The ESR of an healthy LiIon battery is very low.
The most important things to check are:
The protection circuit on the cell. It uses a MOSFET and
on cheap battery packs can easily contribute 50mOhm or
more ESR. If at all possible use a battery pack without a
protection circuit. But only if the battery is not user accessible,
and if regulations allow it.
Battery connector/wires. Can contribute significant ESR.
PCB layout. Often overlooked, but narrow tracks and via’s in the
routing is fatal. Use at least 1.5mm tracks, with multiple via’s in
parallel if the routing does not allow the track to stay on one layer.
The connection to the 60pin is crucial.
The decoupling caps have to be right next to the power pins on
the connector, and if they are on the opposite side of the pcb use
plenty of via’s.
Don’t forget about the ground plane impedance.
Testing without an antenna does not always increase power consumption,
since there is no load to absorb RF energy. Use a lossy 50ohm terminator
to force maximum load on the RF output. Even better is to use a very
directional antenna and point it at a cell that is far away.
If everything is perfect the ripple voltage should be acceptable without
any decoupling caps, and negligible with a single 22uF 1210 ceramic
cap. (If the battery is fully charged.)
Remember to test at minimum temperature with a cell discharged to 3.4V
if you are using electrolytic caps, or maximum temperature if you
are using ceramics.
If your battery pack has a protection circuit use as much decoupling as
you can afford.
Depending on the application you should also check the ripple without
the battery but with the charging circuit connected.
The most bullet proof combination I have used is 2x 22uF ceramics in
parallel with a Nichicon 470uF PW series electrolytic.
On another note, we have noticed that from a cold start the wavecom
powers up but the software doesn’t start if the ESR of the caps is too
high -using something such as an electrolytic for instance - I presume
this is to do with power not ramping fast enough for internal reset
sequencing or some such - can anyone confirm?
I have never seen slow ramping power causing a problem. I have tested
with a discharged 2Ah LiIon starting at 1V, and charging at 100mA. This
results in the voltage sweeping smoothly from 1V to 4.2V in 20-30 hours. The Wismo reset is released at 3.40V and it starts up correctly every time.
Check if the regulator you are using is stable with low ESR caps. Many
low drop regulators will oscillate or spike when driving very low ESR
caps. If you are using a digital scope to check the ripple it should be at
least 1gS/s, otherwise you may miss the very short positive going spikes
at the end of each transmission burst.
I have seen a few applications using LM317’s and jellybean caps, and they
don’t work well. The LM1086 does work exceptionally well, but stick to
National, Fairchild or ST and watch the dropout voltage.
Don’t overlook the ripple rating of the caps. Many caps with suitable ESR
will deteriorate quickly with 2A ripple current, and when using caps in
parallel one always hogs all the ripple.