Best class D amplifier board for 9 Volt supply

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
What are your recommendations for a class D amplifier board running off a 9 volt power supply?

There appears to be a gap between the 1 watt boards (5 volt supply) and the 20+ watt boards requiring >12 volt supply?

This is for an active speaker project with a single full range 8 ohm driver. Looking to obtain around 3 watt of high quality power output with no background noises. The max current delivery of the 9 volt (regulated) supply is 1 Amp.
 
Thank you for your advice, I will narrow my search down to mono TPA3118 boards.

There are many TPA3118 boards to choose from, on places like eBay, that all look alike. Are all of these products the same or are there quality differences? How do you recognize these differences?

Would a board without a heat-sink would be OK for this low power application inside a speaker box?
 
The TPA3118 is a 'pad down' version of the TPA3116, in that it uses the PCB as a heatsink. It should be OK in a speaker cabinet.

... Especially considering the low voltage it's intended to run at :)

Going by the datasheet plots and doing some arithmetic, at 5W into 8ohm, we might be looking at >85% efficiency, so a measly 750mW of heat - that's diddly-squat :D

And that would be using a continuous sinewave; with music signals, it's gonna be idling at least half the time. Idle current's only spec'd in the datasheet at 12V the lowest, and quoted as a maximum 35mA, so let's say 30mA @ 9V, that's a quarter of a watt. In other words, pretty much nothing.
 

Attachments

  • IMG_9922.jpg
    IMG_9922.jpg
    472.5 KB · Views: 188
TPA3118 results on 9VDC

Finally got all the parts to build a test system and make some measurements. The carrier board contains a mono TPA3118 credit card style board (PBTL) and a DC/DC converter module (SIP-8).

The local power supply is the DC/DC converter, with 48VDC input and 9VDC @1A output. It is driven by a bench power supply set at +/-24V in series.

The maximum observed output in a 8 Ohm resister load with 1 KHz sine wave is 12VTT, which means a power output of 2.25 Watts. The power output is limited by the 1A max current of the DC/DC converter. I could not find a suitable 9V converter with higher current rating. My goal is to get 3-4Watt, so I have ordered a 12VDC converter with 20W rating (1.67A, DIP-24) and will try again when parts arrive.

The scope shows input and output signals; the measured gain is 32dB. The measured THD+N at this level is ~0.56% with the residual being mainly broadband noise.
 

Attachments

  • TestSetup.jpg
    TestSetup.jpg
    315.8 KB · Views: 161
  • TestSignal.jpg
    TestSignal.jpg
    219.6 KB · Views: 175
That's right, I want an integral power supply with the amp board. The intent is to power the amp using power over ethernet (PoE), which supplies 48VDC.

As stated, I have done some measurements with the amp board already. There are a couple of things that need to be modified. I want to lower the gain to 12dB or so, and I want to have a balanced input rather than single ended. On my board the negative input is grounded.

I seem to have some stability issues (seen as oscillations riding on the sine wave) which may be related to the fact that the DC/DC converter is isolated and the test equipment, connected to input and output, are grounded. I am not to worried at this point because I can make them go away by adding another ground wire.
 
The only way you can reduce the gain lower than the chip's minimum of 20dB (10x) is by adding some sort of (fixed) attenuator on the input. The former can be achieved by (carefully) removing the 100k R27 (3rd from the bottom in your photo).

"Creating" a balanced input is quite trivial as a concept, but perhaps a bit fiddly in practice. What you need to do is separate the "outside" of C21 (second component above R27) from the ground-plane (while ideally still leaving a bit of copper there, so you can scrape off the solder-resist and solder a wire to it), with the sharp tip of a box-cutter knife or something along those lines.

Just be careful not to get the solder blob all the way to the exposed edge of the remaining groundplane - ask me how i know :D
 

Attachments

  • IMG_0384.jpg
    IMG_0384.jpg
    597.8 KB · Views: 288
@Khron thanks for your detailed response; it confirms my plans nicely!

The only difference is that I want to remove the 2 input capacitors all together and solder in some leaded ones. This would take care of the tricky part to separate the one capacitor from the ground plane and provide an easier way to attach the input wires.

The 12V DC/DC converter should arrive by the end of the day; so, I'll be able to continue my quest shortly.
 
"Where there's a will, there's a way" :D

I used those to "active-ate" the pair of junky "surround" speakers i hacked together a while ago, and since they're on the other side of the room from where my computer & interface are (and the power brick's plugged into another mains socket) i figured running a balanced line couldn't be a bad thing. :rolleyes:

Khron's Cave: #28 Active full-range surround speaker project (BN96-12832 & TPA3118)

I chose to not connect the shield to the amp's ground, to eliminate the chance of ground-loops.

Very nice. :)

That's fair enough, i suppose - i'd prefer to keep those leaded caps as close to the input side as feasible (if not even under the amp, so they're shielded by the amp board's groundplane on the bottom).
Just to minimize output switching noise coupling back into the input, since leaded caps will need to be off-board, no matter how you cut it.

@Khron thanks for your detailed response; it confirms my plans nicely!

The only difference is that I want to remove the 2 input capacitors all together and solder in some leaded ones. This would take care of the tricky part to separate the one capacitor from the ground plane and provide an easier way to attach the input wires.
 
Last edited:
12V supply update

The 12V DC/DC converter does the trick and easily provides the desired 4W output in 8 Ohms. The output now clips at 20Vtt; i.e. 6.25W.

The observed THD+N at 4W and 1 KHz sine wave is 0.15%.

I have removed R27 to lower the gain to 20dB and have added a TY-145P input transformer and passive divider to bring the overall gain down to 10dB. The transformer provides a balanced input into the single ended board (no board modifications needed). The frequency response is flat between 100Hz - 10KHz; sufficient for my application.
 

Attachments

  • TPA3118 balanced 12V.jpg
    TPA3118 balanced 12V.jpg
    366.1 KB · Views: 97
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.