-30°C rated 230VAC to 9V 1A supplies or 802.3at/af PoE splitters?

[Nominal Animal]

I've got some outdoors IP cameras above 68°N, sheltered from water and snow but not from the cold.  They lack PoE support, but consume less 0.8A at 9VDC.  I do have 230VAC mains available.

TP-Link TL-POE10R and other similar IEEE 802.3at/af PoE/PoE+ splitters I've found are only rated to 0°C/32°F (maybe to -10°C/14°F), but my cameras are rated to -30°C/-22°F.

I suspect my best bet is to get a Mean Well GST18E09-P1J wall wart, and shield it from the elements.  It is rated down to -30°C/-22°F, but is not IP-rated (so at best can be considered IP20).

The other option is to create a heated/insulated box, but that might overheat in the spring.

I like the idea of using a more easily available outdoors-rated mains to 12VDC 1A adapter, plus bodging in a linear regulator like Rohm BA90DD0T to regulate it to 9V.  However, making it stable for such a wide temperature range is black magic to me.  Maybe a couple of aluminium electrolytics in parallel?

Any suggestions?  I'd rather not spend too much money on these, as the cameras themselves aren't that expensive.  But perhaps someone here has already solved similar issues?

[langwadt]

do you need "real" POE or can you just do the hack version and use the unused pair in 100Mbit?

I doubt  the 9V is that critical, the first think inside the camera is probably a voltage regulator

[Nominal Animal]

There are some expensive industrial 802.3bt/at/af splitters like Aetek S69-220 that list -40°C to +75°C operating temperature, but at $150 a pop they're not worth it. direktronik.se makes an industrial 30W 12V 2.5A one, 20103219, that they promise -40°C to +85°C operating range; and at 30€ a pop it's on the border of my budget.. but they're sold out everywhere.

do you need "real" POE or can you just do the hack version and use the unused pair in 100Mbit?

I can, yes.  While the mains power is nearby, the Ethernet cabling will be on the order of 15-20m, and I do intend to use good quality Cat6A/Cat7 S/FTP -40°C rated cables.  I think I can count on under 100Ω/km DC resistance.  Let's assume a generous 500mA current, which would correspond to 0.5V drop per 10m, wasting a quarter of a watt or so.  Perfectly acceptable... except I don't like the idea of 20m long cable without regulation, so I'm back at how to ensure the voltage at the camera stays within the safe limits.

I did find Silvertel Ag9712 and Ag9912M IEEE802.3af modules providing 12V 1A+ (15W) down to -40°C, only requiring PoE magjacks, a couple of full bridge rectifiers, ESD diode, 100µF+10µF+1.8µF+10µF filtering, a couple of 2kV 4.7nF (Y?) capacitors, and some ferrite beads for EMI suppression on the input.  This indicates this is definitely possible, even in a tiny form factor molded into resin.  But this is far outside my skill range and experience.  I'd have better success growing fur and moving to live in a nearby tree with binoculars.

I could just use an IP44 -40°C rated 12V 2A mains supply like EAN 6438195018625, possibly with a 5m extension lead, and be done with it, I suppose.  Making a small cylindrical linear regulator down to 9V encased in silicone and plastic at the camera power connector would be just about perfect, soothing all my fears and needs.

I doubt  the 9V is that critical, the first think inside the camera is probably a voltage regulator

I'm currently testing Tapo C520WS cameras, and TP-Link is notorious for claiming nothing but their (0°C to +40°C non-IP-rated) 9V 0.8A supply will work.  If they only added IEEE 802.3bt/at/af support to Tapo cameras, they'd have a much better product in their hands.

[Nominal Animal]

Seeed Studio makes a very interesting variant as well, SenseCAP PoE Splitter DC Jack 12V.  (Waveshare has a similar one, but only for 5V; the 5V 5A version is very interesting for small Linux servers like camera DVRs.)

Many Aliexpress sellers confuse "operating temperature" and "storage temperature".

And indeed, the proper industrial ones all seem to be using aluminium electrolytic capacitors.  I do believe the capacitors, and general thermal expansion/contraction (and difference between the coefficients for FR4 and electronic components, unless allowed for in the design), is what limits the minimum operational temperature.

[TimFox]

If you need a simple thermostat to control a heater to keep your box above minimum temperature, "Klixon" snap-action thermal switches are available with standard setpoints down to -54^o C = -65^o F.

[Manul]

Many Aliexpress sellers confuse "operating temperature" and "storage temperature".

And indeed, the proper industrial ones all seem to be using aluminium electrolytic capacitors.  I do believe the capacitors, and general thermal expansion/contraction (and difference between the coefficients for FR4 and electronic components, unless allowed for in the design), is what limits the minimum operational temperature.

I would say expansion/contraction is rarely a problem. And most ICs not rated for low temperature still work perfectly fine in most applications, cause usually it just means that some of their spec will be slightly off. Like for example max offset voltage of an Op Amp might not fit datasheet value. The real problem is wet electrolytic capacitors, cause ESR can increase by more than an order of magnitude at -40C. Such increase in ESR may cause switching power supply to fail. Someone who likes tinkering might simply replace capacitors with polymer type ones. Should solve 99% of low temperature issues.

[mariush]

Can you put some resistors inside the heat the internals , maybe with a temperature sensor to turn off the resistors once the temperature goes above 5-10c?

If you use passive poe (ex put 12v or 24v DC or something in between on the two pairs that aren't used with 100 mbps ethernet) , then there's dc-dc converters that are rated for -40c or even lower.

Ex

working down to -40c

12v-42v to  9v 1A  : https://www.digikey.com/en/products/detail/recom-power/R-78C9-0-1-0/2652149

9V - 36V in, 3 ~ 18V out, up to 3A (max 27w ) https://www.digikey.com/en/products/detail/omnion-power/APXW003A0X3-SRZ/2640012

19v-40v in, up to 15.5v out , up to 36w  : https://www.digikey.com/en/products/detail/murata-power-solutions-inc/OKI-T-36W-W40N-C/2605210

The first one is the simplest, but would be best to put a voltage higher than 12v on the wires, maybe 16.5v - 20v from a laptop adapter style power supply, or 24v from some cheap psu.


Wondering even if it wouldn't be easy to just use a linear regulator, and let it dissipate 1-2 watts worth of heat all the time...

A 7809  or equivalent (ex LM2940) in TO-220 package can do 1.5A and works down to -40c : here's LM2940 : https://www.digikey.com/en/products/detail/texas-instruments/LM2940T-9-0-NOPB/212655

With 11-12v on the input, it will dissipate (12v - 9v ) x 0.8A = 3 x 0.8A = ~ 2.4w of heat ... with a small heatsink, the chip will stay at under 80-90 degrees Celsius in the summer.

At -30c, those 2-3w of heat would keep it reasonably warm. 

You could also just bump up the voltage on the cable from 12v to a higher amount in the winter.


You could probably even add some sort of automation ... like for example have a couple 1n400x diodes in front of the regulator drop around 2 x 0.75 = ~ 1.5v , so that the regulator will get around 10-10.5v instead of 12v.  Have a mechanical relay short out the diodes, giving the regulator more voltage (and dissipate more heat) when it's too cold.

[wraper]

Do you really need that rating? Something rated for 0^oC will work just fine in 95%+ of cases. 0^oC comes mostly from not bothering with assurance rather than actual ability to function. The major concern for negative temperatures is stuff with batteries and LCD displays. Other than that, you usually need quite a bit colder than -30°C for properly designed stuff (that is not marginal to begin with) to give up.

[Nominal Animal]

Do you really need that rating?

Yes.

My brother stored some 12V Goobay wall warts in a cupboard in a storage shed (dry, no humidity issues) over the previous winter, and most of them failed.  They're sold by Verkkokauppa, a large reputable brick-and-mortar electronics store here.

(We use them to power 12V fans in mosquito traps based on CO2 from fermenting sugarwater and certain mushrooms, "kiljuansa".  Octenon, the best attractant, is pretty expensive, but it is plentiful in certain mushrooms, you see.)

The major concern for negative temperatures is stuff with batteries and LCD displays.

I disagree.  My shoddy investigation indicates electrolytic caps are the ones that cause the fail at freezing.  You can check this easily yourself: stick some electrolytic caps in the 100µF to 470µF range in a small ziplock bag with some desiccant, let the desiccant do its job for a few days, and then stick it in the freezer for a day.  Check how many of them fail (ESR in particular) afterwards.

Everything electronic without LCDs, batteries, or electrolytic capacitors, including the random collection of 12V 3-pin PC case fans, have survived in those same storage cupboards.  I have considered getting some cheap boxed PoE splitters, and recapping them with corresponding aluminium electrolytics, but the ESR is difficult to match across capacitor types, so switcher stability at those low temperatures is not guaranteed.

Plus, it typically does get below -40°C or so for a few days during the winter, so I do already rely on the devices surviving exceeding their rating for a bit.

[wraper]

You can check this easily yourself: stick some electrolytic caps in the 100µF to 470µF range in a small ziplock bag with some desiccant, let the desiccant do its job for a few days, and then stick it in the freezer for a day.  Check how many of them fail (ESR in particular) afterwards.

The vast majority of electrolytic caps are rated for at least -40^oC, how they are supposed to fail in your test? As of ESR, yes it rises at low temperature, however under heavier load where ESR is the most important, they will heat more due to high ESR, which will bring it down.

My brother stored some 12V Goobay wall warts in a cupboard in a storage shed (dry, no humidity issues) over the previous winter, and most of them failed. 

And what it has to do with operation temperature? Storage temperature usually is much wider. Also IME Goobay sells shit.

[David Hess]

I have designed and build circuits which operated down to liquid nitrogen temperatures.  I qualified parts including ICs for operation at such a low temperature.

Aluminum electrolytic capacitors do not work when frozen, but have a lower temperature limit from -25C to -55C.  I used solid tantalum capacitors without problems.

Transistor gain is lower at low temperatures.  This may cause oscillator circuits to not oscillate.

ICs may contain startup circuits which can fail at low temperatures.

[RoGeorge]

Any suggestions?  I'd rather not spend too much money on these, as the cameras themselves aren't that expensive.  But perhaps someone here has already solved similar issues?

Put the power supply, or the entire device, in an enclosure together with a 24/7 powered incandescent light bulb.  If the bulb power is just right, there is no need for a temperature regulator.  This trick is enough for winter times here, continental climate, 45°N (though the method is not used very often).  Don't know if it can be used in much colder weather.

[Nominal Animal]

The vast majority of electrolytic caps are rated for at least -40°C, how they are supposed to fail in your test?

Vast majority of quality caps, yes, definitely.  The whatever-"brand" ones used in cost-cut supplies?  I don't think so.

The equivalent series resistance did not recover, and capacitance dropped; the switcher fails to oscillate.  I don't have the equipment to quantify the changes, though.  I suspect some kind of chemical change in the electrolyte, perhaps crystal formation leading to conductive tendril formation.  Basically, poor quality electrolyte failing at low temperatures.

That said, I could be wrong, absolutely!  Like I said, electronics without caps, batteries or LCDs have survived several winters in the same outdoors-temperature but weather-shielded shed cupboard without observable issues, so I'm making informed guesses (by a materials physicist specialized in molecular simulations) at best.  I mean, practical knowledge beats theoretical musings every day.

And what it has to do with operation temperature?

I thought it explains why I don't trust wall warts to survive freezing.  I would never trust those cheap Goobay wall warts to not catch fire in freezing temperatures, for example.

Many different stores here (not part of the same chain or ownership) sell 'Enjoy' brand 12V 2A IP44 rated outdoors power supplies with -40°C ~ 85°C operational temperature range for 20€-25€, marketed to work with several different brands of cameras and LED lights even in reputable electronics stores like Verkkokauppa, so I am willing to trust them.  (Based on reports from others running the same camera, it works fine from 12V without overheating.)

Its own 9V 0.85A 100-240VAC wall wart is marked TO90085-2C1 (manual states T090085-2B1), and seems absolutely tiny (albeit not "too" light); I definitely don't trust it, having watched DiodeGoneWild and BigClive teardowns of such.  The Finnish sellers note that this has a 3m long DC cable, because you need to place the 9V wall wart indoors.  It has a CEE 7/16 plug, so IP20 rated at best. The camera itself, according to its datasheet, is IP66 rated when mounted, with -30°C ~ 60°C (-22°F ~ 140°F) operational temperature range.

I have designed and build circuits which operated down to liquid nitrogen temperatures.  I qualified parts including ICs for operation at such a low temperature.

Oo, that's about -196°C (77 K). Impressive! 

I used solid tantalum capacitors without problems. Transistor gain is lower at low temperatures.  This may cause oscillator circuits to not oscillate.

Right.

In practice, PoE voltage on the cables is in the 40V to 57V range, and powered devices like these cameras need to use a switchmode DC-DC converter to get to 12V.  It is exactly the malfunction of the converter in low temperatures, either not working, working intermittently, or causing a voltage spike exceeding the safe input range, and eventually breaking the camera when I'm not there to investigate, that I want to avoid.  It is also the main reason why I so like the idea of adding a linear regulator at the camera power connector.

there's dc-dc converters that are rated for -40c or even lower

Mouser sells tiny (23mm×15mm×14mm) Silvertel Ag9912MTB IEEE 802.3af Class 0 modules for about 9€ in singles (link), outputting 12V 1A (12W), requiring 42V ~ 57V at startup, 36V ~ 57V during normal operation.  These are rated for -40°C ~ 85°C operating temperature, and require very few additional components (see attached diagram) to make a fully-featured industrial IEEE 802.3af 12V 1A PoE splitter or powered device.  A single resistor can adjust the output between 10.0V and 12.75V.

It does need a 100mA minimum load for stable noiseless operation, but in my case an external 120Ω 2-5W resistor (or four 470Ω in parallel) in parallel with the camera is quite acceptable.  A better solution would use a power N-channel MOSFET controlled by an instrumentation amplifier over a 0.1Ω shunt resistor, so that without any output load, the MOSFET would dissipate 1.2W.  When the voltage drop over the shunt resistor exceeds 10mV, it would be fully conductive.  Between these, it should be roughly linear, ensuring a constant 1.2W minimum load.

If built in to the camera, none of this would be needed, so designing this in from the get go would make much more sense.

I think that if one adds both the suggested CLC filter and a linear regulator to 9V or 10V, you'd get pretty nice low-noise 1A supply out of it, too, for other applications.

Thus, a valid DIY solution from reputable sources like Mouser/Digikey etc. is available to anyone with sufficient skills.  I'm not sure I myself am there yet, though.

[David Hess]

I have designed and build circuits which operated down to liquid nitrogen temperatures.  I qualified parts including ICs for operation at such a low temperature.

Oo, that's about -196°C (77 K). Impressive! 

The stuff I designed had to work down to -55C for the military temperature range, but the test chamber used liquid nitrogen for cooling and there was no reason not to test at a lower temperature.  Everything worked while submerged in liquid nitrogen.  Obviously I avoided aluminum electrolytic capacitors, but I also avoided Z5U type ceramic capacitors.  X7R ceramics were acceptable at -55C but maybe not lower.  ICs that I remember working correctly were the LT1117 voltage regulator, and various Linear Technology operational amplifiers like the LT1013 and LT1112.

I used solid tantalum capacitors without problems. Transistor gain is lower at low temperatures.  This may cause oscillator circuits to not oscillate.

Right.

In practice, PoE voltage on the cables is in the 40V to 57V range, and powered devices like these cameras need to use a switchmode DC-DC converter to get to 12V.  It is exactly the malfunction of the converter in low temperatures, either not working, working intermittently, or causing a voltage spike exceeding the safe input range, and eventually breaking the camera when I'm not there to investigate, that I want to avoid.  It is also the main reason why I so like the idea of adding a linear regulator at the camera power connector.

Crystal oscillators, especially CMOS gate oscillators, can be a problem because they often operate with very little excess gain.  There have been problems with common microcontroller oscillators not starting at the lower end of their operating temperature range, but see below about gradual turn-on.

Linear ICs, which includes switching regulators, include some variation of a startup circuit which relies on a pinch resistor or leakage to start the current regulating bias circuits.  Unlike a 741 which just uses a resistor, this allows a wide supply voltage range without changing the quiescent current and operating parameters.  On the other hand, a 741 does not have a startup circuit to fail.

At low temperatures, there may not be enough leakage for this circuit to start.  Sometimes an IC is produced which seems to start correctly at low temperature, however if the supply voltage is raised gradually, it does not start.  Here capacitive coupling is allowing the startup circuit to work when the supply voltage is applied abruptly, but not when applied gradually.

Other things happen to semiconductors at low temperatures.  Low voltage ICs will have problems operating at low voltages when transistor Vbe increases.  Lower transistor current gain will increase input bias currents.

As far as the current problem, I might test different power supplies and PoE splitters for low temperature operation.  Some of them are likely to work fine despite what their rated operating temperature range is.