100 kOhm is odd certainly when opamp driven. Also 10kOhm attenuation resistor in series with the potentiometer is unnecessary. A stop resistor of 100...300 ohm should do it.
Effectively you have balanced input converted to unbalanced only to be converted to balanced again. What is gained with double conversion? A cheaper potentiometer for sure 🙂
I think I would leave away the section before the potentiometer and have just the active UNBAL/BAL section. Then a passive balanced input without volume for MA12070. Issue is of course that MA12070 performs best when driven balanced regardless of unbalanced or balanced stuff as sources. So: input of MA12070 should always be balanced. The choice for BAL/UNBAL should be done before that point. That seems an error in the schematic.
It would be OK if Radian would chime in, he uses MA12070 without any stage before it AFAIK.
Effectively you have balanced input converted to unbalanced only to be converted to balanced again. What is gained with double conversion? A cheaper potentiometer for sure 🙂
I think I would leave away the section before the potentiometer and have just the active UNBAL/BAL section. Then a passive balanced input without volume for MA12070. Issue is of course that MA12070 performs best when driven balanced regardless of unbalanced or balanced stuff as sources. So: input of MA12070 should always be balanced. The choice for BAL/UNBAL should be done before that point. That seems an error in the schematic.
It would be OK if Radian would chime in, he uses MA12070 without any stage before it AFAIK.
Last edited:
Correct, cheaper and easier to source stereo volume potentiometer by adding one additional OPAMP.
I´ll change to 10k before potentiometer to 300R and choose a 10k ALPS RK09 potentiometer (should be available but not checked).
Let the customer give advising.
Don´t forget, I´m spending a lot of time to make a PCB for the DiyAudio customers and we need to freeze the design to finish the job.
Regarding PSU:
The question is, are expensive low noise LDO´s really important for better SQ? Seems that no one had an answer so testing and measuring lefts.
JP
I´ll change to 10k before potentiometer to 300R and choose a 10k ALPS RK09 potentiometer (should be available but not checked).
Let the customer give advising.
Don´t forget, I´m spending a lot of time to make a PCB for the DiyAudio customers and we need to freeze the design to finish the job.
Regarding PSU:
- searching solution for the +19 to +26VDC@5 - 8A (silent Switcher from LT/AD?)
- +5VDC@200mA: LT3045 (7.6USD) or simple 7805
- +/-15VDC: TPS7A3901 (TI stock only 2.5USD) or simple 7815/7915
- custom toroidal transformer from TOROIDY (good experience and good quality expected)
The question is, are expensive low noise LDO´s really important for better SQ? Seems that no one had an answer so testing and measuring lefts.
JP
No need to freeze when it is not ready yet. It is not commercial, no need for version 2 with blue LEDs etc. and choices regarding PSUs must be made first. First version should be AOK and also final when looking at costs. Yours truly can be a beta tester.
PSU:
2. As MA12070 itself is not solderable by the average DIYer LT3045 comes to mind but it is indeed expensive. Maybe the odd one can be used: AMS1117-ADJ with Cadj and set at 5V. These cost almost nothing (10 pieces for 1.90 Euro shipping included 🙂). Maybe smaller and closer to the chip counts more than absolute noise numbers when looking with a half eye to the various more noisy switchers on ready made boards.
3. TPS7A3901 seems a very fine choice for this purpose. Small, good and cheap.
4. Custom toroid was also my conclusion but the project will be expensive. I would never use 2 or 3 transformers.
Maybe it is handy to first ask for a price for that Toroidy 100 ... 120VA toroid with let's say 18 or 20V 5A, 2 x 6V 250 mA and 2 x 12V 250 mA. The values can be changed but I think one will use LDO regs anyway so then 2 x 6V is enough just like 2 x 12V. 27.5V is the maximum of the chip, I would play safe with 20 ...26V regulated and then preferably on the conservative side. If those last few Watts are decisive then one picked the wrong chip anyway. With 18V one can also use unregulated voltage so just rectifiers and CLC with large filter caps.
This all goes down to the fundamental choices of wanting getting the best out of the chip regardless of cost or to have a very well performing amplifier. Both choices will cost a multiple of ready made chi-fi. This will stir up debate (again) about switchers versus linear. You know my view. What is the final cost that is acceptable to you and possible builders? 200 Euro?
PSU:
- searching solution for the +19 to +26VDC@5 - 8A (silent Switcher from LT/AD?)
- +5VDC@200mA: LT3045 (7.6USD) or simple 7805
- +/-15VDC: TPS7A3901 (TI stock only 2.5USD) or simple 7815/7915
- custom toroidal transformer from TOROIDY (good experience and good quality expected)
2. As MA12070 itself is not solderable by the average DIYer LT3045 comes to mind but it is indeed expensive. Maybe the odd one can be used: AMS1117-ADJ with Cadj and set at 5V. These cost almost nothing (10 pieces for 1.90 Euro shipping included 🙂). Maybe smaller and closer to the chip counts more than absolute noise numbers when looking with a half eye to the various more noisy switchers on ready made boards.
3. TPS7A3901 seems a very fine choice for this purpose. Small, good and cheap.
4. Custom toroid was also my conclusion but the project will be expensive. I would never use 2 or 3 transformers.
Maybe it is handy to first ask for a price for that Toroidy 100 ... 120VA toroid with let's say 18 or 20V 5A, 2 x 6V 250 mA and 2 x 12V 250 mA. The values can be changed but I think one will use LDO regs anyway so then 2 x 6V is enough just like 2 x 12V. 27.5V is the maximum of the chip, I would play safe with 20 ...26V regulated and then preferably on the conservative side. If those last few Watts are decisive then one picked the wrong chip anyway. With 18V one can also use unregulated voltage so just rectifiers and CLC with large filter caps.
This all goes down to the fundamental choices of wanting getting the best out of the chip regardless of cost or to have a very well performing amplifier. Both choices will cost a multiple of ready made chi-fi. This will stir up debate (again) about switchers versus linear. You know my view. What is the final cost that is acceptable to you and possible builders? 200 Euro?
Last edited by a moderator:
Jean-Paul I think hits the nail right on the head in the last 2 paragraphs.
Nobody really complains about the SQ in the Shenzhen Pillor boards which are just clones of the Merus Eval boards. Aiyima has the A08 and Topping has the PA3s which go for about 90 USD and 150 USD, respectively. SMSL has the DA6 which includes the PSU, remote, digital volume, etc. and measures fairly well on ASR.
The price of the Topping includes a 26V SMPS with <50mV Vpp output (very good!). Then there is always the looping Aiyima A07 and the question of, "Why would I spend more, to get less overhead, and introduce a lot of complexity; when I can get this cheap amp, a cheap PSU, and then do a couple component swaps?"
For somebody looking at a DIY chipamp PCB, they would naturally also consider the plethora of TPA3255 boards on the market. The XKR is a very premium TPA3255 board and a dual channel unit will run about 200 USD w/o power supply. Even then, I think some people would look at the total BOM and decide that is a bit glamorous for a Class D where we have seen so much improvement/change recently. I think the demand we see here for those boards might be a little bit artificial due to supply issues. Outside of DIYAudio I do not see people talking about said boards. Not to mention any future TI releases, future Infineon releases, or the coming of the MA5332 boards (and people seem to like the IRS2092 so a discrete packaged component of that family would probably be a nice seller). There is also the Leaf Audio dual MA12070 with bal/unbal input, remote, screen, and headphone amp for ~150-180 USD. The Sabaj A10A and A20A 2022 versions. These all come with everything to have a working amp but can also be a DIY project if so desired. (If it ain't broke, fix it until it is!)
Any shoppers of a bare PCB would see the 3e ADAU1707 selling points (DSP! even though that chip on a TPA3255 may be a net neutral/negative) , as well as boards from other CN houses such as Aiyima who offer a dual TPA3255 fully balanced bare PCB for 90 USD. Then you have the people building active monitors who would be constantly wondering if the Wondom JAB5 might be easier/more cost effective or perhaps an ICEpower module and discrete DSP. Once you crest the 200USD per speaker/channel hill, the Hypex plate amps start to look like a really good value, too.
It seems nobody around here with the PA3s has been curious/willing enough to just unscrew the case until I got mine; and it has been out for over a year already. If it didn't require a button press to turn it on then I may have never even opened them. This tells me that even audio nerds might feel the reference design meets their needs for a 20-40W dual-chip MA12070 unit. It will be a difficult sell to tell folks that they should pay more for a single chip unit that they then need to house and complete.
The end customer would (presumably) be responsible for all the I/O, the enclosure, etc. I have only built 3-4 amps because by the time I added all the parts, time, etc (especially I/O jacks, knobs, enclosures, power supplies, so on) the build just was not even close to worth the cost.
As a designer, you may think "well, just grab a couple Molex microfit Jr males and crimp them; easy!" But most people don't have lumps of spare interconnects nor parts laying around. To market a DIY board you really need to consider somebody looking at it and buying crimpers, interconnects, ports, enclosures, power supplies, etc. Even with the very very budget-oriented boards (3116D2, et al.) I would bet more of those sales are sitting on shelves incomplete than they are in service... because people actually looked at the cost of completing the amp compared to a ready-made offering.
The beauty of the MA12070 is that you can get nice measurements from a power efficient and lower cost platform than a built-out TPA3255 if no extra power is needed. But most people would need/want the extra power because the MA12070 can only truly handle very efficient (usually expensive) speakers so the cost/benefit argument falls on its head beyond highly efficient designs once you leave the mobile realm.
I think this is very evident just by looking at the datasheet. Infineon targets the chip to battery powered applications because they know it is not very competitive in a hardwired install market. The more boutique you make a low-powered chipset, the less consumer base you will have who actually wants to pay for it. The MA12070 uses a 5V rail because that voltage is already more-or-less guaranteed to be present in any mobile device that can charge via Lightning, USB, or QI. The 26V max is there because most QuickCharge circuitry can take up to 30V and buck/boost to/from 5V in most mobile applications.
Example: The Ali platforms are chock full of Celeron JXXXX router computers. But if you look at a lot of people who actual build Opnsense/Pfsense routers they tend to buy either first party units or repurpose corporate second-hand mini PCs such as the Lenovo Tiny series or HP Elitedesk series. Why? Because when a Lenovo Tiny with a Ryzen processor (TPA3255) is being sold for only a little more than a dedicated ultra low power Celeron JXXX unit (MA12070) it makes sense to be able to use it in more than a single, niche application.
This is one area where I, as an engineer by training (mechanical) (a.k.a taught about cost reduction in school), would look at (for example) the flying capacitors and say "The pins are only broken out of the package so you can put a tiny circuit with a capacitor next to the chip. There is no need nor desire to use large parts and extend the traces. With base designs using technically inferior parts just skirting the thresholds of audibility, and only bothering very few people, this is wasted capital which cannot be recovered in design time, consumer price, nor in production cost."
As a passion project I think you can do whatever makes you happy. But if you plan to sell the product to the community then everything is different.
I would advise this: If you want to make a truly special MA12070 board then internalize the 5V rail (can use the SilentSwitcher above, or even a less expensive part with a lower current rating) and just make the common components have multiple connection options via nested pads.
- Put in traces for ferrite beads, for air coils, and for the big shielded inductors. Slap both electrolytic SMD capacitor pads and through hole pads of different spacing next to teach other on the input paths. Lay it out for the largest components, then nest-in pads for optional smaller components. Have everything ship by default with lower cost parts installed as per the MA12070 datasheet spec. This will reduce your per-board cost to manufacture while providing a modular PCB for DIY customers to adjust as they wish.
- Anybody who wishes to mod the board with pricey capacitors, inductors, etc., etc. would just need to remove a couple SMD components (easy to do accidentally) and swap in their own chosen components. If people want to take detailed measurements across specific parts of the circuit, they are given exposed pads to do so.
- Make the board a dual MA12070 (a whole chip is $5.25 at Mouser) and allow people to have either a dual PBTL or quad BTL board with their own chosen components or just the datasheet specs. This will expand the potential interest because people can at least get a clean 25-ish Watts into 8-Ohms per channel as opposed to 10-15W. Since two inputs can be ignored / left floating in PBTL mode it could be as simple as wicking a solder bridge between two pads on the PCB to disable the extra inputs. A single channel balanced input is 3 pins. A stereo unbalanced input is 3 pins. Relay to the opamps like Topping, have a couple solder bridges, or use jumpers. Designer's choice. Potentiometer? Put in a header. Those who want one can install it. Those who do not can just bridge the input and output.
- Put in a few 3-pin headers to tie the board always on /unmuted or alternately route the control and I2C pins to a DIP8 or DIP16 socket. People can then either have it just be a dumb poweramp, wire out the pins to custom MCUs, or use a PIC like the Attiny series. They can even slap a DIP to SMD adapter in the socket and go wild. There would also need to be jumpers for internal/external clock but that's sunk cost of two jumper positions instead of forcing people to use the mode you laid out.
I am not saying your design is poor. I think you have done a better job than I ever could. But there are already several options for chipamps ranging from expensive and boutique (XKR) to cheap and preassembled (Aiyima, SMSL). What does not exist is easily-modified boards with multiple component pads that could satisfy a whole range of customers ranging from measurement and numbers-driven (me) to those who want to customize every step of their signal path (jean-paul).
I hope this helps you think of ways to make your board appealing to more people. I think that is what you asked for.
Nobody really complains about the SQ in the Shenzhen Pillor boards which are just clones of the Merus Eval boards. Aiyima has the A08 and Topping has the PA3s which go for about 90 USD and 150 USD, respectively. SMSL has the DA6 which includes the PSU, remote, digital volume, etc. and measures fairly well on ASR.
The price of the Topping includes a 26V SMPS with <50mV Vpp output (very good!). Then there is always the looping Aiyima A07 and the question of, "Why would I spend more, to get less overhead, and introduce a lot of complexity; when I can get this cheap amp, a cheap PSU, and then do a couple component swaps?"
For somebody looking at a DIY chipamp PCB, they would naturally also consider the plethora of TPA3255 boards on the market. The XKR is a very premium TPA3255 board and a dual channel unit will run about 200 USD w/o power supply. Even then, I think some people would look at the total BOM and decide that is a bit glamorous for a Class D where we have seen so much improvement/change recently. I think the demand we see here for those boards might be a little bit artificial due to supply issues. Outside of DIYAudio I do not see people talking about said boards. Not to mention any future TI releases, future Infineon releases, or the coming of the MA5332 boards (and people seem to like the IRS2092 so a discrete packaged component of that family would probably be a nice seller). There is also the Leaf Audio dual MA12070 with bal/unbal input, remote, screen, and headphone amp for ~150-180 USD. The Sabaj A10A and A20A 2022 versions. These all come with everything to have a working amp but can also be a DIY project if so desired. (If it ain't broke, fix it until it is!)
Any shoppers of a bare PCB would see the 3e ADAU1707 selling points (DSP! even though that chip on a TPA3255 may be a net neutral/negative) , as well as boards from other CN houses such as Aiyima who offer a dual TPA3255 fully balanced bare PCB for 90 USD. Then you have the people building active monitors who would be constantly wondering if the Wondom JAB5 might be easier/more cost effective or perhaps an ICEpower module and discrete DSP. Once you crest the 200USD per speaker/channel hill, the Hypex plate amps start to look like a really good value, too.
It seems nobody around here with the PA3s has been curious/willing enough to just unscrew the case until I got mine; and it has been out for over a year already. If it didn't require a button press to turn it on then I may have never even opened them. This tells me that even audio nerds might feel the reference design meets their needs for a 20-40W dual-chip MA12070 unit. It will be a difficult sell to tell folks that they should pay more for a single chip unit that they then need to house and complete.
The end customer would (presumably) be responsible for all the I/O, the enclosure, etc. I have only built 3-4 amps because by the time I added all the parts, time, etc (especially I/O jacks, knobs, enclosures, power supplies, so on) the build just was not even close to worth the cost.
As a designer, you may think "well, just grab a couple Molex microfit Jr males and crimp them; easy!" But most people don't have lumps of spare interconnects nor parts laying around. To market a DIY board you really need to consider somebody looking at it and buying crimpers, interconnects, ports, enclosures, power supplies, etc. Even with the very very budget-oriented boards (3116D2, et al.) I would bet more of those sales are sitting on shelves incomplete than they are in service... because people actually looked at the cost of completing the amp compared to a ready-made offering.
The beauty of the MA12070 is that you can get nice measurements from a power efficient and lower cost platform than a built-out TPA3255 if no extra power is needed. But most people would need/want the extra power because the MA12070 can only truly handle very efficient (usually expensive) speakers so the cost/benefit argument falls on its head beyond highly efficient designs once you leave the mobile realm.
I think this is very evident just by looking at the datasheet. Infineon targets the chip to battery powered applications because they know it is not very competitive in a hardwired install market. The more boutique you make a low-powered chipset, the less consumer base you will have who actually wants to pay for it. The MA12070 uses a 5V rail because that voltage is already more-or-less guaranteed to be present in any mobile device that can charge via Lightning, USB, or QI. The 26V max is there because most QuickCharge circuitry can take up to 30V and buck/boost to/from 5V in most mobile applications.
Example: The Ali platforms are chock full of Celeron JXXXX router computers. But if you look at a lot of people who actual build Opnsense/Pfsense routers they tend to buy either first party units or repurpose corporate second-hand mini PCs such as the Lenovo Tiny series or HP Elitedesk series. Why? Because when a Lenovo Tiny with a Ryzen processor (TPA3255) is being sold for only a little more than a dedicated ultra low power Celeron JXXX unit (MA12070) it makes sense to be able to use it in more than a single, niche application.
This is one area where I, as an engineer by training (mechanical) (a.k.a taught about cost reduction in school), would look at (for example) the flying capacitors and say "The pins are only broken out of the package so you can put a tiny circuit with a capacitor next to the chip. There is no need nor desire to use large parts and extend the traces. With base designs using technically inferior parts just skirting the thresholds of audibility, and only bothering very few people, this is wasted capital which cannot be recovered in design time, consumer price, nor in production cost."
As a passion project I think you can do whatever makes you happy. But if you plan to sell the product to the community then everything is different.
I would advise this: If you want to make a truly special MA12070 board then internalize the 5V rail (can use the SilentSwitcher above, or even a less expensive part with a lower current rating) and just make the common components have multiple connection options via nested pads.
- Put in traces for ferrite beads, for air coils, and for the big shielded inductors. Slap both electrolytic SMD capacitor pads and through hole pads of different spacing next to teach other on the input paths. Lay it out for the largest components, then nest-in pads for optional smaller components. Have everything ship by default with lower cost parts installed as per the MA12070 datasheet spec. This will reduce your per-board cost to manufacture while providing a modular PCB for DIY customers to adjust as they wish.
- Anybody who wishes to mod the board with pricey capacitors, inductors, etc., etc. would just need to remove a couple SMD components (easy to do accidentally) and swap in their own chosen components. If people want to take detailed measurements across specific parts of the circuit, they are given exposed pads to do so.
- Make the board a dual MA12070 (a whole chip is $5.25 at Mouser) and allow people to have either a dual PBTL or quad BTL board with their own chosen components or just the datasheet specs. This will expand the potential interest because people can at least get a clean 25-ish Watts into 8-Ohms per channel as opposed to 10-15W. Since two inputs can be ignored / left floating in PBTL mode it could be as simple as wicking a solder bridge between two pads on the PCB to disable the extra inputs. A single channel balanced input is 3 pins. A stereo unbalanced input is 3 pins. Relay to the opamps like Topping, have a couple solder bridges, or use jumpers. Designer's choice. Potentiometer? Put in a header. Those who want one can install it. Those who do not can just bridge the input and output.
- Put in a few 3-pin headers to tie the board always on /unmuted or alternately route the control and I2C pins to a DIP8 or DIP16 socket. People can then either have it just be a dumb poweramp, wire out the pins to custom MCUs, or use a PIC like the Attiny series. They can even slap a DIP to SMD adapter in the socket and go wild. There would also need to be jumpers for internal/external clock but that's sunk cost of two jumper positions instead of forcing people to use the mode you laid out.
I am not saying your design is poor. I think you have done a better job than I ever could. But there are already several options for chipamps ranging from expensive and boutique (XKR) to cheap and preassembled (Aiyima, SMSL). What does not exist is easily-modified boards with multiple component pads that could satisfy a whole range of customers ranging from measurement and numbers-driven (me) to those who want to customize every step of their signal path (jean-paul).
I hope this helps you think of ways to make your board appealing to more people. I think that is what you asked for.
Just when I thought my posts are too long.
Agree on many points but the chip is so good that I would take the challenge even it would cost too much. I just counted what I spent already on the devices I bought. Bummer.
However, none of the existing devices have gotten the maximum out of the IC IMHO. There is always something missing which is understandable in this price range.
High risk of speaking of a design publicly is everyone wanting something else. As one JPS to another JPS I suggest to design under the radar and make and test a device that will be able to shine. A test run of 5 pieces and then present the newborn PCB with chip already soldered or the like and maybe the custom toroid. As you can guess I am in. Risk is good.
Agree on many points but the chip is so good that I would take the challenge even it would cost too much. I just counted what I spent already on the devices I bought. Bummer.
However, none of the existing devices have gotten the maximum out of the IC IMHO. There is always something missing which is understandable in this price range.
High risk of speaking of a design publicly is everyone wanting something else. As one JPS to another JPS I suggest to design under the radar and make and test a device that will be able to shine. A test run of 5 pieces and then present the newborn PCB with chip already soldered or the like and maybe the custom toroid. As you can guess I am in. Risk is good.
Last edited:
Well, I was spoiled in the past by something like requirement specifications.
Only Jean-Paul and you are making remarks, so you´re right, no interest here from the community.
I don´t know the market with all this Class D amplifiers. Nice found with this TPA3255.
BTW, XRK TPA3255 is my layout as you know.
I´m very busy next days but I´ll think about all the remarks and make a last proposal.
JP
Only Jean-Paul and you are making remarks, so you´re right, no interest here from the community.
I don´t know the market with all this Class D amplifiers. Nice found with this TPA3255.
BTW, XRK TPA3255 is my layout as you know.
I´m very busy next days but I´ll think about all the remarks and make a last proposal.
JP
Don’t underestimate practical DIYers that step in when a tested PCB is presented.
Some need to stick out their necks developing and some stick out their necks by building. There is a kind of skewed balance in that 😉
Even it would fail with regards to demand those prototypes can be sold at cost.
Some need to stick out their necks developing and some stick out their necks by building. There is a kind of skewed balance in that 😉
Even it would fail with regards to demand those prototypes can be sold at cost.
Last edited:
Hahahahahahahahaha I didn't even realize that was you! Before I found the PA3s if the dual TPA boards were in stock I probably would have those since they garnered very high praise.
Things really do go smoothly with design constraints since it limits the scope of what is feasible.
I figured if anybody would listen then I would put out my fantasy chipamp board idea with lots of footprint options for commonly swapped components which would allow each member of the community to build their own 'ideal' component stack. Then perhaps there would be common 'recipes' that pop up and folks could enterprise the whole ordeal by selling component kits based on the popular recipes so novices could get some soldering in and feel accomplished while also having a really nice product.
And yeah, I can be super long winded when I try and be helpful in feedback. Apologies. I just hope it is actually helpful.
ETA: jean-paul's idea of covert design is also valid. I think I was partly inspired by seeing how frequently DIYINHK's semi-modular designs are mentioned across multiple forums such as Reddit, ASR, here, AVS, Head-Fi, etc. Kind of a thing that has the potential to please everybody except the ultra hardcore 1 micro-ohm trace impedance matching types.
Last edited:
I'm not commenting because I don't have the knowledge 🙂
MKHunt's thoughts above around configuration and expansion options make a lot of sense to me. I want a board for a battery powered system (MA12070, or MA2304DNS when it's released). The idea of soldering something this small myself as a DIYer or designing PCBs is daunting, so a well thought out board from someone else would suit my first choice. It's fairly niche as you suggest, but I'd expect to have to pay a bit extra because of that.
I'm in the interested but not enough knowledge to comment camp. I've got a 4 way active setup and looking for something lower noise than my 4ch TPA3251 boards for the compression drivers. Ease of implementation is important for me, therefore I'd be wanting:
- Good on/off pop suppression
- Single DC power supply eg meanwell SMPS providing 24v
- Balanced input
I think the interest is there, some of us are just trying to avoid "thread pollution".
The points raised by @MKHunt regarding costs are very relevant. "Bang per buck" should be one of the most important criteria IMO, and personally I would warmly welcome a "balanced input only" design.
I also looked all kinds of options for active multiway speaker quite recently. MA12070 was strong contestant but lack of output power was a problem, there is enough with PBTL but it means multiple pcb:s, multiple psu:s and all that hustle. Could not buy the chinese ready made amps either as they have either power bricks or all kinds of extra features that really aren't too handy with active speaker. Eventually had to bite it and buy some ICEpower modules, ready to roll.
For active multiway speaker: chip per channel for power, 3-4 channels on single psu if at all possible, balanced inputs. At least this is something that is not readily available, suspect there is not too big of a market on this.
For active multiway speaker: chip per channel for power, 3-4 channels on single psu if at all possible, balanced inputs. At least this is something that is not readily available, suspect there is not too big of a market on this.
As DIY i second a modulair system. Here there a so many preferrences. Balanced or unbalanced, linear or smps, output filter setups, PBTL or BTL. I started it here but abandon it because no MS12070 are avaiable at the PCB producer.
https://www.diyaudio.com/community/threads/ma12070-adapter-board.389911/#post-7116496
https://www.diyaudio.com/community/threads/ma12070-adapter-board.389911/#post-7116496
It seems joint effort and testing and then offering a PCB design with chip already soldered is viable. As known I strongly detest modular approach certainly with HF/RF generating designs and any extra joint/contact/wire/PCB track is one too many. In that aspect I would adhere to RF design rules as much as possible and keep stuff compact and all on 1 board except for the main PSU. For the output filter and other items PCB pads can be added for a few options without penalty as already illustrated by JPS64.
WHEN a design is done for the community/afficionados it should surpass ready made devices big time and be more or less easy to build with as least wiring as possible for a tidy build. The main issue is simply how much it may cost. We are exploiting a low cost chip in expensive waters. Therefor a 2 chip design will be a bridge too far I guess. The JPS64 solution with daisy chaining a few boards is a good solution.
The question is if a full fledged design can compete with the higher classed class D devices out there. My personal thoughts (assumption alarm!) are that it will be close or even better with regards to sound quality when looking at the subjective performance of cheap Chinese carton amplifiers. If the latter then it was worth the energy/time/money. Once a member and I did a design and I think we sold more PCBs after the news got out that it was good then was expected when looking at the number of people that were interested at the time of development.
WHEN a design is done for the community/afficionados it should surpass ready made devices big time and be more or less easy to build with as least wiring as possible for a tidy build. The main issue is simply how much it may cost. We are exploiting a low cost chip in expensive waters. Therefor a 2 chip design will be a bridge too far I guess. The JPS64 solution with daisy chaining a few boards is a good solution.
The question is if a full fledged design can compete with the higher classed class D devices out there. My personal thoughts (assumption alarm!) are that it will be close or even better with regards to sound quality when looking at the subjective performance of cheap Chinese carton amplifiers. If the latter then it was worth the energy/time/money. Once a member and I did a design and I think we sold more PCBs after the news got out that it was good then was expected when looking at the number of people that were interested at the time of development.
Last edited:
I am not opposed to a double chip design as a significant portion of the projects are likely to be centered on active filters using DSP, this is not a hugely expensive chip and the cost per channel might be reduced if you have a multi-chip design.
Though the current suggestion from @JPS64 would perhaps be more flexible.
You forget the double amount of not so cheap film caps.
I wonder about the significant DSP crowd.....
I wonder about the significant DSP crowd.....
Probably somewhere between 1/3rd to 2/3rds of boards sold, hard to say. Those that like to tinker and build stuff are likely to accept DSP in various forms as the powerful tool it can be.
DSP, as all other tools, can be used in several ways, the most common being "wrong" and "sub-optimal" but are still likely to do more good than harm.
I bet you would be just as critical if all you knew was 1st order filters, and suddenly someone came along with a second order filter. The right tool for the right job is half the work.
Edit:
At any rate, I will not digress further on the point of DSP.
The design discussion and resulting suggestions seem very relevant to what people want, just wanted to give my vote towards seeing a decent quality per buck design.
A future GB with components and SMD pre-equipped boards will reduce drastically the costs.
They´re a lot of PCB manufacturer who could make this SMD boards, only MA12070 has to be available.
Using Ciiva:
I´ll not organize this GB for some reasons mentionned here in the past.
Passthrough PSU for dual boards version in PBTL is also doubling the power of the power supply!
Very busy this week, I´ll work on the PSU this weekend.
JP
They´re a lot of PCB manufacturer who could make this SMD boards, only MA12070 has to be available.
Using Ciiva:
I´ll not organize this GB for some reasons mentionned here in the past.
Passthrough PSU for dual boards version in PBTL is also doubling the power of the power supply!
Very busy this week, I´ll work on the PSU this weekend.
JP