ESP32-POE-ISO temperature

[Drjaymz]

> But, is it necessary?  Lets have a look at the schematic...

Well, you already answered yourself. Yes. In fact we didn't have this resistor when we switched to TPS2375 and weren't aware of the MPS requirement during sleep. This was causing two issues:

1) It turns out most of the customers use these boards on remote locations where sleep and deep sleep are important to conserve power during power downs. Furthermore, most of the customers evaluating which ESP32 board to consider would always measure power consumption in sleep and deep sleep modes. And the boards in sleep would lose PoE power since the PSE would turn off the power supply completely.

2) Same thing naturally happened when using the on-board reset button.

Of course we would consider your advice for future revisions. I've added notices for future revisions to again empirically measure the load and the temperature and to consider better routing of R41 or/and adding a jumper.

Thanks for your constructive feedback.

I hadn't considered the consumption drop on using the reset button.  I tried it and it seemed ok.  But some PSE may be more finicky.  I only have a fortiswitch and tplink brands and they don't drop the power until a few seconds after you remove it and provided you pull >10mA at any point in the last second or so they are happy.

The irony is that the MPS resistor is wasting as much power as they are saving.  And its not OK anymore to burn 1W for nothing, especially if we have 1000 of these out there.  It will be an improvement if the ESP32 is enough for most PSE's.

In a way the board is perhaps a little too efficient, there is a 3v3 switched where most would use a linear reg, great for battery operation.  Luckily the 5V DCDC isn't too efficient

[Konsulten]

Hi is this valid also for the non ISO version. That R41(R42 on non iso) can be removed so the board no longer heats up as high?
I burnt and browned a plastic casing that was with good ventilation. And later found this post.

[LubOlimex]

Yes, removing R41 will lower heat but still notice what I described previously and consider that your PoE switch might disconnect the board. Probably you can test it empirically, remove R41 and test for a few days to see if that causes disconnects from the switch. If it does, solder R41 back. Notice that the expected behavior largely depends on the PoE equipment. If you expect your project to connect to different switches, it is better idea to keep R41.

You can find more about why this resistor is placed if you Google for "Maintain power signature PoE". This is one of the better sources that describes it:

https://www.poepower.net/mps.html

[Konsulten]

Okay. I now noticed that sy8089 step down had been burnt. Could this overheating have caused this? 3.3V is gone from the board and it does not power the esp32 anymore. Also I found this board did not look like the other Rev K boards. The C25 was distanced from the board with some rubber thingy. This was not on the other 5 boards I ordered at the same time. Maybe a patched version happened to tag along?

Verified on the other boards that removing R41 indeed lowered the temperature alot and also worked fine with my two different Unifi PoE switches.

[LubOlimex]

Without SY8089 you can't power the board...

> The C25 was distanced from the board with some rubber thingy.

Sounds like C25 exploded which is pretty serious hardware fault. It is possible  Did you have USB cable and PoE power attached and enabled at the same time?

You need to at least replace C25, and SY8089 and double and triple check for short circuits on 3.3V and 5V lines before trying to power the board. Even then it is possible that the PoE chips were damaged...

> Verified on the other boards that removing R41 indeed lowered the temperature alot and also worked fine with my two different Unifi PoE switches.

It will work fine until sleep or deep sleep are used. If you don't need sleep or deep sleep it would be fine.

[Konsulten]

No I did not have it connected to both PoE and microUsb. It has been running a couple weeks with BLE discovery with PoE until it just died. I dont think the capacitor exploaded. I would say this board was not made in the same way as the other 5 I had ordered. Here is an image of the two different boards. One with the elevated capacitor.
Do you know why this was done?


Is there some warranty for these? Bought in 2022-11-21

[LubOlimex]

The board from top image clearly has the capacitor damaged. The board was not made this way definately. It also didn't leave our facilities this way since it will never pass optical inspection tests, let alone empirical tests.

What happened is that the 5V line received more than 5V for some reason. I expect that it won't be just that capacitor but also converters around it, since it is pretty important capacitor that is in the middle of the schematic.

You said you didn't have USB attached at the same time as PoE which is good but:

1) did you have any other device connected to the board at the time of the accident;

2) how did you power it at the time of accident?

We have warranty on all products, but it won't apply here. You didn't receive this board this way, something damaged it. Whether it is overvoltage, static electricity, short circuit contact at the bottom, some other influence in the setup might be hard to determine (but it is important to determine what caused it since it will probably happen again) - the board got damaged. You can find our warranty and returns policy for a reference here:

https://www.olimex.com/wiki/GTC#Warranty

[Konsulten]

No other devices. Was used for espresense project. Only PoE from ubiquiti unifi switch.

Yep the cap had been popped
Will check the running time of the device and see if its close to the 1000hours of lifetime of the capacitor

[arumi88]

I found a solution to the ESP32-POE-ISO's heat generation.
- Removed R41
- Add 100ohm/0.25W resistor between +3.3V and GND
This allows heat generation up to nearly 80°C to be suppressed to less than 40°C, and also supports PoE's minimum load of 0.1W.
However, it will affect the operating time of LiPo batteries.

[LubOlimex]

> - Add 100ohm/0.25W resistor between +3.3V and GND

Just remove that resistor. It doesn't do anything related to the MPS requirement. It doesn't solve any issue and only bugs the 3.3V circuit. The TPS2376D would still disconnect even if that 100ohm/0.25W is present since the +3.3V is after the TPS2376D.

[mihaiadrian]

I am sure that instead of 100Ohm resistor, you might have other power consumer that can do job instead without wasting energy. For example, the Olimex 2.8 display with backlight off, pulls 0.1w .   For now , regarding power consumption without R41 I only have the data here, I need to do some more tests with deep sleep:   https://www.olimex.com/forum/index.php?topic=7993.msg38392#msg38392

I have to say I don;t understand "The TPS2376D would still disconnect even if that 100ohm/0.25W is present since the +3.3V is after the TPS2376D "  -   I would expect that a power consumption of at least 0.5 - 1w either on 5v or on 3v to do the job in keeping power on

My only tests with removed R41, was to keep the board powered for a few hours with ping continuously from another host. No packet was lost. 

[LubOlimex]

In normal operation it won't disconnect, it happens only when you are using some low power modes of operation. Hence my advice is to first empirically test or measure the behavior without anything additional (just without the resistor that heats up) and only try additional measures or adding loads and stuff only if it disconnects.

[redbarn1315]

Sorry for resurrecting this topic from the grave but I've also noticed this issue and I would love to see a jumper added if possible that comes installed by default which connects R41 to ground. This would make isolating R41 much easier for projects where MPS using R41 is not required.

[LubOlimex]

Thanks for suggestion but there is no place for PTH jumper and if we place SMT jumper it would be similar soldering effort as to remove the resistor. We would consider it for future revisions tho.