You'd think that there are a lot more modules that to the same thing for a lot less.... What does this module have that others don't?
Compared to what, an 8 dollar dealextreme module? The RN52 has:
- A known, respectable manufacturer that I can go to for support if I run into any problems
- Full production testing
- A guaranteed unique bluetooth MAC address
- A reliable supply chain through several electronics distributors
- A datasheet giving guaranteed minimum performance specs and a reflow soldering profile
- Packaging that a pick and place machine can pull modules from
- A decent user guide
- Upgradeable firmware, and an actual manufacturer's site I can download it from
- FCC/industry canada/CE certifications that I can look up and verify
- A great deal of compatibility testing performed with other bluetooth devices, so it'll just plain work when you need it to
I'm aiming for quality with this design, not cost, I want people who buy the cAMP and this bluetooth module to be happy with it - and I'll stand behind that.
The RN52 and KC Wirefree KC-6112 were really the only two modules that seemed to fit the bill, and the RN52 won out on price/ease of purchasing.
- A known, respectable manufacturer that I can go to for support if I run into any problems
- Full production testing
- A guaranteed unique bluetooth MAC address
- A reliable supply chain through several electronics distributors
- A datasheet giving guaranteed minimum performance specs and a reflow soldering profile
- Packaging that a pick and place machine can pull modules from
- A decent user guide
- Upgradeable firmware, and an actual manufacturer's site I can download it from
- FCC/industry canada/CE certifications that I can look up and verify
- A great deal of compatibility testing performed with other bluetooth devices, so it'll just plain work when you need it to
I'm aiming for quality with this design, not cost, I want people who buy the cAMP and this bluetooth module to be happy with it - and I'll stand behind that.
The RN52 and KC Wirefree KC-6112 were really the only two modules that seemed to fit the bill, and the RN52 won out on price/ease of purchasing.
Does the KC Wirefree come with licenses? That narrows the price gap a bit and the rx is something like 5 db more sensitive too.
KC Wirefree doesn't come with licenses either. The modules as shipped are capable of AAC which can be turned on with a command, however you have to buy the license separately to legally do that. aptX is available also, though it's custom firmware from KC to do that.
I'd love for someone to make a module that has the codecs enabled by default, but I suspect that the licenses are designed so that the company ultimately building the finished product is the one that has to pay the license.
Though SBC at maximum bitrate (372kbps) is actually pretty good - comparable to 192kbps AAC. And provided the host device is is capable of bluetooth 3.0 in order to achieve that, which means pretty much any modern android phone or iPhone, it'll use the high rate. The RN52 supports MP3 also, though I don't know at what bitrate. More reading to do..
I'd love for someone to make a module that has the codecs enabled by default, but I suspect that the licenses are designed so that the company ultimately building the finished product is the one that has to pay the license.
Though SBC at maximum bitrate (372kbps) is actually pretty good - comparable to 192kbps AAC. And provided the host device is is capable of bluetooth 3.0 in order to achieve that, which means pretty much any modern android phone or iPhone, it'll use the high rate. The RN52 supports MP3 also, though I don't know at what bitrate. More reading to do..
Prototype PCB is in. Building 'er up, if this works out the design will hit production with only a couple of extremely minor changes.
A2DP is the underlying bluetooth protocol. Codecs (SBC, AptX, AAC) ride on top of it.
The vanilla RN52 does only SBC. I'm sure it's good enough, especially considering the listening audience of a Boominator system usually has a few drinks in them. I can't help trying to make the best solution I can though.
Prototypes are mostly together, accidentally left a couple things off the digikey order.
The vanilla RN52 does only SBC. I'm sure it's good enough, especially considering the listening audience of a Boominator system usually has a few drinks in them. I can't help trying to make the best solution I can though.
Prototypes are mostly together, accidentally left a couple things off the digikey order.
She's running. Have a hiss caused by an unstable voltage regulator for the audio input/filter section, tacking a big electrolytic onto the rail shuts it up and it sounds good then, gonna have to swap out the regulator for a different one. +5V charging works on my android phone and gf's ipod touch, solar charging also works. Full board controller software is getting there.
I'm happy with the board overall. Making a couple minor changes for the release version of the card - changing from 100V X5R caps to film caps in the audio path, changing the output filter film caps from 63V to 100V ratings to better handle unloaded operation, and a couple other minor things.
Looks great - nice job.
You said that there was some oscillation somewhere - might I ask what reg it was, and what capacitors were around it? Most linear regs (think 7805 and other jellybean parts) actually prefer higher ESR caps, so long-life capacitors as opposed to low ESR types are actually better.
Is that an LT part I see top left (assuming a switcher)? Fancy 😉
You said that there was some oscillation somewhere - might I ask what reg it was, and what capacitors were around it? Most linear regs (think 7805 and other jellybean parts) actually prefer higher ESR caps, so long-life capacitors as opposed to low ESR types are actually better.
Is that an LT part I see top left (assuming a switcher)? Fancy 😉
Regulator is a TPS70933, which is designed for ceramic caps. I've got a 1uF/25V X5R in front of it, and a 10uF/6.3V X5R on its output.
Rather than using an extra op-amp to make a virtual ground, in a few places in the audio path I'm using resistor dividers between Vcc and ground to make thevenin grounds. As a result the PSRR of the analog section sucks, but I was hoping the LDO would keep that quiet. Time to hit up the catalogs and find something a bit better.
There's a few switchers on there. Upper left is a Linear LT3652 MPPT solar charger, upper right is a LM43603 buck converter for generating +5V. Arguably the TPA3116 is a 4-channel synchronous buck converter 😉
Rather than using an extra op-amp to make a virtual ground, in a few places in the audio path I'm using resistor dividers between Vcc and ground to make thevenin grounds. As a result the PSRR of the analog section sucks, but I was hoping the LDO would keep that quiet. Time to hit up the catalogs and find something a bit better.
There's a few switchers on there. Upper left is a Linear LT3652 MPPT solar charger, upper right is a LM43603 buck converter for generating +5V. Arguably the TPA3116 is a 4-channel synchronous buck converter 😉
Last edited:
And that was a quick/painless search..
LTC1844 = pin compatible with TPS70933, stable with 100uF+ output capacitance, and lower noise.
LT1761 = lower noise than 1844. Stability plot only shows 10uF but I don't see any reason it's not stable with more.
And both these chips have a BYP pin, which is NC on the TPS, where you can stuff a capacitor to further reduce the output noise of the part. I'll fire one of these onto the board and see how it goes, I'll probably throw a big electrolytic on the rail anyway for the final design.
Unrelated, but NIC Components has a really interesting line of film caps (NSPH series) which are basically the same size/shape/performance as MLCC's but they're film instead... I reallyreallyreally want to use these in the highpass instead of the bigger 1210 $1/pop films currently designed in, sadly they're not available anywhere yet.
LTC1844 = pin compatible with TPS70933, stable with 100uF+ output capacitance, and lower noise.
LT1761 = lower noise than 1844. Stability plot only shows 10uF but I don't see any reason it's not stable with more.
And both these chips have a BYP pin, which is NC on the TPS, where you can stuff a capacitor to further reduce the output noise of the part. I'll fire one of these onto the board and see how it goes, I'll probably throw a big electrolytic on the rail anyway for the final design.
Unrelated, but NIC Components has a really interesting line of film caps (NSPH series) which are basically the same size/shape/performance as MLCC's but they're film instead... I reallyreallyreally want to use these in the highpass instead of the bigger 1210 $1/pop films currently designed in, sadly they're not available anywhere yet.
The NIC little films are from 2011, so if not available yet, better not expect it will ever be available. Digikey and some others stock Rubycon PML, quite pricey and not yet, LOL, the audio version of PML, MU.
Yeah, tell me about it.
Highpass currently uses 100V rated, 5% X5R capacitors. Didn't really have much else of a choice for a 330nF cap - films are huge/expensive (other than the unobtanium NICs), bipolar electrolytics are +-20%, NPOs don't go that high.
I've currently got these designed in, resistor values changed to keep the same highpass corner. Thankfully I only need 4 at this price point...
ECH-U1C104GX5 Panasonic Electronic Components | PCF1189CT-ND | DigiKey
Part of me is still thinking "keep the X5R's" though. I'm operating them well away from their capacitance/voltage roll-off so they shouldn't distort by any appreciable amount (I'm betting the TPA, output filter, etc create way more distortion) and the temperature/capacitance variation should only cause the highpass corner frequency to shift by a few hertz. At the end of the day, it's a party amp, not an audiophile statement..
Highpass currently uses 100V rated, 5% X5R capacitors. Didn't really have much else of a choice for a 330nF cap - films are huge/expensive (other than the unobtanium NICs), bipolar electrolytics are +-20%, NPOs don't go that high.
I've currently got these designed in, resistor values changed to keep the same highpass corner. Thankfully I only need 4 at this price point...
ECH-U1C104GX5 Panasonic Electronic Components | PCF1189CT-ND | DigiKey
Part of me is still thinking "keep the X5R's" though. I'm operating them well away from their capacitance/voltage roll-off so they shouldn't distort by any appreciable amount (I'm betting the TPA, output filter, etc create way more distortion) and the temperature/capacitance variation should only cause the highpass corner frequency to shift by a few hertz. At the end of the day, it's a party amp, not an audiophile statement..
The bypass pin is for decoupling the reference. Not necessary, but it can lower noise. To be honest, with all the switching you have around the place it probably won't make much difference.
I think that using large ceramics is a bit pointless - maybe use 1uF ceramic on the output (if the OP is 5V, use at least a 10V, 6.3 is only really for 3.3V applications) along with a 47uF or so for bulk.
As for the high pass - I assume it's a sallen-key job or something similar? If it were me, I would just use the input impedance of the TPA3118 and select a capacitor accordingly to make an RC high pass filter. It will work just fine and removes the opamp and other capacitors from the circuit.
Especially for class Ds like this, I really can't tell much difference between film and a GOOD ceramic. I've been using good 1uF 50V 1206 caps on my little TPA amps and they work very nicely. I'm going to change to 1210 on these as there are already 1uF 1210 caps for decoupling so it reduces the BOM count a bit.
I'd say that some 1uF X7R caps will do just fine here. Really no point in spending a dollar each for films. I've done double-blind tests with WIMA SK2 film caps, real cheapo film caps and ceramics. I have a nice composite LM3886 amp with really low distortion, on that amp there was possibly some difference between the ceramic and film, but no difference at all between the two film caps. On the class Ds, there is absolutely no discernible difference.
I think that using large ceramics is a bit pointless - maybe use 1uF ceramic on the output (if the OP is 5V, use at least a 10V, 6.3 is only really for 3.3V applications) along with a 47uF or so for bulk.
As for the high pass - I assume it's a sallen-key job or something similar? If it were me, I would just use the input impedance of the TPA3118 and select a capacitor accordingly to make an RC high pass filter. It will work just fine and removes the opamp and other capacitors from the circuit.
Especially for class Ds like this, I really can't tell much difference between film and a GOOD ceramic. I've been using good 1uF 50V 1206 caps on my little TPA amps and they work very nicely. I'm going to change to 1210 on these as there are already 1uF 1210 caps for decoupling so it reduces the BOM count a bit.
I'd say that some 1uF X7R caps will do just fine here. Really no point in spending a dollar each for films. I've done double-blind tests with WIMA SK2 film caps, real cheapo film caps and ceramics. I have a nice composite LM3886 amp with really low distortion, on that amp there was possibly some difference between the ceramic and film, but no difference at all between the two film caps. On the class Ds, there is absolutely no discernible difference.
Input filtering is as follows...
- differential amplifier with highpass -3dB at ~20Hz, lowpass -3dB at ~30KHz
- sallen key highpass butterworth at 68Hz
- TPA input coupling capacitors, putting another highpass corner at ~20Hz.
I do want to keep the differential amplifier (prevents ground loop ugliness when you're charging your phone/MP3 player) as well as the higher order, "set where I want it" highpass to keep bass out of the Boominator. The same filtering (albeit lower order) could be accomplished done by using the TPA's differential input and picking suitable coupling caps, but it would make the frequency response sensitive to the impedance of the source device, not to mention the input impedance of the TPA has a fairly wide tolerance.
The 'switching going on' shouldn't be a huge issue. The LT3652 effectively filters its own output (buck converter, output inductor feeds +12V) which keeps hard-edged trash off the rails. The LM43603 for +5V USB power puts hard switching edges on +12V, but it's synchronized to the TPA when it's on which will prevent any audible beating from happening between TPA and 43603.
The LDO for the op-amps (3.3V output) is fed from the LM43603 through a RC filter to keep the highest frequency trash out of it, its control loop should hopefully take care of lower frequency content. I put a big cap at the LDO input after the RC to see if that was causing the instability, wasn't that...
- differential amplifier with highpass -3dB at ~20Hz, lowpass -3dB at ~30KHz
- sallen key highpass butterworth at 68Hz
- TPA input coupling capacitors, putting another highpass corner at ~20Hz.
I do want to keep the differential amplifier (prevents ground loop ugliness when you're charging your phone/MP3 player) as well as the higher order, "set where I want it" highpass to keep bass out of the Boominator. The same filtering (albeit lower order) could be accomplished done by using the TPA's differential input and picking suitable coupling caps, but it would make the frequency response sensitive to the impedance of the source device, not to mention the input impedance of the TPA has a fairly wide tolerance.
The 'switching going on' shouldn't be a huge issue. The LT3652 effectively filters its own output (buck converter, output inductor feeds +12V) which keeps hard-edged trash off the rails. The LM43603 for +5V USB power puts hard switching edges on +12V, but it's synchronized to the TPA when it's on which will prevent any audible beating from happening between TPA and 43603.
The LDO for the op-amps (3.3V output) is fed from the LM43603 through a RC filter to keep the highest frequency trash out of it, its control loop should hopefully take care of lower frequency content. I put a big cap at the LDO input after the RC to see if that was causing the instability, wasn't that...
Last edited:
Ah OK, points noted.
That's still very weird about the oscillation... Did you scope around to see what's actually going on?
That's still very weird about the oscillation... Did you scope around to see what's actually going on?