CM5908/5907 Plug-n-Play Bluetooth upgrade board

[earthwormjim]

That module should have provisions for an antenna pin. Looks like if you desolder the 0402 jumper leading to the antenna, and move it over to the left position, that lets you wire an antenna to pin 43.

I'm guessing what you did, is you messed up the impedance matching of the antenna, so you ended up weakening the RF signals. It's no longer a nice fraction of a wavelength.

Normally those little modules (FCC certified ones) have a shield covering everything except the antenna, wonder if that might help?

[jaysterling]

That module should have provisions for an antenna pin. Looks like if you desolder the 0402 jumper leading to the antenna, and move it over to the left position, that lets you wire an antenna to pin 43.

I'm guessing what you did, is you messed up the impedance matching of the antenna, so you ended up weakening the RF signals. It's no longer a nice fraction of a wavelength.

Normally those little modules (FCC certified ones) have a shield covering everything except the antenna, wonder if that might help?

Oh ok, makes sense about killing the impedance matching, so it's just the reduced signal strength that's making the noise go away.

The problem with the external antenna solution is the pick-n-place machines that will be doing the assembly can't desolder components right? So I'd need to try to source some modules that already had the right jumpers soldered on, which I'm not sure is possible. I can't imagine manual rework of each module is cheap. If f I did go that route, I'm also clueless what I'd actually do with the antenna design to fix the problem.

So you don't think moving the chip as far that corner as possible, adding a top ground plane + extend the bottom ground plane up around the chip would do anything? I gotta think this can be solved without an external antenna since any run of the mill BT receiver uses a similar module + chip antenna and they don't have this problem, even without a shield over the module. When you look at one though (I tore apart a cheapo one I had) it'd definitely a 2 layer ground plane with lots of stitching that fills everywhere except right under the antenna, all things I'm not currently doing.

[earthwormjim]

Jay, I'd test out first if I'm right about having to move the jumper, I'm just basing it off your pictures. See if there is continuity between that pin and the antenna already. Where did you get that module anyway? If ordering from Alibaba, you should be able to specify from the vendor, or they may have an external antenna version.

I gotta think this can be solved without an external antenna since any run of the mill BT receiver uses a similar module + chip antenna and they don't have this problem, even without a shield over the module.

Most that are FCC certified, do have a shield over the module.




Sounds like it is time to bust out the copper tape, and see where you might need ground planes or shields.
On my bluetooth project, my mating board had all available space filled in with ground pours.

You could also extend your PCB underneath the module's antenna, and pour in a ground plane. Covering underneath part of the PCB antenna with ground, should weaken RF signals like your added wire does too. You can test with strips of grounded copper tape.

So you don't think moving the chip as far that corner as possible, adding a top ground plane + extend the bottom ground plane up around the chip would do anything

It could very well help, if you can move the module and wire it up with thin leads, test it out. On the power supplies I designed, I ended up putting stuff on flexible magnetic wire, and moving it around while powered up, to see if my EMI emissions changed in real time.

[bradnic]

Jay how close are you to being able to add pre-outs? A lot of premium sound cars have bad amps, and I bet many folks have bypassed the premium amp by now. I have my car harness setup to support both pre-out and the inline premium amp so I'm ok without pre-out support for now. But I would definitely prefer a pre-out arrangement as I'm sure most folks would.

Not something needed for a minimal viable product version of your board, but should be high on the dev list.

[jaysterling]

Sounds like it is time to bust out the copper tape, and see where you might need ground planes or shields.
On my bluetooth project, my mating board had all available space filled in with ground pours.

Gave that a shot today. I didn't have copper tape (ordered some) so I took a shield off some junk I had and soldered on a lead so I could ground it and hold it in various positions. Covering the non-antenna part of the BT module didn't change anything unfortunately, nor did covering just my analog section (as expected, since the noise is there without the audio cable even being plugged in). The positioning below kills it though:



The most crucial area to make the noise go away seems to be the most forward section of that shield, where there is currently no PCB or ground planes, which seems kinda relieving since there are known improvements I can make there. This also kinda makes sense because the unbalanced audio out cable is right behind the shield, and some other audio mixing/EQ circuitry is on the vertical board that it reaches over.

You could also extend your PCB underneath the module's antenna, and pour in a ground plane. Covering underneath part of the PCB antenna with ground, should weaken RF signals like your added wire does too. You can test with strips of grounded copper tape.

All the datasheets I read say to keep copper out of that area right underneath the antenna. They're still effective like that? Then again, it still streamed music with a wire soldered on the antenna... I suppose it's just a trade-off of range vs. EMI in some cases like this?

Still thinking I should extend the PCB frontwards as far as possible and use stitched ground pours on both sides, then shift the BT chip as far away as possible. Hopefully it doesn't require some kind of vertical shield soldered on the board... I've got the copper tape coming tomorrow so I can at least try building up an extended ground plane with that.

[earthwormjim]

Cool, so looks like you found what was picking up the RF emissions; the EQ circuitry and/or the audio cables.

Hopefully you can solve it by relocating the module as you were saying. If relocating or more filled out ground planes doesn't help, you could use a transistor PCB heatsink to act as a verticle shield, similar to what you have now.
They're pretty cheap.

Something like this:

https://www.alibaba.com/product-detail/OEM-High-power-to-220-heatsink_60499402903.html

All the datasheets I read say to keep copper out of that area right underneath the antenna. They're still effective like that? Then again, it still streamed music with a wire soldered on the antenna... I suppose it's just a trade-off of range vs. EMI in some cases like this?

Any point of the antenna that has a ground underneath it, is no longer an antenna, so you effectively reduce the antenna length to some odd value that deviates from 1/4 or 1/8 of a wavelength, similar to what your wire does. But definitely scratch that suggestion, shielding is way way better to do, and you found what to shield.

[george graves]

LDOs are linear regulators

Yep. I was asking why a LDO would be more quite then a non-low-drop-out linear regulator - both should be running an op-amp internally. And both sinking excess current as heat. Even so, unless you have some unstable feedback internally, both should be about the same.

[earthwormjim]

Yep. I was asking why a LDO would be more quite then a non-low-drop-out linear regulator - both should be running an op-amp internally. And both sinking excess current as heat. Even so, unless you have some unstable feedback internally, both should be about the same.

They're not, and I never implied otherwise. They're cleaner than a buck regulator would be. Was pointing out his 3.3V supply is clean, and not to look there for noise issues.

[george graves]

Everyone knows that!

That's good you are using separate regulators, and LDOs are super clean.

Now can you see my confusion? You specifically mentioned LDO and not just jelly bean linear regulators. That's all. So that's why I was asking. No harm, no faul. Carry on! ;)

[yanal]

I really love seeing projects like this, its sad that I longer have enough time on the side for fun things like this. good for you! also check you PM I'll be sending you a message.

[bradnic]

Dealing with premium amp bypass

A lot of premium sound cars have bad amps, and I bet many folks have bypassed the premium amp by now.

Pretty sure this is what Barry was getting at earlier in the thread:

Brilliant. Is this to be used with the premium sound amp or does it have an onboard amp?

As I explained, the current board doesn't add pre-outs, so it doesn't matter whether/not you have the premium amp. It does matter if you've bypassed the premium amp though, since the OEM radio/head unit only supports speaker level outputs.

If you don't have an aftermarket amp you need to connect your OEM radio speaker outs to your speakers and you're limited to 5W per channel. If you do have an aftermarket amp, you need to use its speaker level inputs or not use it all with this version of Jay's board.

[Boggie1688]

Jay,

Do I need to be ordering a RPi2 in preparation for the beta board?

[jaysterling]

Project Update: I spoke with the guys doing the beta test offline but I did not get boards out at the start of May, I didn't get the board design out before going on vacation so that added time. I got the boards last week though and was shooting to send them out today.



Unfortunately, that is not happening either. I assembled the first board on Saturday and immediately had all kinds of problems. The bench supply was cutting power like there was a short, and after a bunch of painful back-tracing the problem it turned out to be the board print itself has a short from a power pad to the top ground plane. It's like an error in the printing itself though. If I probe the remaining bare boards, there are a handful of pads on the top on each that are shorted to ground. And it's not consistent; different pads are shorted on each board, but not one is all-good

So why such fail? Well I didn't pay attention to the isolate parameter on the top ground plane I added. This sets how much spacing it puts between the ground plane and other copper on the top. It defaults to a very small value, and I didn't notice to change it to something more reasonable. It did, however, pass the DFM check from the fab house, which is supposed to catch stuff like this. The min trace-space distance was reported as 4.99 mil on the report, and although it's in red, the test still passed. Anyway, there are tiny little shorts between pads and the plane, so unfortunately, oddly-shaped coasters is all these boards will ever be. I should have paid more attention to the report, and there's no need for the spacing to be so tight, but I would think if the fab house test passes, they're saying they can build it. So I'm working with them to see what they'll do about it.

Regardless, it's going to be another board rev before I'm ready for beta, which means another 2 weeks, so that sucks.

On a positive note, one of the new boards had its ground plane shorts strategically placed so I could at least test audio output through it, and was able to validate the changes I made. I still need the upright RF shield, but my home-made aluminum shield in here did the trick:



Still gotta find someone on allibaba who can supply something like it or could even press that piece custom. Thanks for the tip on that Jim.

So stay tuned, this will happen eventually! I've been working pretty long hours at my normal job lately so that hasn't been helping progress, but it'll get there.

Jay,

Do I need to be ordering a RPi2 in preparation for the beta board?

Nope, this doesn't use a RPi, that's the other version of this project you're thinking of from my other thread. That one's on hold for now.

[Boggie1688]

All good Jay!! Plenty of other things to fiddle with while you iron out the details.

[Appleseason]

How could i get one of those boards?