmitters must implement authentication using the Wireless Power Consortium (WPC) crypto certificates.  And the certification is done by the WPC, not by Apple.

[jbb]

...
I'm trying to find a solution that is always safe.
...

But I'm also curious. Can an off-the-shelf collection of parts assembled using consumer interfaces such as USB charging plugs result in fiery death ?

Well, there isn't really an 'always' safe option.  I guess the closest you could get would be to implement a Qi 2.0 MPP Power Receiver (you want MPP for the attach magnets, right?) which will only work with a properly certified Power Transmitter using secure cryptographic authentication.  That would probably be a complete pain...

As to off-the-shelf death traps: yes. Or at least watery death is possible.  In this case I expect it was a knock-off phone charger that didn't properly isolate the mains from the output (because that costs some money, see).  Arguably, whichever store she got it at shouldn't have sold it... but how is a random shopkeeper expected to evaluate the safety of a product (especially when people will just print fraudulent approval information onto the product)?

[Marco]

And the certification is done by the WPC, not by Apple.

I said thanks to Apple, not by Apple. I'm assuming Apple agreeing to compatibility came with strings attached.

[jbb]

Oh yes, so you did. Sorry.

[jbb]

Back to reading the PRx messages…

I’m sure it’s technically possible to put a sniffer coil right next to the PRx, but I don’t know of a convenient product for it.

How much access do you have to the innards of the PRx module(s) you’re trying out? I can think of 3 good places to tap for comms:

• If you’re really lucky, the comms are done by switching capacitors to ground using discrete MOSFETs. Getting a probe connected to the gate of such MOSFETs would give a really clean comms signal.
• If the comms switches are integrated into the receiver IC, simply connecting an oscilloscope probe (use a 10x probe!) to the switched leg of the capacitor will yield a pretty clear comms signal, but it will include lots of AC switching edges
• Attaching an oscilloscope probe to the PRx DC rectifier output may yield a useable signal; as the comms switches on and off it generally perturbs the DC rectifier output a bit. Setting your scope to ‘AC coupling’ will remove the DC component and could help with readout

[artag]

I can easily strip them down, so if the useful testpoints are outside the control chip it's not a problem.

Not sure I'm looking for a protocol violation though.  Will a Qi2 transmitter ramp up power if it detects a poor connection to a Qi receiver? Or is that something it only does with Qi2 receivers ?

[jbb]

A Qi transmitter listens for Control Error Packets and adjusts its operating point.

I could have a sign error in the following…

If the Control Error (CE) value is negative (or maybe positive, I could have a sign error), this tells the Power Transmitter (PTx) to turn up its power output. This could be done by adjusting the PTx drive voltage or by changing the PTx inverter frequency.

If the CE value is positive (once again, I could have made a sign error) this tells the PTx to turn down its power output.

If the CE value is zero, the PTx will leave its output alone (give or take any Derivative action from a PTx PID controller).

So looking at the CE value can tell you a bit about the health of the link. Once the link is up, we expect to see mostly-zero CE values. This indicates that the PTx can supply the required power.

If they’re all negative (I may have made a sign error) and never go to zero, it suggests that the PTx has hit its maximum output limit - be that a frequency, voltage, current or input power limit - and something’s a bit wrong. Possible wrong things include: poor coupling, inappropriate LC tuning, excessive current draw by your load.

[artag]

I see - useful.
I'll see if I can find a place to sniff the data.