LM51772: Webench Design for a 5V DC does not work

Hi Mark,

If I2C is not needed, you could use LM51770, it should be easier to understand. And you could find very clear explanation on current limit in this apps note. Thank you.  Precise Current Limiter Over Wide Operation Voltages in Buck-Boost Converters

Hello Alicia,

First of all thanks a bunch for quick response! It's Alex here, and I am working with Mark in the same team (device architecture). Here is a complete story a way I see it:

1) LM51772 choice vs LM51770 was due to very strict PCB real estate limitations.
2) Schematic capture was completely copied from your Webench tool on June 23, 2025 (and yes, we did compare Webench schematic produced from Webench link on product page, and separately from Webench tool itself by specifying part and parameters).
3) We ordered x5 boards to prepare for client demo in October and since the rest of design is already verified from previous designs, 5V circuitry is the only "new" part.
4) We received our boards last week and unfortunately 5V supply is not working, at all - problem that we did identify immediately is incorrect lower resistor value in volage divider feeding Enable pin. As of today, September 22, your Webench produces correct value for this resistor, but by the same token generates slightly different values for the rest of this circuitry. Mark has all data captures from June 23rd as an evidence that we did follow your Webench data to the point.
5) We did fix this wrong Enable resistor value and 5V converter turns on now, but starts oscillate immediately.

We need your help and support/advice how to:

a) Fix boards that we have in hand (see schematic that Mark posted) - we absolutely can not miss our demo date. Barnacles, traces cuts etc. are ok, as well as components values change (we need your help ASAP)
b) Your advice and suggestions for next board re-spin, and we need it to be LM51772 (can wait a few days).

It is business related urgency, we hope you do understand.

Cheers,
Alex
 

Hello Alex,

I am sorry for the problems you have getting the device to work. Could you please do me a favor and check your design with the quickstart calculator that you can download from here: LM51772-QUICKSTART-CALC Calculation tool | TI.com

If you fill this in, it gives us the parameters of your power supply design and helps us to debug the issue you observe.

Best regards,
Brigitte

Hi Brigitte,

I filled in the "LM51772_Buck-Boost_Quickstart_Tool_V1_2_0.xlsx" spreadsheet as best I can. Where or how do I upload this?

Mark

Hi Mark,

You should be able to upload to this thread directly and share the link with us. Thank you.

No. It turns out your forum doesn't allow "xlsm" files. I had to zip it first. That would have been nice to know ahead of time.

LM51772_Buck-Boost_Quickstart_Tool-iPCB.zip

Please note that we are not using the secondary current sense defined with the Risns resistor.

Notes about the calculator data.

Since the Webench Design tool was used for this design, I have tried to "retrofit" the parameters in this calculator to best fit the Design Tool specifications.

• I have entered the data according to what the Webench Design tool originally specified. See Webench Design Report above.
• Where the Webench tool specified components that are not available or were too tall or too large etc. we replaced them using the Webench BOM feature and adjusted quantity as needed. See the resulting schematic above.
  I can provide a separate bill of materials if needed. LMK
  
  
• Operating Specifications. Are the intended design specification
  Switching Frequency is calculated from the Rt value based on Rt = 44.2K
  
  
• Filter Inductor. Some items like Cdiff is a calculated entry that can't be changed. Please review carefully the values in this calculator against our supplied schematic above.
  For Cdiff, we used 120pF.
  
  
• OCP, Sense Resistors, Slope Comp. Values are the default values except for "SLOPE" factor. I selected the "recommend" value of 1.
  Sense Resistance, Rs, is ambiguous, because this "Rs" resistor label does not appear in the calculator schematic. It's unclear if this is supposed to be the Rcs or the Risns(out) resistor. Since we are not using Risns, I have entered the Rcs value.
  
  
• Output and Input Capacitors The ESRs are taken from the capacitor datasheet.
  
  
• Soft-start time, tss I entered 3ms and is only and approximation.
  
  
• Compensation Design. The Desired Crossover Frequency is the approximate based value based on the recommended value.
  
  
• Compensation Components. These values are the Webench values.
  
  
• Efficiency/Power Loss Analyzer. We are using the Webench specified MOSFETs. There are 4 different MOSFETs used. They are;
  Q1 = CSD18540Q5B
  Q2 = CSD18512Q5B
  Q3 = CSD13202Q2
  Q4 = CSD17573Q5B
  Their values are taken from the datasheet.

Remember that this design was built and assembled and we are trying to debug the existing design and we ARE NOT trying to design a new circuit.

Please respond ASAP as this is very time dependent!

Do you have any updates for us?

Hello Mark,

Sorry for the late response.

Do you have the possibility to first remove the load from the device to check the converter alone?

When I entered the values of your output capacitance into the calculator, I observed that the softstart capacitor is too small for the recommended softstart time. Please check if the issue changes when you increase the softstart capacitor to 47nF or higher.

Please check the input and output voltage at the same time. And measure as well the switch nodes at the same time.

Best regards,
Brigitte

Thank you Brigette,

We don't have the ability of removing the load. The output is connected to an internal plane.

I'll source a 47nF to try. Specifically which nodes are you referring to?

Hello Mark,

Please increase the capacitor connected to SS pin.

Please measure the VIN, VOUT and SW1 and SW2. Normally SW2 should not switch at all, as the input is a lot higher than the output.

Best regards,
Brigitte

Hello Bridgtte,

I managed to replace Css (was 33nF) with a 47nF.
See Vout image below.

The period increased from ~3.6ms to ~5ms. 

Then I tried increasing Css to 100nF. See below.

The period increased again, to ~11ms. Vout still not working.

You asked me to verify Vin, (Vout above images) and SW1 and SW2

VIN

Vin is ~12.5V

SW1 
 
Images captured with auto trigger and single.

SW2

SW2 is sporadic. Single trigger capture shown.

I though I should add that the 0.95V regulator driven by this 5V regulator does regulate when the 5V is high enough. See image below.
The yellow is the 5V Vout and the magenta is the 0.95V out.

This should at least indicate that the 5Vout is not shorted.

As a follow up. We noticed that in the LM51772 datasheet, page 4, under CDC, It says "In case the current monitor is disabled connect CDC to ground". However, the Design Tool, has connected this pin to a 40.2K resistor to ground. So I have removed this resistor and installed a short to ground in its place, in order to test if this was the problem. It did not fix the issue.

Your prompt response would be greatly appreciated.

Hello Mark,

Could you please check the voltage on VCC2?

BTW, is SW1 pin connected to the SW1 between the buck-side transistors? In the schematic I can see some impedance that is not installed, but no other connection.

Best regards,
Brigitte

Hi Brigitte,

I measured VCC2 and it is 4.97V. 

Yes. That impedance (Z3) is a board etch feature and is not a real component. It is an 0201 copper short. Therefore it's a DNI for the bill of materials. SW1 is connected.

Hello Mark,

Ok, thank you for the information. Can you please measure the COMP voltage and possibly increase the slope compensation?

Best regards,
Brigitte

This is the Comp voltage, measured at Ccomp (C162). C162 is 0.33uF.

Can you suggest some values?

Hello Mark,

Could you please check the FB pin voltage? I am really puzzled why the converter is fully turning off when it reaches 5V and restarts afterwards.

Another test would be increasing the SS again as you have big output capacitors on the converter.

Do you have the possibility to measure the inductor current and the 2 switch nodes at the same time?

Best regards,
Brigitte

Hello Mark,

Can you please check the voltage on the EN/UVLO pin as well?

Best regards,
Brigitte

Hi

This is the FB pin voltage measured between Rfbt (R142) and Rfbb (R140)...

This is the EN/UVLO pin voltage measured between Rent (R148) and Renb (R149). 
Please note that I installed a 38.3K for Renb. I.e. It's no longer 1.96K as shown in our schematic at start top of this thread.

The SS capacitor is currently 100nF. I can search for a 1uF or something in that range?

We don't have the ability to measure the current directly. However I can show the voltage difference across Rsns (R169).
Yellow is measured at SW1 side of Rsns and magenta is the other side (SW2).
White is math function of SW1 - SW2. Ignoring the spike, the delta voltage is ~1.1V.
I measured Rsns to be 0.6R. So the current would be ~1.83A

Here is 5V Vout with Css = 1.0uF

Period is now 106ms. Vout is still not working.

Is there a way we can promote this to TI support that can work with us in real time (Canada EST)?
It is now critical to get this working and this 1 day delay between posts is not working.

Hello Mark,

I am sorry for the delay, but all experts for this device are in Germany. Nevertheless, you can contact the TI support via the phone numbers given on ti.com and check if you get local support.

There is a point which I overlooked up to now.

Do you have a local ceramic cap connected next to the VIN pin of the IC? If not, it is possible that VIN is seeing a lot of noise locally at the VIN pin causing VDET to trigger. So if VDET is not needed, you should disable it by changing the resistor on CFG2 from 20.5k to 8.3k.

Can you measure the VIN voltage very close to the IC VIN and PGND pins?

In addition, I recommend connecting BIAS to GND instead of having only a cap there.

What I as well do not understand is the amount of noise you measure on the output below 2.5V.

BTW, the voltage of the sense resistor measurement is just common mode noise as the voltage there should be a triangular shape. We are only capable of seeing the expected shape with probes that have very high common mode rejection as the signal is flying up and down with the switch node.

Best regards,
Brigitte

Hi Brigitte,

I would prefer to continue working with you. You've been a great help. I'm only concerned with the response delay due to the different time zones.
What time is best for me to work with you?

To answer your questions...
No we do not have a local ceramic cap next to the VIN.

I will change the CFG2 as you recommend. I first have to source this value. I should get that tomorrow sometime.

Here is the Vin measured at pin 36.

Here is the same pin zoomed in (AC coupled)

I will connect BIAS directly to GND and let you know.

Understood w.r.t. sense resistor measurement. I found it strange as well. ;-)

Thank you. Mark

Hello Mark,

The VIN pin needs a local ceramic capacitor. Please try this first. In your measurement it is still not at VDET level, but possibly it is close to your UVLO level.

I hope that the local VIN cap solves your issue.

Best regards,
Brigitte

I tried a different scope to look at the VIN at pin 36. This one is a bit clearer.

I'll try to figure out how to add this here...

Hello Mark,

Maybe at the via? Is this GND next to the via? If yes, I would scratch the solder resist.

Best regards,
Brigitte

Here is a clearer view of 5Vout with the other scope. This might help you identify the noise source/cause?

BTW I removed the BIAS capacitor and installed a wire short.
Now I'm just waiting on the 8.35K resistors to arrive tomorrow.

Hello Mark,

Still no idea where the noise is coming from. Sorry. Please install the additional cap on VIN.

Best regards,
Brigitte

I added 1uF ceramic cap to Vin at pin 36.
I changed Rcfg2 to 8.35K
Here is a picture of the mods.

This is Vin mesured at the new Cvin.

Here is the unchanged 5V Vout

Hello Mark,

Thank you for sharing the changes. I am searching for a shutdown feature that is causing the device to restart regularly. What I can see is that the device is not in hiccup mode, but it fully resets whenever the output voltage reaches a certain level. Today is a public holiday in Germany and I am on the street, so I am not available often today.

As you did the above measurements with 8.35kOhm, we can exclude VDET.

I think the next step is to check if OVP on VOUT kicks in. Interestingly, OVP2 would show this type of behavior, but by default this is at 55V. Can you please measure the VOUT pin and the output voltage at the same time?

Another possibility would be thermal shutdown. Can you please check if the device gets extremely hot? Be careful not to hurt you. If this is the case, please check how you connected the power pad.

Then the cycle by cycle current limit might kicks in. Then a possibility would be to update the filter on CSA/CSB to a higher tau. You mentioned that you cannot measure the current, does this mean you do not have a current probe? A possibility would be to borrow one from somewhere for debugging your design. Would be extremely helpful to debug any power supply.

Another possibility is using a high common mode rejection active probe to measure the voltage on the sense resistor, but these are expensive.

Can you check the inductor datasheet? Is this one possibly going in saturation with the maximum currents allowed on CSA/CSB (In your case this should be 50A at the 1mohm resistor).

Best regards,
Brigitte

Hi Brigitte,

Below is the image of both Vouts. Yellow is taken at TP18 and magenta is measured at pin 18.

I only have a standard DVM for monitoring current and it's limited to 10A. I will see if I can borrow one.

W.r.t. the temperature, I have a temperature probe and the IC measures only ~34°C. When I first started debugging this board I noted that the capacitors were getting very hot. To hot to touch. I even mentioned it at the start of this thread. They are no longer hot. My probe tells me ~32°C. I can't recall exactly when that stopped but I do recall that the IC was not hot at that time.

The inductor is a coilcraft XAL7070-102MEC. It's stats are:

Have a wonderful day off!

Thank you again. 

Mark

Just as a reminder, this 5V is driving a 0.95V power supply. See schematic below.

This is a TI TPSM82816SIER step down power module.

This 0.95V is driving a PCIE switch will consume between 2A and 3A, depending on the ambient temperature.

One more note. Yesterday, I added a 3Ω resistor load to the 5V Output connector, to see if it was possible that the PS was too lightly loaded. However, this did not change anything.

Mark

HI Mark,

i just reviewed the schematic you share in the first part of this thread.

I shows a component Z3 marked with DNI.

Can you please check and confirm if this is assembled or not or replaced with any other component.

Best regards,

 Stefan

Hi Stephan,

As mentioned earlier in this thread, Z3 is not a real component. It's a copper short, etched on the board directly. 
So it is connected.

Mark

Hi Mark,

sorry if I asked questions which has been checked already. I am just going through the whole thread again and try to dive into this issue.

I am just puzzled on an item above where you measured the voltage on the current sense resistor attached to CSA/CSB (we call this R_CS to differ to R_ISNS) = R169.

The schematic and calculator sheet shows that this should be 1mOhm but you said you measured 600mOhm.

Can you please double check this again. The 600mOhm would be far to much and if you have a voltage larger the 50mV across this resistor the device will detect over current and reset.

Measuring the voltage above this resistor is very critical and required an setup to avoid picking up noise as much as possible (short wires - shortest possible ground loop with scope ground clip (Tipp and barrel method  /   ground clip wire wounded around the clip, ....)

Can you please check this again?

The device does ramp up to 5V and the reset:

- is there anything switched on additionally one the 5V has been reached?

- from the other scope plot it looks like there is a 0.95 V supply generated out of the 5V but this has already reached and settled much earlier.

Best regards,

 Stefan

Hi Stefan,

This is my error. I remeasured as you requested, usimg my DVM  and I measure 0.1R. Sorry

The following is the measured voltages at Rsns. Hopefully this is a bit cleaner. Unfortunately I'm limited by my setyup and can't properly implement tip and barrel method. So here I have the scope probe gnd clips connected to the same GND and located very close to Rsns. 
Single trigger. The math cursors shows ~840mV delta.

HI Mark,

there is sill a large mismatch of between 0.1Ohm and 1mOhm. Not sure what your DVM is capable to measure but if you really have a 100mOhm as sense resistor the over current can trigger.

On the other side I would expect a triangular waveform on the sense resistor (which will represent the inductor current). Also here I am not sure if this is a measurement issue or something else.

Can you add to the plot above also add SW2 as additional channel or if not possible provide a scope plot it exactly the same condition with SW1 and SW2.

Best regards,

 Stefan

Hi Stephan. Yes you're correct. But since that's the limit my DVM can measure as you mentioned.
This Rsns part is a Stackpole Electronics CSNL1206FT1L00, RES 0.001 OHM 1% 1W 1206.

FYI I have a 2 channel scope (200Mhz) and another 2 channel scope (350Mhz).

It's not clear to me what you're asking me to measure. 

Hi Mark,

make the same input output and load configuration as you have made the screen shoots above (yesterday).

Then probe SW1 and SW2 - i would like to see the wave form of  SW1 and SW2 - this will provide the info which voltage is applied on the inductor and also provide the di/dt  on the inductor and voltage (dv/dt) be expected on the current sense resistor.

Please use the 350Mhz scope (for power stages the resolution of a scope is essential - otherwise information can get hidden easily).

Best regards,

 Stefan

Here is screen capture of SW1 (yellow) and SW2 (cyan)

This scope does not have the same math function (CH1-CH2) as the one above.

Hi Mark,

still struggling to understand what is going on here.

Can you please do / try to other things:

- zoom in in the above scope plot to the falling edge of SW2, so that we can see ~ 10 switching cycles on SW1 (5 left and 5 right of the falling edge)

- as the device seems to shut down when it reaches 5V (try to set the output voltage to a lower level, e.g. 4.7V or 4.5V) what happens then?

Is there anything enabled when the output voltage reaches 5V which would draw a lot of current ?

Best regards,

 Stefan

Unfortunately it seems that I have just shorted out this device with my probe. Urgh!

Hi Mark,

for me it looks like there are issues on the the Boost side - so the MOSFETs on the SW2 side.

I assume you get shoot throughs which are related to a mismatch of the MOSFETs as different types are used for high side and low side.

Unfortunately there are different footprints for both so they cannot be replaced with same types.

Can you measure the dead time of the gate signals for the high side and low side MOSFET of the Boost side.

Additionally you can already place gate resistors (unfortunately it looks like there are no footprint placeholders in the schematic / Layout).

With that you should also be able to compensate (at least to a certain degree the mismatch of the MOSFET gate charge which could lead to dead time violations and shoot throughs).

Best regards,

 Stefan

So I damaged the board I was testing, and I have to wait for the replacement boards to arrive, this week I hope. I will then have to redo the previous component changes. This would be a good time to review these changes to see if they can be “fine tuned”. I.e. I don’t think leaving Css at 1uF is good. Would you please review and LMK if there are others I should change at the same time. (Comp….?)

Once I’m back up and running I will measure the high side MOSFET gate of the boost side as you requested, I will let you know then.

Since there is no place for gate resistors, is there something more I can do to test these MOSFETs? Is it possible for you to simulate this?

Hi Mark,

I agree Css does not need to be 1uF. Would reduce to 100nF

I reviewed the quickstart calc sheet you provided above:

- the schematic shows 2 times 1000uF but in the quickstart just 1000uF are entered?

- Compensation looks OK.

There are several things which could be optimized in the schematic but would need a layout update:

- adding gate resistors

- adding placeholder for snubber.

When doing an update I recommend to ask here for an schematic an layout review. We can help here as well.

Best regards,

 Stefan

Hi Stefan,

I'm back with new replacement boards. I made the following changes to bring the original design up to the changes we previously made in this thread.
Rcdc (R128) = 0Ω
Rcfg2 (R130) = 8.35K
Css (C157) = 0.1uF
Cbias (C159) = 0Ω
Added CVin = 0.1uF

The following are measurements made at the MOSFET Boost side (SW2), as you requested.

Channel 1 (yellow) is taken at M3 gate (Q4-3) and channel 2 (magenta) M2 gate (Q3-4).

I forgot to mention Ren (R149) = 38.3K

Hi Mark,

to see the dead time you need to zoom in to see the time between one signal going down before the other went up (so more in the range of 10ns/div).

It looks like the device gets an reset and restarts when it is in BUCK mode. So also Buck side measurements would be good to seen (Q6 and Q5).

Have a review of the layout would also be good - esp. the routing on the power stage and current sense signals.

Best regards,

 Stefan

It's difficult to trigger on any specific pulse. So I can't identify which pulse these are. But here are a coupe more zoomed in.

I hope this is more informative. Is this what you were expecting to see?
Next... Q5 and Q6