TIDA-00367: SMPS testing

Update:

The following tests conducted with a 12V automotive halogen lamp load

@60W V=0V (SMPS shuts down, the sound of a cricket chirping at an interval of about 2.5s )

@55W V=19.82V (The same sound as in the case of a NO LOAD, but it is at a slightly reduced level)

@21W V=20.05V (the same sound as in the case of NO Load)

@ 5W V = 20.14V (the previous sound exists, in addition a kind of braking noise which is slightly more prominent)

NO LOAD V= 21.7V (the sound of a square wave, similar to a mosquito buzzing)

Update:

Tried capturing the MOSFET gate waveform with an oscilloscope: Please find attached images.

Funny to say, the weirdness as the scope probe was connected between the MOSFET Gate and Ground, the weirdness vanished. (That said, there is still a RESET happening but that's in the order of a few minutes.)

Could this probably be a decoupling issue ?

Thanks,

Manu

Update:

Following the tip with situation improving when the scope probe was introduced at the FET Gate, tried looking at the electrolytic capacitors, which looked okay, but nevertheless the electrolytic caps and the decoupling capacitors were replaced.

There was no visible change at all, with those changes.

The thought switched to common mode noise:

Looking at the 1000pF capacitor connected in between the primary GND and the Output GND, the capacitor seemed to be fine, measuring the capacitance.

Adding another 1000pF in parallel to it, reduced the issue very much, but the issue still does not go away completely.

What else can I look at, which contributes to common mode noise, if that's the culprit ? Is there a possibility of transformer degradation, worsening the Common mode noise situation ? Any thoughts, someone ?

No one's reading this thread ? Any suggestions ?

Thanks,

Manu

Feeling a bit silly replying to my own post, but hoping that someone would reply:

#1. Measured between C2 GND and T1 GND Pin 11. The spikes are recurring, I guess these spikes are the one RESET ing the C2000 controller eventually. The voltage is between 2 GND ends!. It does look a bit strange, doesn't it ?

#2.  The voltage waveform at the output of T1 Pin 8

#3. The MOSFET Drain - PRI GND waveform

#4. The FET Drain waveform with the LOAD connected and BLDC motor running.

#5. The same as #1, but after removing C1(1000pf) across the main rectifier D2, with the LOAd connected and running. The noise amplitude seems to have reduced in comparison to #1, but there appears to be more noise ?

Can someone please have a look ?

Thanks,

Manu

Feeling a bit silly replying to my own post, but hoping that someone would reply:

#1. Measured between C2 GND and T1 GND Pin 11. The spikes are recurring, I guess these spikes are the one RESET ing the C2000 controller eventually. The voltage is between 2 GND ends!. It does look a bit strange, doesn't it ?

#2.  The voltage waveform at the output of T1 Pin 8

#3. The MOSFET Drain - PRI GND waveform

#4. The FET Drain waveform with the LOAD connected and BLDC motor running.

#5. The same as #1, but after removing C1(1000pf) across the main rectifier D2, with the LOAd connected and running. The noise amplitude seems to have reduced in comparison to #1, but there appears to be more noise ?

Can someone please have a look ?

Thanks,

Manu

Update:

The problem mostly fixed by adding a Common Mode Choke at the SMPS output. Attached picture. So, the culprit was indeed Common Mode noise as identified in the previous post. That said, even though the problem has nearly been fixed, once in a while, some noise does really pass through.

Any suggestions to to filter it out ?

Maybe increase the number of turns, or maybe even use copper wires instead ? Any suggestions ?

Thanks,

Manu

Some more update:

I was able to get rid of the noise as well as the audible noise when I introduced a Load of ~0.55A.

Dummy Load schematic

With this load in parallel to the BLDC controller, all the odd behaviour with the BLDC setup vanishes away for good.

Looking at the UCC28630 datasheet, it appears that the audible noise as well the weird behaviour occurs most likely

when the UCC28630 is running between the P0-P1 region. Also, when the SMPS is operating in that region, there is a sound of a mustard seeds bursting open in hot oil. It is with that noise, the BLDC controller goes berserk and reset's itself.

The various modes appear to be controlled by Vcs, which is the the same as CS Pin Voltage ?

Can someone clarify, please ?

To increase Vcs can I simply increase the CS resistor value and expect that the UCC28630 moves operation away from P0-P1 region to P1-P2 and avoid the dummy Load altogether ?

Any thoughts ? Anyone reading this thread ?

Thanks,

Manu

Hi Manu,

I will contact designer to help answer this question, thanks.

BR

Jason

Hi Jason,

Much appreciated, awaiting feedback on the subject.

Thanks,

Manu

Hi Manu,

I have forward your questions to Designer S, Ramkumar (ramkumar.s@ti.com), or you can contact him by email directly, many thanks.

BR

Jason

Hi Manu,

Do you have a single output  or multi output?

Incraseing the Current sense resistor will help reduce this situation but not completely. It is better to put a 100 to 200mW dummy load at the output just to keep the system from moving into OVP.

Reading through your earlier post, I assume your motor driver board might have been misbehaving because of high ripple at the output when the system going into OVP.

Best Regards,

Ram

Hi Ram,

Thanks for the response.

I have the exact same transformer design, what you provided previously (in the other thread, link with the first post of mine. Sorry, about the multiple posts. Have been tinkering to find the problem and hence), without any changes at all. 5 and 4 turns on the secondary respectively. Have a TLV1117 connected to the 12V (4T) tap. A LED is alone connected to the TLV1117.

Tried putting a 0.33 ohms Current Sense resistor instead of the 0.12 ohms, but that did cause SMPS shutdown. The SMPS will run, I guess for about 1 second or less and then would shutdown. I guess that was OVP kicking in ?

A dummy load 200mW appears not to be sufficient at all. Tried different loads, with the 0.12ohms CS resistor. The driver needs a load of about 0.5A (empirically, the motor driver board works okay with 0.54A lamp load and not either at 0.35A a different lamp load) (at 21 - 20.6V of course).

Another thing to notice is that, adding a common mode choke with 2 capacitors also helped in reducing the issues, but that did not help completely though.

Wondering whether a design change in the transformer, in adding a shield and or a shield winding would help as in the UCC28630 EVM board ? To reduce the common mode noise ?

But in the given context, given that the primary sandwiches the secondary, shield would be necessary between both sides of the secondary winding ?

Just thoughts, would like to know what your thought on the following:

The Yellow color is the insulation tape. Brown color is Copper wire / foil. Orange color is copper wire that's used as a shield winding along with the Bias winding, as in the EVM design.

Having 2 shields reduce the efficiency ? A small mistake in the drawing that the top shield should have been between the next lower layer ?

Or is that superfluous altogether ?

Thanks,

Manu

Hi Jason,

Much appreciated.

Continuing the discussion with him. Issue still not solved, looking for a proper solution on the issue.

Thanks,

Manu

Hi Manu,

Let's contact designer again, sorry for waiting.

BR

Jason

Hi Jason,

Wondered whether he is active on the forum, hence tried sending him an email with the address that you gave me earlier. Will wait for his reply either way.  When one really needs to find a solution to a problem at the earliest, it seems the waiting has been for ages altogether. :-(

Thanks,

Manu

Hi Ram,

Any thoughts ?

Best Regards,

Manu

An update with a bit more clarification on the situation:

#1. BLDC running with a 10W lamp load attached. SMPS makes no noise. BLDC runs okay. This setup has a common mode choke in between the SMPS and the setup

#2. BLDC running *without* the 10W lamp load. SMPS makes noise somewhat like water flowing in a river (with a bit more high freq). BLDC motor runs. The setup has the same common mode choke in between. The controller runs, but the noise transients does cause abnormalities, such that it cycles through settings. The controller has a rotary encoder. The transients simulate the encoder being rotated/changed, which causes the settings to change by itself.

Measured the voltage waveform across the Current Sense Resistor in both cases.

Case #1

Case #2

Case #2, but expanded waveform

AFAICS, running the SMPS at 11kHz is causing the problem (audible noise, as well as electrical)

If I switch RCS from the current 0.12ohms to 0.1358 ohms based on these empirical tests, the SMPS will switch running from 11kHz to 30kHz ?

(And this change probably solve the issue at hand ?)

Can someone kindly answer ?

Thanks,

Manu

Hi Manu,

I am taking a look at this. Will get back to you shortly.

Bets Regards,

Ram

Hi Ram,

Thank you. Much appreciated.

Best Regards,

Manu

Update:

Changed RCS from 0.12 to 0.15 ohms (the noise increased to a kind of a square wave).  The same abnormality persists, but the rate of abnormality has reduced, but still exists.

(Without the lamp load) Attached waveform across RCS.

Yeah the frequency increased from the previous 11kHz. But this change brings in yet another oddity, that it shuts down the SMPS completely. unable to recognize a pattern there. But I guess, the controller assumes Over Current Protection probably ?

Thanks,

Manu

Hi Ram,

Accidentally came up on this document from Ray Ridley: 

Looking from the description given on Page #7 High noise and Low noise configuration;

In TIDA-00367,

W1 starts at Pin 1, ends at Pin 2; Pin 1 is the switching node

W2 starts at Pin 8, ends at Pin 9; Pin 8 is connected to the main rectifier, Pin 9 is connected to the tap and a small rectifier

W1 resumes at Pin 2 ends at Pin 3; Pin 3 is VBulk

W3 starts at Pin 10, ends at Pin 11; Pin 10 and 9 are bridged together, Pin 11 is GND

According to Rays document, W1's VBulk and W3 GND are quite close, which he appears to state as a High Noise configuration, that's what I think. Is my understanding correct (considering that we have a sandwiched configuration, which appears to makes things a bit more complex), to what he states there ?

If that's true, could that be the source of the noise and thus the problems ? Or is it something else ?

Can you please comment ?

Thanks,

Manu

Hi Manu,

What is the primary magnetizing inductance and the leakage inductance in your design?

Best Regards,

Ram

Hi Ram,

The design was taken exactly as you detailed here.

https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/234/TIDA_2D00_00367-Transformer-spec.docx

Thanks,

Manu

Hi Ram,

Any thoughts on this ?

Thanks,

Manu

HI Jason,

I see, ok. Will wait for his answer, after his vacation.

Thanks,

Manu

Hi Ram,

Like the desert longs for a rain, I have been waiting for a reply.

Thanks, reply much appreciated! Waiting for your email.

Thanks,

Manu

Hi Manu,

Not sure if you have reached out to Ramkumar and your question has been resolved. If yes, I will close this thread. Thanks

Hi Rixhard,

Talking to Ramkumar on the topic. Working on the issue.

Thanks,

Manu