Maybe 3 metal-film resistors in parallel, each with nominal value of 1.8 ohm? That would be pretty close. OK, I'm obviously too tired to calculate...
OK AndrejaKo, take a rest and will try again later (tomorrow). You'll need, seven 1.8 ohm in parallel (0.257 ohms) to come close to the value.
Edit:When I mentioned frequency shift in my previous post, I meant the signal where you show the duty cycle, not the big noise spikes. Although the duty cycle looks a lot closer to the one rf-loop posted, I was wondering if there was jitter or a gradual change in frequency. On the first duty cycle capture I see a range from 43-50KHz, on the last capture I see a frequency below 38KHz. Also as you pointed out the duty cycle varies quite a bit.
If everything is working correctly and the converter is working in CCM the duty cycle is a function of the voltage in and voltage out approximately like this: DC = Vout / (Vin + Vout) If you work that out with Vin = 8.4V it comes out pretty close to 50%. Of course when you are using the battery it'll change some. But it shouldn't be changing back and forth from 40% to 60% as you reported.
The frequency is controlled by C6 which is charged and discharged by an internal current source and sink. The current values are fixed so the voltages in and out of the IC shouldn't affect the frequency. However, the charge current is increased to cause a short oscillator cycle if too much current is detected by the current sense circuit (which checks the voltage across R40). So if everything is working correctly the frequency should be pretty steady, going back and forth from 38KHz to 50KHz doesn't seem right to me.
In conclusion, I think something is still wrong. You replaced the IC and things got better, maybe because of slight differences in characteristics between the two ICs, but still there seems to be an underlying problem. If that gets fixed, maybe the old IC will also work correctly. Like I said before, there is only a handful of external components, and I think I would start with R40.
Note: The ON Semi application note for this IC states that you can disable the current sense circuit by connecting pin 7 to VCC. So for a test, it may be OK to short R40. I wouldn't do it for too long and would be watching for other components overheating. Up to you, if you want to risk it!