TPS62125 in Inverting Buck-Boost Configuration, need help with Tina simulation.

Chris thanks for the reply.

I do have other voltage rails I could tied into, 5v and +12 boosted from battery.  All running on ti parts for low Iq.  I initially wanted to be able gate this supply off separately, but there are options.

Thanks for introducing the tps62135.  The  peak voltage I expect the FETs block off is -10 V,  marginal if the tps62125 is used. It comes down to how much margin I feel comfortable and static current drain +5uA, I guess there are better low hanging fruit to gain that back.  Do you happen to know another part that would out do on the two figures?  Thanks!

Chris,

How's going? Want to check in and pick your brain again.

We got a test board back, and are trying the inverting boost with tps62135.  The schematic is attached. 

The inductor is Murata LQH3NP series 1uH part number LQH3NPN1R0NJ0L.

Resistors are 0603 thin film. R1 = 5.64M and R2 = 350K, for a 2uF feedback current cost (accuracy is not critical).

Output caps are parallel 0805 10uF with minimum 20V+ rating.

Input cap is a standard 0603 X7R good for input voltage of 5V+

Output PMOS discharge switch for shutdown is tied to Vin, so that it is off and not loading for this test.

The output load is a 2K resistor to ground drawing a max of 6mA from output at  -12V

The issue on immediate functionality is that, this circuit take 30 seconds to come up to a stable point (~-10V) and the end point transition has a significant overshot as shown in the attached waveform snapshot.

The engineer tested this circuit used only 3V input supply, though the 3V is stable and low impedance from a power supply.

Please share your thoughts.  Appreciated.

Best

FI

Hello Chris,

Things did not improve, here are some observations:

Testing the implementation at 5V input did not improve the outcome.  The photo shows the supply at input and output trace over about 30s.

The SW node waveform reveals long switching interval. The interval is longest at initial startup and becomes shorter and shorter as the output is coming up.

The initial startup SW switching interval is shown here with a 1nF SS: ~680us

The initial startup SW switching interval is shown here with a 10nF SS: ~410us

In a steady state condition after startup, the switching interval becomes about 20us

The input waveform is pretty stable driven by a power supply at 5V.  The average input current was at 17mA for a 2.2Kohm (~5mA) load at output at steady state. 

In addition here is a zoom-in on a switching pulse at SW node.

Please advise on what's maybe happening, we need to decide whether we need to change our design asap.

Thanks

FI

Hi,

i did a startup measurement on the TPS62135 with following conditions:

Vin: 5V,

Vout:-12V (same resistors as above)

Load: 2k

Cout: 22uF

Cff: 50nF

Blue: Vin, Light Blue: Vout   Green: Icoil

With this conditions i couldn't see any issues. Could you generate the same plot?

Thanks,

Michael

Michael, your results are encouraging, thanks for the support. Your traces looks good and the part switches nicely. Just to make sure the comparison is strict.

Can you confirm the connections in your test are exactly the same as the given schematic in the earlier post, MINUS the Cff (I assume it is Css for softstart)? And if Cff is for soft start, could you confirm with the SS pin open, 1nF and/or 10nF? Really appreciated. Also is your version of the IC the straight up tps62135 or the tps62135(1) version?

Putting some expectation on this on going communication: The engineer who is hands-on with this block now is at his night time, I am designing and facilitating around the clock.

Hello Chris,

Here is what I have for now.  I will populate and update once I have better information

Here is a waveform trace of the SW and output at comparable scale minus input for now.

Again, here is the schematic

Here are snapshots of the layout, dark green is ground

Top Layer

Inner Layer 1

Inner Layer 2

Bottom Layer

Immediate BOM information

Please let me know if you see something jump out. 

FI

We don't have a current probe for the scope.  We can only obtain voltage trace right now. Let me know the best way to present this information.  The engineer still had no luck getting it to work as intended just now. He made another board with a 47nF CSS,mimicking Michael's test, the behavior is still the same taking 30s to come up.  He is assembling information per the above list, but in general, the caps are all rated at 25V X7R or X5Rs.  He did try lower value feedback resistors though yielding the same result.  The trace you pointed out is not long, pretty much ends at the edge of the capture, it should be 0.4mm if not any bigger, I can widen that in the coming revs, but the important thing is that, the test is done by soldering wire directly at the input cap.  Thanks again and let's figure this out.  Best.

Our engineer would like to get a close up picture of Michaels setup just to compare and see anything is wrong with our board.  Thanks again.

Hello Chris and Michael,

Made another board, same un-functional behavior, setup exactly like Michaels with a 47nF Css.  Engineer wants to have a close up on Michaels board.

Here is the updated scope capture

Here is the updated BOM

Hi,

below a picture of the board which i used for the measurement. 

I tried to use same values and same connections as on your schematic. The device is starting as shown below.

Which inductor do you use? on the schematic is a 1uH but on your BOM is a 2.2uH

Regards,

Michael

Hello Micheal and Chris,

Just want to close the loop, I have received, assembled and tested two boards, both are working fine.  Though I am trying to come up with better topologies for lower Iq and in-rush current, the inverting boost tps62135 per this design yields (120uA no load) and in rush current (> 1A).

Again, thank you very much for you support!  Best!

FI