Please tell me the following thing.
That 1000 times of condenser is added to the purpose which improves a pulse-skipping behavior in a data seat.
If it's more than 1000 times, would a defect form?
Thank you for your post.
Don't make the capacitor from SW to VC too large (I call the capacitor Cblank but that may not be the best name).
As mentioned in the datasheet excerpt you show above, the capacitor from SW to VC should be around 1000 times SMALLER than the VC capacitor.
I put an image below that shows the results of an LTspice simulation where I sweep Cblank. I also put an image of the block diagram from the LT3476 datasheet. The LTspice simulation shows that Cblank affects the VC signal, especially at the time when the LT3476 makes the decision to turn off the switch. I talked to my colleague who has the most experience with the LT3476. He said the LT3476 works better - pulse skips less - at low duty cycles and low LED currents with the capacitor from SW to VC, just as mentioned in the datasheet excerpt. I don't know the exact mechanics of why the cap from SW to VC makes the LT3476 run better, but you can see the effect of the capacitor on VC in the LTspice simulation results.
You can see from the LTspice simulation that the cap from SW to VC tends to pull up VC. When VC is pulled up, the LT3476 does not want to turn off the switch - that is why it is bad to make the cap from SW to VC too large.
Thank you Very much
Retrieving data ...