TPS54620 Bode plot difference between actual measurement value and SwitcherPro simulation with Vout=1V condition

K. Sasaki14

Other Parts Discussed in Thread: SWITCHERPRO, TPS54620

Our customer observed Bode plot is different widely between SwitcherPro and EVM measurement result with Vout is 1V condition.
When EVM is defauly setting (Vout is 3.3V), the difference of Bode plot is small.

They modified EVM component value for 1Vout setting. Please refer to the attached file about component value.
The actual measurement result of that EVM is different from SwitcherPro result widely. We think actual measurement result with 1Vout condition is different from WEBENCH result too.

We would like to know the reason why the only 1Vout setting result is different from simulation widely.
Is there possibility that simulation model is not match with actual device at low output voltage condition?

Then phase curve drops suddenly at Gain is 0dB point in actual Bode plot. Do you have any opinion about this cause?

TPS54620 Bode plot.pdf

Best Regards,
Kohei Sasaki
.

over 8 years ago

0 JohnTucker over 8 years ago

TI__Guru**** 191445 points

I would be willing to bet that is some type of measurement artifact. We see those occasionally. First thing I would try is add 100-220uF electrolytic capacitor to the input.

0 K. Sasaki14 over 8 years ago in reply to JohnTucker

Guru 10200 points

John-san,

What influence would be removed by adding electrolytic capacitor to the input?
We can measure the bode plot with default EVM without any problem.
The bode plot of 1V output is the same as their customer board.

Best Regards,
Kohei Sasaki

0 JohnTucker over 8 years ago in reply to K. Sasaki14

TI__Guru**** 191445 points

When I see those type of rapid changes, they cannot be explained by the pole / zero placement of compensation or LC filter. If you are powering the EVM from a long cable, you shoiuld use some bulk capacitance at the input to the EVM. Try it to see if it works.

0 K. Sasaki14 over 8 years ago in reply to JohnTucker

Guru 10200 points

John-san

I checked again with electrolytic capacitor at the input to the EVM. However it was not improve.
Then it was improve by changing comp components value.

Best Regards,
Kohei Sasaki

0 JohnTucker over 8 years ago in reply to K. Sasaki14

TI__Guru**** 191445 points

Can I see the actual schematic and the resultant loop response?

0 K. Sasaki14 over 8 years ago in reply to JohnTucker

Guru 10200 points

John-san,

I will attach my test result.

Best Regards,
Kohei Sasaki

20151106 TPS54620EVM measurement.pptx

0 JohnTucker over 8 years ago in reply to K. Sasaki14

TI__Guru**** 191445 points

In that case you have rolled the loop response way back so you are not passing any signal at the higher frequencies where you were seeing issues.

0 K. Sasaki14 over 8 years ago in reply to JohnTucker

Guru 10200 points

John-san,

Yes. I adjusted to Fco becomes low.
Then does the gain and phase curve change steeply depend on comp component value?

Best Regards,
Kohei Sasaki

0 K. Sasaki14 over 8 years ago in reply to K. Sasaki14

Guru 10200 points

John-san,

Please refer to the attached file P.1.
The bode plot Gain and Phase curve of this TPS54620 circuit has rapid change as point 1 and 2. What is the possible cause about this?
You said adding input bulk capacitor, however it didn't improve though I add a 220uF electrolytic capacitor.
The bode plot improved by changing compensation components value as P.2. Thus I think this relates to compensation circuit.
Could you give me your opinion about rapid change of Gain and Phase curve on attached file P.1?

20151125 TPS54620EVM measurement.pptx

Best Regards,
Kohei Sasaki

0 JohnTucker over 8 years ago in reply to K. Sasaki14

TI__Guru**** 191445 points

For plot 1 it is normal to to see rapid gain and phase changes due to the power stage characteristics. Slope compensation causes the circuit not to act as pure ideal current mode control, so you can get rapid slope changes. The second peak I think is just noise or parasitics. I can usually get rid of those by tuning the input impedance (bulk cap). In the second case, you have rolled off the loop such that none of those noise or parasitic effects are passed. They are attenuated below the measurable threshold.

Simulation, hardware & system design tools

Simulation, hardware & system design tools forum

TPS54620 Bode plot difference between actual measurement value and SwitcherPro simulation with Vout=1V condition