This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

TLV627432: Hardware troubleshooting, output unstable

Genius 16199 points
Part Number: TLV627432
Other Parts Discussed in Thread: TCAL9539

Tool/software:

Hi Experts,

Can you advise this customer's report about TLV627432YFPR?

I am having trouble getting my 1.8 V power rail to start up successfully. I am following the datasheet design guidelines quite carefully, including the layout. I'm using a 2.2 uH inductor (Samsung CIGT201610EH2R2MNE). Nominally, my output capacitance (local to the IC) is 10 uF X5R, as recommended. However, there is additional load capacitance (~35 uF) some distance away on the board. This is in service to an EFR32MG24 RF MCU I'm using under this supply. There is also a series 30 nH inductor in the supply path between the regulator and the 35 uF capacitor to isolate the supply domains at RF frequencies. Lastly, I have two 4.7 uF ceramic capacitors at the input voltage.

I have the voltage selection inputs configured for 1.8 V, but I am seeing the output read 3.4 V. Is this consistent with instability due to load conditions, or is there some other possible issue? I tried removing MCU capacitors and shorting across the 30 nH inductor, but the outputs did not change. During testing I increased the input capacitance to 100 uF to ensure the input is very stable. That did change the output from 3.4 V to around 2.3 or 2.4 V. I also tried to configure the device to give 3.3 V out, with no change to the outputs I described above. 

Finally, I completely removed the MCU from the load (and associated 35 uF decoupling capacitors) by removing the 30 nH inductor. This made the output settle to the desired voltage. Obviously, this isn't a useful configuration. 

Is it likely that the outputs I've described above are a result of instability due to load conditions? Or is there some other possible reason? Can you comment on how inductance in the supply path might change the stability of the regulator? 

Please let me know if schematic or layout screenshots would be helpful in troubleshooting.

Regards,
Archie A.

Here is the project shared: 1070.Attachments.zip

  • Hi Archie A,

    Thank you for using E2E,

    I reviewed the schematic and layout and didn't find any concern.

    Can you please share me more information.

    1. Is there any other device connected to this 1.8V rail?

    2. Can you please share me the scope shoot of VIN, SW node voltage & VOUT during the abnormal VOUT condition.

    3. Can you please try to replace 30nH inductor with a ferrite bead and redo the measurements.

    Thank you & regards,

    Moheddin.

  • Thank you, Moheddin.

    1) There are a few other things connected to the 1V8 supply, but all are much lower power and have smaller decoupling capacitors compared to the MCU I referenced. U27 is an I2C expander (TCAL9539) interfacing the MCU. This is powered from the 1V8 supply. There is one resistor divider generating a cheap mid-rail reference voltage (schematic page 3, R70 and R71). The other connections are mostly 10k Ohm pull-ups for open-drain output buffers, e.g. R14, R16, and R17. Lastly, there are some level shifting logic gates (U15, for example) which apply outputs from the MCU to other voltage domains. During power up, I expect the MCU and I2C expander to have IOs configured as inputs. The MCU does have an indirect control mechanism for disabling the power domain, but I removed that path in testing to confirm no interaction between the MCU and the regulator's EN pin. The EN pin measures stable at the input supply voltage (3.7V) during testing. There are provisioned outputs to allow the 1V8 supply to power other boards, but those are not in use now. 

    For 2 and 3, I don't own an oscilloscope, but I expect I can find one to use temporarily. I will try to get the requested screenshots in the next few days and reply accordingly. 
    Regards,
    Archie A.
  • Hi Archie A,

    Thank you for the detailed explanation. Unless we have the requested measurements, I cannot conclude the cause for this incorrect VOUT.

    Please update me once you have the measurements.

    Thank you & regards,

    Moheddin.

  • Hello Moheddin,

    The screenshots are attached. All screenshots are taken with the o-scope AC coupled. When DC coupled, the oscilloscope showed values very close to the earlier stated measurements. Ripple and ringing on each signal were fairly low in amplitude (e.g. < 30mV), with no obvious switching events or other large transients.

    We also checked the signals on the enable and voltage select pins, and all were stable at their expected values. 

    8877.captures.zip

    Thank you.

    Regards,
    Archie A.

  • Hi Archie A.,

    I will look into it and provide you the feedback by end of the day.

    Thank you & regards,

    Moheddin.

  • Hi Archie A.,

    the shared waveforms does not help, all the measurements are done using AC coupling which only shows the noise and ripple,

    Please provide me something like below, in DC coupling. All VIN to GND, SW to GND & VOUT to GND in same scope shot.

    1. VIN = 4V, VOUT= 1.8V, IOUT = 0A

    1. VIN = 4V, VOUT= 1.8V, IOUT = 0.2A

    3. Can you please try to replace 30nH inductor with a ferrite bead and redo the measurements.

    Can you also try this?

    Thank you & regards,

    Moheddin.

  • Thank you, Moheddin.

    Customer responded:

    I don't believe that AC coupling is an issue with the screenshots. The scope (Rigol DS1104Z) has a high-pass corner frequency below 10 Hz in AC coupling mode. I don't think switching behavior in the screenshots you showed would be attenuated. I made some effort to ensure I was not missing any switching behavior by using AC coupling. Furthermore, with AC coupling, switching behavior from other regulators (e.g. my 5V regulator) was clearly visible on the scope. The Vin sawtooth ripple is shown clearly in the images I sent, for example. All voltages were stable close to their DC operating points; the switching/AC behaviors on top of these static voltages was small enough that AC coupling was the only way to capture them in detail. 

    Rather, I think the issue is the IC is not operating as intended, i.e. it is not switching at all. I checked the enable and voltage select pins, and all voltages were static at the expected voltages. Is there some other condition which would cause the IC to latch the output close to Vin? Perhaps an issue with device damage or overvoltages? I have tried multiple boards with the same result, so it seems unlikely to be an ESD issue. 

    I am also wondering now if there is a parasitic path elsewhere on the board driving the 1V8 node to the Vin voltage. I will study this possibility, but no such path seems obvious to me at present. 

    Regards,
    Archie A.

  • Hi Archie A.,

    The possible reason for the VOUT close to Vin is when the VOS connection is open or when the HS-FET is damaged due to overvoltage's spikes at VIN.

    Can you solder the device on the EVM and do some measurements to make sure if the device is working or damaged ?

    Thank you & regards,

    Moheddin.

  • Hello Moheddin,

    Apology for the delayed response.

    We have been investigating other possible causes for the issue, and found the issue related to another chip on the board. We are now able to get the 1V8 supply to come up correctly. It seems the regulator is working as intended.

    Thank you for your great technical support!

    Regards,
    Archie A.