Posted by: @nixem
Hi all, just wanted to make this thread for the Infineon MA12070 and MA12070P amplifier boards. If anyone has any experience feel free to share!
I am interested in the P version, running direct i2s from an adau board.
At the moment, I am not sure about running multiple PBTL mode boards (3-4) on a single smps power supply, and if any special considerations need to be made, like multiple i2s amp boards on 1 PSU would need decoupling capacitors. Any ideas appreciated, I do not have much experience in this area.
So, many smps power supplies are single rail. Some considerations to have is peak transient amperage that may be called on the power supply with so many boards. For a linear transformer, you can just put a slow start to help suppress the instant call for power so it doesn’t slam down stream when you turn on the power supply. With an SMPS, it depends on nuances related to the design on if it has something to help arrest the transient hit of power for peaks or even turning on the setup.
And with bridging, it may request up to 640W instant upon being turned on. Then, you have to remember the SMPS is not fully efficient, so if you spec’ed the power supply too close to the draw on the 4 boards, pop goes the power supply. I would recommend between 20-50% more than the amount you plan on drawing regularly from all of the amplifiers, which 20% should somewhat keep the PSU safe, but then you need to also figure out the possible transient peak pulls of the amplifiers under load. Those peak transients, if they last over a certain amount of time or pull too many amps, you could pop the power supply with the surge. Think of power like water, if you have too much power go down a tube just to hit a wall, that wall has to absorb A LOT of momentum. In the same way, if the transient amps pulled exceed anywhere in the chain, you can pop something.
So, try to figure out the transient power pulled and the terms for when over current protection kicks in, so that if it does have enough capacity to handle that many amps, that it won’t kick the OC protection every time you turn it on or when you crank up the volume.
That means if you got a 1000W power supply (56% over the potential 160W per board bridged), you will leave headroom for the power supply of 20%, then you will leave room for inefficiencies of the amps (which could be an additional 6-40%, I do not know your amplifier boards and haven’t seen if the board has been benchmarked for efficiency, which means you may pull even more than the output levels, depending).
Decoupling capacitors or input buffer boards can sometimes help alleviate what I discussed for the initial power slammed into the device (they use input capacitor filters for CPUs on motherboards to help combat the transient effects, so this should help with that concern to a degree as well).
Now, decoupling capacitors or input buffer boards/power filter boards (do not confuse the latter with rectifier boards which also contain a large number of filter capacitors as a rectifier shouldn’t be needed in this situation but is required after the toroidal transformer) can have additional benefits of further smoothing PSU ripple, which you may have if you do not use an SMPS where they test for ripple which can effect audio components (all should test for ripple, but there are different acceptable amounts of ripple depending on the use case). So I will not discourage that practice of putting in a filter bank before providing the power to the amplifier board, just on certain considerations other than your question.
But the main thing is making sure the power supply is suitably rated for enough power that it can handle that much load onto the supply, that the ratings for over current protection are enough that it can handle the transient loads demanded of the power supply, etc.
There is also the question of the power from your mains having dips in voltage that can disrupt things, but that is less likely to cause the catastrophic destruction found in over current. Look for a power supply that can has under voltage protection and know that filter banks that are large enough can help with this to a degree.
Direct i2s is pretty cool. I would need to research more the chip’s handling of the signal. ( https://www.infineon.com/cms/en/product/power/class-d-audio-amplifier-ic/integrated-class-d-audio-amplifier-ics/ma12070p/ ). That said 93% efficiency on the chip at full rated load @8 Ohm, but that doesn’t mean there is not other inefficiencies of the board on which it is deployed. Just to give another data point.
I hope some of this made sense and if I misstated anything, anyone else can feel free to correct me. This is my current understanding, but that does not mean I cannot be wrong on certain points.