TPS92520-Q1: Current accuracy issue of tps92520 to be continued

zhang lin

Intellectual 310 points

Part Number: TPS92520-Q1

Tool/software:

Hello, expert.

I have read other posts about the current accuracy of TPS92520 on the forum.

QA:I would like to consult again about the current accuracy problem of TPS92520:

For 16:1, it refers to ILED_REF_DACx, 1023:63=16:1, is this understanding correct;

QB:For the content of the manual: ANALOG ADJUST SETTING AND CURRENT SENSE AMPLIFIER (CSPx, CSNx);

I understand it like this: For different values of ILED_REF_DACx, the final LED current accuracy is different. The larger the value of ILED_REF_DACx, the higher the accuracy of the LED. Is this understanding correct?

① When ILED_REF_DACx = 1023, the range of V(CSPx-CSNx) is 167.5mV, 173.0mV, 178.5mV, and the LED current accuracy is -3.18%~+3.17%;

② When ILED_REF_DACx = 512, the range of V(CSPx-CSNx) is 83mV, 88.5mV, 94mV, and the LED current accuracy is -6.21%~+6.21%;

③ When ILED_REF_DACx = 63, the range of V(CSPx-CSNx) is 6.5mV, 12.5mV, 18.5mV, and the LED current accuracy is -52%~+48%;

Because our driver needs to match different LED loads, under the condition of not changing the hardware, the current error will be very large in the case of small current. Is there any detailed parameter for the LED current range corresponding to different ILED_REF_DACx values? Thank you.

3 months ago

0 fhoude 3 months ago

TI__Mastermind 23326 points

Hello Zhang,

QA: Yes, this is correct how we got to 16:1, but we can do better than that because this is a synchronous part and don't have the negative effects of DCM (Discontinuous Conduction Mode) of a asynchronous buck when driving with constant current source.

QB: I am not sure what you mean by different LED loads (do you mean the number of LEDs???). This is constant current buck so the output voltage doesn't change the regulation of the output current. It may change the duty cycle of switching but it doesn't affect the current regulation loop. In my other posts I talk about how the sensed current (Vcsp-csn) across current sense resistor will have different accuracies based on the offset voltage error, especially at low current points. At low current setpoints the Vcsn-csn is small and the offset voltage is a larger percentage of the signal thus the worse accuracy. I don't know what you mean by "Is there any detailed parameter for the LED current range corresponding to different ILED_REF_DACx values?" Can you expand what you are trying to ask me here.

-fhoude

0 zhang lin 3 months ago in reply to fhoude

Intellectual 310 points

Well, let me just talk about the problem we are facing. The TPS92520 has different precision at different output currents. We have current accuracy requirements. Is there a good solution to solve the output current precision problem? Because the driver has been developed. If possible, can you tell us how other customers do it or how it is done on the production line.

0 zhang lin 3 months ago in reply to zhang lin

Intellectual 310 points

The table lists the 4 cases of ILED_REF_DACx = 1023/512/192/63, and I want to know the range of V(CSPx-CSNx) under other numerical conditions (for example, in the table), and use this range to evaluate the error of LED current.

+1 fhoude 3 months ago in reply to zhang lin

TI__Mastermind 23326 points

ILED_REF_DACx outputs a scaled version of the V_DAC(FS) voltage = 2.45V as the reference to the input of the current sense amplifier. There is a circuit shown in the block diagram that takes that reference voltage and divides it by 14 at the Vcsp-Vcsn. In other words that larges signal at Vcsp-Vcsn is 2.45V/14 = 0.175V = 175mV (this is ideal). If you change the current sense resistor (R_CS) then you change the max current it supports. For example RCS = 0.1 ohms support 1.75A. Keep in mind that you have current limits for the high side and low side fets that limit the max current but that is not the loop but just for protection.

I have at least three other posts that go into detail about how to calculate accuracy and how this part regulates, see below:

https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1524526/tps92520-q1-led-current-accuracy-improvement-by-offset-compenstaion/5861371?tisearch=e2e-sitesearch&keymatch=TPS92520-Q1%2520accuracy#5861371

https://e2e.ti.com/support/power-management-group/power-management---internal/f/power-management---internal-forum/1441055/tps92520-q1-higher-accuracy?tisearch=e2e-sitesearch&keymatch=TPS92520-Q1%252525252525252520current%252525252525252520accuracy

https://e2e.ti.com/support/power-management-group/power-management/f/power-management-forum/1330731/tps92520-q1-how-to-improve-led-current-accuracy?tisearch=e2e-sitesearch&keymatch=TPS92520-Q1%2525252520current%2525252520accuracy

There are a couple of ways to calibrate our devices if you need tight accuracy at lower currents.

You can test the IC at the end of production and back calculate the Vos (offset voltage of current sense amplifier) and then store that information in the MCU and use it in the equations. This works pretty well for lower current setpoints, but only address the offset voltage error and doesn't address other sources of error like gain.
The other option is to have a lookup table where you test the ic at various conditions at the end of production and use that test data to set your current at various setpoints with calibration factors.

-fhoude

Power management

Power management forum

TPS92520-Q1: Current accuracy issue of tps92520 to be continued