TPS92515: Current sense comparator offset error

Hello,

On the TPS92515 the offset error is +/-2.5mV.  The Iadj voltage can be lower (see figure 28 in the datasheet) however the accuracy percentage goes up because of the offset.  Note that the TPS92515 has a turn-on calibration which can cause offset voltage steps of 5mV during power cycling if it is near a calibration step.  At low Iadj levels this can be noticed when turning the circuit on and off in some cases.  The TPS92515A does not use the turn on calibration however the error amplifier offset is higher.

Calculating Fsw starts by calculating duty cycle (equation 9.2.1.1 in the datasheet design example) and off-time (section 7.3.3 equation 7 in the datasheet).  Note that this part is not a fixed frequency part, it will vary if Vin or Vled vary.

Then ton = toff * (duty cycle/(1 - duty cycle)).

Fsw = 1 / (toff + ton)

Hi Irwin, 

Thank you for the quick response. I'm a little unclear on what you mean by 5mV steps during turn-on calibration. Could you please explain in more detail?

Also, suppose my circuit uses VIADJ of 0.13V, leading to13mV of sense voltage, would this cause an issue in the current setting accuracy?

Regarding fsw, based on the equations in the datasheet, which parameter needs to be assumed, is it toff?

Hello,

The TPS92515 has a turn on calibration.  If it senses that the error amplifier offset is off it will adjust in 5 mV increments to put the offset in the +/- 2.5% range.  If it happens to be close to +2.5 mV or -2.5 mV offset it can shift 5 mV during a power cycle.  As an example if it is near the +2.5 mV edge it can power up near a -2.5 mV offset if the calibration circuit adjusted it's power on calibration.  This can happen due to temperature change, operation change, etc.

If you are at 13 mV of sense range and the error is +/- 2.5 mV the peak current threshold can be 10.5 mV to 15.5 mV.  Note that if you are running discontinuous this is not a linear change in average LED current.  Depending on the design it's possible it will be limited by minimum on-time as well.

If your input and output voltage are fixed you can calculate duty cycle and derive toff from the switching frequency and duty cycle.  You can also calculate toff how the datasheet shows it and figure out the switching frequency afterwards.  If it's not where you want you can change Roff to adjust toff to get closer to what you want.  Note that when calculating your design when changing toff the average current will change so Rcs will need to be adjusted.

Best Regards,

Hello Irwin,

Thank you for the response. Incase the analog dimming function is not used, and if DCM is not used, then with such a low sense voltage of 10mV, would lead to very low current, using the formula [(VIADJ/10)/R - I_L_pk-pk/2],  I_L_pk-pk determined by t_OFF, V_LED and L of the design. If I use the formula with the offsets (V_CST_offset), the current would be higher than with the earlier formula. Is my understanding right? What is the value of CST_Offset to be used?

Also, if I used DCM, how do I calculate the ILED?

Hello,

If the offset error is +/- 2.5 mV (TPS92515 range) you will have a range of 7.5 mV to 12.5 mV, this is not possible without using the analog dimming function, from above "In case the analog dimming function is not used,", that is the only way to get the peak current sense comparitor to trip at 10 mV, by setting Viadj to 100 mV.  The equation "[(VIADJ/10)/R - I_L_pk-pk/2]" is the way to calculate CCM current, Peak current minus half of the current ripple is the average LED current, if I_L_pkpk/2 is greater than VIADJ/10/R the equation is not valid since it is in DCM.

I peak is still the peak current trip threshold.  you calculate toff as you would normally for CCM mode.  Then you use V = L * (di/dt) to calculate dt for toff(DCM).  You can calculate ton the same way or use duty cycle and toff(DCM).

As an example, if Vin is 24V and Vled is 6V and the inductor value is 100 uH and current sense resistor is 0.25 ohm the calculations would be:

di = 0.01V/0.25 ohm = 0.04A

dt(off) = L * di / V.  For the off time V = 6V so dt = 100 uH * 0.04A / 6V = 0.667 us

dt(on) = 100 uH * 0.04A / (24V - 6V) = 0.222 us

The voltages used are what is applied to the inductor, off is the LED voltage, on is the input voltage minus the LED voltage.

Again, as an example:

Take your original toff calculation say is was 2 us.

Calculate the duty cycle of when current is present in the inductor, ton is 0.222 us, toff(DCM) is 0.667 us, toff(CCM) is 2 us:  duty(of current in the inductor) = (ton + toff(DCM)) / (ton + toff(CCM) = (0.222 us + 0.667 us)/(0.222 us + 2 us) = 0.40 (40% duty cycle of when current is present in the inductor.

Now take the peak current divided by two (area under a triangle) and multiple by duty:  0.04A / 2 * 0.4 = 0.008A average current in the LED.  These equations only work for DCM.

Note that in this example ton is 0.222 us or 222 ns.  minimum on-time typical for the TPS92515 is 195 ns and maximum minimum on-time is 275 ns.  Once the TPS92515 reaches minimum on-time the LED current will not go lower, it has to have a minimum on-time every switch cycle.  In this case some TPS92515 may be limited by minimum on-time if they are on the higher end spec of minimum on-time.

If you don't understand the above information I can calculate the values of your design.

Best Regards,

Hello Irwin,

Thank you for the detailed response. I understand it better now. 

Could you please let me know what is the V_CST_offset value to be used in the below equation? This would provide worst case ILED, if I am not wrong.

Hello,

The offset error is +/- 2.5 mV

Best Regards,