General Suggestions on TDMA Noise in portable products

TDMA noise is a problem in many portable and office audio products. This discussion is taken from a correspondence with a customer, and it contains a few general recommendations on how to improve TDMA noise in a general audio application. This particular application is for a subsystem - an amplifier with both headphone (HP) and speaker (class-D) outputs integrated into a single package. Please feel free to add comments or suggested troubleshooting methods to this discussion post.

The first step in TDMA noise troubleshooting is figuring out how the noise is getting into the system. There are three common injection/coupling paths to check regarding TDMA noise:

Inputs

The TDMA noise could be coupling into the system along the input traces leading to the amplifier device. To check this, try disconnecting the amplifier's input capacitors from the PCB and AC GND the inputs of thedevice (AC GND the input pins as close to the device as possible). If the TDMA noise goes away, then the noise is likely coupling onto the inputs and transmitting through the amplifier.

In this case, ensure that the input traces are buried under a GND plane. Also, ensure that the shielding for the audio portion of the product is independent of the antenna and RF portions of the phone (or receiver). Finally, make sure that the input traces are not crossing or overlapping with any other board level traces that could carry the TDMA noise – particularly supply traces or RF traces.

Headphone only OR class D only

If the TDMA noise is heard on the HP only or the class D only, it is likely that the inputs to the amplifier are not to blame. In this case the noise could be getting into the system by either directly coupling onto the output traces, or by coupling onto a power supply line that only connects to the HP or class D of the amplifier.

In this case, ensure that the output traces of the affected output are buried under a GND plane if possible, and that they are not exposed to any RF traces. Also, see if the noise can be improved by increasing the value of decoupling capacitors for the supply lines to the affected output (such as HPVDD, or VBatt – assuming these are supplied via separate references).

Regarding the decoupling capacitors, please ensure that all decoupling caps are connected to the top GND layer. Sometimes capacitors are grounded to an inner layer using vias. This is not recommended as it can increase the AC impedance between supply and ground, which can make TDMA noise issues worse.

Headphone AND class D (but not input traces)

If the TDMA noise is present on both the HP and class D amp, but does not seem to be injected through the input traces, check the supply references that are used by both systems (primarily, check Vbatt. Try improving the decoupling on this line by increasing the value of the decoupling cap, and ensure that it is well grounded in the system). If this third situation is the case, the signal could also be coupling onto both sets of output traces, or onto the inputs after the input capacitors (not likely – but possible). Try shielding these areas from the antenna and see if the noise goes away.

Lastly, related to all of these points, check how the amplifier's GND balls are connected to the system. The best performance TDMA noise performance is usually achieved by connecting the GND ball(s) to both the inner GND layers and to the top GND layer. This provides the lowest impedance AC path from the supply lines to system GND, and should improve TDMA noise in almost all cases.