Synchronous AM demodulation (again...)

Hey there

I am the new one and I am from Germany. Currently I am in the USA writing
my Bachelor Thesis.

I have the following issue:
I have an AM modulated signal with a carrier voltage about 100V to 500V and
a message voltage of about 250mV. This signal is out of a coil which is
part of an RFID reader (more MCID - magnetic coupling... that's why the
voltage is so high). The carrier has a frequency of 134.4khz and the
message 4khz.

My job is it to demodulate the signal digitally (if possible) and do some
work with it to get the information out.

My idea to get this signal into a DSP was just to take a transformer to put
the voltage down to 5V which makes the message really small. Therefore I
thought on taking a 16bit DSP and do some oversampling (I thought on 1 to
2MHz concerning the carrier frequency to improve the resolution and noise
ratio).

I read a ton of articles in this forum and the internet and it's a bit like
looking for a needle in a haystack.
I think the best way to do this is this
http://www.dspguru.com/dsp/tricks/magnitude-estimator way, isn't it?.
I am using a C5515 from TI so PLL would be possible. But I don't know if I
need PLL because I am producing the signal for the coil with the dsp (a
134.4khz rectangular signal works fine). So I know the frequency and I can
produce a internal signal with this timer, too. And a correct knowledge
about the carrier is a good thing I think.

Therefore I am looking for books to reference which are explaining this...
Otherwise I would have to invent it new... I cannot just use something out
of the internet without saying why and profing it with a source (book).
I hope someone can help me :)

Greetings

> I am from Germany.
de.sci.electronics would be an alternative.

> demodulate the signal digitally (if possible)
> ... get the information out.

Have look at the datasheet of the old
Telefunken / Atmel U2270B reader IC.

> take a transformer to put the voltage down to 5V 
Due to the high voltage a simple diode is easier,
its demodulated then. A 4pole Butterworth
/ Bessel LP-filter follows. Then a schmitt-trigger
to get it digital. The manchester decoder is
done in the 8 bit microcontroller.

MfG  JRD

AlanWrench wrote:
> Hey there
> 
> I am the new one and I am from Germany. Currently I am in the USA writing
> my Bachelor Thesis.
> 
> I have the following issue:
> I have an AM modulated signal with a carrier voltage about 100V to 500V and
> a message voltage of about 250mV. This signal is out of a coil which is
> part of an RFID reader (more MCID - magnetic coupling... that's why the
> voltage is so high). The carrier has a frequency of 134.4khz and the
> message 4khz.
> My job is it to demodulate the signal digitally (if possible) and do some
> work with it to get the information out.
> My idea to get this signal into a DSP was just to take a transformer to put

[...helpless bleat skipped...]

Microchip has a bunch of good appnotes on RFID system design. Start from 
there.


Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

>> I am from Germany.
>de.sci.electronics would be an alternative.
>
>> demodulate the signal digitally (if possible)
>> ... get the information out.
>
>Have look at the datasheet of the old
>Telefunken / Atmel U2270B reader IC.


I had a look at this IC. The U2270B could redundantize the DSP and
everything. But this is not what my requirements list is telling me for my
thesis. Therefore I don't think that I can use this IC. But I have an eye
on this.

Why not digitally? It seems that you will tell me that it will not work to
demodulate the signal inside a DSP 'cause you write nothing about it. If
so, tell me why not.

>
>> take a transformer to put the voltage down to 5V 
>Due to the high voltage a simple diode is easier,
>its demodulated then. A 4pole Butterworth
>/ Bessel LP-filter follows. Then a schmitt-trigger
>to get it digital. The manchester decoder is
>done in the 8 bit microcontroller.
>
>MfG  JRD
>

Yep, it is possible. Actually this is the actual version of the RFID
reader. But it should be digital with a DSP therefore I can not use a 8bit
microcontroller or a didoe envelope detector. By the way: it is biphase
coded.

And I can not see any problems with a transformer and 100V to 500V. At
least there are transformers for higher voltages. I could imagine a problem
with the small message signal, though.

AlanWrench wrote:


>>Have look at the datasheet of the old
>>Telefunken / Atmel U2270B reader IC.
> 
> 
> I had a look at this IC. The U2270B could redundantize the DSP and
> everything.
> Why not digitally? 

What is the point? Digital is going to be about 100 times more expensive 
and 10 times more complex.

> It seems that you will tell me that it will not work to
> demodulate the signal inside a DSP 'cause you write nothing about it. If
> so, tell me why not.

Sure you can make a right-angled toothbrush to clean teath from behind a 
corner. It could work but for some reason they don't make curved 
toothbrushes.

> And I can not see any problems with a transformer and 100V to 500V. At
> least there are transformers for higher voltages. I could imagine a problem
> with the small message signal, though.

You didn't try to accomplish anything yet; that's why you don't see any 
problems.

VLV

>
>
>AlanWrench wrote:
>
>
>>>Have look at the datasheet of the old
>>>Telefunken / Atmel U2270B reader IC.
>> 
>> 
>> I had a look at this IC. The U2270B could redundantize the DSP and
>> everything.
>> Why not digitally? 
>
>What is the point? Digital is going to be about 100 times more expensive 
>and 10 times more complex.

True. But to be honest it wasn't my idea, I just choose the project.


>> And I can not see any problems with a transformer and 100V to 500V. At
>> least there are transformers for higher voltages. I could imagine a
problem
>> with the small message signal, though.

>You didn't try to accomplish anything yet; that's why you don't see any 
>problems.

Yep I didn't. Unfortunately I don't have the time to test several ways. The
transformer would have been one thing but because of that I wrote in this
forum to get this idea ... kind of accomplished or destroyed ;)

I am looking at the microchip site and see if I can find something
helpfull. I also will see if I can use this IC or not.

Anyhow: Can someone tell me a book about the demodulation with complex
numbers? And can someone explain me the problem with the transformer?

Cheers

On 10/06/2010 07:37 AM, AlanWrench wrote:
>>> I am from Germany.
>> de.sci.electronics would be an alternative.
>>
>>> demodulate the signal digitally (if possible)
>>> ... get the information out.
>>
>> Have look at the datasheet of the old
>> Telefunken / Atmel U2270B reader IC.
>
>
> I had a look at this IC. The U2270B could redundantize the DSP and
> everything. But this is not what my requirements list is telling me for my
> thesis. Therefore I don't think that I can use this IC. But I have an eye
> on this.

If you have a solution that's well established in the analog world and 
you want to implement it in the DSP world, it is smart to fully 
understand the analog implementation of the solution.  With luck just 
simulating the analog implementation in DSP will be a pretty good 
solution, but even if that doesn't get you there, when you do the 
mathematical analysis of the analog solution you may find your "DSP 
answer" staring you in the face.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

On 10/06/2010 08:04 AM, Vladimir Vassilevsky wrote:
>
>
> AlanWrench wrote:
>
>
>>> Have look at the datasheet of the old
>>> Telefunken / Atmel U2270B reader IC.
>>
>>
>> I had a look at this IC. The U2270B could redundantize the DSP and
>> everything.
>> Why not digitally?
>
> What is the point? Digital is going to be about 100 times more expensive
> and 10 times more complex.

If the prof wants it done with a DSP then that's the "right" solution, 
regardless of how it may survive in the commercial world.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

On 10/06/2010 04:19 AM, AlanWrench wrote:
> Hey there
>
> I am the new one and I am from Germany. Currently I am in the USA writing
> my Bachelor Thesis.
>
> I have the following issue:
> I have an AM modulated signal with a carrier voltage about 100V to 500V and
> a message voltage of about 250mV. This signal is out of a coil which is
> part of an RFID reader (more MCID - magnetic coupling... that's why the
> voltage is so high). The carrier has a frequency of 134.4khz and the
> message 4khz.
>
> My job is it to demodulate the signal digitally (if possible) and do some
> work with it to get the information out.
>
> My idea to get this signal into a DSP was just to take a transformer to put
> the voltage down to 5V which makes the message really small. Therefore I
> thought on taking a 16bit DSP and do some oversampling (I thought on 1 to
> 2MHz concerning the carrier frequency to improve the resolution and noise
> ratio).
>
> I read a ton of articles in this forum and the internet and it's a bit like
> looking for a needle in a haystack.
> I think the best way to do this is this
> http://www.dspguru.com/dsp/tricks/magnitude-estimator way, isn't it?.
> I am using a C5515 from TI so PLL would be possible. But I don't know if I
> need PLL because I am producing the signal for the coil with the dsp (a
> 134.4khz rectangular signal works fine). So I know the frequency and I can
> produce a internal signal with this timer, too. And a correct knowledge
> about the carrier is a good thing I think.
>
> Therefore I am looking for books to reference which are explaining this...
> Otherwise I would have to invent it new... I cannot just use something out
> of the internet without saying why and profing it with a source (book).
> I hope someone can help me :)

Have you googled "synchronous AM"?  Have you read up on synchronous AM 
demodulation in the ARRL handbook (that's always the go-to place for any 
communications technology that may intersect with amateur radio)?

Monolithic ADCs often get noisier as their maximum sampling rate goes up 
-- do you gain anything using a really fast ADC sampled at 1MHz vs. a 
slower ADC sampled at 100kHz?

Your signal is, indeed, teeny compared to your carrier.  Is there any 
way that you can subtract out your carrier before the ADC step, or is 
that "not digital enough" for your prof?

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

 > By the way: it is biphase coded.

Manchester goes by many names ...

> The U2270B could redundantize the DSP and everything. 

Thats a version were the U2270 is replaced by standard
components:
http://www.embeddedFORTH.de/temp/r.pdf  ( text is in german )
The MC68HC908 8 bit controller generates the 125kHz via timer.
And samples the analog signal via 8 bit A/D. The schmitt
trigger and some ALC-typ adjusting of switching thesholds is
done in software.
The filter is analog, the MC68HC908 was not fast enough
for that. Could be software, but probably not a reason to
switch to DSP, rather another faster controller.

> But this is not what my requirements list is telling me for my
> thesis. 

In theory a fancy demodulator could be useful to extend the
range of the reader by detecting very weak signals. But these
RFIDs are powered by the reader. If they do not get enough energy
they will not send a weak signal but will be unable to send any
signal.
Adjusting the frequency to the LC-component tolerance of the RFID
can be usefull to improve the strengh of the signal, but isn?t
related to demodulation directly.

> And I can not see any problems with a transformer and 100V to 500V. At
> least there are transformers for higher voltages. I could imagine a problem
> with the small message signal, though.

A 200V signal with 1V load modulation is easily handled by the
si-diode circuit. 0,25V not so. Schottky isn?t that usefull at high
voltages, so 0,5V would be hard too.
So you could scale the signal down and feed it via A/D-converter
to a DSP.
8 bit would not be enough, probably 12 bit or more. Samplerate
2x125kHz = 250kHz, some analog circuit that makes sure it samples
at the peaks would be usefull.
Remains only the real world problem: would a marginaly
improved performance be relevant relative to the more complex
circuit ?

MfG  JRD