Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Step Down 16khz noise output on VIN

Status
Not open for further replies.

player80

Full Member level 2
Joined
May 31, 2013
Messages
122
Helped
0
Reputation
0
Reaction score
1
Trophy points
1,298
Activity points
2,415
Hi,

I'm using a TPS54528 voltage regulator to get an 1.05V output, the input is 5V.
However once I apply some load (eg. 1.5 Ampere) I get a 16khz spike on VIN (the closer I go to the chip pin the bigger the spike will be).

So since it's 16khz it will result in a high noise, and since some other regulators are also attached to the 5V rail their capacitors will also pick up that noise it seems.

The 16khz spike is converted to audible noise by the ceramic capacitors, it gets stronger if I press a screwdriver against the component.

Does anyone have an idea how to dampen the noise around the regulator? I have added a ferrite bead to it but it seems like the 16khz spike is just converted to more noise that is backfiring.
 

Hi,

What about R13? Does it have the correct value? I can't check this because I don't have the schematic.

it gets stronger if I press a screwdriver against the component.
Don't press a screwdriver against the component.

Klaus
 

I don't see any meaningful debugging here. Do you have an
oscilloscope?

I might suspect that the load dependence of this "noise"
may have to do with the source you are powering the
converter with, and maybe you are dropping it into UVLO
due to inadequate input filter capacitance / quality, or a
current delivery limitation of the upstream supply. The
chip switching frequency is evidently fixed at 650kHz so
this is not subharmonic related. There appears to be no
user accessible frequency compensation, the part says it
is "constant on time" (hysteretic, PFM, I guess) with a
"quasi-constant-frequency mode" (whatever that is
supposed to mean). Heavy load should increase fSW
and I would not expect it to be down in the audible
range (but here a 'scope would be real handy).

How high is your output inductor's saturation current?
If you lose the L, from saturating, you will -slam- the
input filter and source with uncontrolled current, and
maybe start all this off.
 

I don't see any meaningful debugging here. Do you have an
oscilloscope?

I don't see any meaningful debugging here. Do you have an
oscilloscope?

I might suspect that the load dependence of this "noise"
may have to do with the source you are powering the
converter with, and maybe you are dropping it into UVLO
due to inadequate input filter capacitance / quality, or a
current delivery limitation of the upstream supply. The
chip switching frequency is evidently fixed at 650kHz so
this is not subharmonic related. There appears to be no
user accessible frequency compensation, the part says it
is "constant on time" (hysteretic, PFM, I guess) with a
"quasi-constant-frequency mode" (whatever that is
supposed to mean). Heavy load should increase fSW
and I would not expect it to be down in the audible
range (but here a 'scope would be real handy).

How high is your output inductor's saturation current?
If you lose the L, from saturating, you will -slam- the
input filter and source with uncontrolled current, and
maybe start all this off.

thanks for the hints, I will go through it within the next days (christmas/family time...)

I'm using the DR73-1R5-R inductor 1.5uH 6.52A 0.013ohms.

Should I cut the power trace and add a shunt resistor for measuring the current?

Is the assumption that the spike of the voltage on the input is caused due to a spike of the current on the output? (the voltage seems to be okay on the output) 1.05V.
 

attached a picture of the voltage at the SW output pin

105v_oscilloscope.jpg

according to this one:

https://www.ti.com/lit/ds/slvsay4c/slvsay4c.pdf

it doesn't look right...

could this be caused from the inductor?

In that case VIN was 5V
I have also changed VIN to 12V the output looks better on the scope (of course referencing 12V) but there's still some noise.

well can anyone recommend another part to provide 1.05V (from a 5v rail) at 3-4A which is not generating such a noise? Maybe the Texas Instruments part is just bad in general.

The entire logic circuit is okay and works properly (also the 1.05v output looks very stable) but I don't want that annoying 16khz noise. It's not very loud but if I go close with my ear I can hear it.
 
Last edited:

TPS54528 has no feature to reduce the switching frequency, in so far it's some kind of irregular operation.

The SW waveform suggest insufficient or too distant bypass capacitors at the VIN pin, did you follow the layout suggestions? By nature of the buck converter, the output current is commutated between VIN and GND, you need sufficient bypassing directly between these pins. Also, do you have an appropriate connection of the exposed pad?

There may be other issues, e.g. related to the feedback loop. Wee should look at the layout and schematic to identify possible problems.
 

There's not much of a "design" left anymore.
I have loosely soldered the inductor directly on the SS pin and cut the traces from the PCB; there's still some noise there.

The initial design I have copied it from another reference design (non TI) to power an ASIC, that board had several issues with the power supply and I just thought I had bad luck but the copied version just had the same problems... during development however I focussed on my FPGA design rather than the power system.

I have ordered the TI reference design now (and recommended 1.5uH coilcraft inductor). The DR73-1R5-R was used in that particular reference design.

I have tried to swap in some other inductors, 4.7uH .. the noise seems to be a bit different, 22uH the noise became much louder. I'm limited with inductors at the moment so I'm waiting for the mouser package.

I have also tried to add more capacitors to VIN but it did not improve the situation.

The whole circuit basically meets the TI reference design (the layout is more or less freestyle due to the cut traces..)
 

There's not much of a "design" left anymore.
I have loosely soldered the inductor directly on the SS pin and cut the traces from the PCB; there's still some noise there.
??? I presume you mean SW rather than SS pin.

Unfortunately the observed problem hasn't been yet clearly reported. The oscilloscope screenshot shows some high frequent (MHz) ringing, you talked however about 16 kHz noise in post #1 which can't be assessed on a 50 ns/div waveform.

Ringing at SW pin isn't primarily caused by the SW wiring, it's more likely copied from VIN because VIN to SW switch is closed at that time.

High frequent ringing will be mainly caused by not realizing the blue marked current path between red GND and VIN pins through bypass capacitors as short as possible.

tps.PNG
 

Sorry I mean SW pin yes.

I don't think there's a problem on the layout since I have tried several variations.

It should be a problem of:
a) possibly wrong components (inductor, different inductors returned slightly different tones)
b) issue with the TPS54528 in that particular design/usecase.

I talked with some other engineers they said:
- it's very likely an issue of the wrong inductor / low quality inductor (for that use case)
- they would not care if there's some low acoustic noise.

What I have observed about my initial circuit (before loosely adding some components and cutting traces):
the ground was really bad, however I have completely fixed that (manually drilled holes to the regulator and filled it up with solder - and there's no change at all on the noise.

The overshoot and undershoot spikes are firing back to the power supply, it's strong on the VIN pin and it gets weaker after the first few capacitors and disappears after the ferrite bead (from VIN to the 5V rail).

The next step is:
I'm waiting for my mouser package where I have ordered several regulators, and different inductors for testing.
However not sure when the package will arrive, Mouser has some weird delivery times to my place sometimes it takes up to 2 weeks (I'll update the post once all the items arrived)
 

Hi,

I don't think there's a problem on the layout since I have tried several variations.
the ground was really bad, however I have completely fixed that
And the fact, that you did not yet show us your PCB layout .... lets me assume: There is a good chance that the problem is caused by the layout.
All sounds like a PCB problem, although you think it is not the case...

Klaus
 

I'll update this discussion once the parts arrive, everything else is speculation at the moment.
Once the TI EVB arrives I can also swap in the current parts to check if they are okay.
 

I'll update this discussion once the parts arrive, everything else is speculation at the moment.
Once the TI EVB arrives I can also swap in the current parts to check if they are okay.

ok the parts arrived, the EVN did not produce any noise (initially).

After replacing the inductor on the TI EVN with the inductor recommended in the 3rd party design (DR73-1R5-R) the EVN also started to produce the same noise. So we'll just swap in the new inductor now and it should be okay.
 

Sorry I was a bit too quick, I did not apply the load correctly - the noise is still there and also on the TI EVBs.

Originally I used 2 TI parts TI54528 (5V to 1.1V) and TPS54335 (12V to 5V).

I have replaced the TPS54528 and with an MP2145 part and got rid of the audible noise in that area of the PCB.
Since I could not get an EVB for the MP2145 part I cut it into a single layer PCB and soldered it as a dead bug - and there's no noise coming from it.

Now I figured out that the TPS54335 also creates some audible noise. This part even supports changing the switching frequency, but even with a switching frequency of 1 MHz I still hear the noise in particular on the EVB.

So my conclusion at the moment:
the noise is caused by the SW - VBST connection on the TPS54528
on the TPS54335 the noise is caused by capacitor between PH - BOOT.
Using an MLCC in such a high frequency environment is not really suitable?

The switching regulator is absolutely okay from the functional side - the output voltage is absolutely clean, the input voltage is also okay.
 
Last edited:

Hi,

If there is audible noise, then most probably the regulator is not continously (clean) switching.
* either it is in hickup mode (high load current) or in burst mode (low lowd current).

Jumping load current or jumping input voltage may also cause discontinous operation.

*******
cut it into a single layer PCB and soldered it
If you are an experienced SMPS PCB designer (and you are aware of current paths and HF behaviour) you may create a suitable single side PCB layout.
If you are not that experienced in this field, then a single side PCB layout may work "by accident". But
* The function will not be reliable (may fail at different input or output conditions, varied temperature, after weeks, months...)
* most probably will violate EMI / EMC regulations

Klaus
 

Hi,

If there is audible noise, then most probably the regulator is not continously (clean) switching.
* either it is in hickup mode (high load current) or in burst mode (low lowd current).

Jumping load current or jumping input voltage may also cause discontinous operation.

*******

If you are an experienced SMPS PCB designer (and you are aware of current paths and HF behaviour) you may create a suitable single side PCB layout.
If you are not that experienced in this field, then a single side PCB layout may work "by accident". But
* The function will not be reliable (may fail at different input or output conditions, varied temperature, after weeks, months...)
* most probably will violate EMI / EMC regulations

Klaus

That breakout board was just for testing the MP2145, it's better than nothing... I'll add that in the next PCB version.
Right now the only way to solve this problem is to replace the components with non TI versions for me..
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top