I have a homemade PCB that consists of a microcontroller connected to a CC2571 ANT+ transceiver. The CC2571 and micro are powered directly from the battery.
The ANT transceiver has an internal processor and the radio is configured to transmit at 4Hz.
It wakes every 250ms and does some brief radio activity. This includes a transmit and receive that cause a brief current drain of around 30mA for a few ms. In the off period, the radio consumes a few uA.

This current spikes are putting a lot of noise on the supply rails of my PCB and I'm looking for a way to reduce it. When I look at the voltage off my 3V battery rail I see drops of around 100mV when the transceiver is active every 250ms.

I also have a 2.5V regulator that powers some sensitive analog stuff and hanging off that there is load of around 7mA being switched on and off at approx 100Hz. This periodic switching puts even more noise on the 3V rail.
On the 3V rail I have a high quality 100uF cap. Any tips on how the to reduce the noise on the supply rails?

 

Try bigger filter caps on the regulator outputs.

 

Try bigger filter caps on the regulator outputs.

The transceiver is powered directly from the 3V battery rail, but yes I can try a bigger cap on the 2.5 reg output

 

Try it on the battery too.

 

Try it on the battery too.

How could I isolate the supply to the transceiver?
I guess an LDO would only give isolation if I'm above the dropout voltage and I want the supply of CC2571 to be the same as the micro

 

Capacitors are the easiest. Next step would be chokes. Does the supply noise cause problems?

 

Capacitors are the easiest. Next step would be chokes. Does the supply noise cause problems?

Putting a 100uF cap on 2.5 reg output has reduced the modulation on the battery rail down to around 25mV ripple.
The 100mV spikes are still present from the CC2571 switching

 

Try connecting a 100μF electrolytic plus a 0.1μF ceramic cap directly across the power and ground pins on the CC2571.

 

What kind of battery?

 

Assuming it is a coin cell, here is some info on the caps.
http://www.ti.com/lit/wp/swra349/swra349.pdf

 

Putting a super capacitor on the battery.

 

I'm trying to learn so please forgive me if this is off the mark, but what about putting an inductor in line with the input to the 2571 to smooth the spikes, and add a cap across the source pins but with the cap + lead between the inductor and the 2571. Did I explain that clearly? My thought is the cap will maintain current to the 2571, and the inductor will smooth spikes to the rest of the system.

 

I'm trying to learn so please forgive me if this is off the mark, but what about putting an inductor in line with the input to the 2571 to smooth the spikes, and add a cap across the source pins but with the cap + lead between the inductor and the 2571. Did I explain that clearly? My thought is the cap will maintain current to the 2571, and the inductor will smooth spikes to the rest of the system.

The problem with standard inductors is that they can resonate with the line capacitance in the system and can actually increase the noise or around that resonant frequency. Noise spikes can also trigger an oscillation around that frequency.
To solve that, lossy inductors such as ferrite beads are sometimes used since they do not resonate with any capacitance in the system.

 

Putting a 100uF cap on 2.5 reg output has reduced the modulation on the battery rail down to around 25mV ripple.
The 100mV spikes are still present from the CC2571 switching

Electrolytics are not so good with sharp transients - but you still need a few of them.

Put disc or chip ceramics in parallel with the electrolytics - anything from 0.1uF to 0.47uF, these should be as close as possible to chip supply pins. Every chip should have at least one nearby.

Others have suggested chokes - I'm thinking ferrite beads, less volt drop and they may be enough inductance for sharp spike suppression. They can be literally a ferrite bead on a bit of link wire, or a manufactured component that looks similar to a resistor.