TDA7294 Amplifier - Pin 8 short to case

I'm working with the ST TDA7294 audio amplifier https://www.st.com/content/ccc/reso...df/jcr:content/translations/en.CD00000017.pdf

Following the application circuit, Pin 8 and Pin 15 are both shorted together, and connected to -Vs (-35VDC in my case).

Pin 8, however, is shorted to the chip back-plate which in turn is bolted to the heat sink and case; this results in the entire case going to -Vs. Should there be a ground connection to the case, the result if fire works.

Is the application circuit correct or am I misinterpreting it somehow?

As a second question, what is the purpose of C9, the 100nF cap in parallel with C8, the 1000uF cap. Won't C8 render C9 essentially insignificant?

Thanks.

You need to use a mica insulator kit between your TDA7294 and the heatsink.

Quercus said:

the 100nF cap in parallel with C8, the 1000uF cap. Won't C8 render C9 essentially insignificant?

Click to expand...

Very large capacitors are not built to work well at high frequencies. There is inductance in the leads and internal connections which turns the cap into a inductor at high frequencies. So it is typical to put a big and a small cap in parallel. This way you get the big capacitance at low frequencies and the small cap working at high frequencies. This power amp may oscillate at 200khz.

Quercus said:

Pin 8 and Pin 15 are both shorted together, and connected to -Vs (-35VDC in my case).

Click to expand...

Yes 15, 8, and case are connected together. If your heat sink is not connected to case all is fine. Or use an insulator.

Why on earth would pin 8/-Vs be connected to the back plate at all? Wouldn't it be more practical to connect the back plate to ground?

Thanks for the explanation of the caps.

Quercus said:

Why on earth

Click to expand...

Because when you build a IC the most negative voltage is often connected to the bottom of the silicon, which is on the heatsink.

ronsimpson said:

Because when you build a IC the most negative voltage is often connected to the bottom of the silicon, which is on the heatsink.

Click to expand...

You build a semiconductor on a base substrate which is typically (not always) the most negative point, and it is bonded to the metal tab of the package. If it is s single rail device, the most neg point is ground and all is easy. If it is a dual rail system then it gets complicated as the mot negative point is -Vs and you don't want that on the heatsink which is often screwed to the chassis ground.

Thanks for the explanation.

One more question back to the caps... I'm building a stereo amp so there will be two chips. The C8 and C9 caps, 1000uF and 100nF respectively attached to -Vs (and their +Vs counterparts), will be in parallel with the same caps used for the second channel IC. Should each chip get its own C8 and C9, (resulting in parallel caps of the same value), or should there be one set of caps that supports both ICs? The current prototype I've got uses separate caps for each chip, mounted as close to the chip as possible, and seems to be working fine, tho I haven't done much testing yet.

Thanks a lot for the help here.

bountyhunter said:

-Vs and you don't want that on the heatsink which is often screwed to the chassis ground.

Click to expand...

This is true for transistors. The collector is connected to case. So in most applications the heat sink (with out insulator) is connected to a signal, or supply and not ground.

If the amplifier was AC coupled and -Vc was connected to ground all would be fine.

This all happened because the data sheet was not read. (I/we have all done this)

Quercus said:

The current prototype I've got uses separate caps for each chip, mounted as close to the chip as possible

Click to expand...

Yes! [insert happy face here]
Long wires are inductors. You don't want that on the supplies.

Thank you everyone for the help!

Quercus said:

One more question back to the caps... I'm building a stereo amp so there will be two chips. The C8 and C9 caps, 1000uF and 100nF respectively attached to -Vs (and their +Vs counterparts), will be in parallel with the same caps used for the second channel IC. Should each chip get its own C8 and C9, (resulting in parallel caps of the same value), or should there be one set of caps that supports both ICs? The current prototype I've got uses separate caps for each chip, mounted as close to the chip as possible, and seems to be working fine, tho I haven't done much testing yet.

Click to expand...

Your current assembly is not just a good way, it is the only way. Audio power amp chips are notorious for breaking into oscillation for reasons ranging from power supply lead inductance to the phase of the moon. Lotsa decoupling, close in, and always separate from everything else equals success. I once rescued a product already in production that was failing test. The electrolytic cap used for decoupling (which worked fine) went obsolete, and the only replacement that fit did not have the high frequency performance. A 100 nF cap tacked on the bottom of the board across the amp chip power pins killed the oscillations, and was much better than scrapping a few hundred units and re-spinning the pc board for a larger electrolytic.

ak