LM5118: finalisation of design for production

Hi Michel,

Thank you for using E2E forums, I am not sure what is the question could you please elaborate on your question?

Thank you 

BR,

Haroon

hi,

thank you for replying to my post.

i asked multile questions i will try to better describe my questions please see the questions whritten below:

please use my first post for the puctures of the scopes and circuit.

1: at 1.5A the inductor is already heating up say about 40 degreas celcius. as are all the capacitors (ceramic and electrolytic) in the output circuit. frequency is about 290 Khz. the powersupply is rated for 3A.

  - will this couse more heat?

  - where does this heat come from? all the devises are rated for this current and frequency.

2: on my scope i saw that the Q2(second fet to ground) is switched on in buck mode Vin:30V at the beginning of the cycle under any load is this normal? why is it? and how does it work?

   - when i see a standard buck converter there is no FET to ground and it does not need one also. why is this switched on for a moment?

   - i ask this becouse it didnt make any sense to me to switch it on and i am afraight that i might have made a mistake on my board design.

3&4 the feedback curcuit... resistor devider from VO to FB and GND / the circuit with 2 caps and 1 resistor between FB and COMP.

- can i simply change the resistor devider circuit with ather resitors to get different output voltages?

- i ask becouse i want  to make a 12V version as well 

- i ask becuose i want to place a trimming potentiometer to fineley adjust to exact 24.0V

5: at testing i also did an output short test and saw that after a few seconds the output diode got really hot. output voltage was very low like 0.8V or so and current 7.5A. i will be placing a 3A slow fuse at the output in my final design for the case of an output short.

- question is this current limit of about 7.5A is really high for a 3A rated powersupply. can i change this? how?

6: i placed a TVS diode at the output just in case of a voltage peak. this TVS diode is placed away from the output diode and caps. and starts conducting at about 30V.

- can this couse problems in the long term say when caps are degrading and not all spikes are captured by the capacitors?

Hello Michel,

Please let me recommend that you check out your design with the Excel calculator: https://www.ti.com/lit/zip/snvu065

What is your input voltage range?

Here are some articles for layout optimization on buck-boost converters:
https://e2e.ti.com/blogs_/b/powerhouse/posts/four-switch-buck-boost-layout-tip-no-1-identifying-the-critical-parts-for-layout
https://e2e.ti.com/blogs_/b/powerhouse/posts/four-switch-buck-boost-layout-tip-no-2-optimizing-hot-loops-in-the-power-stage
https://e2e.ti.com/blogs_/b/powerhouse/posts/four-switch-buck-boost-layout-tip-no-3-separating-differential-sense-lines-from-power-planes
https://e2e.ti.com/blogs_/b/powerhouse/posts/four-switch-buck-boost-layout-tip-no-4-routing-gate-drive-and-return-paths

They are written for 4 switch buck-boost converters, but should be of good help for LM5118 as well.

Best regards,
Brigitte

hi,

input voltage range is about 18 to 35V say a battery voltage of a 24V vehicle.

i placed different resistors at uvlo to switch off at 18V 63K & 4K7 this works well now.

those aricles are very usefull this will be implemented in the fnal board. and the calculater is a great toll to check my design afterwards.

i have been doing more tests and found that the output diode was not recovering fast enough. i think i made it too hot by soldering becouse after replacing the heat went down a good bit.

question 1 is probably the diode i have to do more tests so for now done:)

question 2 i think it is becouse it is still in buck/boost transient mode at 30V please correct me if i am wrong.

question 3&4 feedback circuit is easy to fix witht he excel sheet so thats fixed for me

question 5 current limit is set by the limit resistor i saw in the datasheet could have known without asking.

question 6 the TVS diode in the output circuit is it safe to use or can it couse problems?

i have a new question after testing.

question 7 in boost mode Vin 18V and Iout 1.5A the converter starts but the output voltage never reaches the 24V it is making a noisy high pitch noise. the converter is not 'starting up'. when no load is attatched it starts up easy. with a small load it takes about 1 second to 'start up'. can you give me a clue as to what is going on?

thanks in advanse

BR.

michel datema

Hi michel,

Please add some waveforms for you questions.

hello guys,

i am now just doing more tests but am not learning anything from them. it behaves different than i expect.

 MR Feng JI. what waveforms do you want to see specificley? where do you want me to place the probes? please note that in my first post i put a lot of waveforms with the probes around the inductor.

with todays tests i placed the probes on the output diode again. i see a lot of spiking the diode is still heating up more and more.

when the circuit is under no load it easily starts up in boost mode. the freqency is varying a lot from 300 to 900Khz only stabilizing under a load.

finally while testing i broke of the input electrolytic capacitor at 27.5V by mistake. this fryed the lm5118 ic. 

i will be replacing it but i am either having the worst of luck or a bad design.

can anybody explain to me the varying frequency?

why the chip fries if the input capacitor braks off?

what i can do check about the spiking at the optput diode or is it normal?

what can be the couse of the buck startup problem?

Hi Michel,

Please specify the problem you met, we are not able to solve every problem you met when debug.

Best Regards,

Feng Ji

is it normal for the IC to change the frequency varying from 300 up to 900 khz when there is no load on the output?

is spiking on the output diode normal? i use 300 khz and BSC123N08NS3G diode

when a input capacitor braks off the ic goes dead... what can be the reason for this?

Hi Michel,

The frequency will not change from 300kHz to 900kHzwhen there is no load on the output. Do you have waveforms for this issue?

What does spiking on the output diode mean? BSC123N08NS3G is power MOSFET

It will be difficult to tell why when a input capacitor braks off the ic goes dead.

Best Regards,

Feng Ji

hi,

Feng Ji said:

The frequency will not change from 300kHz to 900kHzwhen there is no load on the output. Do you have waveforms for this issue?

Feng Ji said:

What does spiking on the output diode mean? BSC123N08NS3G is power MOSFET

yes i put the wrong part noumber  this is the diode i use: SBRD10200TR

i made a waveform of the output diode at 1.5A load (50% of the capacity) here the diode is heating up quite a bit. please note the high voltage drop...

i replace this diode a few times hoping on better resouts but none yet.

the probe and gnd from the scope were directley on the diode poles

with the spiking i mean the yellow line under tha wave. it does not happin every wave but some times.

the startip problem is as follows:

the devise has to switch on in boost mode due to UVLO. at about 18~19V under no load it switches on instantley and no problems occur.

when i switch on with a load attatched the devise only switches on at 27 or so volt.

i will place some scopes: yellow line is the gate of input mosfet // blue is gate of boost mosfet

boost: VI20V

buck:VI30V

startup problem:

please tell me your opinion how can i change this startup problem?

should i find another output diode? 

Hi Michel,

can you share your layout - i would like to check and ensure that the issues you have described are not generated due to layout issues.

Thanks,

 Stefan

hi this is the layout of this test board.

i placed a lot of components apart to be able to reach them easily.

in the mean time i learned a lot about the layout also the official version will be mutch different.

please comment on it:)

8540.schematic.pdf

HI Michel,

i see several issues in the layout which might trigger the issues you have observed:

- the Current sense (R2) should be connected to CS and CSG with a Kelvin connection (dedicated wires, both routed in parallel with a very low area in-between)

- the AGND should be connected via the Thermal pad to the common GND

- All AGND related components should be connected as close as possible to the AGND pin e.g.  RT (R9) has a very large ground loop a

- Vcc can GND connection would be better to directly connect to PGND

To the scope plots above:

Plot 1 (Spiking): why do you have a negative pulse here? The spikes are from parasitic of the circuit - most come from layout and reverse recover charge of the Diode

Startup problem - might be related to the to high sense resistor as mentioned above - the device might detect an over current as beside the normal load also the charging of the output caps needs to be done.

Last plot: this is the typical waveform when the SMPS operates in DCM (which could be the case with no or low load)

Can you be more specific on this point and add some scope plots showing that : frequency changes from 300khZ to 900kHz

Best regards,

 Stefan

hi stefan

Stefan Schauer said:

- the Current sense (R2) should be connected to CS and CSG with a Kelvin connection (dedicated wires, both routed in parallel with a very low area in-between)

i placed the wire but i did not see the cupper poore pooring over it thats my mistake can my tests still be relyable? or should i place 2 seperate wires to test this.

Stefan Schauer said:

- All AGND related components should be connected as close as possible to the AGND pin e.g.  RT (R9) has a very large ground loop a

i made the notes for the final board thanks does it make test unrelyable?

Stefan Schauer said:

Plot 1 (Spiking): why do you have a negative pulse here? The spikes are from parasitic of the circuit - most come from layout and reverse recover charge of the Diode

do you think the forward voltage is high on this? datasheet sas 0.7V this seems to be more. the diode is getting hot under 1.5A output load. do you think this is normal?

Stefan Schauer said:

Startup problem - might be related to the to high sense resistor as mentioned above - the device might detect an over current as beside the normal load also the charging of the output caps needs to be done.

i can do a test with a 0.01Ohm or 0.005Ohm both available. i was playing with the excel sheet and found this out:

buck 7.4A // boost 15.89A isnt that already really high? 

Stefan Schauer said:

Last plot: this is the typical waveform when the SMPS operates in DCM (which could be the case with no or low load)

i see 1.09Mhz really high. it makes a high pitch noise in this mode.

the load is here 1.5A@24V only the output voltage of 24V is not reatched.

the noise scares me a little all parts are heating up in this mode that not working in DCM mode

input voltage is here above 18V and under 26V 

this is the last scope plot in my previous post

Stefan Schauer said:

Can you be more specific on this point and add some scope plots showing that : frequency changes from 300khZ to 900kHz

i think you already anwered this can this be the DCM mode? or am i just reading the scope wrong in this case? its regarding the 1Mhz at the last picture of previous post

stefan.

i just checked my sense resistor it was a 0.005Ohm resistor

i placed the 0.015 per webbench designer and the area where the converter would not start went from 19.0 to 19.3V

with a 0.01 Ohm resitor there is no problem at all!

is it safe for me to sticke with the 0.01Ohm sense resistor?

Hi Michel,

is this right that you first had 5mOhm in - just wanted to ensure as this looks strange.

To check if the 10mOhm is save you need to check on the peak current limit you get then, with 10mOhm this gives

14.7A in Buck

30.0 A in Boost

So you need to check your circuit if this is able to handle this current, esp,

- MOSFET

- Inductor (I DC and I saturation)

- shunt power rating

Best regards,

 Stefan

hi stefan

thank you for replying. this is what i gound:

the mosfet can handle 55A so that works

about the inductor i dont really understand whats going on there can you teach me how to read this?

https://www.we-online.com/components/products/datasheet/7443632200.pdf <- datasheet :)

i can not understand how an inductor of this size would not fit into a small 3A buckboost converter 

Shunt is 3W 10mVx30A is 0.3W this is more than good enough.

BR.

michel

Hi Michel,

The MSOFET will be OK.

Also for the inductor you should not seen any issue for normal operation, here the max inductor current  and also the RMS current will stay below 4.5A. So you should not have an issue with the selected inductor.

The current limit given in Step7 of the Excel sheet are representing the max current just limited by the shunt and over current protection.

So in a failure cause (short at the output) you can reach that current before the controller detects it and shut off. 

In this case the inductor can already go into saturation or gets overloaded due to the to high current.

I am not sure for which sense resistor the above screenshot of the Excel quickstart calculator was done, but you may can check again.

Best regards,

 Stefan

Hi guys.

It looks like it has been fixed. I had the problem of the diodes heating up fast. They were bad i think. I tryed to find another diode thats a good replacement and found the pds760-13 had a lower voltage drop but the spiking is gone so i think that solved the problem. After that i placed the 15mohm shunt back and there was also no startup problems any more. Also it is using 3W less from the power supply now under the same load.

It is time for long period testing so see it it will be robust enough.

I will ofcourse update you on the resoults but for now i am already very happy with the resoults