That is a pretty neat job so far, I'm afraid I probably won't stand a chance against those ... How am I ever going to make it?
I'll only sell populated/tested PCBs, there's way too much fine pitch SMT on this board to offer it as a bare PCB or as a kit. Soldering the LFCSPs require either a reflow oven or hot air and a bottom heater, and a fair bit of experience.
Here's my options for how to get these built for others:
- Use an outfit like Seeed Studio to build the boards and facilitate orders, collect a royalty. Haven't really dug into how to do this.
- Set up a Kickstarter account and have buyers fund a build. Use the money to buy all the parts and get the SMT work done. I'd probably use a local assembly shop if I go this route.
I still have to get an idea of how many people want to buy a board like this, what the cost is going to be, etc... right now I'm more concerned with getting the prototype built and working.
Here's my options for how to get these built for others:
- Use an outfit like Seeed Studio to build the boards and facilitate orders, collect a royalty. Haven't really dug into how to do this.
- Set up a Kickstarter account and have buyers fund a build. Use the money to buy all the parts and get the SMT work done. I'd probably use a local assembly shop if I go this route.
I still have to get an idea of how many people want to buy a board like this, what the cost is going to be, etc... right now I'm more concerned with getting the prototype built and working.
Done.
Powered it up on a bench supply, so far all the power supplies are working, the SigmaDSP's 1.8V regulator works, and ISP is working on the micro.
Building the full test setup now with a fused battery, resistor loads, LEDs, pots, switches, etc. Fingers crossed that I get audio out of it over the weekend.
Powered it up on a bench supply, so far all the power supplies are working, the SigmaDSP's 1.8V regulator works, and ISP is working on the micro.
Building the full test setup now with a fused battery, resistor loads, LEDs, pots, switches, etc. Fingers crossed that I get audio out of it over the weekend.
Attachments
Fairly boring. No parts or sikscreen, and a relatively unbroken ground plane.
No updates... St. Patricks Day on Saturday last weekend ate into the verification, and operating a router table hungover the following day and losing half a fingertip took care of the rest of it.
I'll spend a few hours of continuous work at it tomorrow afternoon and see where I end up.
No updates... St. Patricks Day on Saturday last weekend ate into the verification, and operating a router table hungover the following day and losing half a fingertip took care of the rest of it.
I'll spend a few hours of continuous work at it tomorrow afternoon and see where I end up.
Update: WE HAVE AUDIO! SigmaDSP is running, class D amplifiers are running. Every part of the card has now been tested with no issues found.
Just have to finalize the SigmaDSP and AVR code as well as decide some functional things (what all the switches do, timing of things like automatic shutoff, etc..) and I'll be done.
Just have to finalize the SigmaDSP and AVR code as well as decide some functional things (what all the switches do, timing of things like automatic shutoff, etc..) and I'll be done.
They'll be sold one way or another.
If only a few people want them, I could assemble the rest of the prototype PCBs - I have 9 more PCBs left, I'll probably sell 8. Only thing is I have to order parts in lower quantities and build/test these by hand (there's a lot of parts on this card!) so I'll probably charge $175-$200 for a card.
If there's enough interest (I'd say 25+ cards) I could redesign the board to make it production ready, build a proper production test setup, and get a local SMT shop to build them. Might be able to touch $100-125 per card if I go this route, depending on quantity.
I'm also releasing the design files under CC BY-NC-SA, so anyone else can build the board if they want. I've been saying I'm going to do this for a while, but the Eagle files need to be cleaned up and I haven't got around to it. It'll get done.
If only a few people want them, I could assemble the rest of the prototype PCBs - I have 9 more PCBs left, I'll probably sell 8. Only thing is I have to order parts in lower quantities and build/test these by hand (there's a lot of parts on this card!) so I'll probably charge $175-$200 for a card.
If there's enough interest (I'd say 25+ cards) I could redesign the board to make it production ready, build a proper production test setup, and get a local SMT shop to build them. Might be able to touch $100-125 per card if I go this route, depending on quantity.
I'm also releasing the design files under CC BY-NC-SA, so anyone else can build the board if they want. I've been saying I'm going to do this for a while, but the Eagle files need to be cleaned up and I haven't got around to it. It'll get done.
Highly unlikely that someone without expert soldering skills can make it work though.
Wouldn't hurt to have the option.
Would rather solder it myself (mostly because I have the proper equipment available and like the DIY aspect of it).
I'd much rather put tested boards in the hands of people than a kit of parts.
Soldering the LFCSPs on this card is difficult. I used a BGA rework station for them, and when one had a solder bridge underneath, I remounted it using a bottom heater and a hot air pencil. Even with the proper equipment it's an exercise in frustration, and I don't want to put anyone who's not familiar with the process through the task.
Also, I'm pretty sure that counting/bagging parts for a kit would cost more than running the board through an assembly house. And with everything made with 0603 parts that are easy to drop, lose, get mixed up with each other, etc.. I wouldn't even recommend it. I got a few parts mixed up when building this card and had to reach for a LCR meter on more than one occasion.
But if anyone really wants to try, I'll mail out a blank prototype PCB. You'll need to buy all the parts - cost is about $100 CAD per card on Digikey, and you'll need an Atmel AVR programmer of some sort to program the microcontroller.
Soldering the LFCSPs on this card is difficult. I used a BGA rework station for them, and when one had a solder bridge underneath, I remounted it using a bottom heater and a hot air pencil. Even with the proper equipment it's an exercise in frustration, and I don't want to put anyone who's not familiar with the process through the task.
Also, I'm pretty sure that counting/bagging parts for a kit would cost more than running the board through an assembly house. And with everything made with 0603 parts that are easy to drop, lose, get mixed up with each other, etc.. I wouldn't even recommend it. I got a few parts mixed up when building this card and had to reach for a LCR meter on more than one occasion.
But if anyone really wants to try, I'll mail out a blank prototype PCB. You'll need to buy all the parts - cost is about $100 CAD per card on Digikey, and you'll need an Atmel AVR programmer of some sort to program the microcontroller.
U should put up some sort of list where people that are interested can sign up. That way you'll have an idea of how to go about the selling and production.
^- I'll get to that.
Haven't had much time to spend on this project, but I'm getting pretty close to done. Don't have the Atmel booting/controlling the DSP yet but the rest of the Atmel code is done. And I'm close to finalizing the DSP code.
I'm dedicating 3 DIP switches to selecting one of 8 speaker configurations. Here's my list (which isn't final by any means)
- 4ch biamped signature Boominator (2ch driving woofers, 2ch driving tweeters)
- 4ch biamped classic Boominator (2ch driving HP-10Ws, 2ch driving KSN-100x's)
- 4ch biamped NA Boominator (2ch driving GW-1058s, 2ch driving GT-100x's)
- 2ch signature Boominator (HP-10Ws and KSN-100x's, using resistors on piezos)
- 2ch NA Boominator (GW-1058s and GT-100x's, using resistors on piezos)
- 4ch 4-ohm Boominator (4-ohm HP-10Ws and KSN-100x's, front/back driven separately)
- 2ch biamped, classic Half-boominator (1ch driving HP-10Ws, 1ch driving KSN-100x's)
- 2ch biamped, NA Half-boominator (1ch driving GW-1058s, 1ch driving KSN-100x's)
Now, I have absolutely no idea where to set up any of these operating points - relative woofer/tweeter gain, crossover frequencies, etc. Anyone any good at speaker modelling?
Haven't had much time to spend on this project, but I'm getting pretty close to done. Don't have the Atmel booting/controlling the DSP yet but the rest of the Atmel code is done. And I'm close to finalizing the DSP code.
I'm dedicating 3 DIP switches to selecting one of 8 speaker configurations. Here's my list (which isn't final by any means)
- 4ch biamped signature Boominator (2ch driving woofers, 2ch driving tweeters)
- 4ch biamped classic Boominator (2ch driving HP-10Ws, 2ch driving KSN-100x's)
- 4ch biamped NA Boominator (2ch driving GW-1058s, 2ch driving GT-100x's)
- 2ch signature Boominator (HP-10Ws and KSN-100x's, using resistors on piezos)
- 2ch NA Boominator (GW-1058s and GT-100x's, using resistors on piezos)
- 4ch 4-ohm Boominator (4-ohm HP-10Ws and KSN-100x's, front/back driven separately)
- 2ch biamped, classic Half-boominator (1ch driving HP-10Ws, 1ch driving KSN-100x's)
- 2ch biamped, NA Half-boominator (1ch driving GW-1058s, 1ch driving KSN-100x's)
Now, I have absolutely no idea where to set up any of these operating points - relative woofer/tweeter gain, crossover frequencies, etc. Anyone any good at speaker modelling?
Update:
Had a bunch of other stuff going on around the house, finally got back to working on the card over the weekend.
The card now works 100% standalone - I'm successfully booting and controlling the SigmaDSP from the AVR, volume control pot etc. are all working. Total # of bodge wires, cut traces, etc. added to the card is zero! Gotta say I love it when something works the first time around.
I've got a couple things remaining to finish up - finish the DSP code with all the features I want added to it, tidy up the AVR code a bit, etc.
Right now the DSP flow is as follows:
Input -> DC block -> highpass (defeatable) -> 9 band parametric EQ (defeatable) -> volume control -> biamp crossover (defeatable) -> output
If you want to write code for the SigmaDSP to add more features than this, you'll have to get your hands on SigmaStudio. Unfortunately, Analog Devices' software EULA prohibits anyone from using the program unless they're working on an active design as a student or as a commercial product designer - I'm hoping that they change their minds on this one and allow unsupported use by electronics hobbyists.
Fortunately, most of the parameters for the default program will be adjustable without SigmaStudio, by altering the AVR code to change the parameters that get sent to the DSP on bootup. Things like moving equalizer bands up and down, adjusting the treble/bass scale factors for biamping, etc. is pretty easy but changing filter slopes, frequencies, Q, etc... involves calculating coefficients for biquads.
End result is that the card, at least in its current form, is a bit of a hacker's delight - it's not exactly plug and play. I'll do my best to get a couple of default modes/presets working. But if you want to do proper equalization of a boombox design, make the best use of a biamped arrangement, etc... at the very least you'll be buying an AVR programmer and tweaking the code.
Had a bunch of other stuff going on around the house, finally got back to working on the card over the weekend.
The card now works 100% standalone - I'm successfully booting and controlling the SigmaDSP from the AVR, volume control pot etc. are all working. Total # of bodge wires, cut traces, etc. added to the card is zero! Gotta say I love it when something works the first time around.
I've got a couple things remaining to finish up - finish the DSP code with all the features I want added to it, tidy up the AVR code a bit, etc.
Right now the DSP flow is as follows:
Input -> DC block -> highpass (defeatable) -> 9 band parametric EQ (defeatable) -> volume control -> biamp crossover (defeatable) -> output
If you want to write code for the SigmaDSP to add more features than this, you'll have to get your hands on SigmaStudio. Unfortunately, Analog Devices' software EULA prohibits anyone from using the program unless they're working on an active design as a student or as a commercial product designer - I'm hoping that they change their minds on this one and allow unsupported use by electronics hobbyists.
Fortunately, most of the parameters for the default program will be adjustable without SigmaStudio, by altering the AVR code to change the parameters that get sent to the DSP on bootup. Things like moving equalizer bands up and down, adjusting the treble/bass scale factors for biamping, etc. is pretty easy but changing filter slopes, frequencies, Q, etc... involves calculating coefficients for biquads.
End result is that the card, at least in its current form, is a bit of a hacker's delight - it's not exactly plug and play. I'll do my best to get a couple of default modes/presets working. But if you want to do proper equalization of a boombox design, make the best use of a biamped arrangement, etc... at the very least you'll be buying an AVR programmer and tweaking the code.
Your project has scopes of a very wide application , the audio input part could be placed in another board and used over a wide range of applications as pre-amps .
would love to see a design with some higher power class d amps - (at a later stage)
what i am really waiting for is a battery drain time test with all of it running
a) with solar charging
b) without solar charging
where battery consumption is an issue please aim to use high efficiency drivers , p-audio is not the best performer in this category , there are some beyma 8 inches with al coils in the usspeaker site you could consider , plus the 5 to 6 in co-ax from beyma , bms , for bass you can really boost efficiency with a compact tapped horn config , one should not generally expect deep bass from this sort of a devise .
Congratulations on the great work so far .
suranjan
would love to see a design with some higher power class d amps - (at a later stage)
what i am really waiting for is a battery drain time test with all of it running
a) with solar charging
b) without solar charging
where battery consumption is an issue please aim to use high efficiency drivers , p-audio is not the best performer in this category , there are some beyma 8 inches with al coils in the usspeaker site you could consider , plus the 5 to 6 in co-ax from beyma , bms , for bass you can really boost efficiency with a compact tapped horn config , one should not generally expect deep bass from this sort of a devise .
Congratulations on the great work so far .
suranjan
I have been planning same kind of an amplifier but with an FPGA as the filtering element. Those circuits that you have chosen seem to be good candidates for this however. At least the implementation should be more straightforward. Really nice job!
The only thing that bothers me is that the filter gives analog output. TI has chips that take in I2S data but I don't know if that matters. They make the DA conversion just inside the side at some point (PWM DAC) I think. So is there more AD/DA conversions in this implementation or not? And does it matter after all? 🙂
Btw. are there any chips that would act as bluetooth audio receivers with no programming efforts? I have now in my "boominator" (not the real one) a Belkin bluetooth receiver but it would be nice to have everything on one PCB.
The only thing that bothers me is that the filter gives analog output. TI has chips that take in I2S data but I don't know if that matters. They make the DA conversion just inside the side at some point (PWM DAC) I think. So is there more AD/DA conversions in this implementation or not? And does it matter after all? 🙂
Btw. are there any chips that would act as bluetooth audio receivers with no programming efforts? I have now in my "boominator" (not the real one) a Belkin bluetooth receiver but it would be nice to have everything on one PCB.
I had a few different ideas before I settled on this. I was considering a PIC32, dsPIC, AVR32 or ARM acting as a combined board controller/DSP, and a separate audio codec. The AVR+SigmaDSP architecture ended up being just plain simple/elegant, great power consumption numbers, very little board space, etc.
I looked into Bluetooth audio, decided it wasn't worth it. There's lots of chipsets out there capable of "bluetooth thing hanging on your ear" audio transmission, but ones capable of A2DP are harder to come by. Even if I could get my hands on one, I'd still have to convert PCM coming out of it to a suitable format/sample rate to feed the SigmaDSP chip. And thirdly, if you mount the "single PCB" in a metal box then bluetooth stops working, unless you run an external antenna out of the box - at that point there's no advantage to having it on the PCB itself.
It's a lot easier to spend 10 bucks and get an A2DP adapter off eBay. There's one on there sold by various companies that looks like a USB thumbdrive, it plugs into a USB port for power and has an analog audio out on the back - I've got one coming in the mail, we'll see how it works out.
As of now, the card is "done enough to use" - the next step is to get the design files cleaned up and posted (AVR/SigmaDSP code won't be finalized, I'll warn), and start working on my boominator cabinet.
I looked into Bluetooth audio, decided it wasn't worth it. There's lots of chipsets out there capable of "bluetooth thing hanging on your ear" audio transmission, but ones capable of A2DP are harder to come by. Even if I could get my hands on one, I'd still have to convert PCM coming out of it to a suitable format/sample rate to feed the SigmaDSP chip. And thirdly, if you mount the "single PCB" in a metal box then bluetooth stops working, unless you run an external antenna out of the box - at that point there's no advantage to having it on the PCB itself.
It's a lot easier to spend 10 bucks and get an A2DP adapter off eBay. There's one on there sold by various companies that looks like a USB thumbdrive, it plugs into a USB port for power and has an analog audio out on the back - I've got one coming in the mail, we'll see how it works out.
As of now, the card is "done enough to use" - the next step is to get the design files cleaned up and posted (AVR/SigmaDSP code won't be finalized, I'll warn), and start working on my boominator cabinet.