TPS61042 Switch Rds(on)

If you are talking about the FET that is used to implement dimming, between LED and RS pins, then its effect will be captured by increasing the value of your expected Vout in the equations.  It is part of the load and not the boost power stage so it will cause an increase in the required output voltage and therefore output power demanded.  The effective duty cycle increases.

No,l'm not using the LED FET. I'm trying to figure out whether the switch FET's on resistance would need to be taken into account. I made an error in the original post by quoting 2 Ohms, this should really read 0.6 Ohms. If you look at the spec for the TPS61042 you'll see 0.6 Ohms as the max at Vi = 3.6V with a typical of 0.3 Ohms. The typical figure matches the graph (Figure 8) that shows how the typical figure increases to 0.550 Ohms at 1.8V. Assuming the max figure scales by the same amount, this yields approximately 1 Ohm for the switch worst case. So 500mA at 1 Ohm is 0.5V, i.e. 27% of 1.8V, so it will have a significant impact on switching speed, which effects max power deliverable at low voltages. The question is whether the conditions that cause Rds(on) for the switch to increase also cause the switch current to increase (this can be between 0.4A and 0.6A). If that were the case then the energy transferred per cycle would increase, this would negate the lower switching speed effect and the max power transferrable at low voltages wouln't be affected by the Rds(on) of the switch.

So, can I neglect the Rds(on) of the switch when working out the minimum deliverable power at low voltages?

Then, yes, you are correct in that the datasheet equations ignore the effect of the power FET's Rdson to simplify them.  Your estimates for the max Rdson seem reasonable.  So, for a good approximation you could simply replace Vin in the fsw(load) equation with Vin - Ilim*Rdson, for example, at Vin=1.8v and with Rdson=1ohm, Vin-Ilim(max)*Rdson(max) = 1.8V-0.6A*1ohm = 1.2V worst case. 

OK, thanks. However it doesn't really answer whether the peak current limit I(LIM) actually rises under the same conditions as Rds(on) increases. Obviously it won't exceed 0.6A (plus a bit for the 100ns delay). I guess we would need to know the relationship between these internal to the chip to know the answer.

As an aside, are there any Spice models (or other simulation models) available for the TPS61042? This would be really handy for at least estimating whether a circuit would work.

Mark. 

Actually as a note to the designers of the chip there is another feature I would like to see that would be handy. Currently the CTRL pin has a Vih abd Vil of 1.3V and 0.3V. If this threshold voltage were accurately specified (using the bandgap reference), and possibly with a little bit of defined hysteresis,  then we could easily add a low voltage lock-out by simply using a resistor divider on the CTRL line.

Mark.

From the block diagram, the current limit circuit measures the voltage across an internal resistor (not shown) to ground.  If that fixed resistance only increases slightly with temperature and the current limit comparator is simply a comparator, then the current limit value trip point  should remain fixed and possibly get lower with temperature. 

You are correct in that the variation in the resistor value is the primary reason for the variation in ILIM.