Synchronous AM demodulation (again...)

> Okay - I think I will miss something to answer but I'll do my best.

Read more of the datasheets and applicationnotes and you will
get a clearer picture what really affects performance of the reader.
It might be advisable to breadboard a U2270 from scratch
http://www.embeddedFORTH.de/temp/marc4_4.pdf ( text german )
just to get a feel for it.
There are books too. The one i can recommend:
http://www.amazon.de/RFID-Handbuch-Klaus-Finkenzeller/dp/3446220712
has a translation by Wiley:
http://www.amazon.com/RFID-Handbook-Fundamentals-Identification-Communication/dp/0470695064/ref=sr_1_1?ie=UTF8&qid=1286432749&sr=8-1

> Sure, there are a bunch of ways to get rid of the carrier. 

Well again: suppose you scale down the signal by a resistor divider
or a another coil that monitors the field. And the signal
goes to a 12 bit flash-A/D that is triggered by some additional
analog circuitry so it samples the 125kHz carrier at the peaks.
250k samples/sec now. Since the signal will probably be not
symmetric one would process the positive and negative peaks,
( two 125k sample streams ) seperatly. The digital signal
consists of the large DC-part ( "200V" ) and a small ripple that
looks like the Manchester-coded signal. A highpass/bandpass-filter
would remove the DC and the schmitt-trigger convert it binary.
One can do that for both streams ( contain the same signal )
and combine the result.
The A/D is obviously not needed: sample & hold will do.
After filtering the 8 bit A/D of a slow 8 bit controller
would do rest of the job.
That would not be limited to a 1V signal like the si-diode but
could handle the 0,25 signal too.
But as already said: another approach is to get signal up
to 1V by adjusting the frequency/coupling.

MfG  JRD

On 07-10-2010 @ 07:58:23 AlanWrench  
<stud.magraf@n_o_s_p_a_m.googlemail.com> wrote:

> Doesn't make it better ... Everyone of use was in the situation to be the
> newbee
I wasn't. At least I'd never tried to break the wall with my head.
Insted I was looking for a hammer.

> and no one would like a post like this ...
I would like.
It's very motivating information
which I couldn't probably find by myself.

-- 
Mikolaj

In article <C7KdnZdeHa1TpDDRnZ2dnUVZ_g6dnZ2d@giganews.com>,
 Vladimir Vassilevsky <nospam@nowhere.com> wrote:

> AlanWrench wrote:
> 
> > What would be a good signal conditioning before going into the ADC? 
> > We had some discussion today and we figured out two plans:
> 
> >  1) I will wire a transformer just to see how the signal at the output
> > looks 
> >     like
> >  2) When I have my stuff I will use the diode envelope detector and go
> > after 
> >     that into a ADC and from there to the DSP. If the whole thing is
> > working 
> >     with this and I have time I will see if the transformer version works
> > or 
> >     not
> > 
> > I am still looking for something to read like a book about the analog thing
> > (envelope detector, other solutions) ... anyone has a good idea for that?
> 
> Idiot

Hey  Vlad:

Please turn back on your "Am I being a douchebag?" filter.

Cheers,

Ken Prager

On 10/06/2010 10:45 PM, AlanWrench wrote:
>>
>>
>> AlanWrench wrote:
>>
>>> What would be a good signal conditioning before going into the ADC?
>>> We had some discussion today and we figured out two plans:
>>
>>>   1) I will wire a transformer just to see how the signal at the output
>>> looks
>>>      like
>>>   2) When I have my stuff I will use the diode envelope detector and go
>>> after
>>>      that into a ADC and from there to the DSP. If the whole thing is
>>> working
>>>      with this and I have time I will see if the transformer version
> works
>>> or
>>>      not
>>>
>>> I am still looking for something to read like a book about the analog
> thing
>>> (envelope detector, other solutions) ... anyone has a good idea for
> that?
>>
>> Idiot
>>
>>
>
> A very kind and smart way to greet new guys on the board ... Actually I
> don't know why I should be a idiot or rather why you call me one.
> Maybe because I use the time during waiting for my stuff to see why a
> transformer is bad or I am a idiot 'cause I am looking for books to read
> and inform myself (besides the ARRL Handbook and Understanding Digital
> Signal Processing).
>
> But okay. It's your opinion though.
> Cheers and thanks for helping me so far.

Don't feel like you have to encourage him when he gets this way.  Either 
he's forgotten what it's like to be a student, or he was really good at 
engineering and forgets all of the things that he was really bad at.

Think of it as an opportunity to be noble.

-- 

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

>> Okay - I think I will miss something to answer but I'll do my best.
>
>Read more of the datasheets and applicationnotes and you will
>get a clearer picture what really affects performance of the reader.
>It might be advisable to breadboard a U2270 from scratch
>http://www.embeddedFORTH.de/temp/marc4_4.pdf ( text german )
>just to get a feel for it.
>There are books too. The one i can recommend:
>http://www.amazon.de/RFID-Handbuch-Klaus-Finkenzeller/dp/3446220712
>has a translation by Wiley:
>http://www.amazon.com/RFID-Handbook-Fundamentals-Identification-Communication/dp/0470695064/ref=sr_1_1?ie=UTF8&qid=1286432749&sr=8-1
>
>> Sure, there are a bunch of ways to get rid of the carrier. 
>
>Well again: suppose you scale down the signal by a resistor divider
>or a another coil that monitors the field. And the signal
>goes to a 12 bit flash-A/D that is triggered by some additional
>analog circuitry so it samples the 125kHz carrier at the peaks.
>250k samples/sec now. Since the signal will probably be not
>symmetric one would process the positive and negative peaks,
>( two 125k sample streams ) seperatly. The digital signal
>consists of the large DC-part ( "200V" ) and a small ripple that
>looks like the Manchester-coded signal. A highpass/bandpass-filter
>would remove the DC and the schmitt-trigger convert it binary.
>One can do that for both streams ( contain the same signal )
>and combine the result.
>The A/D is obviously not needed: sample & hold will do.
>After filtering the 8 bit A/D of a slow 8 bit controller
>would do rest of the job.
>That would not be limited to a 1V signal like the si-diode but
>could handle the 0,25 signal too.
>But as already said: another approach is to get signal up
>to 1V by adjusting the frequency/coupling.
>
>MfG  JRD
>
>

I know it sounds dumb but I have to ask if I got this right:
Do you mean that I should take/need 2 ADCs to monitor the signal?

I added something to my original plan: A analog circuit for peak detection
after the diode envelope detector to get the peaks. But I have to do some
research on this, never heart/used something like this before but it sounds
like that it makes sense ;)

> Do you mean that I should take/need 2 ADCs to monitor the signal?
No, you have only one signal so one A/D with 250kHz samplerate will
do for the 125kHz carrier if you hit the peaks.

Thats a fairly typical industrial synchronous demodulator:
http://www.seekic.com/uploadfile/ic-circuit/200972452849422.gif
It uses a zero-crossing detector ( much easier then a peak-detector )
to generate a digital signal that controls the switch.
In this implementation the amplifier is switched inverting/
noninverting. The following RC-lowpass smooths the rectified
sinewave to DC.

But depending of the way the carrier is generated i would not
take it for granted that the sinewave is symmetric/clean.
Therefore combining both peaks like in the synchronous demodulator
may be counterproductive.

MfG  JRD

>> Do you mean that I should take/need 2 ADCs to monitor the signal?
>No, you have only one signal so one A/D with 250kHz samplerate will
>do for the 125kHz carrier if you hit the peaks.
>
>Thats a fairly typical industrial synchronous demodulator:
>http://www.seekic.com/uploadfile/ic-circuit/200972452849422.gif
>It uses a zero-crossing detector ( much easier then a peak-detector )
>to generate a digital signal that controls the switch.
>In this implementation the amplifier is switched inverting/
>noninverting. The following RC-lowpass smooths the rectified
>sinewave to DC.
>
>But depending of the way the carrier is generated i would not
>take it for granted that the sinewave is symmetric/clean.
>Therefore combining both peaks like in the synchronous demodulator
>may be counterproductive.
>
>MfG  JRD
>

Not directly an answer for my question but interesting to see how a
demodulator analog works. Why the switched capacitor?

And is the idea to digitalize after the envelope detector and a
lowpassfilter inside the dsp dump or not? I'm not sure right now 'cause I
need something like a Schmitt-Trigger after the filter to have clear 1 and
0 values ... Therefore I could just use a microcontroller ... Dump
requirements -.-

>> Thats a fairly typical industrial synchronous demodulator

> Not directly an answer for my question 

If your problem is "can?t use a diode because 1V signal is
needed and the spec says 0,25V is available" then the obvious
next step is a synchronous demodulator that replaces the diode
by an analog switch.

> Why the switched capacitor?

There is no switched capacitor. Its a switched amplifier gain
+1 or -1. Works as a full-wave rectifier.
The advantage of analog circuits is that they have standard
symbols and are easier to grasp then flowcharts for software.
Implementation is most often possible analog or digital.

> Therefore I could just use a microcontroller ... Dump
> requirements -.-

If this is your Bachelor Thesis what do you intend to do
afterwards ? If you go into industry the obvious step would
be to join a company that builds this type of circuits. But
then it should better work, be technically reasonable.
If on the other hand you go for academia a DSP will not
do, thats old 80ies technology. You will have to throw
in at least a FPGA and a Kalman filter.

MfG  JRD