Drive speakers with motor bridge

Talk to Disaster Area's engineers.

Yeah, I just watched it again a few weeks ago. I don't think they ever made a video or movie or even a coloring book from the fourth book of the trilogy.

I kept running into snags that bothered me too much to settle on an approach until I found this part from Monolithic Power, MP6519. It is referred to as an H-Bridge Current Regulator, but it seems very much like a class D amplifier. It accepts a PWM input which is filtered and used to set the threshold which is compared to a sawtooth waveform to generate the higher frequency PWM driving the output. The ground leg current is multiplied by the input PWM average in this process which results in an average current proportional to the average voltage on the input.

The current sense is set by a resistor making it available for other purposes such as sensing the open load case. The chip detects significant over current conditions and shuts down the chip with a flag indicating the fault.

Working from 2.5 to 28 V it is suitable for a wide range of speaker impedances. Around $1 it will require two to drive the two speakers, but requires no other expensive components. A couple of ferrites should reduce the high frequency harmonics adequately.

They show a compensation circuit. I guess I'll need to dig into that to get it right.

Using one of those newfangled "mixed-signal programmable ICs" you can design your own Class D controller:

Haven't looked into this device much but you could probably do something with these parts, internal analog integrator block, PWM modules, dual 2 amp H-bridge drivers with 12 volt supply, i2c module, dual analog comparators, programmable band gap, temp sensor, and a bunch of other logic:

Where to buy? How to program?

They're one-time programmable devices, you can grab the programmer/emulation tools and order programmed or unprogrammed chips direct from the manufacturer shop in quantities of 5000 or under, no BS. I like it (no financial affiliation, I don't have stock in the company currently but I might at some point...)

the mid-range programmer/emulator is only $59:

If I wanted to work with a 5V supply, there are tons of amps that would do the job. The driving signals are coming from an FPGA, so any logic I want is at my finger tips. It's the high side FET driving that is such a PITA. But thanks for the suggestion. I've not found a use for the Greenpak stuff. Visiting their website has become a chore to find info rather than marketing crap. They can't just explain what products they sell in engineering terms. The marketeers have to interject their spin on my application.

I'm actually pretty burnt out on this after digging around for weeks looking for a good part. This Greenpak devices would be adequate, but it's not clear if it can measure the output current. The only real advantage I can see is the cost which is not a huge factor.

It has a 12 volt secondary supply for the output stage. Yeah it can measure the output currents it has two uncommitted analog comparators...P channel high side not good enough?

No idea what kind of bandwidth it has though it's really for driving brushed DC motors. TI has a lot of integrated class D amps, nothing from them good enough?

Do you need the full hi-fi audio frequency range? You might be surprised how fast you can PWM the shutdown pin of a synchronous buck converter

How does it measure the output current? Is there a way to put a resistor between the FET source and ground of the output H bridge?

What does the pFET have to do with it?

Class D amps tend to fall in two camps, neither of which tend to support an output current measurement. One is the low power, microscopic flip chip type of device that doesn't work above 5V. The other is oriented to high output gadgets where space is no issue and don't work down to 5V necessarily. TI seems to only have parts that don't cover 7.5 V at all.

You sometimes find motor drivers that do the job but have other issues. The MP6519 doesn't seem to have any significant issues. I wish I knew more about EMI. The ferrites I've added do ok above a couple MHz according to LTspice. Adding capacitors makes it roll off faster but not at a lower frequency. Make them too large and a big peak occurs around 1 MHz. Add more capacitance and the peak gets taller.

One really nice aspect of the MP6519 is that it allows a PWM input for the signal level setting. This PWM input is separate from the speaker PWM so can be done to suit the FPGA. MPS has a two channel H-bridge chip that specifically says it can be used as a class D amp, but again... no way to measure current. It does seem to separate the FET grounds, but I just don't know if that would be suitable. So, I'll stick with the chip that gives the current measurement as part of the spec.

Can you shut it down fast enough to generate the negative portion of the waveform? Some of the drivers show design suggestions of them operating as switchers. A buck or boost regulator is really just a half bridge with a educated controller.

The LTC3623 datasheet has an example of how to use it as a class D amp, in this case injecting the audio signal to the error amp through the "ISET" pin. Page 25. There's an output ground connection for the low-side switch independent of the rest of the IC ground:

Oh, it conveniently gives an an "IMON" output, the current out of the IMON pin is 21uA * output current.

I can see why the cap to ground from the speaker is there, but I wonder why the cap to the 12V rail?

Less of a pop at switch on?

It's just this minus the resistors, there's no DC return current to the virtual ground so they're unnecessary:

(if you were using an unbuffered virtual ground like that to bias op-amp inputs etc. there would be DC bias currents and such that would need a return path to the supplies)