Re: [Freetel-codec2] RX chip CMX994

Mon, 16 Feb 2015 14:45:07 -0800 

Daniel
Good you have been getting out there to get a better understanding of desktop studies. 


The distance required between two antennas to acheive diversity gain 
(that is, the fading envelope of the n antennas to be uncorrelated ) is 
all about establishing a different pattern of scatterers.



In the urban mobile environment, just a half wavelength is often enough 
to get diversity gain.
But, at a decent transmit site, it takes alot of distance to establish a 
different path to the mobile.
Cellphone towers get away with small spacings because  in general they 
are not hundreds of meters about the average terrain- maybe 30-40.   On 
a mountain top, you are unlikely to be able to achieve any reasonable or 
worthwhile space diversity gain to a mobile.  It seems counter intuitive 
when you only have to go a meter in the mobile but can't achieve space 
diversity gain  on a mountain top, but it is all about the local 
scatterers, and of course on a mountain top, there are none, so you rely 
on establishing a different ray path to the mobile.



I'm excluding space diversity RX for the mobile end (while very very 
effective)  because it is a pain
HOWEVER, perhaps for those that want  it, they can certainly do it. 
Build in two receivers into your box.... much is shared of course.


Suggestions. hmm
what you really want to build is some sort of channel measurement box.

in a simple form, take the RSSI of the receiver out (high bandwidth at 
least 1kHz) and take that to a laptop capable of logging the samples at 
a sample rate of at least 2ksps.  there are many usb devices that will 
do that, you dont need much resolution... 8 bits....
...better,  record the I and Q out out of the receiver into a sound 
card  (with AGC OFF) ...



... better, have two receivers (one antenna)   each with RSSI to a dual 
channel ADC USB device and log that, two x RX 200kHz apart listening to 
a TX eg 200kHz carriers apart so you can observe diversity gain for 
various frequency spacing and antenna spacing setups.



----even better - build a bi-static chirp channel sounder that sweeps 
over a whole band



HPOL turnstiles are fine, but very hard mobile to get nice low radiation 
angle. a pain. Try circular using an Lindenblad antenna at the base.




On 17/02/2015 3:27 AM, Daniel Mundall wrote:

Hello Glen,


I totally agree with the flat fading, at the moment it appears that we 
are getting anywhere from 10-30db of attenuation during fades, but 
when we are within say 5km of the TX I still have around 10db during 
the worst fades. So I think there's more at play than just no signal. 
Which is kind of pointing towards a demod bug, likely to do with the 
rapid amplitude or phase shifts. Seems to do fine if the fades are 
slow, like with say walking. At the moment moving samples on the edge 
(35km away) of the coverage seem to be demoding better than the ones 
are are say 1km away.



I've been scheming about diversity: And my current idea is dual TX 
chains but closely spaced say 10khz apart, and separated by say 10λ 
horizontally. Another idea is to use cross popularization instead of 
horizontal space. The data going into these channels could be 
interleaved and some redundancy added, or it could be an exact copy of 
the other chain.



I think it would be nice to stay within a single 25 or 15khz channel 
if possible, makes spectrum planing way easier, and for that matter 
filtering on the RX side.



With the above method there would still be a weak link with the 
uplink, you're dealing 3db less power, but we should be able to regain 
that with either digitally combining the two RX antennas using an FEC 
method, or by using something more complex like MRC(which I'd like to 
avoid).



At the end of the day it would only be an extra chain at the repeater 
which doesn't add much cost in the big picture.



Anyway plan to get a test setup for this done up in the next day or 
two so I can see how much the horizontal space diversity helps. See 
how many of the errors are correlated.



Do you have any suggestions about this idea, or other test I should 
run while I'm doing them?



Oh and by the way, do you know anything about turnstile antennas 
configured as horizontal polarized omni antennas, are they any good? 
I've been thinking of building one for testing H-pol. ( 
http://www.g8seq.com/vhf_antennas.php )



Thanks,
Daniel VA7DRM


Daniel Mundall


On Mon, Feb 16, 2015 at 1:59 AM, glen english <g...@cortexrf.com.au 
<mailto:g...@cortexrf.com.au>> wrote:


    Hi Daniel

    Yes.. you can improve the demod to cope with multipath, but for a
    flat fading channel like 2m or 70cm, it does not matter how good
    your demod is if you have no signal when you stop at the traffic

    lights..... that is why the system needs some sort of diversity. 
    It doesnt matter what type, space, frequency, site just some type.

    Unless you go for a linear radio (which would be a no no for
    compatibility and upgrade and all sorts of reasons)  you are
    pretty much limited to some sort of dual site TDM (single
    frequency) or dual frequency single site.

    circular polarisation can go about 50% of the way and will make a
    big difference... but in most cases you wont have much choice of
    existing antennas.

    glen




    On 16/02/2015 2:33 PM, Daniel Mundall wrote:

    Hello Glen,

    Thanks for your thoughtful email, and I totally agree, I'm
    probably not experienced enough to make a radio yet.
    But I am familiar with SPI and have used it on other projects
    FlexScada. (http://flexscada.com/)
    As for the C-Bus, no where could I find it mentioned that it was
    SPI, but by the looks of it it seemed to be, which was why I was
    asking about it.
    And as you suggested, I am starting with the dev boards connected
    to some STM32's once that's all working maybe move to a single PCB.

    Anyway, I appreciate your patience and time answering my questions.
    Daniel VA7DRM

    PS. I've been doing over the air testing with David's Octave GMSK
    modem, we're having some fun with multipath now. :) Nothing that
    can't be fixed, but it's helping us prefect the demod. Multipath
    fades driving seem to be anywhere from 5-20Hz. Interesting how on
    the fringe the fading is much less (I assume that when the path
    losses are higher there are just less paths even making it to the
    RX.)

    Daniel Mundall

    On Sun, Feb 15, 2015 at 7:09 PM, glen english
    <g...@cortexrf.com.au <mailto:g...@cortexrf.com.au>> wrote:

        Hi Daniel

        Cbus/SPI is easy, you dont need a library BUT

        - but... if you need to ask about "what is SPI/Cbus, do I
        need a lib ?"  then you do not, in my opinion,  have enough
        experience to make this radio work properly..... these are
        very basic things and skillsets you MUST have.
        I'll be honest , while I think your endeavours are great, I
        dont think you have the skillsets to execute a PRODUCT,
        without a lot more rubber on the road.....  Certainly, this
        is a very good way to learn about the technology, but if you
        are planning to build and sell a radio, or have a radio club
        project, you need much much more understanding and experience
        to be able to execute such an endeavour successfully. Please
        dont take this the wrong way....Maybe I have misunderstood
        your intentions..

        Get yourself the eval board, or get some chips and build your
        own... you will need to learn everything and understand
        everything about each building block. This probably means
        bulding and prototyping each block of your radio one by one,
        and connecting it all together before even thinking about
        integrating it into one housing....

        There will be many issues for you to overcome that are not
        documented anywhere.

        What do I mean ?

        >>>>>I mean that the datasheets for a chip (this CMX chip and
        others) , and the application notes only ever have about HALF
        of what you really need to get the thing working. The other
        half is required to be learnt by the designer ! this requires
        alot of work and intuition.

        DC offsets are no issue....they are servoed out.

        I am certainly happy to continue to answer questions adlib.

        cheers

        On 16/02/2015 1:28 PM, Daniel Mundall wrote:

        Thanks, Glen.
        I agree we want to keep things simple.
        Since you've used this IC, maybe you could answer two of
        David's concerns. Is C-Bus a problem to interface to, or is
        it just something like SPI? Are there already lib's out there?
        And also has DC offset been an issue with your devices?
        Thanks,
        Daniel

        Daniel Mundall

        On Sun, Feb 15, 2015 at 6:19 PM, glen english
        <g...@cortexrf.com.au <mailto:g...@cortexrf.com.au>> wrote:

            Daniel , External VCO might be 80 cents.. but then it is
            something that has to be tracked in production.

            Don't set lofty goals for your 1st  version, just use
            the internal VCO.

            Don't promise too much..

            Over deliver.

            David's route is different and not comparable- you are
            talking about a standalone receiver.

            A crystal filter, for this application  type and
            performance, in my opinion, is to be avoided .
            They are expensive, need specific tuning/ production
            checks to get the shape right (otherwise you are wasting
            your time) .
            There is a place for crystal filters in receivers , but
            I can't think why it would be here.

            glen


            On 16/02/2015 1:08 PM, Daniel Mundall wrote:

            Hello Glen,
            Thanks so much for your email!

            I've been wondering which way to go with that, how much
            cost do you think an external VCO adds to the BOM?
            The dev boards I ordered should be here soon and I'll
            have to play around with that.
            David on the other hand is playing with a much more
            simple approach, if he can get it to work well, we may
            go that route. He has a basic mixer and xtal filter and
            then dealing directly with IF.

            Anyway thanks again for the warning.
            Daniel

            Daniel Mundall

            On Sun, Feb 15, 2015 at 12:24 AM, glen english
            <g...@cortexrf.com.au <mailto:g...@cortexrf.com.au>> wrote:

                Hi Daniel

                I think the CMX994 is a good choice for your radio,
                overall. But don't
                expect anything magnificent out of the receiver for
                1st 2nd and 3rd

                adjacent channels though- this should be strictly 
                assumed as a portable

                radio receiver, or 'average' mobile radio receiver.

                The reason for this is that the internal VCO is
                rather noisy, and as an
                integer N loop with a small step size, you are
                limited to quite narrow
                PLL bandwidths. (<2 kHz)  This has the effect of
                not correcting for the
                rather noisy VCO away from channel.

                An external VCO would improve it, and external LO
                again improve it.
                BUT !
                it is more than adequate for a basic receiver that
                can 'do all' at a
                good price.

                You would not do better for the same cost.

                I am using this chip on another project, but with
                an external local
                oscillator- a DDS . (Not a portable application)
                (and at a higher cost)

                footnote
                The advantage of an integer-N PLL is that is is
                relatively free of PLL
                generated spurs, whereas a fractional N PLL can
                operates at loop
                bandwidths, typically for this application , up to
                100kHz, which often
                translates into a very good adjacent phase-noise
                performance- at the
                expense of many (calculable) spurs. The integer N
                PLL will also often
                have  much lower power consumption.

                regards

                glen english VK1XX









                
------------------------------------------------------------------------------
                Dive into the World of Parallel Programming. The Go
                Parallel Website,
                sponsored by Intel and developed in partnership
                with Slashdot Media, is your
                hub for all things parallel software development,
                from weekly thought
                leadership blogs to news, videos, case studies,
                tutorials and more. Take a
                look and join the conversation now.
                http://goparallel.sourceforge.net/
                _______________________________________________
                Freetel-codec2 mailing list
                Freetel-codec2@lists.sourceforge.net
                <mailto:Freetel-codec2@lists.sourceforge.net>
                https://lists.sourceforge.net/lists/listinfo/freetel-codec2




            
------------------------------------------------------------------------------
            Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
            from Actuate! Instantly Supercharge Your Business Reports and 
Dashboards
            with Interactivity, Sharing, Native Excel Exports, App Integration 
& more
            Get technology previously reserved for billion-dollar corporations, 
FREE
            
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk


            _______________________________________________
            Freetel-codec2 mailing list
            Freetel-codec2@lists.sourceforge.net  
<mailto:Freetel-codec2@lists.sourceforge.net>
            https://lists.sourceforge.net/lists/listinfo/freetel-codec2



            -- 
            -

            Glen English
            RF Communications and Electronics Engineer

            CORTEX RF
            &
            Pacific Media Technologies Pty Ltd

            ABN 40 075 532 008

            PO Box 5231 Lyneham ACT 2602, Australia.

            au mobile :+61 (0)418 975077  <tel:%2B61%20%280%29418%20975077>  



            
------------------------------------------------------------------------------
            Download BIRT iHub F-Type - The Free Enterprise-Grade
            BIRT Server
            from Actuate! Instantly Supercharge Your Business
            Reports and Dashboards
            with Interactivity, Sharing, Native Excel Exports, App
            Integration & more
            Get technology previously reserved for billion-dollar
            corporations, FREE
            
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
            _______________________________________________
            Freetel-codec2 mailing list
            Freetel-codec2@lists.sourceforge.net
            <mailto:Freetel-codec2@lists.sourceforge.net>
            https://lists.sourceforge.net/lists/listinfo/freetel-codec2




        
------------------------------------------------------------------------------
        Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
        from Actuate! Instantly Supercharge Your Business Reports and Dashboards
        with Interactivity, Sharing, Native Excel Exports, App Integration & 
more
        Get technology previously reserved for billion-dollar corporations, FREE
        
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk


        _______________________________________________
        Freetel-codec2 mailing list
        Freetel-codec2@lists.sourceforge.net  
<mailto:Freetel-codec2@lists.sourceforge.net>
        https://lists.sourceforge.net/lists/listinfo/freetel-codec2



        -- 
        -

        Glen English
        RF Communications and Electronics Engineer

        CORTEX RF
        &
        Pacific Media Technologies Pty Ltd

        ABN 40 075 532 008

        PO Box 5231 Lyneham ACT 2602, Australia.

        au mobile :+61 (0)418 975077  <tel:%2B61%20%280%29418%20975077>  



        
------------------------------------------------------------------------------
        Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
        from Actuate! Instantly Supercharge Your Business Reports and
        Dashboards
        with Interactivity, Sharing, Native Excel Exports, App
        Integration & more
        Get technology previously reserved for billion-dollar
        corporations, FREE
        
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
        _______________________________________________
        Freetel-codec2 mailing list
        Freetel-codec2@lists.sourceforge.net
        <mailto:Freetel-codec2@lists.sourceforge.net>
        https://lists.sourceforge.net/lists/listinfo/freetel-codec2




    
------------------------------------------------------------------------------
    Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
    from Actuate! Instantly Supercharge Your Business Reports and Dashboards
    with Interactivity, Sharing, Native Excel Exports, App Integration & more
    Get technology previously reserved for billion-dollar corporations, FREE
    http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk


    _______________________________________________
    Freetel-codec2 mailing list
    Freetel-codec2@lists.sourceforge.net  
<mailto:Freetel-codec2@lists.sourceforge.net>
    https://lists.sourceforge.net/lists/listinfo/freetel-codec2



    -- 
    -

    Glen English
    RF Communications and Electronics Engineer

    CORTEX RF
    &
    Pacific Media Technologies Pty Ltd

    ABN 40 075 532 008

    PO Box 5231 Lyneham ACT 2602, Australia.

    au mobile :+61 (0)418 975077  <tel:%2B61%20%280%29418%20975077>  



    
------------------------------------------------------------------------------
    Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
    from Actuate! Instantly Supercharge Your Business Reports and
    Dashboards
    with Interactivity, Sharing, Native Excel Exports, App Integration
    & more
    Get technology previously reserved for billion-dollar
    corporations, FREE
    http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
    _______________________________________________
    Freetel-codec2 mailing list
    Freetel-codec2@lists.sourceforge.net
    <mailto:Freetel-codec2@lists.sourceforge.net>
    https://lists.sourceforge.net/lists/listinfo/freetel-codec2




------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk


_______________________________________________
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2


--
-
Glen English
RF Communications and Electronics Engineer

CORTEX RF
&
Pacific Media Technologies Pty Ltd

ABN 40 075 532 008

PO Box 5231 Lyneham ACT 2602, Australia.
au mobile : +61 (0)418 975077


------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
_______________________________________________
Freetel-codec2 mailing list
Freetel-codec2@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/freetel-codec2




























					Reply via email to