Hi all,

I once again submit a question to the allmighty piclist :)

I have a circuit which provides about 100mA of current at 3.3V from a single NiMH cell using the LTC3400 boost switcher IC. I'm using a 6.8uH inductor, the feedback resistors are 1.05Meg and 604K. The shutdown pin is tied high with 100K to Vin (I've also tried 1Meg). I've tried two different caps for the input and output, both tantalum SMD types: 10uF 20V and 47uF 6V.

The following configuration (config A) works well as tested on about 5 PCBs:
LTC3400B (the version which doesn't allow burst mode)
100K on Shutdown pin to Vin
10uF 20V caps on input and output

I made a run of 8 PCBs with the following configuration (config B):
LTC3400
100K on Shutdown pin to Vin
47uF 6V caps on input and output
The idea behind this is that I'd like to take advantage of burst mode's low quiescent current when the PIC and other stuff which will be connected to this will be in sleep mode.

Config B usually doesn't run at all. I see about 850mV to 1.2V on the output depending on load. One of the PCBs also shows some square-wave oscillation on the SW (switch) side of the inductor. It sounds like the circuit is in shutdown (which it should not be with a pull up to Vin)

I then made the following changes to one PCB (config C):
LTC3400
100K on Shutdown pin to Vin
10uF 20V caps on input and output

Config C runs but shows only 2.9 volts on the output. When I measure the FB node (middle of the voltage divider) I get 1.24V which is what it should be, but when you divide 2.9 volts by 604/(604+1050) you get 1.06 volts not 1.24, so the LTC3400 must be leaking current into the divider (the spec on the FB pin's leakage current 1 nanoamp typical which shouldn't be able to make 200mV difference).

Putting a scope on the switch node for Config C shows squarewave bursts which can be either almost continuous or very rare depending on whether I use a NiMH cell or a bench power supply at 1.6 V.

To make a long story short, It seems as though burst mode is causing much higher leakage current on the feedback pin. Has anyone seen this before or have a suggestion on how to fix it?

Thanks very much,

Sean



-- 
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist