|Thread Links||Date Links|
|Thread Prev||Thread Next||Thread Index||Date Prev||Date Next||Date Index|
Currently the standard allows adjusting Peak-to-average ratio if system parameters per equation 33-4 are knows.
It includes old equation 33-4 that was function of Rch and Pclass and include the K factor that is function of E2EP2P_Iunb as function of Rch.
This is a good feature that helps optimizing the system performance.
If you don’t use LLDP, than you can use the default values generated by equation 33-12 and 33-4 at worst case conditions.
The peak power is limited to 5% duty cycle 50msec maximum averaged on sliding 1 sec window. In any case the average DC and the RMS current must be kept per the limits of Pclass/Vport at all times. As a result, in average, it will have less effect on the PSE power budget.
In addition, I will propose in next week to update equation 33-4 and 33-4a so 33-4 will be used for Class4 only with 1.11 peak/average ratio and above class 4 it will be 1.05 ratio. This will reduce further the peak power and peak current and as a result will reduce ILIM_MIN with the P2P_Iunb effect.
One area to be concerned about if peak-to-average ratios are reduced relates to how Type-3 / Type-4 PD’s will form power requests in LLDP.
For example, a PD that might have 10% power bursts for 10msec or less will be forced to form the LLDP Power Request based more on peak power than average power. This in turn burns more power budget in the PSE.
No idea how many PD’s will fall into this category, but lower peak-to-average ratios assure that more PD’s will encounter this issue.
I would like to have your opinion on the following plane:
We have numbers for ILIM_2P_MIN and ICUT_2P for Type 3 and 4.
ILIM_2P_MIN may reach to 1.2A for Type 4 and 0.875a FOR Type 3.
Technically we can work with this numbers.
These are worst case numbers based on worst case model and also the fact that we allow ripple current Ipeak, in the PD according
Practically I would like to reduce this numbers in order to reduce stress from transformers.
We have several ways:
1. Reducing Ipeak by reducing the Ppeak_PD/Pclass PD factor from 1.11 for Type 3 to a lower value e.g. 1.07 range and for Type 4 to reduce it to ~1.03.
2. Reducing PD Vdiff budget to lower value So E2EP2P_Iunb will be lower (this move has the highest effect)
3. Using statistical analysis that will help reducing E2EP2P_Iunb (preferred to be our last move only after our spec is stable).
I prefer to start with 1 due to the fact that in reality, PD behavior has much less AC ripple that we assumed in 802.3at.
My goal is to reduce currents by at least 5%.
I'll appreciate your inputs.
Chief R&D Engineer
Analog Mixed Signal Group
1 Hanagar St., P.O. Box 7220