High Temp - CPU and PHY - ws-12-250-DC

[dersonik]

Hi
I have some issue with CPU and PHY temp.
This is new switch.
Switch is stable just wondering about temp.

[jonwang]

The fan shows its off. Are you nearby that you can confirm the fan isn't moving?

[sirhc]

The fan shows its off. Are you nearby that you can confirm the fan isn't moving?

Yea the fan would not be moving with a low board temp as we are currently controlling the fan soly on board temp alone.

We are going to change this on the next release so that we will also take into account CPU and PHY temp.

We are guessing this is because the switch is in a COLD environment and the board is so cold but the CPU and PHY are warm.

The CPU and PHY are rated to run at 125C all day long so he is not in danger.

Also the CPU and PHY sensors are built into the chip and the manufacturer says to take their readings with a grain of salt as I think they have a +/- 30% variance which is why we had used our board sensors.

With the new firmware will always kick LOW fan on if CPU or PHY is > 100C

[dersonik]

I think LOW fan is not enough.
Should be medium. LOW is not enough - checked in practice when temperature was 120 - LOW cooled down max - 115

[sirhc]

I think LOW fan is not enough.
Should be medium. LOW is not enough - checked in practice when temperature was 120 - LOW cooled down max - 115

It is very possible the CPU and PHY sensors in that SOC are just way out of whack.

The manufacturer of the SOC said those sensors are +/- 20% - 30%.

The next firmware release will have new fan controls so let's see what it does for you but you're fine for now as the SOC is rated to operate at 125C all day long.

Another thing I have seen is bugs (spiders) get in the switch and make a nest in the heat sinks which make them run hot. Nothing like opening up an RMA unit on the bench and hundred of baby spiders run all over the place.

[IntL-Daniel]

I have found this old topic and want to share my current experience with the NEW unpacked WS-12-250-AC switch. I have replaced other NENONIX switch on my site with the new one and see attachements below.

So while the other WS-12-250-AC switch has the highes CPU/PHY temp from 44 to 74°C, the new one has ~ +20°C more - from 70 to 95°C. As I read above, it seems that in the newer switch has different CPU/PHY sensors than before and now I have these with +- 30% accurancy? I was thinking about bad mounting of heat sink inside, or maybe missing?


So should I be affraid of something with this strange CPU/PHY Temp state on the same product?

Edit: Curently I have found that both mentioned screenshots are from two switches from the same batch - production 03/2021. So for me it souds more like problem of heat sink on CPU (?) and the reason for RMA? @sirhc ?

[sirhc]

I am sure your switch is fine.

The temp sensors for the CPU and PHY are built into the chip and are notoriously inaccurate. No they are NOT new or differnt as they are built into the SOC chip, we simply read them.

The board temp is an external sensor we put on the board and is much more accurate.

If your concerned the heat sink is lose then open the chassis and verify the heat sink on the SOC chip is secure.

[IntL-Daniel]

Thanks for reply. And can I open the case (break the sticker) without lose of warranty?

Edit: BTW I though that this annoucement since fw v1.5.3 means changes in these sensors? If not, then there is really something different because as you can see on all of my screenshots the PHY and CPU temp is exactly the same WHILE on older switches (even on the same board Rev F) I can see always different values on PHY a CPU (see another screenshot example).