TPS92692: Not starting properly in boost configuration

Gerry,

Please include your schematic along with all the design parameters such as input voltage range and output voltage and current range so I can check your design first.

Thanks Tuan

Here is the schematic. Please note that C7 has been changed to 82n

How do I send the excel design file?

Gerry,

You are switching at 800KHz so the following is what needs to be changed:

From the schematic, you should change the following:

1.  C7 to 0.1uF

2.  C9 to 0.1uF

3.  R12 to 0.05 Ohm

4.  R2 to 150k

5.  R13 to 3.01k

6.  For your power level you should add another 10uF in parallel with C2

7.  The inductor should have saturation current of 3.5A

8.  R11 to 45.3k

You might find out that at 800KHz switching frequency you will have high switching losses on T1 depending on your PWM dimming duty ratio but you can try this first.

Thanks Tuan

I used the TPS92692-Q1 BOOST LED DRIVER COMPONENT CALCULATOR spreadsheet to get the values for my design except I made a mistake on C7. How did you come up with the values you suggest.

I used the same spread sheet but there are parameters such as Soft-Start and that we prefer for it to Soft-Start slower so we do not trip the IS sense.  The IS resistor is recommended to be less than what the calculated value is.  It calculates 0.082 for R12 and you used exact values...Should use lower and I choose 0.05 for power loss reason also.  The value for Ccomp, I recommend 0.1 uF for low BW and stability first and you can recompensated if necessary.

Please follow my recommendations first.

Thanks Tuan 

Ok that makes sense, thanks

I had to order some parts so I will try these changes tomorrow and let  you know the results

Gerry,

We'll wait for your results.

Thanks Tuan

The changes seem to have got the converter up and running at full power but I have not confirmed the operation with an oscilloscope yet. Thanks for your help and I will keep you posted 

The converter is still not running correctly. It does start at close to full power but the switching waveform is not what is expected.

Then after about 30 seconds, it goes into an unstable state where the led is blinking.

I have oscilloscope screen captures but I cannot see how to upload them here.

Gerry,

Please do a search on E2E if you need help with how to load an image to it.  You can also attach a file to it.

Thanks Tuan

This is the first condition. The waveform at TP10 is shown

This is the second condition:

Please change R11 to 54.9K...Thanks Tuan

I didn't have a 54.9K so I used a 56.2K. It didn't seem to make much difference.

Gerry,

It looks like you could have noise on IS because you do not have an RC filter there.  Please see below section in the data sheet.  We if you can dead-bug that 100 Ohm and 1nF on the board.

Thanks Tuan

This mod will take a bit of work. Will you be available later today and/or tomorrow?

Why do you think the issue is with the slope compensation?

How does the slope compensation affect the duty cycle and switching frequency?

Gerry,

It's a current mode controller so if it reaches more than 50% duty cycle then it can have what they call sub-harmonic oscillations.  The slope compensation helps with this.  I'll try my best to help you with this so just reply here to E2E and I'll check every so often.

Thanks Tuan

The calculated duty cycle from the spreadsheet is less than 50% so the slope compensation shouldn't be a factor. I will try the filter anyway. Thanks for helping out, I appreciate it.

Gerry,

Make sure the cap is right at the IS pin of the IC to ground close to there.

Thanks Tuan

Yes that is what I am doing,

That didn't help

Actually it did seem more stable even though the duty cycle and frequency are still not correct. Also the inductor got very hot. Tomorrow, I am going to try and run it at a lower current level using the analog dimming and fixed resistors

Gerry,

The duty cycle will vary a bit because of the spread spectrum but it should be close in duty cycle from one pulse to the next.

With a boost regulator it should be close to 90% efficiency.  But the power loss should be on the main FEt and diode and not the inductor.

This part is a very mature part that has been used in industry with many programs.

Thanks Tuan

Are you a Texas Instruments support engineer?

I am TI’s application engineer for the TPS92692

Can you explain why the part in my design does not run at the correct frequency even from startup? I can understand that the duty cycle can be affected by feedback but not the basic switching frequency. I have disabled the spread spectrum and the switching frequency is still unstable. I chose this part because it was a mature part and seemed to have a good design tool. I do not understand what is causing it to not function correctly. Because it is not running at the correct frequency, the inductor current is going too high. My client needs to have a demo system running yesterday and I am at a loss as to why the part is behaving this way.

My understanding of how a boost convertor should work is the switching frequency is constant and the duty cycle is varied for output control. This part in this design is not behaving this way.  

Gerry,

Is the duty cycle more stable now or is it one narrow and one wide like you have before?  
Please obtain an evaluation board for the TPS92692 to help you test and evaluate your design if there is more issues here.

Thanks Tuan

The duty cycle is not stable. What appears to be happening is the switch is being turned on for too long so that the inductor current reaches the IS threshold, at this point, the switch is turned off briefly which causes the inductor current to fall and then the switch is turned on again briefly.

I looked at the evaluation board but it was not possible to get it to simulate our led configuration. I will attach more screen shots later but as far a I can see, the basic problem is the switching frequency. How is is possible that it runs at the wrong frequency?

Gerry,

You should be able to use our evaluation board with some minimum modifications for your application.  We will wait for your scope captures.

Thanks Tuan

I managed to get it working well enough using analog dimming and reducing the maximum output current to 1A. I am working on getting some boards ready for a marketing demo. I will get back to trouble shooting in a few days.