Well, let's take a look at the main needs?
First, power. The #1 thing is to roll off the load for pitches the woofer won't output. That function can be done with a capacitor series to the woofer. And sure, we could also right-size the amplifier input cap.
EQ stuff. Well, what can RockBox not do?
RockBox, widely available on multiple platforms for portable music, also has Compressor, Replaygain, and 10 band EQ.
The remaining need is for whatever output caps and rockbox don't get done.
When I try to figure out what the remaining needs are, I come up with. . . a basic preamplifier. In fact a preamp with a buffered input (for rockin bass, of course). We could also fine-tune the bandwidth of the preamp to avoid outputting subharmonic below pitches that could be passed by the speaker.
There's the problem. This system needs to 1) roll off extraneous fundamental (power waste beyond speaker capacity), and then 2) use psychoacoustic bass enhancer for bass extension.
One could also incorporate a power amp; however, Sure has some power amplifiers capable of wide voltage range needed for safe use of solar power (8~24vdc).
SO, the needed circuit is actually buffer > preamp > psychoacoustic bass enhancer > roll off the fundamental. (the circuit can also be used to get big bass into small rooms, big clear deep bass out of little speakers, etc. . .)
So, howabout a simpler board with preamp+psycobass?
First, power. The #1 thing is to roll off the load for pitches the woofer won't output. That function can be done with a capacitor series to the woofer. And sure, we could also right-size the amplifier input cap.
EQ stuff. Well, what can RockBox not do?
RockBox, widely available on multiple platforms for portable music, also has Compressor, Replaygain, and 10 band EQ.
The remaining need is for whatever output caps and rockbox don't get done.
When I try to figure out what the remaining needs are, I come up with. . . a basic preamplifier. In fact a preamp with a buffered input (for rockin bass, of course). We could also fine-tune the bandwidth of the preamp to avoid outputting subharmonic below pitches that could be passed by the speaker.
There's the problem. This system needs to 1) roll off extraneous fundamental (power waste beyond speaker capacity), and then 2) use psychoacoustic bass enhancer for bass extension.
One could also incorporate a power amp; however, Sure has some power amplifiers capable of wide voltage range needed for safe use of solar power (8~24vdc).
SO, the needed circuit is actually buffer > preamp > psychoacoustic bass enhancer > roll off the fundamental. (the circuit can also be used to get big bass into small rooms, big clear deep bass out of little speakers, etc. . .)
So, howabout a simpler board with preamp+psycobass?
All true, all true. Equalization, compression, fake bass and whatnot can be done by preprocessing the MP3s, or perhaps with MP3 player plugins.
Main problem is still domestic manufacturing cost. I'd have to make something kittable like the AMP6 - make everything through-hole and buildable by the end user, which gets rid of manufacturing costs altogether. Thinking about it more, I might make a stab at a through-hole design that tries to be the "single board you need" for a Boominator. Maybe something like:
- TA2020 amplifier IC, or some other through-hole class D amp IC. I'll make a proper effort into finding a reliable Tripath source if I have to.
- Decent front-end into the TA2020, with a differential receiver and highpass.
- Microcontroller based MPPT solar charger, also doing low battery protection. I'm fairly sure I can make an 8-pin ATTiny85 and a couple power management bits do the job without taking up too much board space.
- Probably leave the 5V DC/DC optional. Mounting a 12V cigarette lighter jack to a sheet of plywood and plugging a cheap USB adapter into it is a lot easier/cheaper than trying to wood mount a USB jack anyway.
Clipnipper might take up too much board space, but maybe it can be made into an add-on board you stack on top of this one. A clipping 'idiot light' might be easier to implement. Have to ponder this some more.
Main problem is still domestic manufacturing cost. I'd have to make something kittable like the AMP6 - make everything through-hole and buildable by the end user, which gets rid of manufacturing costs altogether. Thinking about it more, I might make a stab at a through-hole design that tries to be the "single board you need" for a Boominator. Maybe something like:
- TA2020 amplifier IC, or some other through-hole class D amp IC. I'll make a proper effort into finding a reliable Tripath source if I have to.
- Decent front-end into the TA2020, with a differential receiver and highpass.
- Microcontroller based MPPT solar charger, also doing low battery protection. I'm fairly sure I can make an 8-pin ATTiny85 and a couple power management bits do the job without taking up too much board space.
- Probably leave the 5V DC/DC optional. Mounting a 12V cigarette lighter jack to a sheet of plywood and plugging a cheap USB adapter into it is a lot easier/cheaper than trying to wood mount a USB jack anyway.
Clipnipper might take up too much board space, but maybe it can be made into an add-on board you stack on top of this one. A clipping 'idiot light' might be easier to implement. Have to ponder this some more.
Are you guys really stuck with ta2020 chip? There are some really nice offerings from TI that are superior spec wise like TPA31xxD2 series, yes smd but boards can be made tiny and BOM is less in the end. The mental hurdle of going beyond thru hole parts has to be embraced. People will adapt and get magnifying goggles and tiny iron tips. I jumped in on my first smd amp build and although it is a pain to work with small parts, I am still having fun and getting better. What I like is how small it keeps everything and how it opens up all the newer chips that are surface mount up as diy choices. The psychoacoustic processor and equalisation is probably a smd chip. It's all becoming smd as default - similar to how radios used to be tubes and dead bug free-space wiring transitioned to transistors and pcb's and then to IC's.
We have looked at both thru-hole and SMD components. Primarily the latter.
If you read this thread from the beginning, or just the last page, you would know that gmarsh design here is done with SMD output chips that measure 7x7mm and has 48 pins.
There are no chips at all by any producer that fulfilled all our demands but there are 2 which came close. Given that one of these isn't in production, the choice was simple.
Last edited:
Making a SMT kit would be a disaster, for many reasons...
Soldering fine pitch SMT devices with thermal pads can't be done with any degree of confidence unless you've got experience and good gear, which the great majority of people don't have. SMT components are easily dropped/lost, or even destroyed during assembly. Kitting parts on my end (cutting/labelling taped parts) would be time consuming, frustrating and error prone. The odds of the customer ending up with a functioning board are way too low, and I won't be able to make any guarantees that a kit would function.
You might have no trouble putting together such a kit and you might disagree, but you're not most people. The only way to make a SMT board reliable would be to get it built on a part shooter, but again, low volume SMT assembly costs a lot of money.
Sadly, through hole is really the only way to go for such a project.
Just a thought, but I think the primary concern is wide voltage range. The batteries "bump" down to 8v on bass beat, but the sturdy solar panels in brightest sun go up to 24v at most. That's a really wide voltage range, and the source of most of the breakage.
RockBox has compressor (with a volumax setting "make up gain"), 10 band EQ, and Replaygain so we've got a lot of functions covered. But talented EQ still doesn't substitute for a buffer on the amplifier input.
As far as playing loud, and despite a wide voltage range, I favor Bi-Amp as the solution because then the tweeter amp does not clip. That might not be the very smartest way possible, but in this case, I like the features of effective and doable. It is also, easily, higher quality audio.
Kits? Not so necessary. Looks like the majority of boominator constructors have trouble soldering, so we probably need Boards that don't require a lot of soldering.
The audio quality of existential boards is dreadful--a sort of loudly blaring silence that is somewhat embarrassing. But there's a fix for that. No need to burden designers with production labor.
For fabbing complex boards, we could ask Sure and a few others if they're willing to do it. After all, who better than production specialists to produce boards? In trade for a momentary hijink of working out those details, then actual production of the boards would go smoothly. And the best feature of that option is boards that definitely exist.
Back to the hard part: I hope for something much more suitable and more powerful or at least more dynamic than the lackluster TA2020. My experiences with that chip have almost been a waste of time because it is just not as lively as music and because it falters, which is disappointing, and music shouldn't ever do that. Supposedly more efficient than a minimal bias aB but with TA2020 doing so much less work, it seems ironically inefficient at loud replay. Double-Yuck! Do we have some more significant amplifier options to use instead?
RockBox has compressor (with a volumax setting "make up gain"), 10 band EQ, and Replaygain so we've got a lot of functions covered. But talented EQ still doesn't substitute for a buffer on the amplifier input.
As far as playing loud, and despite a wide voltage range, I favor Bi-Amp as the solution because then the tweeter amp does not clip. That might not be the very smartest way possible, but in this case, I like the features of effective and doable. It is also, easily, higher quality audio.
Kits? Not so necessary. Looks like the majority of boominator constructors have trouble soldering, so we probably need Boards that don't require a lot of soldering.
The audio quality of existential boards is dreadful--a sort of loudly blaring silence that is somewhat embarrassing. But there's a fix for that. No need to burden designers with production labor.
For fabbing complex boards, we could ask Sure and a few others if they're willing to do it. After all, who better than production specialists to produce boards? In trade for a momentary hijink of working out those details, then actual production of the boards would go smoothly. And the best feature of that option is boards that definitely exist.
Back to the hard part: I hope for something much more suitable and more powerful or at least more dynamic than the lackluster TA2020. My experiences with that chip have almost been a waste of time because it is just not as lively as music and because it falters, which is disappointing, and music shouldn't ever do that. Supposedly more efficient than a minimal bias aB but with TA2020 doing so much less work, it seems ironically inefficient at loud replay. Double-Yuck! Do we have some more significant amplifier options to use instead?
Last edited:
Just ordered bare PCBs for the shiznit V2 prototype. Trying a couple of new things with the design, and crossing my fingers that they'll work out. If everything measures/tests OK, I'll start work on the production version.
I gave up trying to make a through-hole card, and decided to go for a SMT board instead, sold as a 100% complete and tested unit. It'll be made here in Canada, and it'll probably cost a few times more than a throwaway chinese amp off ebay, but it'll be a damn good product that I'll 100% stand behind.
I gave up trying to make a through-hole card, and decided to go for a SMT board instead, sold as a 100% complete and tested unit. It'll be made here in Canada, and it'll probably cost a few times more than a throwaway chinese amp off ebay, but it'll be a damn good product that I'll 100% stand behind.
Even then I reckon it's actually not much more than cheapo chinese amps when you consider that it replaces a TA2020 amp, a TK2050amp , a dc-dc converter, a battery indicator, and saves the trouble of having to switch between amps to get the best efficiency. And all the thing that can go wrong in that process. Even saves the trouble of having to turn it on and off.
Prototype PCBs are on the way from Seeed, and I'm keying in the last of the digikey order right now for the rest of the parts. Also trying to track down a couple of cheap 4 ohm loads capable of 100W.
I'm trying something new on the design - a DC/DC converter that'll convert 12V battery input to anywhere from 12V to 26V depending on the audio listening level. So it'll be power efficient at low listening levels, but capable of going damn loud if you want. Everything seems to work in simulation, but I'll wait for real hardware to show up to verify that. Fingers crossed..
I'm trying something new on the design - a DC/DC converter that'll convert 12V battery input to anywhere from 12V to 26V depending on the audio listening level. So it'll be power efficient at low listening levels, but capable of going damn loud if you want. Everything seems to work in simulation, but I'll wait for real hardware to show up to verify that. Fingers crossed..
^- this. They're the main chips.
TPA3118D2 has the same die as the TPA3116D2, but heatsinks to the PCB instead of to a top-mounted heatsink. With proper cooling it's capable of the same power as the 3116.
On my board, I've got a ridiculous # of thermal vias under the 3118 to heatsink the chip to the bottom side ground plane, and I'm sticking a big thermal pad to the bottom of the PCB to cool it to the chassis. Hoping that's sufficient for cooling.
The LTC3787 operates in burst mode, which should give very little power consumption, and interference between the 3118 and 3787 should hopefully be inaudible. If burst mode becomes audible on the class D output, I could put the 3787 into pulse skipping mode, perhaps even sync the 3787 and 3118 in that configuration - hopefully the power dissipation won't be too bad in that scenario. Guess we'll see what happens..
TPA3118D2 has the same die as the TPA3116D2, but heatsinks to the PCB instead of to a top-mounted heatsink. With proper cooling it's capable of the same power as the 3116.
On my board, I've got a ridiculous # of thermal vias under the 3118 to heatsink the chip to the bottom side ground plane, and I'm sticking a big thermal pad to the bottom of the PCB to cool it to the chassis. Hoping that's sufficient for cooling.
The LTC3787 operates in burst mode, which should give very little power consumption, and interference between the 3118 and 3787 should hopefully be inaudible. If burst mode becomes audible on the class D output, I could put the 3787 into pulse skipping mode, perhaps even sync the 3787 and 3118 in that configuration - hopefully the power dissipation won't be too bad in that scenario. Guess we'll see what happens..
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.