Infineon MA12070 Class D

Why don't you want to split the 5V supply line into two lines - one digital and one analog? It is possible to feed both lines with separate linear power supplies. The consumption of MA12070 is very low.
Probably could do, but doesn't seem to get any mention in the datasheet, unlike high end DAC datasheets. And the measured specs are based on a single 5v supply, so good enough for me :)

But first we need to see some test of your analog version and compere it with brzhifi and tweaked brzhifi

My ears are my main test instrument, so you can't rely on my results. But I will get the analog one tested once I'm done with my new DAC.

Anyway, I don't have any commercial MA12070 amps with me. I looked at a few and I don't agree with their implementation plus they don't have features I want

My MA12070 integrates with my DSP/Source/Master clock, so I need mute/engage/MCLK/i2s/i2c access. The commercial ones have Bluetooth/tone control etc which I don't need.

I've also not seen a commercial implementation that follows datahseet recommendations. For example, the flying capacitors are critical in setting the voltage for the various power steps. The datahseet dedicates a few pages to them, so it must be important. The specs are 20uF/25v minimum and the common MLCC size for that is 1206/1210. Every Chinese amp has I've seen has 0806 or smaller caps. I think they are all under specced, except for the Shenzhen Pillor one, which seem to have decent sized MLCCs

https://community.infineon.com/t5/K...-Ceramic-Capacitor-MLCC-KBA223633/ta-p/250035

So doing my own board enables me to do the 'bigger is better' DIY thing 🤣
 
Last edited:
  • Like
Reactions: 1 user
Disabled Account
Joined 2002
Datasheet says to use 10 µF 25V (or higher) for the flying capacitors. The recommended part is GRM21BZ71E106KE15L which is a 10 µF 25V X7R 0805 cap. Some amplifiers have 2 x 10 µF X5R 0805 in parallel to compensate for the capacitance loss with DC bias applied. The minimum value is 4 µF at 0.5 x PVDD. In the first edition of Aiyima A8 these caps made noise, something they solved in the second revision. I use one every day and it is dead silent except when it is supposed to make noise. It also has an IC for the PMP modes FWIW.

BTW some brands produce amplifiers with MA12070 that are very sensitive to power supply connections and they even put stickers on the packaging telling the right sequence (first DC then AC and the reverse at pulling power). Like some already experienced the chip then burns at the power supply pins. I suspect it has to to with the 5V ramp up time but that is only a suspicion.
 
Last edited:
I've always questioned why the 0806 GRM21BZ71E106KE15L was recommended. 0805 10uf X7R MLCCs are rare and expensive compared to more common 1206/1210 ones.

image (1).png


I did a sim when I designed my amp and the 0806 capacitance falls significantly (green line) to about 2.5uF @13v (1/2 PVDD), that's why 2 are required to make the minimum 4uF. Just a size up, (1210 - blue line), the performance is much better that you need only one unit.

My concern is the quality of the 0805 capacitors used in Chinese boards, especially when X7R is relatively expensive. Probably X5R substituted, where capacitance peaks at about 80c, and probably only marginally reaching 4uF at 25-30c normal working temps, even when 2 x 10uF are used.

But as long as the amp stays quiet and perform up to expectations, then its good enough. Also good to know that the Aiyima A8 has proper implementation to select the right PMP mode.
 
Last edited:
I have high hopes that the I2S version will sound good enough to be put in my Behringer DCX crossover instead of the AK4393 DACs.
3 of them might even fit into the case.
AD1896 came in so if programming the Attiny fails I'll be covered.
AK4393 still sounds pretty good, but you will be bypassing all the post DAC nasty filters and op-amps.

The I2S MA12070 was designed exactly for this situation; a static I2S format and datarate. Just need to lift the i2s signals from the board get the initial settings right.
 
Can one use telnet to see what, if any, messages are displayed during the boot process? It would be very helpful to see what the calculated clock rates are, as well as all the register values.
Telnet is something I've not heard of for a long time :p

I believe you can program an Arduino to read the register via i2c using the wire library and use the 'serial print' command to output it on the Arduino IDE on the computer. But I'm terrible at microcontrollers and takes me ages to get a script working. Normally getting some simple syntax wrong.
 
Anybody else have this DC offset?
Yes I switched every L+R I could and speakers output.
Not a big thing just wonder.
Also check if you have one inverted output, I had one but havent double checked it all the way.
Balanced input.
Just thinking about layout on pcb @altus fidelitas

 

Attachments

  • s-l1600 (2).jpg
    s-l1600 (2).jpg
    307.6 KB · Views: 113
Last edited:
Disabled Account
Joined 2002
2 and 9 millVolt or 0.002 and 0.009 Volt can be ignored as it can be called zero offset. Probably will be even less with better filtering. I wouldn't even worry till about 20 mV just for the sake of desired "perfection" as even 50 mV won't harm a loudspeaker.

Speak0A and Speak1A should both be + AFAIK. I have the 1.3 so judgement from your 1.5 picture.

Balanced inputs are nice but less so without balanced volume control. Only if you are truly faithful you can connect a power amplifier directly to a device with software volume control. It can go well for some time, many did it ...and burned woofers. I am curious how most balanced people (I see what I did here) do this as 4 channel potentiometers are not the best. Probably relay based stepped attenuators are about the best one can build. Gustard does this in some DACs.

Unbalanced to balanced active conversion with unbalanced volume control in front of it is way easier and cheaper to realize. Better than the single cap to GND ChiFi way of doing it.
 
Last edited:
Disabled Account
Joined 2002
Yes I did. Tripath, by failure.

But sudden 100% volume is not funny either. Power blackouts/brown outs and then severe power on/off phenomena sometimes repeatedly.

It is a choice. Probably depending on loudspeaker costs and a wish to prevent any failure.
 
Last edited:
Disabled Account
Joined 2002
Solve it is a question of gambling. Often I am the person in my circle that is called when stuff is already broken. Gamblers usually are not able to repair. Gamblers always try to reduce real life risks by waving them away. “I do this for years” not realising it is a question of “when” and not “if”. Acting before thinking also is in the pattern.

British cottage industry amplifiers that had no speaker protection as that was bad for sound quality. Boy oh boy that costed some woofers.
 
Last edited: