@JoeAH 

PoE negotiation happens when the cable is plugged in. 

When you change the port, you have to reconnect the port or try to disable the port and enable it again to let it re-neogitate the power. 

  @Clive_A 
Thank you for your reply. I tried both, but it don't re-neogitate the power class. I plugged the CAT-cable out, wait some seconds and plugged in again. I also disabled, applied, enabled and applied again. I did also both together, diabled, applied, disconnect cable, enabled, applied, connect cable. But it don't work, regardless which class I try to set. Do you have the possibility of recreating the issue?
Best regards
Joe 

  @JoeAH 

JoeAH wrote

  @Clive_A 
Thank you for your reply. I tried both, but it don't re-neogitate the power class. I plugged the CAT-cable out, wait some seconds and plugged in again. I also disabled, applied, enabled and applied again. I did also both together, diabled, applied, disconnect cable, enabled, applied, connect cable. But it don't work, regardless which class I try to set. Do you have the possibility of recreating the issue?
Best regards
Joe 

Will look into this. 

  @Clive_A 

Clive_A wrote

  @JoeAH 

JoeAH wrote

  @Clive_A 
Thank you for your reply. I tried both, but it don't re-neogitate the power class. I plugged the CAT-cable out, wait some seconds and plugged in again. I also disabled, applied, enabled and applied again. I did also both together, diabled, applied, disconnect cable, enabled, applied, connect cable. But it don't work, regardless which class I try to set. Do you have the possibility of recreating the issue?
Best regards
Joe 

Will look into this. 

I have something that needs to be confirmed. 

A diagram of the network where your PD is specified.

Do you ground the PD? Grounding can be done in various ways. So I am asking if you have use the proper grounding tools or it is metal case and you have placed it to the wall with metal nails or screws?

We assume that your PD might be grounded, which, in standard PoE, if the PD is grounded in any way, it will bypass the management from the  PoE controller on the switch. 

  @Clive_A 

Hello,

for our test conditions we have only the TP-Link SG3452XMPP and our PD in the network:

We use as PD the Si34071. The schematics are like the same as for the 24 V eval-kit:

www. skyworksinc. com/en/Products/Power/Evaluation-Kits/si34071-isolated-24v-class-7-evaluation-kit

We just modified the PoE Class to 8 in Autocalss mode:

Our device is not connected to the wall or has extra grounding than from the network shielding. I even don't understand why the grounding bypasses the PoE management? Because the switch detects correct the wrong class and the overload. If it detects so it should shut down the affected port. That seems to me more like a software issue than a hardware issue.

  @JoeAH 

JoeAH wrote

  @Clive_A 

Hello,

for our test conditions we have only the TP-Link SG3452XMPP and our PD in the network:

We use as PD the Si34071. The schematics are like the same as for the 24 V eval-kit:

www. skyworksinc. com/en/Products/Power/Evaluation-Kits/si34071-isolated-24v-class-7-evaluation-kit

 

We just modified the PoE Class to 8 in Autocalss mode:

 

Our device is not connected to the wall or has extra grounding than from the network shielding. I even don't understand why the grounding bypasses the PoE management? Because the switch detects correct the wrong class and the overload. If it detects so it should shut down the affected port. That seems to me more like a software issue than a hardware issue.

You can consult the IEEE by email for the PoE standard, and regarding the question of why it would bypass the PoE management. The IEEE should have related docs. As far as we know, that is a thing. We are not the PoE standard law maker. So, we are not able to explain that for you.

Is the wire shielded? Try a UTP cable. 

So the PD is not a device but a bare metal PoE controller? 

Have you tried a regular device like an AP that uses standard PoE?

Dear @Clive_A,

thank you for the hint to the IEEE. I didn't find there a explaination about a classification bypass. Could you please provide me with a link or dokument and page number?

What I found by reading the IEEE Std 802.3™ ‐ 2022 (https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9844436) is on page 5738 in capter "145.3.6 PD classification" the explaination how the negotiation should work. It's decribed as following:
"Depending on the number of class events produced by the PSE, the assigned Class is equal to or lower than the PD requested Class. The PD shall conform to the assigned Class, regardless of its requested Class."

Furthermore, in chapter "145.2.8 PSE classification of PDs" (p. 5696) it's stated:
"When a single-signature PD requests a higher Class than a PSE can support, the PSE assigns the PD Class 3, 4, or 6, whichever is the highest that it can support. When a dual-signature PD requests a higher Class than a PSE can support, the PSE assigns the PD Class 3 or 4, whichever is the highest that it can support. The minimum output power a PSE supports depends on the assigned Class."

Also read chapter "145.2.10.8 Overload current" (p. 1340):

"If the current on either pair of a pairset exceeds ICUT-2P for longer than TCUT, the PSE may remove power from that pairset. The cumulative duration of TCUT is measured using a sliding window of at least 1 second width."

Or "Table 145–16—PSE output PI electrical requirements for all PD Classes" (p. 5704).

The switch should not assign to a Class that is higher than the current limit. When the current limit is reached (overload) the switch should turn off the port. It's not working as described in the IEEE rules.

-----------------------------------------------------------------------------------------------------------------------------

Test with other device:

I tested with an aruba access point, which is a class 4 device, but the class or power limitation does also not work with this. 

I also tested with a poe device that has a TI PD IC, but the class or power limitation does not work with this.

I will search for other devices to do more tests.

The PD is an audio amplifier which should detect the assigned class and limit in this way the output power.

I also tried with no autoclass but this makes no diffence.

Thank you for your patient on trying to find a solution.

  @JoeAH 

JoeAH wrote

Dear @Clive_A,

thank you for the hint to the IEEE. I didn't find there a explaination about a classification bypass. Could you please provide me with a link or dokument and page number?

What I found by reading the IEEE Std 802.3™ ‐ 2022 (https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9844436) is on page 5738 in capter "145.3.6 PD classification" the explaination how the negotiation should work. It's decribed as following:
"Depending on the number of class events produced by the PSE, the assigned Class is equal to or lower than the PD requested Class. The PD shall conform to the assigned Class, regardless of its requested Class."

 

Furthermore, in chapter "145.2.8 PSE classification of PDs" (p. 5696) it's stated:
"When a single-signature PD requests a higher Class than a PSE can support, the PSE assigns the PD Class 3, 4, or 6, whichever is the highest that it can support. When a dual-signature PD requests a higher Class than a PSE can support, the PSE assigns the PD Class 3 or 4, whichever is the highest that it can support. The minimum output power a PSE supports depends on the assigned Class."

 

Also read chapter "145.2.10.8 Overload current" (p. 1340):

"If the current on either pair of a pairset exceeds ICUT-2P for longer than TCUT, the PSE may remove power from that pairset. The cumulative duration of TCUT is measured using a sliding window of at least 1 second width."

 

Or "Table 145–16—PSE output PI electrical requirements for all PD Classes" (p. 5704).

 

The switch should not assign to a Class that is higher than the current limit. When the current limit is reached (overload) the switch should turn off the port. It's not working as described in the IEEE rules.

 

-----------------------------------------------------------------------------------------------------------------------------

Test with other device:

I tested with an aruba access point, which is a class 4 device, but the class or power limitation does also not work with this. 

I also tested with a poe device that has a TI PD IC, but the class or power limitation does not work with this.

 

I will search for other devices to do more tests.

 

The PD is an audio amplifier which should detect the assigned class and limit in this way the output power.

 

I also tried with no autoclass but this makes no diffence.

 

Thank you for your patient on trying to find a solution.

What I wrote is to email the IEEE commission regarding the grounding that could bypass the PSE control, or how this theory works. 

We do not provide this information as we are not the IEEE maker. From what we learned based on the field experience, the grounding could cause such a problem. It is not our job to investigate why grounding would cause such a problem as we are only providing what IEEE has. 

In your test with the Aruba, the results you provided seem to indicate that the AP is not able to properly use the class. Based on the IEEE, they contradict. 

I assume that there may be a problem with the current PD. The PoE STD from the IEEE was a new one in 2022. I am not sure if vendors are following this up. 

Dear @Clive_A,

Thank you for the reply and clarification. Then please also change the description in the manual of all your managed PoE devices:

Best Regards
Joe