Filters on Idle Outputs

[Mark Kelly]

I can't see that I need a filter on an output which is permanently idle.

By my understanding, the output devices are switching at 50% duty cycle (see illustration in www.national.com/an/AN/AN-1497.pdf) so there is no net current.

I am using a four channel amp (TAA4100A) in a three channel configuration (three phase motor drive). The fourth channel will be permanently mute and I'd like to be able to leave out the extra filter components.

[Mark Kelly]

I guess the question I'm asking is "Can anyone see a good reason to include the filter components on the unused channels output?"

[Pano]

it's a very good question. I don't have a good answer, but maybe someone here will.

FWIW, with the Tripath chips most of the "harm" done to the output transistors comes from boounce and overshoot in the filter. Thus the protection diodes. Maybe no worries without?

[Pafi]

"Can anyone see a good reason to include the filter components on the unused channels output?"

Maybe. The inductive load current can help decreasing switching losses (and radiated noise). But I don't know if it's significant in this particular case.

[theAnonymous1]

Looking at the sad dead carcus of a Motorola DCP501 board, the TA3020 used for the fifth channel only has the source to drain diodes connected on the unused channel.

The TAA4100 is different than the TA3020 with the MOSFETs being internal, so I don't know how that effects things. Maybe all you need to do is remove the output inductors.

Here are pics of the bottom of the TA3020 single channel and two channel.....

Single channel
Two channel

[Mark Kelly]

OK

After some experimenting I can state that the answer is no.

In the absence of the LC filter the TAA4100 goes in to rail to rail oscillation at about 50 kHz (presumably the switching frequency).

I'm prototyping with a board with limited ability to add different parts (the AMP9 from 41 Hz) so the easiest solution is to re-instate the existing filters but I shall do some further experiments with a new board I'm laying out myself.

[Pano]

It should certainly be in rail to rail square wave oscillation, - that is how it works - but not a 50KHz. More like 500~700Khz

Was that just a typo?