TPS54620: Device not working, oscillation and high current drain

Hello Marco,

I can not see your schematic as the link is blocked by TI site.

Can you attach the schematic instead of a link?

Regards,

Yitzhak Bolurian

Hello again,

It is also recommended to increase the input voltage gradually, starting at Vin=0, and check at which point it starts to draw high current (500 mA).
Please try this and advise.

Regards,

Yitzhak Bolurian

Hi Marco,

Nothing obvious is wrong with the schematic.

Your soft start time is set to about 8mS, so your Cout charging current should be only 1/8 Amps.

Please try to raise the input voltage slowly and see what voltage level triggers the high current consumption. 

Regards,

Yitzhak Bolurian

Thanks, I'll try, but do you think noise from chip (maybe oscillations?) and high current consumption are related?

Hi Marco,

We have no evidence of that yet. 

Let's try the gradual increase in Vin first.

Regards,

Yitzhak

Hello,

from 0 to 4.5V: no chip noise and no current comsumption

At 5V noise start and current consumption is about 50 mA with no load

At 6V noise is stronger and current consumption becomes very high (more than 100 mA) with no load.

Noise change frequency with input voltage.

Hello Marco,

Now we know that the circuit draws high current only after switching gets started.
You may be right, this could be caused by stability issues and oscillation.

You are having a two stage output filter on both stages.
That can lead to a lot of phase shift that may lead to stability issues.
As the first debugging step, I would suggest that you by pass L10 and L12 with short circuit/wires in parallel to those and see if the oscillation is gone.

Regards,

Yitzhak

Hello, 

I removed L10 and replaced it with a 0 ohm resistor. The problem is still here, noise and high current consumption starting from 5 V.

We could try to change switching frequency to a lower/higher value and see what happens. Or we could change the compensation network to lower the compensation pole. R147 takes a role in that network and is easy to change.

What do you think about? I have no more idea, please give me some other debug steps.

Regards

Hello Marco,

Yes, changing the compensation seems to be the next step to stabilize the loop.
Please give this a try.

Regards,

Yitzhak

Hello, I'll try as soon as possible.

A lower frequency compensation pole and zero should be desidered, right?

Hi Marco,

Yitzhak will check it and reply you soon.

Hi,

removing that cap helps a bit, I mean the chip is stable for 1-2 seconds after power on than starts to oscillate, but with less current consumption.

Finally I can meaure voltage on the output, so this is a nice improvement, 5V and 2.5V are there but unstable.

I noted that only the first device U55 make noise, while U56 not. Maybe only U55 is unstable, and the oscillations propagates on the output of the second device. I don't know why, anyway I'll play a bit with compensation network.

Thanks, 

regards.

Hi Marco,

Please share your test progress when you got any updates.

For now, I will close this thread.

Hello,

the loop is still unstable, I changed the passive components according to the tool, with a good phase margin, but still noise, oscillations and current consumption. The noise comes always for the first chip U55 but not from the other one.

I used 470pF for C262, 1K for R147 and 100 nF for C261 with no luck. 

I don't know to do, really a strange situation, maybe a defective chip?

This is my only opinion at the moment, however is not available on the market actually, so is difficult to change and try.

Hello Marco,

So the second stage is working and stable now?

If so, can you swap between the two ICs?

You can order a TPS54620EVM-374 EVM and run your stability test or swap between the ICs as well. I have checked and it seems to be available.

Regards,

Yitzhak

No, the second stage is not stable, that chip doesn't make audible noise but I see oscillations with the scope.

I don't know if it's unstable itself or the oscillations of the first stage propagate to the output of the second.

It's not so easy to swap between the two ICs, without to damage them during desoldering process.

Hello Marco,

You should disconnect the EN of the second stage from the PWRGD of the first stage before you check the second stage.

Also the new compensation values that you have selected are a bit atypical, and set the Fco frequency to less than 2 KHz!

More typical compensation values could be, 4.7pF for C262, 41.2K for R147 and 5.6 nF for C261 and 27 pF for C258.

Sorry that I can not help you with the IC availability.

Regards,

Yitzhak

Hello,

just a summary of the situation: actually I removed U55, so only U56 is mounted on PCB.

Now U56 is making noise, so there is a common problem to both stage, a design issue, this is the first evidence.

I can observe 1 V p-p oscillation at 10 kHz on the output of U56 that should be 2.5V. Is a perfect sinewave centered at 3 V with that audible frequency.

Current consumption with no load is about 200 mA (abnormal). Compensation value: R151=6.34 K, C272=100nF, C273=68pF, C269=not mounted

According to the tool and datasheet they are good values, so the loop should be stable.

With a higher switching frequency same problem and IC draw more current (300 mA) with no load.

There is a design issue somewhere and I don't find it, is an important university project I have to solve it.

Here a picture of the PCB. I don't see any big mistake, could be better of course.

Hello, sorry for the late, I've been busy with other jobs.

In the top layer there is a different GND plane, it's isolate from dGND of the TPS converter, so I have to use vias.

The input voltage is absolute clean. Yes, I know the max value for EN is 6 V, but in the design is floating, there isn't a pull up to VIN.

I can't still understand why it doesn't work.