I've been tweaking mine with pleasing results. The next weak spot (after those X5Rs) is the power supply to the analog rails (a familiar theme 
). On the board I have - which looks to based largely on Infineon's reference design - this supply (5V) is provided by a buck regulator (TI TPS62175) in a miniscule 10pin package. A look at the datasheet does not reveal how much output ripple can be expected. There is a post-filter prior to the Merus 5V supply pin (inductor L2) but without knowing the component values of the Ls and Cs its not possible to estimate how much reduction of the output ripple is going to occur.
I think the ideal upgrade would be to install an LDO between the switcher and the amp chip - the output voltage of the TPA can be increased to 6V to give the LDO some headroom to operate. Anyone got a suggestion for a suitable 5V LDO?
). On the board I have - which looks to based largely on Infineon's reference design - this supply (5V) is provided by a buck regulator (TI TPS62175) in a miniscule 10pin package. A look at the datasheet does not reveal how much output ripple can be expected. There is a post-filter prior to the Merus 5V supply pin (inductor L2) but without knowing the component values of the Ls and Cs its not possible to estimate how much reduction of the output ripple is going to occur.I think the ideal upgrade would be to install an LDO between the switcher and the amp chip - the output voltage of the TPA can be increased to 6V to give the LDO some headroom to operate. Anyone got a suggestion for a suitable 5V LDO?
I would try LT3042 or LT3045
Regulateur de tension lineaire RF LT3042 PSRR | Module d'alimentation, batterie inversee Ultra faible bruit, 20V 200mA, Protection du courant, CJMCU | AliExpress
Don't know though if the implementation on this board complies with the capabality of this chip...
Regulateur de tension lineaire RF LT3042 PSRR | Module d'alimentation, batterie inversee Ultra faible bruit, 20V 200mA, Protection du courant, CJMCU | AliExpress
Don't know though if the implementation on this board complies with the capabality of this chip...
Any comments on the sound quality so far compared with other amps? I am sure you have quite a bit in your arsenal...I've been tweaking mine with pleasing results. The next weak spot (after those X5Rs) is the power supply to the analog rails (a familiar theme
I think the ideal upgrade would be to install an LDO between the switcher and the amp chip - the output voltage of the TPA can be increased to 6V to give the LDO some headroom to operate. Anyone got a suggestion for a suitable 5V LDO?
Oon
Happy to give subjective impressions of this, but bear in mind I've not been listening to it stock, it has the following tweaks.
a) Shorted out the 0805 10uF input caps with 0R resistors
b) Taobao trafos used to generate + and - phases
c) Extra decoupling on MA12070 pin17 AVDD
The extra decoupling caps on pin17 are a 1800uF/16V Nichicon HZ and a Panasonic 15000uF/6.3V
With these tweaks this amp is a very enjoyable listen, comparable to my current reference TDA8932 amp with regulated supplies : Transformer-fed TDA8932 25W/8R mono amp kits
Without the addition of those extra caps the amp lacks a bit in the dynamics department - with them in circuit the subjective noise floor drops and recorded ambience becomes more apparent. When I added the caps the sound became more 3D and more detached from the speakers.
Compared to other amps - well even out of the box I preferred it to all other unmodded amps I've played with (TPA3116, TPA3255, TDA8932). I tried for a long time to get TPA3255 sounding as good as this but failed in the end - how its better is the HF is sweeter and less 'forward' - noticeable mainly on choral works.
I'm ordering a couple more of these - one analog and one with I2S input. On the new one I plan to ditch the caps and install a good regulator for that critical pin17 - thanks @yopi for your suggestion, I'll look into it. After that I'd better get cracking on software to see whether tweaking the power profiles brings any further enhancement.
a) Shorted out the 0805 10uF input caps with 0R resistors
b) Taobao trafos used to generate + and - phases
c) Extra decoupling on MA12070 pin17 AVDD
The extra decoupling caps on pin17 are a 1800uF/16V Nichicon HZ and a Panasonic 15000uF/6.3V
With these tweaks this amp is a very enjoyable listen, comparable to my current reference TDA8932 amp with regulated supplies : Transformer-fed TDA8932 25W/8R mono amp kits
Without the addition of those extra caps the amp lacks a bit in the dynamics department - with them in circuit the subjective noise floor drops and recorded ambience becomes more apparent. When I added the caps the sound became more 3D and more detached from the speakers.
Compared to other amps - well even out of the box I preferred it to all other unmodded amps I've played with (TPA3116, TPA3255, TDA8932). I tried for a long time to get TPA3255 sounding as good as this but failed in the end - how its better is the HF is sweeter and less 'forward' - noticeable mainly on choral works.
I'm ordering a couple more of these - one analog and one with I2S input. On the new one I plan to ditch the caps and install a good regulator for that critical pin17 - thanks @yopi for your suggestion, I'll look into it. After that I'd better get cracking on software to see whether tweaking the power profiles brings any further enhancement.
LT3042 regulator board is on order : https://item.taobao.com/item.htm?spm=a230r.1.14.135.100277a8KyQMLE&id=594969223329&ns=1&abbucket=20#detail
Two new amp boards have arrived - I have taken a quick look at the IIS (digital input) version of the board using the MA12070P and it looks like it might be rather tricky to get up and running.
The first reason for the difficulty is that - while its advertised as having I2S input - I2S isn't the default input mode. Meaning you can only use it with an I2S source if you program it over I2C to accept I2S. The second reason is - it looks like the 'P' version of the board is populated identically with the non-P. The 10uF input coupling caps are still fitted, despite the fact that the 'P' version has no analog input pins, only digital ones. I can't imagine how it will work with the digital inputs AC coupled.
While I heartily recommend the analog input version of this board, until I hear my I2S version singing sweetly I can't recommend it to anyone else.
The first reason for the difficulty is that - while its advertised as having I2S input - I2S isn't the default input mode. Meaning you can only use it with an I2S source if you program it over I2C to accept I2S. The second reason is - it looks like the 'P' version of the board is populated identically with the non-P. The 10uF input coupling caps are still fitted, despite the fact that the 'P' version has no analog input pins, only digital ones. I can't imagine how it will work with the digital inputs AC coupled.
While I heartily recommend the analog input version of this board, until I hear my I2S version singing sweetly I can't recommend it to anyone else.
Trying to connect the LT3042 to the AVDD (pin17) I've run into a problem. The buck reg also feeds the DVDD (pin55) and I haven't yet been able to figure out the tracking to separate these two +5V supplies. Even if I do manage to figure out how to separate them my DMM is indicating there's an internal resistor between them which will thwart attempts at isolating one from the other. So for now I think adding caps to these rails is the best solution.
On the digital input board, I have a 'Bus Pirate' on order to allow me to set the I2C registers to accept I2S input (rather than left justified which is the default).
On the digital input board, I have a 'Bus Pirate' on order to allow me to set the I2C registers to accept I2S input (rather than left justified which is the default).
is the emc filter at the outputs , as shown in the schematics , on the board?
The I2S board is working now and sounding very promising. There's another road-bump even beyond those I have already mentioned to surmount but seems worth the trouble so far. Stay tuned!
Let me answer you some of the questions of this thread which came up because I worked with it a while ago.
The MA12070 (analog) and the MA12070P (digital) are sharing the same pinout besides the analog / digital inputs.
Therefore the evaluation board is the the same for both analog and digital version.
However it is wrong, that the digital inputs are AC coupled because there is an additional pin-header on the evalboard to access the I2S pins directly.
Another comment in this thread is also correct. The default audio-input format is left-justified with 2*32 bit framing.
This audio-format can be used directly without any further configuration over I2C.
To use I2S you have to access via I2C and reconfigure the format - that's right.
Very nice feature of this amp is, that it can handle I2S format without any additional master-clock (the standard I2S BCLK is sufficient). For this you can just short-circuit the BCLK and MCLK (pin 21+32) pin of the amp together. The amp will auto-recognize it and adopt its internal PLL.
The MA12070 (analog) and the MA12070P (digital) are sharing the same pinout besides the analog / digital inputs.
Therefore the evaluation board is the the same for both analog and digital version.
However it is wrong, that the digital inputs are AC coupled because there is an additional pin-header on the evalboard to access the I2S pins directly.
Another comment in this thread is also correct. The default audio-input format is left-justified with 2*32 bit framing.
This audio-format can be used directly without any further configuration over I2C.
To use I2S you have to access via I2C and reconfigure the format - that's right.
Very nice feature of this amp is, that it can handle I2S format without any additional master-clock (the standard I2S BCLK is sufficient). For this you can just short-circuit the BCLK and MCLK (pin 21+32) pin of the amp together. The amp will auto-recognize it and adopt its internal PLL.
Interesting comments, thanks @maggusxy. Yes I did discover about the need for BCLK to be used as MCLK but with this board, even that insight isn't sufficient to get the board playing audio. That's for a couple of reasons - first there's a missing resistor (R4 on the PCB) which needs to be present to provide the MCLK to the Merus IC. Having fitted that resistor I got sound but it was crackly. What I hadn't reckoned with was that there's a pin whose level seems undefined - one which on the analog version changes between master/slave operation on MCLK. The digital (P version) only works in slave mode but the PCB supports both modes as it needs to work with the analog (non-P) chip too.
Incidentally in talking about the AC coupling, I'm talking about the Chinese-sourced board, not the official evaluation module.
Incidentally in talking about the AC coupling, I'm talking about the Chinese-sourced board, not the official evaluation module.
Last edited:
About which pin are you talking?
The namings of the pins are a little bit weird.
There is a pin called CLKM/S (Pin 31), which must be pulled low constantly. Don't know what's the sense behind this.
Your actual MCLk input is PIN 32 -> and this can be connected together with Pin21.
I also had some problems with cracking sound at the beginning.
I tried to connect the MA12070P to an ADAU1701. And it just didn't work. But then I found out finally that the ADAU1701 pulls the IO pins only to approx 0.6V low-voltage level which doesn't seem to be enough for the Merus-AMP to recognize a stable 0-level.
I tried then an STM32 to connect to the I2S interface and this worked without any further problems.
Can you maybe check the voltage high/low levels on the I2S bus? Maybe it works with an additional level shifter.
Which evalboard are u using by the way? I have the big-one with the USB-I2C bridge. There is a nice software-tool from infineon where you can try out all the I2C settings inside the computer. There is even live-level monitoring and a digital signal limiter integrated.
The namings of the pins are a little bit weird.
There is a pin called CLKM/S (Pin 31), which must be pulled low constantly. Don't know what's the sense behind this.
Your actual MCLk input is PIN 32 -> and this can be connected together with Pin21.
I also had some problems with cracking sound at the beginning.
I tried to connect the MA12070P to an ADAU1701. And it just didn't work. But then I found out finally that the ADAU1701 pulls the IO pins only to approx 0.6V low-voltage level which doesn't seem to be enough for the Merus-AMP to recognize a stable 0-level.
I tried then an STM32 to connect to the I2S interface and this worked without any further problems.
Can you maybe check the voltage high/low levels on the I2S bus? Maybe it works with an additional level shifter.
Which evalboard are u using by the way? I have the big-one with the USB-I2C bridge. There is a nice software-tool from infineon where you can try out all the I2C settings inside the computer. There is even live-level monitoring and a digital signal limiter integrated.
Yes I was talking about pin31 which selects the sense of the MCK. It needs a low on the 'P' version but seems undefined in practice - I wasn't able to trace the track to its destination. Setting it low gave me undistorted, uncrackly sound. Although not documented on the DS, if pin31 is high then it outputs an MCK from the internal oscillator on pin32, it seems to be close to 2.82MHz.
The board I'm using is the Taobao one I linked to in post #32. Visually it looks identical to the one on sale at Audiophonics and the one @daniboun has shown.
The board I'm using is the Taobao one I linked to in post #32. Visually it looks identical to the one on sale at Audiophonics and the one @daniboun has shown.
Last edited:
Ok, well.... yes as I said I was using the original evalboard
https://www.infineon.com/cms/de/product/evaluation-boards/eval_audio_ma12070p/
https://www.infineon.com/cms/de/product/evaluation-boards/eval_audio_ma12070p/
This is very interesting. I was not aware of it. Did you also check if there's a signal coming out of the WS pin?
Maybe there is some undocumented master-mode capability?
I only discovered this when 'scoping my I2S pins to see if there was a clue to the clicking I was getting. The MCK/BCK pin was the only one I recall looking weird but I could easily have missed strangeness on the others. Next time I'm playing with the 'P' version I might just have another try....
So far I'm impressed with how the 'P' version sounds. On 44k1 I found it sounded a bit 'larger than life' and too forward - I started finding it a little bit fatiguing on a longer-term listen. Running 2XOS (upsampling with SoX in Foobar) though that 'hyper-real' feel seems to have gone. But I'll need to listen many more hours to be sure.