TPS92692-Q1: Voltage range issue

Hello Andreas,

33V input should not be a problem if designed correctly so something is going on. Can you tell me what the LED string voltage and current is set to? What switching frequency are you using? Could you share a schematic of your prototype for me to look over for any issues? Do you have any FET drain and LED current waveforms taken while it is working well and while it is flickering?

Thanks,

Clint

Hi Clint,

Will send a schematic and additional information as soon as possible.

Kindest regards,
Andreas

Hi Clinton,

Thank you very much for the reply.

I attached the schematic, PCB layout and the TPS92692-Q1 Buck-Boost LED driver calculations. As you can see in the schematic the TPS92692 should drive 3 strings of 8 LEDs in series. The 3 currents are balanced via some simple current mirror(s).

I tried to look at the FET drain and LED current waveform but the prototype is built on a single layer aluminium PCB and as soon the LEDs come on it’s hard to do any measurements. I will check if I can disable the LEDs and use a resistor instead.

We use Cree XPGBWT-L1-0000-00GE2 LEDs at around 1000mA.

Hope that helps.

Kindest regards,

Andreas

Ultima 180 Schematic.pdf

Ultima 180 MCPCB.PDF

Copy of TPS92692_BuckBoost_Design_Calculator_Andrew.xlsx

XLampXPG2.pdf

Thanks,

I will try and look over this either this afternoon or tomorrow. I will let you know if I have any thoughts then.

Regards,

Clint

Hi Clint,

Thank you so much. Your help is greatly appreciated.

Cheers,
Andreas

Hello Andreas,

There is nothing obvious here unfortunately. Layout could have something to do with it as metal core boards are difficult with switching regulators due to the switching edges. But if it were the only factor then things would not work better at lower input voltage but likely the other way around.

So we can investigate more if needed, but the one thing I notice that may significantly contribute is the diode D1. It looks to be quite a bit overspecified at 150V, 20A. You would think that isn't a bad thing, but the reality are that the parasitics are much greater. Oversizing the diode can actually cause a lot more noise in the system than you would like. This would make some sense since it happens as you are increasing the input voltage and the diode reverse voltage will be Vin+Vout, so increasing the reverse voltage would generate more noise during reverse recovery. If the noise lasts long enough it will go past the leading edge blanking time of the IS pin and it will hit current limit randomly and cause flicker.

So for now my recommendation would be to change the diode to something like a 75V to 100V rating with 3A to 5A current rating. If that allows you to increase the input further then you will know for sure. Let me know.

Regards,

Clint

Hi Clint,

Thank you for your feedback.

I will change over the diode and ask to build up a new prototype as the current model is of quite bad quality.

Kindest regards,
Andreas

Hello Andreas,

Sounds good. I would try a new diode first just so you are checking one change at a time. But as I mentioned, MCPCBs can be difficult. Grounding and ground return paths make a huge difference. Are you able to make the metal core ground and connect it to the ground planes on the top layer? Or do you have the option of two layers? Ground is the hardest part with metal cores. But if using a different diode helps great. Otherwise look at some layout guidelines (some are in the datasheet) and be as liberal with ground planes/traces as possible. These things are hard to diagnose remotely, but I'll try and help if I can.

Regards,

Clint

Hi Clint,

Thanks for the hint.

Just out of curiosity, do you have any (reading) recommendations for a beginner in that field of switching regulators, etc?

I am quite new to that all as my expertise is more on the embedded world and automated testing. :)

Kindest regards,
Andreas

Hello Andreas,

I don't have anything specific in mind to look over regarding switching regulators. But the datasheets are usually pretty good and you can learn a decent amount from some of them. There are probably some app notes on ti.com and I have also searched and found some material on university websites before as well.

Regards,

Clint

Hi Clint,

Thanks for the reply.

Just a short update.
We tried to change over the diode but that didn't work out as we do not have a pre-heater for the Aluminium PCB. We are looking into purchasing one as we use quite often Aluminium PCBs. Will keep you updated on that but it will take a while.

In the meantime I started to play around with TINA and the TPS92692 SPICE model.

Cheers,
Andreas

Hi Clint,

Sorry for the late reply.

We changed the diode and the voltage range went up to almost 30V. Unfortunately everything below 12V did not work.

I believe it's a combination of bad sample (craftsmanship) and wrong component selection.
The board is an aluminium PCB and it's really hard to re-work. We used a hot air gun but it's tricky.

We will get some more samples, I purchased a hotplate to preheat the PCB and ordered an evaluation board.
I would like to ask if it is possible to modify the evaluation board so that the input voltage range will be 9V... 33V and if so do you have any recommondations?

Kindest regards,
Andreas

Hello Andreas,

Like I said, MCPCBs can be tough, I suspect that has something to do with it. Are you able to ground the metal core and via to it? I know that's more difficult, but grounding is usually the issue with that type of board and I suspect it has some effect here. 

As far as the EVM goes I believe it will work to higher voltages. It's designed for load dump so at least 36V input transients, and the compensation looks to be enough to allow for a wide range.I can find out for sure tomorrow, but it's likely the EVM would work fine at higher input voltages. The range in the table is just a typical for automotive type applications.

Regards,

Clint

Hi Clint,

Thank you so much for your help.

Do you have any recommendation for the load that I could use?
At the beginning I would like to use one string of 8 LEDs. ULED = 3V and ILED = 1A.

Kindest regards,
Andreas

Hello Andreas,

Do you mean a load to use rather than the LEDs themselves? If so we sometimes use standard 1V Vf diodes and put 3 in series for each "LED". That way you can test without blinding yourself and you can probe things more easily if they are on the same board as the driver. Let me know if I misunderstood.

Regards,

Clint

Hi Clint,

Thanks for the reply. A bit of a misunderstanding.
Wanted to know it if is all right to start with a string of 8 LEDs (ULED = 3V, ILED = 1A).

I plan to connect the EVM with one string of our prototype PCB using cables. In that case I can turn the prototype PCB away from me.
As I mentioned we are very interested if the input voltage range as we require 9V... 33V.

Kindest regards,
Andreas

Hello Andreas,

Yes, that is fine. I usually use an LED load board that I can turn away as well. You should also not have a problem with that input voltage range. With the buck-boost your only concern is Vin+Vout. If that ever exceeds 65V you will need to change where you are sensing the current. In your case it shouldn't be a concern.

Regards,

Clint

Hi Clint,

Thanks for your reply.

I am looking forward to receive the evaluation PCB which should happen next week.

Do you have a spice model for the TPS92692?

We do quite often a simulation run as part of our documentation.

 

Kindest regards,

Andreas

Hello Andreas,

It looks like there is a spice model under "tools and software" in the product folder on ti.com. I hope that helps.

Regards,

Clint

Hi Clint,

Thank you for your reply.

Next week I should have some more time to look into our prototype.
I am quite new to the company and there are currently lots of distractions.

Just a few observations that I made:

- Our prototype is missing any capacitor at the output towards GND. But we have caps between output and LED-
- At the input we use an array of 10uF capacitors and an electrolytic 100uF capacitor.

The prototype is flickering with maybe 2 Hz independent on input voltage.

Kindest regards and have a great weekend,
Andreas