TPS922052: Dimming performance

Hi Edward,

Could you share more details about how the current "externally managed shut" solution is realized? Through discrete gate driver + MOSFET or integrated shunt FET control solution?

Before the recommendation, I'd like to know the reason why an alternate resolution is required? Does current accuracy or dimming resolution of existing solution not satisfy the requirement? if so, what's the required current accuracy and dimming resolution @ dimming frequency?

Best regards,

Felix  

Hi Felix, yes we use a UCC27517DBVR to drive a FET to shunt the output, we power the UCC from the VDD of LM3406 which is about 7V so we get a pretty decent performance from it. It runs off of its own PWM channel from the micro so we can bring the controller on (with PWM) independent of controlling the shunt, both of these PWMs are sync'd so we can manage turn on and shunting to reduce power consumption at lower dim levels.

This works great, but now we really want to be able to software control the current to dial it in for different product configurations.

We've built a prototype with TPS92515HV however the current is very much dependent on the true inductor henries and that we're finding too much variance with inductor tolerance and also the way inductors change characteristics with temperature/load.

I'd like the current to be <+/-3%, I'd like >10:000:1 dimming depth at 4KHz.  I/d like to be able to dial in currents (from another PWM signal or analog signal or a digital bus) in the range of  500mA to 1400mA, though above 1000mA less accuracy is ok for us.

Could I also request the latest version of the TPS92205x_Cal_Tool thank you

Hi Edward,

Thanks for your explanation. Not sure if you have heard about the TPS92402/TPS92401 that is integrated solution to replace the discrete shunt control solution.

As for the latest calculation tool, i will have a check and send it you if have.

Best regards,

Felix

I"m just driving COBs for interior lighting applications so I'm not sure those devices are that helpful.  

Looking at that old TPS92205x calculator it seems to be a bit wrong, the way it's calculating the sense resistor seems to be based on the ADIM scaled current rather than the full power current. Would be handy to have the password so I can check its workings!

A few questions on TPS92205x, I have read the data sheet and I'm still not really sure about these;

1) Can you run it without an output capacitor, we've been designing solutions with larger inductors and without output capacitors to get us the best dimming control.  It doesn't seem like that's the intention with this device.

2) Is this device designed to DIM better than when you just use the enable pin to control a converter like LM3406, as we find with deep dimming you end up with too few converter cycles inside the enable window and you loose any granularity of control, is TPS92205x meant to mitigate this issue with its internal functionality to get that headline performance figures for dim depth.  I'm trying to understand what the 150ns control means, and do I need to run the converter in the MHz range to benefit from this?

3) Could we add a shunt to TPS92205x, possibly not as it would cause chaos with its internal fault detection.

I have got an EVM on it's way, hopefully arriving today!

Thank you

Just tried out the TPS922055 EVM with one of the 0.1 ohm resistors lifted to reduce the standard drive current, the DIM/EN performance definitely isn't good enough of our application, I'm finding when I set the current at 50% duty cycle (1KHZ), and then bring the EN up from off, I get no output until 0.22% (1KHz) duty cycle?  Maybe I need to strap these pin or something - more reading to do!

EDIT - I think I have to be careful with the way I manage the EN pin to stop it going completely off, maybe I need to manage the ADIM pin as well to bring it up from off - will look at it more tomorrow - the latest calculator would be helpful.

Hi Edward,

Before answering your questions about the TPS92205X, I'd like to clarify with you about your requirement.

So from my understanding your requirement is shown below. Right?

• 7V input voltage
• One LED connected to output, 3V output voltage?
• Output current 500mA to 1400mA, accuracy +-3%
• Dimming frequency 4KHz
• Dimming resolution 10000 : 1 (about 13 bits)

And your system has pre-stage dc/dc constant current LED driver and LED load with shunt FET as shown below (simplified block diagram). correct?

Normally, as I know in order to achieve high PWM dimming frequency and resolution the shut dimming will be used, which means the pre-stage dc/dc only provide the constant current and the dimming control is realized through controlling the shunt FET on/off. 

So I'm a little confused that you mentioned that you want to select a led driver (the pre-stage DC/DC CC) that supports the high dimming frequency and resolution. If so, what the shunt FET will be used for?

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Come to the TPS922055,

1. yes, the TPS922055 can run without output capacitor. However, TPS922055 is not recommended to be used in such system to support shunt dimming.
2. Yes, TPS922055 has better dimming performance than LM3406 as Inductive Fast Dimming (IP) is introduced. Yes, in order to achieve high pwm dimming frequency, higher switching frequency is required. And for the TPS922055 there are 4 dimming methods supported. With the hybrid and flexible dimming method the 10000:1 at 4kHz pwm dimming can be realized.
3. as i said, TPS922055 is not designed to support shunt dimming. Actually we have the special led driver support shunt dimming that is TPS92640/1.

Best regards,

Felix

Hi Felix,

My comment about 7V was relating to the fact you can get 7V out of the VCC rail on the LM3406 which is very handy for then powering the shunt FET driver to control the shunt.  My applications are all 48V in and 12 to 36V out.

I think I need to get my head around how the TPS92055/2 actually works, I think if I manipulate the two inputs in the right way I can get what I need out of it, at least to an adequate level of performance.

I saw the  TPS92640/1, the number of FETs put me off, I'm also considering LM3409 which can also do shunt dimming (with two FET total), but it would be nice to use something more modern - one of the designs I did at the start of my career was using that part, not that that discounts it!

Hi Edward,

Suggest two approaches to realize the requirement below.

1. Use the dimming functions of TPS922055/2 to realize the dimming performance without the shunt FET (as I said TPS922055/2 could not cooperate with shunt dimming well)
2. Use the LM3409 or TPS92515 as the pre-stage constant current source to the LED loads with shunt FET. Control the shunt FET to realize the dimming performance.

Design Requirement:

• 48V input voltage
• 12V to 36V output voltage
• Output current 500mA to 1400mA, accuracy +-3%
• Dimming frequency 4KHz
• Dimming resolution 10000 : 1 (about 13 bits)

Best regards,

Felix

Hi Felix, thanks for your feedback.  I just can't get the TPS922055 to turn on nicely from dark, I'm running at 200Hz PWM a the moment, I thought if I kept a small 200ns pules on the EN pin I could keep it in hybrid mode, and then bring it up nicely, but of course I can't get it truly dark, and then as soon as I reduce the EN pules to get the LED dark the converter seems to shutdown and I have to crank it up to 2100ns to get it to come back to life.

I guess this is made for people less fussy about fades up from dark than my project is.  It's a shame as its a really nice looking device otherwise.