🙂 An answer on the last lines of the post before.
And back to MA12070 which is still not in stock.
And back to MA12070 which is still not in stock.
Last edited:
I changed to OPA1656 on my Aiyima A8 and it was very bass thin.
So I put on that bass boost on the bottom and there it comes!
Maybe some impedance problems.
A little clearer I think the OPA1656, it have something that I like from also TPA3255.
So I put on that bass boost on the bottom and there it comes!
Maybe some impedance problems.
A little clearer I think the OPA1656, it have something that I like from also TPA3255.
Mmmm. Fishy. The NE5532 were OK but only 1 x OPA1642 was an improvement straight away. No bass boost anymore so I can’t try that out anymore.
“Just doing something” is a guilty pleasure and I very well see the benefits. Next is to replace the 2 other opamps for OPA1642.
Since I am more critical with the device I noticed (probably) the flying caps to make mechanical noise.
Ah the device will be compared to a McIntosh MA252 in a few weeks. Curious if it can be compared at all 😉
“Just doing something” is a guilty pleasure and I very well see the benefits. Next is to replace the 2 other opamps for OPA1642.
Since I am more critical with the device I noticed (probably) the flying caps to make mechanical noise.
Ah the device will be compared to a McIntosh MA252 in a few weeks. Curious if it can be compared at all 😉
Last edited:
I added a sub. My speakers were never designed to be used without one.
I can say that the MA12070 does not lack bass now that there is a sub... but yeah, I could not really eval before/after since my cabinets are tuned to drop off like a rock below 90Hz.
One thing I can say for sure is that the epads should not be left disconnected. I soldered one amp's epads (using pb/sn paste) and left one amp's epads bare. The one with soldered epads dissipates a LOT more heat through the PCB. Since the PCB is getting warmer, it means the actual silicon is cooler (since heatsink on top does almost nothing) and that amp actually plays slightly louder at the same input level. This measures out to be from +0.5dB to +2dB. Enough to be annoying.
I think these lil guys get toastier than we realize. The text on the application notes also just 'clicked' for me. 125C on a 4-L board (I don't think many boards are actually 4 layers) from a 24V supply ( I think most run 24-26V from commercial offerings?) and only at 11WPC. Can see that even with a 4-layer board the dissipation is rather poor. I'm sure they are designed to run hot, but since it is not NAND memory where an electron pierces the barrier for each read/write (easier/less wear when running warm) there seems to be no harm in running them cooler. I really should mill those PA3S cases and flip the heatsinks to the PCB side.
Has anybody else tried a proper vs improper heatsink mount? Maybe somebody with a thermal camera? 😀
I can say that the MA12070 does not lack bass now that there is a sub... but yeah, I could not really eval before/after since my cabinets are tuned to drop off like a rock below 90Hz.
One thing I can say for sure is that the epads should not be left disconnected. I soldered one amp's epads (using pb/sn paste) and left one amp's epads bare. The one with soldered epads dissipates a LOT more heat through the PCB. Since the PCB is getting warmer, it means the actual silicon is cooler (since heatsink on top does almost nothing) and that amp actually plays slightly louder at the same input level. This measures out to be from +0.5dB to +2dB. Enough to be annoying.
I think these lil guys get toastier than we realize. The text on the application notes also just 'clicked' for me. 125C on a 4-L board (I don't think many boards are actually 4 layers) from a 24V supply ( I think most run 24-26V from commercial offerings?) and only at 11WPC. Can see that even with a 4-layer board the dissipation is rather poor. I'm sure they are designed to run hot, but since it is not NAND memory where an electron pierces the barrier for each read/write (easier/less wear when running warm) there seems to be no harm in running them cooler. I really should mill those PA3S cases and flip the heatsinks to the PCB side.
Has anybody else tried a proper vs improper heatsink mount? Maybe somebody with a thermal camera? 😀
Maybe it is because of the modifications!? Or do you assume the PA3S to be equal to the evalution board!? The board in the picture has 4 Ohm load and probably it is measured with a sinus testsignal.
The ones I have are not even lukewarm when playing at moderate volume levels. Haven’t measured output power but at 11W music power per channel I would need new windows. As I will be modding the A8 again I can measure temp.
Aiyima A8 has a quite large heatsink. There is no valid reason to change anything in that regard I think. The chip also is designed to run without heatsink and it protects itself too.
The ones I have are not even lukewarm when playing at moderate volume levels. Haven’t measured output power but at 11W music power per channel I would need new windows. As I will be modding the A8 again I can measure temp.
Aiyima A8 has a quite large heatsink. There is no valid reason to change anything in that regard I think. The chip also is designed to run without heatsink and it protects itself too.
Last edited:
Oh PA3s is in no way a 4 layer board. It;s 2 layers and maybe 1.5oz copper. I've finished mods to the DAC so I'm just brainstorming if there's aaaanything else I can do to improve small signal SNR before I figure out where to permanently attach the boards to the speaker boxes.
I don't think the mods would really do anything because they're (the mods) just bypassing the input select relays and then hard-setting the volume pot to 68.75% by replacing it with resistors. The rest of my stuff is just because I want to understand the entire board / look at it on the scope, and how it all works before deciding what I really want to change.
I just have a little urge to tinker and haven't really used the mill much lately. Sadly I have no FLIR and no thermocouples I trust below 180C. Some SOCs can feel quite cool through the package but have a junction temperature of 80C. I think the louder amp being a little more warm is because it is better dissipating heat through the PCB (and the silicon is running cooler because the the heat energy is working on a larger thermal load). Similar to how an empty water heater will feel quite cool until the elements burn out.
I don't think the mods would really do anything because they're (the mods) just bypassing the input select relays and then hard-setting the volume pot to 68.75% by replacing it with resistors. The rest of my stuff is just because I want to understand the entire board / look at it on the scope, and how it all works before deciding what I really want to change.
I just have a little urge to tinker and haven't really used the mill much lately. Sadly I have no FLIR and no thermocouples I trust below 180C. Some SOCs can feel quite cool through the package but have a junction temperature of 80C. I think the louder amp being a little more warm is because it is better dissipating heat through the PCB (and the silicon is running cooler because the the heat energy is working on a larger thermal load). Similar to how an empty water heater will feel quite cool until the elements burn out.
By coincidence I got a brand new Sabaj A10a again yesterday. I felt it needed a second chance. However in unmodded state it is less good than the modified Aiyima A8. Will let it run in today and maybe things will change slightly.
This time I promise to make pictures when I will do work on it.
This time I promise to make pictures when I will do work on it.
How does the efficiency of the Infineon chips compare with TI chips of similar power ratings? Obviously this is implementation dependent so people won't be able to give specifics, but I'm just trying to get a rough feel for how much difference the multi-level switching makes. For my use case I expect I'll often be playing at lower power so I'm guessing efficiency in the 0.5W - 5W range will be important.
The challenge as a relative beginner is less around finding the link to the datasheet and more around how best to interpret it. To me the numbers look very similar across both manufacturers, and I suspected the ma12070 may have the slight edge at lower power due to topology. However I struggle to know whether I'm reading the datasheets correctly sometimes, and there are likely real-world implementation factors I'm not considering that people on here will be more aware of.
Using this as a reference point, how might I accurately compare at say 0.5W?
https://www.ti.com/lit/ds/symlink/tpa3244.pdf?ts=1667153620130&ref_url=https%3A%2F%2Fwww.ti.com%2Fproduct%2FTPA3244
If I'm reading the chart on page 10 correctly, that looks like 45% at 0.5W, 60% at 1W.
When I look at page 29 in the Infineon datasheet, the resolution doesn't appear good enough to get accurate insight around 0.5-1W. It does however state 80% efficiency at 2W (which seems to compare to about 75% for the TI chip at 8ohm).
I realise these two amps aren't perfect for a like for like comparison, but I was wondering whether anyone had real world insights on whether they think the multi-level switching has a noticeable impact on efficiency.
Using this as a reference point, how might I accurately compare at say 0.5W?
https://www.ti.com/lit/ds/symlink/tpa3244.pdf?ts=1667153620130&ref_url=https%3A%2F%2Fwww.ti.com%2Fproduct%2FTPA3244
If I'm reading the chart on page 10 correctly, that looks like 45% at 0.5W, 60% at 1W.
When I look at page 29 in the Infineon datasheet, the resolution doesn't appear good enough to get accurate insight around 0.5-1W. It does however state 80% efficiency at 2W (which seems to compare to about 75% for the TI chip at 8ohm).
I realise these two amps aren't perfect for a like for like comparison, but I was wondering whether anyone had real world insights on whether they think the multi-level switching has a noticeable impact on efficiency.
Last edited:
https://e.video-cdn.net/share?video...er-id=E9W5foaMoUuxrto5-fGaPP&channel-id=37319
Is quite good at explaining. Its all about the switching losses
Is quite good at explaining. Its all about the switching losses
That video is funny, they tell you a 1k sine wave is annoying, then they play it for the first 30 secs of the video. On narrow bandwith laptop speakers it's excruciating. Thanks for sharing - I think I have a (very basic) grasp of the theory, I first became aware of multi-level switching via PKN amplifiers (PA world), but my feel for how much real world difference this makes, particularly at this power range, is poor. Anyway, sorry for clogging the thread with vague open-ended queries 😀
With the same Vdd, switching freq of output stage and loss in transisors it should make a big difference.
But with different values of those parameters the comparission between amps is difficult.
Switching freq and loss is related to cost. So the whole idea is to get same performance at lower cost.
But with different values of those parameters the comparission between amps is difficult.
Switching freq and loss is related to cost. So the whole idea is to get same performance at lower cost.
In analog all constructors see the link between supply rail voltage and heat generated in transistors. (I not, you have a lot of smoke)
So lower supply voltage generates less heat. Here is the supply voltage adapted to the input signal in a somewhat coarce way.
Lets say as a excample -20 volt and +20 volt or -10 volt and +10 volt instead of only -20 volt and +20 volt
So lower supply voltage generates less heat. Here is the supply voltage adapted to the input signal in a somewhat coarce way.
Lets say as a excample -20 volt and +20 volt or -10 volt and +10 volt instead of only -20 volt and +20 volt
It looks like this device can accept SPDIF 5.1, so I can't picture how this would work. Is there a chipset that converts 5.1 into multiple I2S data streams to feed the MA12070P amplifiers? Seems it would be easier to use MA12070 amplifiers instead. 🤔
I have no experience of this so please take with a pinch of salt, but I believe you'd need to decompress the 5.1 input, maybe you can use something like the Cirrus Logic CS49834 to output to separate i2s. There's no datasheet shared publicly so I'm not sure I'm right here.
There is probably some processing - filtering, EQ etc. - happening in the digital domain. I suspect this makes MA12070P easier, because that avoids a DAC in between the DSP and the amps.
There is probably some processing - filtering, EQ etc. - happening in the digital domain. I suspect this makes MA12070P easier, because that avoids a DAC in between the DSP and the amps.
To answer your previous question, the TPA3116D2 is more efficient than the MA12070: @5W, it;s about 75% effieicnt for the TPA3116S2 If we look at the MA12070 at 4-Ohms, the TPA3116D2 is more efficient above ~10W. Both are remarkably similar, except the MA12070 efficiency flatlines much faster.
The TPA3255 is less efficient. This is partly why I am confused about people heatsinking TPA amps and saying the MA12070 does not need to be heatsinked. I guess under 10W neither would need a sink but both will live a whole lot longer and have lower distortion with a proper heat dissipation method.
I have spoken before about dielectric resistance in silicon and how it degrades with heat. Touching the plastic case means nothing. To measure the heat, you MUST measure at the heat dissipation method which on the ma12070 is the epad. Depending on the plastic case, it could easily be 100C but feel like 40-45C to your finger on the plastic. The 12070 will make more heat at any load above 2.5W than the 3116D2 with both running on a 24V supply due to lower efficiency. Power in = (Power out / Efficiency %)
The TPA3255 is less efficient. This is partly why I am confused about people heatsinking TPA amps and saying the MA12070 does not need to be heatsinked. I guess under 10W neither would need a sink but both will live a whole lot longer and have lower distortion with a proper heat dissipation method.
I have spoken before about dielectric resistance in silicon and how it degrades with heat. Touching the plastic case means nothing. To measure the heat, you MUST measure at the heat dissipation method which on the ma12070 is the epad. Depending on the plastic case, it could easily be 100C but feel like 40-45C to your finger on the plastic. The 12070 will make more heat at any load above 2.5W than the 3116D2 with both running on a 24V supply due to lower efficiency. Power in = (Power out / Efficiency %)