TPA3120D2: Competition for your T-Amp?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Has anyone thought about making a board for this new TI chip? It looks fairly simple to implement. It would make a nice "mono block" style amp in BTL configuration. Perhaps give your low powered T-amp a run for the money?

I would give it a go myself, but I would like to see what someone who actually knows what their doing can come up with; i.e. BWRX (hint hint ;) ).

http://focus.ti.com/docs/prod/folders/print/tpa3120d2.html

An externally hosted image should be here but it was not working when we last tested it.
 
Ex-Moderator
Joined 2005
There are some things I don't particularly like about this chip: the price, the package isn't the most DIY friendly because of the thermal pad on the underside, the osciallation frequency is only 250kHz and cannot be changed, the gain setting resistors are not accessible, and the input stage is very similar to what the Tripath chips use... but the circuit is simple enough and I've got some samples on the way so I can't let them go to waste ;)

We will definitely want to use this as a mono BTL amp to get the most out of the chip and the configuration. This will be open source and I'll make the schematic and layout files available so everyone can tinker and play around with them. And since it is open source any comments, suggestions, etc. are welcome. If there is enough interest I could do a group buy for the PCBs, but I do not have the time to do one for parts as well.

The TPA3120D2 is available at Digikey with the following qty/price breakdown:
QTY/UNIT PRICE
1/6.40
25/5.12
100/4.32
 
BWRX said:
There are some things I don't particularly like about this chip: the price, the package isn't the most DIY friendly because of the thermal pad on the underside, the osciallation frequency is only 250kHz and cannot be changed, the gain setting resistors are not accessible, and the input stage is very similar to what the Tripath chips use... but the circuit is simple enough and I've got some samples on the way so I can't let them go to waste ;)

We will definitely want to use this as a mono BTL amp to get the most out of the chip and the configuration. This will be open source and I'll make the schematic and layout files available so everyone can tinker and play around with them. And since it is open source any comments, suggestions, etc. are welcome. If there is enough interest I could do a group buy for the PCBs, but I do not have the time to do one for parts as well.

The TPA3120D2 is available at Digikey with the following qty/price breakdown:
QTY/UNIT PRICE
1/6.40
25/5.12
100/4.32

Did I ever mention your my hero Brian?:D

I'm staring at one of these IC's right now, and while I think I can handle soldering the pins, I'm not so sure I can do the thermal pad with the (lack of)equipment I have. What is the best way for a DIYer to solder a thermal pad like this? IR, hot air, toaster oven? I was thinking as a last resort for myself I could always just use an extremely small amount of thermal epoxy on the pad and then solder the pins the old fashioned way.
 
Ex-Moderator
Joined 2005
theAnonymous1 said:
What is the best way for a DIYer to solder a thermal pad like this?

The best way is probably using solder paste and some sort of oven. The next best way is a big old soldering iron. I don't have the proper tools at home either, but I may be able to coerce some people at work to let me use the reflow oven. If that doesn't work the thermal epoxy may be the next best solution.
 
Ex-Moderator
Joined 2005
Since I'm tired of winding my own inductors I'm looking into the surface mount types offered by Coilcraft and Pulse (if you know any other good manufacturers let me know). Because of the low switching frequency and bridged output we will want them to be around 20uH with a saturation current of at least 5A, preferably twice that. That basically narrows down the type to a drum core or laydown toroidal.

The MSS1278-223ML (22uH 6.2A) from Coilcraft is cheap, magnetically shielded, and has a reasonably small footprint. Pulse has some inductors worth considering too.
 
Ex-Moderator
Joined 2005
Those look very nice as well. Good find. It's also the only one Digikey has in stock with those specs :)

I browsed through the Coilcraft surface mount power inductor selection again and noticed the SER2800 series. They look like the SER1360s on steroids. This is more like what I was looking for except they only come in an unshielded package, are unfortunately quite a bit larger (but not quite prohibitively so), and cost around $3.50-4 each.
 
Administrator
Joined 2004
Paid Member
I figured they would be a good choice because Digikey is a reliable source and I would have to place an order through them for other parts like resistors and capacitors anyway.

I sure don't want them to break anyones sound though.:smash:

Would there be any benefit to using an opamp to reverse phase on one of the inputs instead of connecting the source ground to the right channel for BTL operation? This would add a bit more complexity because of the power supply requirements, but it could serve a dual purpose as a buffer as well.

EDIT: Nevermind, I guess it's best to keep this as simple as possible. I could always add the buffer externaly on its own board.
 
Ex-Moderator
Joined 2005
theAnonymous1 said:
Would there be any benefit to using an opamp to reverse phase on one of the inputs instead of connecting the source ground to the right channel for BTL operation?

If you grounded an input the output voltage would be at about half the supply voltage. The amp would still work fine, but you would not gain any advantages of bridged operation except for not needing output DC blocking capacitors.

So for bridged operation we will need to come up with out of phase signals for each half bridge. A single op amp to invert the signal is the simplest way but there are better ways to do this. The best one I can think of is to use a differential output op amp (like TI's THS4131) to take a single ended or differential input signal and output a differential signal. The THS4131 can even run off of 5V and has a common mode voltage pin so we can reference the differential outputs to the TPA3120's bypass pin voltage. The THS4131 could also be given its own separate supplies, but the inputs to the THS4131 will still need to be capacitor coupled. This chip also costs around $5/each, which is pretty expensive.
 
If you are ordering PCB's made with vias under the pad as reccomended by TI, the thermal pad is rather easy to deal with.

Simply tin the back of the IC very lightly, then saturate the vias in the pad area of the PCB with solder. With the pad area of the PCB still hot, apply the chip while heating the back side of the PCB pad with your iron. The solder will wick to the pad area while you align the pins.

Remember to do this as quickly as possible to avoid heat damage, but I have never suffered a casualty to heat.






-Matt
 
Ex-Moderator
Joined 2005
You made it sound easier than it really is :) Depending on how may vias are in the area and how large the copper ground pour in the area is (and the weight of the copper used on the PCB) it can take a LOT of heat to get it up to the temperature where the solder will flow nicely.

Edit: Luckily the thermal pad for the TPA3120 is only about 5mm x 3mm, so it should be easier to solder than something like the Tripath TA2024.
 
BWRX said:


If you grounded an input the output voltage would be at about half the supply voltage. The amp would still work fine, but you would not gain any advantages of bridged operation except for not needing output DC blocking capacitors.

So for bridged operation we will need to come up with out of phase signals for each half bridge. A single op amp to invert the signal is the simplest way but there are better ways to do this. The best one I can think of is to use a differential output op amp (like TI's THS4131) to take a single ended or differential input signal and output a differential signal. The THS4131 can even run off of 5V and has a common mode voltage pin so we can reference the differential outputs to the TPA3120's bypass pin voltage. The THS4131 could also be given its own separate supplies, but the inputs to the THS4131 will still need to be capacitor coupled. This chip also costs around $5/each, which is pretty expensive.

Ohhhh, so the schematic assumes a differential input, not one channel referenced to the source ground. Don't I feel stupid.:eek:

An op amp would be the most convenient for this, but I have no problem with going with the THS4131(SOIC) if you think it's a better solution.

And about the power pad... this chip is small and I can't imagine it would need to dissipate gobs of heat, so wouldn't just putting some thermal grease between the chip and the PCB be sufficient?
 
Ex-Moderator
Joined 2005
The OPA3162 and the THS4131 have the same pin outs and are very similar spec wise. They're probably ever so slightly different internally but since the OPA chip is marketed under the Burr Brown name they can charge more! Regardless, either could be used if that's the sort of chip we'd like to use.

I haven't been doing much with this because it's been a busy week, but I did create an eagle library part for the TPA3120 based off of the recommended footprint in the datasheet. I made the pads for the legs a little longer so there's more area to solder to.
 

Attachments

  • tpa3120.zip
    1.6 KB · Views: 67
BWRX said:
The OPA3162 and the THS4131 have the same pin outs and are very similar spec wise. They're probably ever so slightly different internally but since the OPA chip is marketed under the Burr Brown name they can charge more! Regardless, either could be used if that's the sort of chip we'd like to use.

I haven't been doing much with this because it's been a busy week, but I did create an eagle library part for the TPA3120 based off of the recommended footprint in the datasheet. I made the pads for the legs a little longer so there's more area to solder to.

Yeah I did some searching after I posted and found out about the similarity between the two chips. I just thought the timing of TI sending me documentation for it was interesting. Gotta love TI.:)

Take all the time you need Brian. I'm in no hurry and it would be quite rude of me to ask for progress on a project I can't add any real contributions to.
 
Ex-Moderator
Joined 2005
Hi Anonymous.

Any differential op amp (that can handle 24V across its supply pins) could be used with the single supply of the TPA3120. There is one problem though, and that has to do with the single ended input signal being referenced to ground and the common mode input voltage range of the op amp. The common mode input voltage range for most of these differential op amps is around 1V inside the rail voltages, which means the diff. op amps will not like being powered from the same supply as the TPA and seeing a common mode input voltage that is the same as their negative supply rail. It appears the only way around this is to bias the op amp inputs at half the supply rail and capacitively couple the input signal or to provide the op amp with it's own separate bipolar supply centered around ground (which will eliminate the need for coupling caps).

One nice thing about the diff. op amps is that we can reference their output voltage to any DC level set by whatever voltage is at the Vocm pin (for the TI chips anyway). The input stage of the TPA is biased at approx. AVCC/8 according to the datasheet, so we can eliminate the op amp to input stage coupling caps by referencing the op amp's outputs to the AVCC/8 bias voltage. The outputs should then stay centered around the bias voltage of the TPA input stage.

At this point though, 3 supply rails is getting more complicated than we would have hoped for.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.