TI releases new PWM amp chip: Digital modulator w/feedback

[Amphiprion]

Found this in my in-box yesterday. TI has released the first PWM audio amp chip with a non-open loop digital modulator.

http://www.ti.com/corp/docs/landing/tas5706/index.htm?DCMP=hpa_amp_tas5706&HQS=NotApplicable+EM+tas5706-em

Not that DIY'ers are quite up to this kind of work, but I thought they'd be the types of folks most interested.

ETA: Actually, is it around the modulator or just the output stage?

[Dennis H]

Sounds like they may be using feedback from the output stage which would be new for TI but not new for other PWM guys.

• Closed-loop architecture
• Feedback improves PSRR performance, enables reduced power-supply costs

The chips they use in the TacT and Panasonic amps don't use any feedback so the power supply has to be really stiff.

[dwk]

Sounds like they may be using feedback from the output stage which would be new for TI but not new for other PWM guys.
.

I think the key is that this is still a *digital* modulator. I believe all the other PWM architectures involving feedback use an analog modulator.

[servicetech]

Anything that handles 40W of audio power w/o a heatsink scares me.

[joecarrow]

40W of audio without a heatsink excites the heck out of me! Think, no need to carry around a pound or two of aluminum, no fan, plus you're not wasting energy heating something up excessively. Pair this with some of the new lithium batteries (like the ones that run power tools), and some neodymium drivers, and we're getting into a whole new world of boom boxes- if you will, doom-boxes :twisted:

[servicetech]

How long to you think before that IC has a big hole in the top of it from thermal stress?

[kevmurray]

Anything that handles 40W of audio power w/o a heatsink scares me.

This amplifier delivers 40watts of power, it doesn't dissipate 40watts. More like 4watts dissipation.

[Amphiprion]

And it sinks all that heat to a ground plane which serves as the heatsink. I've seen the Cirrus 30W equivalent of this chip running full out on one of their eval boards, and the load resistors get hotter than the chip.

[kevmurray]

It's a nice little chip, but there are much higher power solutions. Especially if you take the LM4651 and build an external bridge out of high power FETs, and of course a huge power supply. Just an option .

http://www.national.com/opf/LM/LM4651.html

[servicetech]

Reading the specs sheet it looks like it's intended for TV speakers, etc. If VCC was lower it would work great in battery powered sound systems.

[Amphiprion]

DTV and iPod docking stations are probably 80% of the market for these guys. They also find themselves in battery powered applications as well.

The TI chip is interesting in its configurability - 4x10, 2x10 + 20, 20+20, 40 (W). The competing Cirrus chip is 2x7 + 15, or 2x15, or 1x30. It doesn't support the quad output channel mode the TI does. The Cirrus P/N is the CS4525.

The Cirrus part will run off as low as 8V supplies; the TI part needs 10V. Some LiPo's and a boost converter would be interesting package with either of these.

[kevmurray]

Reading the specs sheet it looks like it's intended for TV speakers, etc. If VCC was lower it would work great in battery powered sound systems.

LOL, you're takin me back to the boom box days in the school yard. They (LM4651) would definitely make the batteries last much longer at 90% efficiency. They are intended to mate with the LM4652 though they could be much more with a custom output stage. :twisted:

[Amphiprion]

I have received the TAS5706 EVM module from TI and it is impressive. It comes as a motherboard/daughterboard combo with the TAS5706 on the daughter card (which is very very teeny).

The rated power/distortion performance is there, no doubt about it. And the feedback does keep distortion curves looking more like a traditional class A/B amp than class D. 20WPC @ 10%THD on 18V supply into 8R0 dummy load was met. At 22V supply it would do 25W at less than 1% THD. I cranked the supply voltage up to 27V (the max I was willing to try since I didn't want to fry the EVM) but the chip started current limiting and cutting out / resetting. I'm not sure which, as I didn't bother to see if there was a register I could access via I2C to see what was going on. In any case, I probably wouldn't go over 24V with this chip. Still, it appears they are making it on a 30V process and probably using dual dies to keep the digital small. I haven't cracked one open yet but I do have some samples I plan on playing with.

I think the EVM may have a problem with heat. They obviously tried to make it as small as absolutely possible - it had chintzy header pins for speaker connectors I desoldered those and soldered on some short lead wires to banana plugs instead. The board is so small it gets quite hot when running continuous 20W output on both channels. I believe they use 2oz copper on their board, and I'm betting it's necessary. This shouldn't be an issue in any design using a normally-sized PCB with microcontroller/ADC/etc on it as well to get some more heatsink area.

Anyway, I thought I'd update you guys on this part. This could be used for a nifty DIY project for active speakers with active crossovers and PWM amplification. I might just have to do that (<-not really going to happen because I don't have the time )

[Hank]

Toys. Okay, for powered outdoor speakers for your lazy, hazy summertime evening on the deck, yes.

[Amphiprion]

I can't even look at the Pd vs Pout charts for AB amps anymore without being horrified. These little doo-dads are the way of the future man. Some of them aren't so little, either:

http://focus.ti.com/docs/prod/folders/print/tas5261.html

[iampivot]

Someone already made a DIY project out of it;

http://www.diyaudio.com/forums/showthread.php?s=&threadid=119917