1 Watt Audio Amp the LM4890

Posted in June 22nd, 2004

This is one cool part. It operates in bridge mode, thus you can get a watt with a low voltage power supply. However, prior experience has shown it to be a little critical when it comes to layout and proper capacitor selection. However, it is much much better than the old LM4860.

The other great thing is that it is already in use, in cell phones. And thats a good thing. I don’t like debugging new parts that no one has used yet.

If you want to see part of my analysis of a data sheet, you can read more. Otherwise, it may well be depth that no one gets any too excited about.

Here are they key notes:

Gain is set via Rf and Ri Avd= 2* (Rf/Ri) due to bridge mode.

Tjmax is 150 deg C… but since our duty cycle is short, and we are running with a lower Vcc, we should be ok. I also don’t have the space to tack in a bunch of copper on the pcb. I do need to run thermals to see what we can safely handle.

Power supply bypass 10UF tant or alum elec with a ceramic in parallel.

Shutdown pin must not float, as such, an external pullup is needed to keep it from flaking out during micro reset when all the IO lines are HiZ.

Cbypass is recommended to be 1uF…. but they are spendy

Cin range is 0.1uF to 0.39uF should allow for a low ramp rate to avoid click and pop. I’m not too concerned about LF roll off as the output transducers low frequency response is abysmal.

I also need to drop off the input voltage, as the micro’s pwm output is 5V pk-pk, so I need a resistive divider to drop that. I must go in and tweak the frequency response.

Most likely I do not need C4, as my gain is low… otoh, I ill include the footprint to capture oscillation problems if needed. C4 can be 15pF to 100pF, again I’m not too concerned with the frequency response, as my transducer will handle it. The main thing is I don’t want some hideous oscillation beyond the audio range to bite me at somepoint. The basic GBWP is 2.5Mhz, and since I’m looking at a gain of 2, C4 can be pretty small, as I will still be a long way for oscillation.

Right now it appears the following make sense
Ri = 20K
Ci = 0.1uF ceramic
C4 = 33pF ceramic as its stock
Rf = 20K
Rshutdown pullup is suggested at 20K
Cbypass 1uF ceramic with a 0.1 tant
C power bypass 10uF electo, 0.33 tant, and 0.05 ceramic

Game plan is a 3:1 divider off the PWM output, thus with a 2X gain, we should have plenty of power to spare, and we can always shift Rf upward for more power.

PCB layout
star ground is key, and I will track back to Vbat
Power supply C bypass should be wider 0.020 same deal with ground return. I will try for 0.025 The rest os the traces can easily handle 0.015.

Transducer output tracks must be wide, I will shoot for 0.1