LM5008 with very low/no load

Salut Sylvain,

Thank you for your interest in TI parts.

The LM5008 will continue to switch at very low, and even no load current, but it may enter a hiccup-like mode because the bootstrap cap (C4 in figure 1) can discharge too much. This is because at very light loads, the OFF time can be very long as compared to at nominal loads. During that long OFF time, the boot cap discharges, and only gets recharged during the next ON time. See the paragraph in the datasheet on page 11 called "MINIMUM LOAD CURRENT". This type of operation is not a problem for the device, but your 5.5V bus will collapse repeatedly. It will recover again when the load is increased.

As stated in that paragraph, you can choose your feedback resistor divider (R1 and R2 in figure 1, page 2) to be low enough value so as to provide at least 1mA of load current. This implies that:

I_FEEDBACK ≥ 0.001 = 5.5/(R1 +R2), so

(R1 +R2) = 5.5/0.001 = 5500 ohms (or less).

This should provide the required minimum load of 1mA even when your system load drops to zero.

Given that at no load, the switcher still provides at least 1mA output current (even though your system is not using it), the measured efficiency will be extremely low, at the limit it will be 0% because the switcher will be consuming power and producing 1mA @ 5.5V, but your output power will be zero (so Eff = (Pout)/(Pin) = 0). But that is the same for any converter, any topology, at no load. The key is how much energy dissipation is wasted at no load, and in this case it will be quite low.

Since the catch device (D1 in figure 1) is a diode, the current in the output choke (L1) will fall to zero during every switching cycle at light loads. So there is less circulating current in L1 and C2 and the internal N-Channel FET as compared to a switcher where the catch device is a controlled FET. That circulating current is usefull in many cases, but at no load it really amounts to a lot of current flowing for no benefit (it just heats up the devices). Just note that a "non-synchronous" (i.e. catch device is a diode and not a synchronous FET) switcher that can reduce the switching frequency and tolerate very light loads is probably the approach you want to take to achieve very light load efficiency, or more importantly, power dissipation. You will need to evaluate the dissipation at no load in your design to determine if it is acceptable.

I hope this helps.

MC.

Hello Sylvain,

Thanks for choosing TI part. At low loads LM5008 will work with reduced frequency as it goes in DCM. But it will regulate fine. As mentioned in the datasheet, you should provide feedback resistors of 2.5 kOhm each or lower to maintain 1mA minimum load.

Hope that helps.

BTW if you can not affort to use low enough feedback resistors to maintain >=1mA or load, consider using LM5008A , as it has a special precharge circuit built in to operate down to zero load.

-Vijay