LM25069 Start Up issue when back-fed voltage applied

  I forgot to mention in the initial post that the waveforms presented are without the power limit set resistor installed (R1190,1330 ohm). When the issue was first observed it was thought that is possibly had to do with limiting MOSFET power. This resistor was removed and all subsequent testing was without it.

Uploaded LM25069 spreadsheet calculator with circuit design parameters included.

Hi George,

I just returned to the office and am looking into this. Thank you for using our tools as it greatly helps us with troubleshooting.

Thanks,

Alex

Hi George,

Do we know what type of ultracapacitor they are plugging in? The calculator tool looks like it will work, but it is only using 270uF in its calculations. If this is trying to startup with a ultracap load with mFs of capacitance (or larger), then the startup time will be much longer, and the timer will be timing out.

Thanks,

Alex

Hi George,

I saved your post but had to delete it from the forum because it contained some confidential information. Please be aware that this forum is external and thus can be viewed by anyone including google search results. Let me know if you are concerned about the information you have posted here from your own company and I will work to have it removed.

I did not view the end customer schematic beyond the "Prioprietary and Confidential" title since it contained end-customer information which TI may not have permission to view without their consent.

From your description alone, it sounds like there is a total of 200F capacitance. Where are these in relation to our Hot Swap, and what path is causing them to be isolated from the hot swap output, yet they are backfeeding voltage through the BSC009NE2LS body diode when off?

 Alex,

  The customer did provide permission to send you the schematics. Yes, cap voltage is being back-fed thru the MOSFETs body diode.  Is it possible for you to provide your direct email address via my direct email address. I'd like to have our customer join the discussion. .... Or possibly set up a 3-way call?

Regards, George

Hi George,

First I will study the schematic and get back to you. I am currently catching up with work as I have been out of the office and will be out again this coming Monday and Tuesday. Feel free to invite the end customer to join this forum discussion and include any additional information.

If we do not resolve the issue next week, then I will gladly support a call with you and the customer.

Thanks,

Alex

Alex,

  I've uploaded the doc "CAP_BD_GateDrive.docx". This doc provides waveform captures for various power-on scenarios and includes waveform captures for the gate of Q1 on the CAP board. As you can see from the waveforms the P-Ch MOSFET (Q1) is "off" when the LM25069 is attempting to start..... so the UltraCAPs are not being charged while HotSwap MOSFET is being turned on during the LM25069's initial start-up sequence or subsequent restarts.

  From my testing the ..... issue is related to the LM25069 being back-fed by the CAP bd. Temperature is also a factor, increasing temperature results in a greater possibility of the LM25069 to fault on the power-on sequence.

Regards,

George

Alex,

  Any update on this issue?  While the CAP board's Q1 gate waveform captures appear to show that Q1 is off during the LM25069's power-on sequence, to be 100% sure I plan on reworking the CAP board to allow monitoring of the current thru Q1 at power-on. I will post the results by 11/14. Would it be possible to have a 3-way call with the customer early next week?

Regards, George

Hi George,

I see now that Q1 in the customer's schematic is a SUB75P03, which is a P-Channel MOSFET and that the LTC4144 is an ideal diode (ORing) controller.

So this is my understanding of the first waveform shown in page two of the CAP_BD_GateDrive document (I believe this is the most illustrative of the problem):

Initially, the Q1 SUB75P03 is on with 5V enhancement since channel 2 is 5V and channel 4 is 0V and it is a P-FET. This means when the LM25069 is turning on, it is initially charging the ultracaps for a period of about 5ms. During this time, the LM25069 will begin regulating to its power limit and begin charging the fault timer. Since the power limit is set to 10W, this would mean ~1A (or larger) current regulation during initial startup. At this point, the Q1 FET's gate pulls up by LTC4144 and thus turns off the connection between the LM25069 and the ultracaps. This should allow the LM25069's output to continue to rise and turn on.

However, roughly 10ms later, the Q1 gate goes low. This would mean the FET is now on again and the LM25069 is trying to charge the ultracaps. It continues to charge the ultracaps with ~1A current (or larger) for 80ms (the fault time) and then shuts off.

My concern is why is the Q1 FET turning back on? Can you try replacing the Q1 FET (SUB75P03) with a Schottky diode (which should emulate the SUB75P03 + LTC4144) and see if the LM25069 powers up?

Thanks!

Alex

Alex,

 I left MOSFET Q1(SUB75P03) in the circuit but added a schottky diode ahead of it. This allows the LTC4144/SUB75P03 operation to continue "as is" but blocks the back-feeding of voltage to the LM25069. The system always powers on successfully with the schottky diode in the circuit. The back-feeding of this voltage is definitely a component of the failure to power-on.

Regards, George

Alex,

  I just realized that the above configuration could be masking the issue because once the CAPS are charged there's no discharge path. I will rework the board to replace Q1 with the diode and allow the CAPs to discharge back to the +12V rail but block the +12V from charging them.

Regards, George

Alex,

  I reworked the CAP board removing FET Q1 and replacing it with a Schottky Power Rectifier (MBRS340). This allow the CAPs to discharge into the +12V rail but blocks the +12v from charging the UltraCAPs.   The unit still fails to power on when the +12V rail is being back-fed by the UltraCAPs. When system's 12V power is turned off the UltraCAPS discharge back to the hot-swapped +12V rail keeping the system alive until the CAPS discharge to ~5v at which point all on-board regulators turn -off. The CAPs continue to discharge but a a very slow rate, if the system's main +12V is turned on with the CAPS back-fed voltage decayed to the 3.75-4.2v range the LM25069 will fault when attempting to turn-on it external pass FET.  I'm not saying that the range 3.75-4.2v is the only range that will cause the failure .... rather this is the range I tested with.

Regards, George

Regards, George

Hi George,

First I would like to say thank you for using our LM25069 device and having the patience to test and debug this issue.

Studying the customer's schematic, it does not appear there is anything connected to the +12V line other than Q1 and a small input capacitor of 22uF.

The back-fed voltage keeping the LM25069 above its POR should prevent the insertion timer from running (as you observed), but I do not see why this would prevent a startup.

I decided to test this on a new LM25069 EVM this morning:

This is the setup I used.

CH1: Input

CH2: Gate

CH3: Output

CH4: Current In or TIMER

For the output I used a 1N4004 Rectifying Diode with a 4V power supply to appply backfed voltage and keep the LM25069 above its POR.

Close up of diode at output:

Then I took waveforms of normal startup, and then startup with the backfed voltage applied.

Normal startup:

Notice the inserstion timer is running during normal startup, then the device successfully comes up.

Backfed voltage:

Notice the insertion timer does not run, and the device still starts up.

To narrow down if it is an interface issue between the LM25069 board and the customer, versus simply keeping the LM25069 above its POR on your circuit, can you try attaching the MBRS340 to the end of your board, backfeed with a 4V power supply as I did on the EVM and then try starting up?

If this test passes, then I recommend to investigate the end customer's board in more detail. Perhaps there are more devices connected than just Q1 and the 22uF. If this test fails, then I recommend investigating your board/design to see what differences between this and the EVM is causing it to fail this test.

Thanks!

Alex

Alex,

  Thanks for testing with the LM25069 Eval board and simulating the back-fed voltage. From my observations not only does the back-fed voltage a component of the failure but also temperature.  With your setup .... can you repeat the test with the back-fed voltage applied but also increase the temp of the LM25069 device. From my testing it doesn't need to approach anywhere near it's upper temp limit but does need to be increased above room temp.

Regards, George

Hi George,

I can run the test in a temp chamber on Monday.

-Alex

Hi George,

I ran the test at 60C and did not see any startup issues.

Thanks!

Alex

Alex,

 Thanks for testing at an elevated temp. I've also done some additional testing using an external supply to provide the back-fed voltage instead of the CAP card. With that configuration I also can't get it to fail start-up. My testing points to something on their CAP card being the issue. I plan on doing some additional testing in the next couple of days, looking more closely at the CAP card  circuits.

Thanks, George