Hot!RF getting into AF

Page: 12345 > Showing page 1 of 5 - Powered by APG vNext Trial
Author
laurie21
Junior Member
  • Total Posts : 103
  • Reward points : 0
  • Joined: 2009/07/21 12:59:44
  • Location: Vancouver B.C. Canada
  • Status: offline
2017/12/07 15:46:20 (permalink)
0

RF getting into AF

I am having problems keeping RF out of the audio section using an RN-52 module. I have the RN-52 EK board and it is what i have followed in laying out my own boards. Also i am using an 18F2680 for the control and function of the module. The pic and all digital works as it should. The audio out and in works but it still has some RF wine included in the AF signal path. I have watched for ground loops and floating gnd areas on the board when drawing it out. I am now considering adding a metal shirld over all but the modules. Many experiments later i know for sure it is a shielding problem but what confuses me is that i have a coard which is very similar in layout to the EVAL board and it does not do this mine does.
 Is there anyone willing to chat or PM me about this and lend some advise knowing what i have already tried?
#1

97 Replies Related Threads

    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/18 11:52:46 (permalink)
    0
    I have fashioned cans from .005 brass. Covered both master and slave units. I find no difference in the amount of RF getting into the signal's AF. I still think the RF wine is coming from the antenna because placing my finger close to the antenna dramatically changes the amount of the injection. Next i am going to change the can's size to only include the audio parts of the board.
    #2
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/19 16:47:23 (permalink)
    0
    Well the cans are doing little to stop the injection. I have taken the cans off and found placing one of my fingers in one spot on the board completely stops the injection. I have tried caps to bleed the noise to grnd. No luck.
    #3
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/21 10:07:13 (permalink)
    0
    Placing my finger in one spot on the board almost completely stops the RF from getting into the AF signal chain. Seeing this, i made another can to shield just the area where i place my finger. It does not stop the wine at all. But now placing my fingers on other areas of the board diminishes the injection but does not stop it.
    #4
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/26 00:58:54 (permalink)
    5 (1)
    What i have observed is, using the scope to measure the audio signal path, I can measure the offending wine.
    It is also on the 5vDC rail as well so tried an experiment. Scope looking at audio sig. Finger on and off the antenna. The offending wine goes up and down. But scope watching the dv5v rail, finger on and off the antenna, offending wine stays at the same amplitude. If i assume my wine is in the power supply rails I should be thinking about including an inductor. Next step a value for the inductor.
     I have made enough postage-stamp sized can's to mail a letter around the world and that approach has not been fruitful. Interesting (results different every time a can was added), Headaches (cross-eyed), Dangerous (sharp edges), Frustrating (did little to rid the prob),   yes, but not fruitful. A new approach is needed.
    #5
    qɥb
    Monolothic Member
    • Total Posts : 2516
    • Reward points : 0
    • Joined: 2017/09/09 05:07:30
    • Location: Jupiter
    • Status: offline
    Re: RF getting into AF 2017/12/26 02:01:39 (permalink)
    0
    laurie21
    ...
    It is also on the 5vDC rail as well so tried an experiment. Scope looking at audio sig. Finger on and off the antenna. The offending wine goes up and down. But scope watching the dv5v rail, finger on and off the antenna, offending wine stays at the same amplitude. If i assume my wine is in the power supply rails I should be thinking about including an inductor. Next step a value for the inductor.

    My conclusion from that test would be that the noise is NOT coming in via the digital power supply.
     

    PicForum "it just works"
    #6
    Peter Sikora
    Super Member
    • Total Posts : 132
    • Reward points : 0
    • Joined: 2003/11/07 12:46:46
    • Location: Brisbane Australia
    • Status: offline
    Re: RF getting into AF 2017/12/26 02:45:52 (permalink)
    0
    How are your grounds organized? Preferably they should be separated between Analog, Digital and RF power, in a star pattern going back to a single point. If Faraday cage (can) doesn't help then it is probably not radiated energy. I would try running RF section from separate isolated power supply to see what effect that has before jumping to conclusions. If the problem goes away then you know for sure it is not radiated emf.
    #7
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/27 00:52:50 (permalink)
    5 (1)
     Hi Peter. The board is 2.5"x 4" and employs a ground plane so each gnd connection is connected directly to the plane. I did my PCB layout in this way as it follows the same pattern of layout as the evaluation board for the Bluetooth modules. I am currently going under the assumption that it is the power supply. I am thinking the 3.3v regulator is breaking down and allowing the osc to occur on the supply rails. It therefore would show up everywhere, <as it is doing>. Putting a can over the audio signal section should have stopped or at least diminished the amplitude of the offending wine if it was an injection from the RF antenna. I have tried an inductor on the supply lines and that has not made any difference either. So, now i am changing the regulator, (150mA to 500mA) to see if it is that which is breaking down and causing the prob's. Stay tuned.
    #8
    Peter Sikora
    Super Member
    • Total Posts : 132
    • Reward points : 0
    • Joined: 2003/11/07 12:46:46
    • Location: Brisbane Australia
    • Status: offline
    Re: RF getting into AF 2017/12/27 02:18:20 (permalink)
    0
    I see this problem come up with switch mode power supplies in a similar way. A single ground plane never works well if there is an ADC converter involved trying to sample the output. If you have more than one PCB available I would try building up the analog section separately on its own and power up from a separate power source. Even taking scope measurements try and pick a ground point as close to the analog section as possible. I have been caught out thinking there is noise on the supply but was actually seeing noise on the ground plane. Good luck and keep posting the results. I will be very interested to see what you find. Can you post a pic of the board?
    #9
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/27 19:22:29 (permalink)
    3 (1)
    Yes. I made the boards so each section is like a module to itself on the board. Then i built each section and tested it. That is why i know all the <audio bus> , <digital control bus> and <communication bus> are correct and all is working as it should. At first i was using a charge pump (MAX682) to get the main power supply bus up to a steady 5v for most of the IC's. To test if that was the source, i changed to an analogue power supply. But no change at all. Also I will say that the audio is dead quiet even with the digital control and comm bus's working. But soon as the Bluetooth module makes the link between master and slave the wine is solid and steady. The problem is compounded by the fact that 2 modules are being used each on it's own pcb. Whatever problem one has, the other is sure to have as well. And once it is in the audio signal path.......... I'm just surprised the whole thing doesn't go into full feedback breakdown. We could go together, it on the board and me on the floor.
      I don't know what to try next. So far i have been trying to discover just the source of the injection. From looking at the scope i can see the beat of 1mS but where it comes from is the needle.
    #10
    Peter Sikora
    Super Member
    • Total Posts : 132
    • Reward points : 0
    • Joined: 2003/11/07 12:46:46
    • Location: Brisbane Australia
    • Status: offline
    Re: RF getting into AF 2017/12/28 03:26:15 (permalink)
    0
    Did you try powering the Analog section from a completely isolated power source ( 9V battery + Vreg) while the Bluetooth is running? That's what I would try next. But it will only work if you have separate ground planes I must add.
    post edited by Peter Sikora - 2017/12/28 03:28:39
    #11
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/28 14:11:26 (permalink)
    0
    NO i haven't tried using 2 different power supplies for audio and RF. Now that i have taken the 150mA reg off the board and replaced it with a 500mA reg i can and will try it. My grounds are all on one plane though so it will not be completely separate.  First i am going to try more C filtering between the reg and the module to see if it is backing up from the mod.
    #12
    Peter Sikora
    Super Member
    • Total Posts : 132
    • Reward points : 0
    • Joined: 2003/11/07 12:46:46
    • Location: Brisbane Australia
    • Status: offline
    Re: RF getting into AF 2017/12/28 19:54:35 (permalink)
    0
    The single ground plane will most likely be the source of your problem. There is no point feeding in a separate power source if you can't separate the ground planes. Split the ground planes and make sure they go back to a single point or build up another board with only the audio section populated and see how that goes in close proximity to the Bluetooth module with both ground returns tied together at the power supply.
    #13
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/28 22:28:42 (permalink)
    3 (1)
    Have built the audio and power and CPU control and it all works as it should. It's only when the Bluetooth module connects that the wine gets in. I have powered the audio and cpu control etc. and placed it all close enough to touch and only when the amp was almost touching the antenna did i get any injection. The thing is, the EK board i got to do initial experiments with has one gnd plane for power, digital, and audio. The module's <audio gnd> is taken to board ground plane right at the agnd pin. The headphone amp is also just connected to board gnd plane. It just doesn't make sense.
    #14
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/28 23:11:22 (permalink)
    3 (1)
    qɥb
    laurie21
    ...
    It is also on the 5vDC rail as well so tried an experiment. Scope looking at audio sig. Finger on and off the antenna. The offending wine goes up and down. But scope watching the dv5v rail, finger on and off the antenna, offending wine stays at the same amplitude. If i assume my wine is in the power supply rails I should be thinking about including an inductor. Next step a value for the inductor.

    My conclusion from that test would be that the noise is NOT coming in via the digital power supply.
     


    What else would you try?
    #15
    Peter Sikora
    Super Member
    • Total Posts : 132
    • Reward points : 0
    • Joined: 2003/11/07 12:46:46
    • Location: Brisbane Australia
    • Status: offline
    Re: RF getting into AF 2017/12/29 03:42:55 (permalink)
    0
    The question is what changes on the blue tooth when it connects. I don't have any hands on experience with the module in question. It may be that it remains in a low power state until paired and transmitting/receiving data. So the kit you used for development is RN52 Bluetooth® Audio Evaluation Kit DS50002153A ? Can you attach your board design?
    post edited by Peter Sikora - 2017/12/29 04:11:51
    #16
    Peter Sikora
    Super Member
    • Total Posts : 132
    • Reward points : 0
    • Joined: 2003/11/07 12:46:46
    • Location: Brisbane Australia
    • Status: offline
    Re: RF getting into AF 2017/12/29 04:53:47 (permalink)
    4 (1)
    If you have a look at the Eval board, they use lots of via stitching and that is probably the reason that board works without noticeable problems. They have inadvertently got multiple paths back to the battery ground from RN53 and U4(TPA6112). Then they have U1(FT232RQ) sucking its ground after RN52 and back to battery ground on the other side of the board. Also B+ distribution is not real good on that board either. They are tapping B+ off after the Audio Amp and then on to the Vreg so any change in load will imprint on the Audio via the resistance of the track going back to the battery. Likewise any time the Audio Amp draws power for a base beat that will be imprinted on the Vreg. Likewise when the Bluetooth draws current from B+ it goes straight back to B+ of the Audio amp. I think they are lucky that demo board worked at all.
    #17
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/29 12:00:31 (permalink)
    3 (1)
    Actually Peter i am using the USB from comp to supply power not the batt conn. The path is the opposite of what you describe. In this case power comes in and goes directly to the FT232 and Reg. Output of the reg goes to RN52 and to TPA6112. The other V+ lines go to pwr sw and tap points. I certainly have as many via's between top and bottom of board continuing the gnd plane. My V+ similarly has separate V+ lines feeding the RN52's 3.3v reg, CPU, and the audio amps from a common (star) point at the entrance to the board. My RN52 has it's own 3.3v reg right at the power input pin of the RN52.
     I was thinking i could cut the gnd connections from the RN52 to separate it from the rest of the modules. It would separate the gnd returns but would not leave any gnd for the RN52.
    #18
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/29 12:20:25 (permalink)
    3 (1)
    On the board there are 21 control lines as well as the audio feeds in and out of the RN52 and the power feeds. It is impossible to have a continuing gnd plane on the board. However any break in the gnd plane on one side is covered by a gnd plane on the other side. They are connected by the via's to get a continuing low impedance gnd plane over the entire board.
    #19
    laurie21
    Junior Member
    • Total Posts : 103
    • Reward points : 0
    • Joined: 2009/07/21 12:59:44
    • Location: Vancouver B.C. Canada
    • Status: offline
    Re: RF getting into AF 2017/12/29 21:48:25 (permalink)
    0
    One important thing i missed was not keeping the differential lines closer together. They are not bad but should be closer to get the most equal injection of any noise into both - / + feeds to get max rejection ratio. So i replaced the foil with 36awg silver wire-wrap wire. Mistake corrected, deep breath, then power-up. Nop, wine is still there.
    1- Being able to place an amp in such close proximity to the RN52's antenna and have little induction. Yet on the same board such a mess. This points to a power supply problem.
    2- BUT, eliminating much of the wine by placing my finger on the antenna indicates the source to be injection to be from antenna.
     Since the audio is not bothered by the digital i am thinking, leaving placement of the Module at the bottom of the board with antenna hanging over the edge, but keeping everything else at the top as far away as possible. That way i can put a Faraday cover on each individually and lay the board out with separate Bluetooth module ground plane and a digital / audio ground plane. Keeping control and audio lines between them together maximizing the area of, and number of ground plane foils on top and bottom.
    3-  I still measure the same wine at the audio/digital 5v+, with the cans covering the audio/digital. Back to it's being injected in the power rails due to transient switching of the RF section when in use.
    Both probably. I'm determined to find out.
     
    #20
    Page: 12345 > Showing page 1 of 5 - Powered by APG vNext Trial
    Jump to:
    © 2018 APG vNext Trial Version 4.5