Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [802.3_4PPOE] PD Load turn on



Hi all,

 

Von and Voff were specified to create the UVLO (Under Voltage Lock Out) function (similar to what we can find in any PWM controller chip).

 

The reason for it is that in remote power feeding systems, when the DC/DC input capacitor is on its way to be fully charged and the PSE current output is in its way to drop to low value, the DC/DC starts to work and to consume power that cause the voltage across the capacitor to drop due to discharging the cap by the load and due to voltage drop on the cable (which is unique to remote power feeding systems that is described by the quadratic equation which was used to proof the need for Von and Voff).  

 

When this process happen, the DC/DC may not successfully POWER_UP and will see start-up oscillations (or motor boating as we called it).

As a result Von (<42V) is set to be higher than Voff (>=30V), creating hysteresis) and the difference Von-Voff defined that amount  of energy required to allow successful POWER_UP under all relevant system parameters: Von, Voff, 802.3af power level, 180uF PD input cap and 50msec PD inrush max time.

 

Von was specified to be <=42V (<Vpse_min=44) and Voff was specified to be (a) >Vpse_min/2+margin and (b) less than (44-0.45*20)=35V resulting with 30V nice to work number that addresses stability point and PD design margins.

 

 

As a result:

- Voff originally was defined below Von to address successful POWER_UP with no oscillations at and for POWER_UP.

- Due to the fact that in 802.3af Load current=350mA was < Iinrush by at least 50mA, it was sufficient to complete startup at 50msec ~t=180Uf*(44-30)/50mA=50msec.

-Once PD was turned on at Von, it must operate down to Vport_PD min for Type 1 PD. Similar concept for Type 2 with different Vport_PD min values with the knowledge that in 802.3at the load may be higher that Inrush (600mA> 450mA) so there is the 80msec  delay requirement before PD can consume current > 350mA.

-After POWER_UP, Voff stays as the PD Off voltage during POWER_ON state. PD Off voltage was needed also to be defined for POWER_ON state to be above the instability area (=44/2=22V for Type 1 and 50/2=25 for Type 2 and to allow some fast load transients without affecting the PD operation.

 

Please see the following sources for more details:

1.       http://www.ieee802.org/3/af/public/may00/lehr_1_0500.pdf  

2.       http://www.ieee802.org/3/af/public/mar01/darshan_2_0301.pdf slides 13-18.

3.       http://www.ieee802.org/3/af/public/jan02/darshan_1_0102.pdf  Slide 3 item 6.

4.       IEEE802.3-2012 clause 33.3.7.1:

The PD shall turn on at a voltage less than or equal to Von.  (This is POWER_UP state)

After the PD turns on, the PD shall stay on over the entire Vport_PD range. (This is POWER_ON state)

The PD shall turn off at voltage less than Vport_PD minimum and greater than or equal to Voff. (This is POWER_ON state)

 

The PD shall turn on or off without startup oscillation (Here Von and Off is specifically related to startup pase which is POWER_UP state.)

and within the first trial at any load value when fed by VPort_PSE-2P min to VPort_PSE-2P max (as defined in Table 33–11) with a series resistance within the range of valid Channel Resistance. (Here we can see the full picture and intent).

5.        In clause 33.2.4.4, PSE state machine, variable legacy_powerup: We can see the importance of knowing when Startup is done.

So to summarize this point:

-It is clear that Voff applies to START_UP and POWER_ON state. Von is relevant to POWER_UP. (At Power on: Vport_PD operating range is relevant and Voff)

-For 802.3af it is allowed to consume 350mA during startup since there is sufficient energy for meeting PD T_inrush_max. The reason: Iinrush_min=400mA > 350mA maximum DC power. There is no shall to prevent applying load during startup and implementation wise, it should be left as implementation specific since there is a requirement to successful START_UP without oscillations that covers this issue per 33.3.7.1.

-If overload occur during this time, STARTUP will fail. Reason: Iinrush_min=Ipeak_PD_max. So it is clear that PDs need to consume total max 350mA maximum during STARTUP however it is designer RESPONSIBILITY to meet all requirements even if one of the requirements if used not correctly, will cause system to fail e.g. when PSE Iinrush=400mA and PD consumes 400mA peak so net current to charge Cpd=0.

-It is not practical to force PD load (DC/DC converter and its load) to consume zero current for the whole START_UP phase.

-Cautionary note that explains the consequences if (Iinrush-Iload) is not greater than [Cpd*(Vpse_min-Voff(PD))/0.05sec +Margin] during startup, successful startup may not happen etc.

Reviewing Kousi proposal:

Suggested Solution

Add the following to section 33.3.7.3

"PDs shall not draw more than the maximum current allowed by a PSE during inrush as outlined in section 33.2.7.5"

Yair: This may not be sufficient since the problem is: what if PD consume Ipeak=400mA during START_UP. We need to specify that during START_UP, PD load current shall not be >350mA. (PD input current is PSE_I_Inrush which is limited by the PSE  and going to Charge Cpd but PD internal load need to be limited to 350mA during Startup time.) OTHER ALTERNATIVE IS DO NOTHING. The reason: it is covered by 33.3.7.1 That requires successful startup without oscillations and it is the responsibility of the PD designer to meet it.  

Add the following to section 33.3.7.1 at the end of second paragraph.

"The turn off voltage Voff applies both during power on and after power on"

Yair: This is correct and accurate. It doesn’t affect PDs in the field. Voff was specified for POWER_UP and POWER_ON states. See above analysis.

Yair

 

 

From: Koussalya Balasubramanian (kobalasu) [mailto:kobalasu@xxxxxxxxx]
Sent: Saturday, February 28, 2015 3:36 AM
To: STDS-802-3-4PPOE@xxxxxxxxxxxxxxxxx
Subject: Re: [802.3_4PPOE] PD Load turn on

 

EXTERNAL EMAIL

Jeff

 

Thanks for your comments. This is related to the fact that some PDs turn on their DC-DC (on top of their PD Cap) during inrush – so sometimes the cap may not be fully charged at the end of inrush leading to PD drawing more current leading to problems.  As per current standard it already gives a Voff that applies to DC-DC turn on.  But like I have pointed out, some people in the field have complained in the past that Voff applies only after PD has fully turned on, while several in the room said that is not the case.  That is all am trying to clarify with my suggested solution.

 

Thanks

Kousi

 

From: Jeff Heath <jheath@xxxxxxxxxx>
Date: Thursday, February 26, 2015 12:12 PM
To: Koussalya Balasubramanian <kobalasu@xxxxxxxxx>
Cc: "STDS-802-3-4PPOE@xxxxxxxxxxxxxxxxx" <STDS-802-3-4PPOE@xxxxxxxxxxxxxxxxx>
Subject: RE: [802.3_4PPOE] PD Load turn on

 

Hi Kousi,

 

This is a tricky problem thanks for tackling it. 

 

As I was not in the room when this got discussed I’m very likely missing some detail on the problem you are trying to solve.  This may be a benefit as I am just reading what is being proposed as a PD designer that was not ‘in the room’.  For instance I am not sure where you were planning on inserting the new text.  Sans that context your proposed wording may run into a problem of interpretation which could make existing compliant PDs non-compliant. 

 

 

PDs shall not draw more than the maximum current allowed by a PSE during inrush as outlined in section 33.2.7.5"

 

One valid reading of the proposed text could be that the PD must always stay under the PSE current limit during inrush.  Many PDs let the PSE handle the current limit during inrush which is currently allowed. 

 

I’m just guessing but is it the case that the turning on of the PD load causes the PSE to go back into current limit (after the PD has reached 99% of its final value)?  This could cause the PSE to go back into current limit and ‘fold back’ the voltage, which can cause the voltage at the PI to go below the Voff voltage. 

 

We have seen PDs that quickly charge up small capacitors, then turn on a load with no soft-start and expect the PSE just to deal with the consequences including the negative resistance of a DC to DC converter.  Depending on how the PSE has implemented Ilimit and foldback, it sometimes worked and sometimes did not.

 

If this is the case, maybe stating it explicitly would help.

 

“After the voltage at the PD PI has reached 99% of its final value, the PD shall stay under the minimum PSE current limit as defined by…”

 

BTW, the existing language of “Cport is charged 99% of its final value” has always been a bit troubling to me because the PD may not be know what the final value is, and Cport is an internal component not testable at the PD PI.  As far as I know the PSE is allowed to change its final voltage during inrush for instance.  I do NOT think there are problems in the field with this pedantic reading of the specification but it is kind of unclear.  If we are touching this, maybe we could think about referencing the PD PI for instance instead of an internal capacitor.  This is why I included PD PI language in my suggested text.

 

 

Regards,

 

Jeff Heath
Design Center Manager

Linear Technology Corporation

 

 

paper:

911 Olive Street

 

Santa Barbara, California 93101

voice:

805.965.1571

fax:

805.965.1701

computer:

jheath@xxxxxxxxxx

 

www.linear.com

 

 

From: Koussalya Balasubramanian (kobalasu) [mailto:kobalasu@xxxxxxxxx]
Sent: Thursday, February 26, 2015 11:09 AM
To: STDS-802-3-4PPOE@xxxxxxxxxxxxxxxxx
Subject: [802.3_4PPOE] PD Load turn on

 

Team

 

In Jan interim meeting, Jean and I presented about clarifying PD load turn on behavior that will help in making sure non-compliant behaviors are avoided.  The team directed us to submit this as a maintenance request.  Here is what we are planning on submitting – given the maintenance request will come back to 4PPoE group, we would appreciate any comment from 4PPoE group before we submit it.

 

Problem Statement:  

- PDs in the field turn on their Load during Inrush.  This leads to PD cap not charging up fully (even if C<180uf and PSE is following inrush rules from 33.2.7.5).  This may lead to operational problems after inrush.  There is a Voff requirement in PD table 33-18 to ensure power supply remains turned off for V<30V, but customers seem to read this as applicable only "after power on" not during "power on" - hence they turn on their DC-DC during inrush causing problems.

 

Suggested Solution

Add the following to section 33.3.7.3

"PDs shall not draw more than the maximum current allowed by a PSE during inrush as outlined in section 33.2.7.5"

Add the following to section 33.3.7.1 at the end of second paragraph.

"The turn off voltage Voff applies both during power on and after power on"

 

Thanks

Kousi