nRF24L01+ PA LNA - Seems that the PA (power amplification) portion of the module is failing?

Discussion in 'Wireless & RF Design' started by Mahonroy, Sep 13, 2016.

  1. Mahonroy

    Thread Starter Member

    Oct 21, 2014
    200
    5
    Hey guys,

    I have designed several circuits all of which use a nRF24L01+ PA LNA module (picture of it below).
    Occasionally we would come across a problem where the module will only transmit when you touch the module or antenna with your finger. This generally means that the transmission is actually occurring, its just extremely weak (and possibly gets amplified by your body slightly) which makes it temporarily work when touching it. Sometimes we could just replace the module and everything would be fine. Other times we would reduce the power output slightly and then it would be fine. A couple times I experimented by replacing the 220uF capacitor with a 1000uF capacitor and then it would be fine... definitely a very frustrating problem.

    The modules have always received transmissions perfectly fine.

    Well, as time goes on, and more designs are made, I now have an extremely high reject rate with these modules (5 out of 6 exhibit this problem now). And now, even replacing the modules or reducing the power doesn't solve the problem. Some of my designs seem to exploit this problem easier than others.

    The problem looks to be that the PA portion of the module is malfunctioning/overloaded which causes this extremely weak transmission to occur (this is only speculation). My different designs seem to play a role in causing this malfunction (also still speculation).

    For all I know, touching the chip seems to remove noise somehow which allows the transmission to work... but the transmission is still extremely weak in this state, so it still seems its the PA.

    I am at a complete loss on how to troubleshoot this thing, so figured I would see if any of you guys could lend some advice, or if you know what might be the problem? I have an oscilloscope and am willing to do whatever it takes to fix the problem, I just don't know what to do exactly, or what else to try.

    Here are a couple things that I am aware of and have tried:
    • Supposedly these modules are very susceptible to noise on power. In all of my designs, I generally use a MCP1700 LDO voltage regulator (3.3V), a bulk capacitor between 220uF and 1000uF close to the nRF24L01+, and decoupling capacitors close to the nRF24L01+ as well. After speaking with the engineers who designed the module, they recommended I try this power filter below. I got a batch of boards made to try it out and it didn't change a thing.
    nrf_troubles1.jpg
    • There are a lot of counterfeit Nordic nRF24L01+ modules floating around that can have a number of problems. The supplier of my modules assured me they are genuine Nordic modules, but just to be sure, I desoldered the nRF24L01 chip, and solder a real Nordic chip in its place. I get the exact same problem still.
    • These modules don't have a shield covering the electronics so I was wondering if that could be a problem. I rigged up one of these modules in its place and I get the exact same problem... only transmits when I touch the antenna or the shield.
    nrf_troubles3.jpg
    • I took an old design that had a working nRF24L01+ module on-board, and a design that currently was not working. I desoldered both modules and swapped them, and the design that originally was working continued to work. The other device barely worked (had extremely low successful transmissions). This sort of reinforces the idea that different designs exploit this problem - whatever it may be.
    • I also experimented with soldering the nRF24L01+ module upright and this didn't change a thing. The modules already overhand the board anyways, but I was out of ideas to try.

    Here is a picture of the module I have been using:
    nrf_troubles2.jpg

    The designs generally consist of a circuit like this:
    nrf_troubles4.jpg

    And the RF modules always overhand the board like this for better reception:
    nrf_troubles5.jpg

    Thanks for taking the time to check it out, and any help or advice would be greatly appreciated!
     
  2. Mahonroy

    Thread Starter Member

    Oct 21, 2014
    200
    5
    I have some good news.

    I decided to test the RF modules grounding, so I scraped off a patch of the RF modules black solder mask near the PA and connected an additional bridge to the circuit boards ground. Sure enough this fixes it. I’ve done this to two of them so far… both didn’t work before and now they transmit just fine.

    I am trying to completely understand what this means. Does this mean that the modules ground pin is too far away from something, and this causes a loss of power during transmission? Was the ground plane interfering/blocking the transmission, and the bridge makes that not happen somehow? Is there a bad ground loop I am not aware of? So it seems related to my layout and how I did the ground, I am just trying to understand it.

    What do you guys think?

    Here is the schematic:
    nrf_problem_1.jpg

    Here is the layout:
    nrf_problem_2.jpg

    Here is the solution:
    nrf_troubles9.jpg

    What can I do differently with the layout so I don't need that ground bridge?

    Thanks again, any help/advice is greatly appreciated!
     
  3. Greebo

    New Member

    Thursday
    4
    0
    Hi Mahonroy! Thanks for these inputs from your side. Last night I discovered that ALL of my 20 new nRF24L01-PA-LNA units do exactly this. They work fine when I touch the PA. So tonight I am trying the better grounding exercise as you described. Hope they start working!
     
  4. Mahonroy

    Thread Starter Member

    Oct 21, 2014
    200
    5
    Hi Greebo,
    Which library are you using to drive the RF module? In one of our cases, it ended up having to do with the CE pin... the datasheet says minimum 10uS pulse to initiate the transmit, however, the way the PA's are wired up is that the CE pin is tied directly to the RX enable of the PA. Something to keep in mind, because we had to extend the duration of the CE pin in order to get it to transmit properly.
     
  5. Greebo

    New Member

    Thursday
    4
    0
    I am using the standard RF24 Arduino library.
    Went through the whole grounding exercise last night. Added an LC filter to the supply. Decoupled the supply and even made the little mode you showed by cleaning a section on the ground plain and fixing ground there. Still the same. As soon as I touch the PA it goes. Will have a look at the CE pulse length to make sure it is long enough just in case my radios are wired as you say. Thanks for the info! Will let you know should I achieve success.
     
  6. Mahonroy

    Thread Starter Member

    Oct 21, 2014
    200
    5
    Yeah thats one thing I've noticed is there are a few different & separate things that can cause these things to fail and only transmit when you touch the module. Its weird that the failure mode is always the same between these things too.

    I would check the code to see how the CE pulse is set up, if its either a standard delay, or if it constantly monitors the status register to tell when its done transmitting. You also might want to try a 4.7K pullup resistor on CE as well.

    If you have a spectrum analyzer, another good exercise to do to make sure its even transmitting (e.g. it could be transmitting jumbled data), is to put the nRF chip in "constant carrier mode", then watch the signal strength on the analyzer. If its a strong signal, then you know its either firmware related, or something is causing the data to get scrambled on its way to the PA.
     
  7. Greebo

    New Member

    Thursday
    4
    0
    Hi Mahonroy,

    I found the problem. I have a feeling that many others have this problem. The CE pin on the header of the radio is not connected to the CE pin (pin 1) of the nRF24L01 chip. Somewhere I saw in a forum that someone had the same problem. He later mentioned that it looks as if the CE pin is not connected to the chip but never posted a followup. So I ignored it.
    I first started debugging by looking at the status register when you transmit and realized the MAX_RT and TX_DS bits never get set. I can understand the TX_DS since I have no other radio to ACK but the MAX_RT should at least get set unless the radio is never instructed to start the transmission. So I measured my radio's CE on th header to the nRF24 chip and open circuit! I did a wire mod on the radio to connect the CE and the radio started working immediately.

    When you touch the PA (while the SW is waiting for either the MAX_RT or TX_DS to be raised) you "trigger" the TX enable on the PA and the transmission happens. I did not do any range test (I think the other mods might help with this) but at least it is acting the same way as my other radios. Guess that is what you get for buying the cheap Chinese copies. The units I bought are the YJ-15008+PA from "keyes". Well that is what is written on the radios. Hope this helps someone somewhere!
     
  8. Mahonroy

    Thread Starter Member

    Oct 21, 2014
    200
    5
    Thanks for this extra information. Can you elaborate a little more on this CE not being connected issue? I am ordering RF modules from the same manufacturer, just with the ceramic antenna instead. How is the CE pin on the header not connected to the CE on the nRF24L01 chip? Was it a bad solder joint?
     
  9. Greebo

    New Member

    Thursday
    4
    0
    I have no idea why it is not connected. But what I can say is that all my units 10+ have the same problem. So I think it was a bad bunch. Below is how I fixed it. Don't mind the black wire running from ground, was another experiment. 20170113_224428.jpg
    First clear the coating from the track to pin 1.
    20170113_224923.jpg
    Then attach wire from header pin CE to track.
     
Loading...