IQ demodulator PFD

In message , dated Thu, 31 Aug

2006, colin writes

If I've understood you correctly, I don't think there is even a difficult way of doing it. You want to generate a 2.200455 GHz signal by one stage of mixing? Even with quadrature mixing, I think it's pretty ambitious.

--
OOO - Own Opinions Only. Try www.jmwa.demon.co.uk and www.isce.org.uk
2006 is YMMVI- Your mileage may vary immensely.

John Woodgate, J M Woodgate and Associates, Rayleigh, Essex UK

Hello Colin,

At the risk of having misunderstood you there could be two methods:

a. Do the full Hilbert thing after the mixer. 90 phase shifters and all. Doesn't have to be very precise since you'd only have to see in which sideband the RF shows up. IOW check where is it stronger. So you probably could get away with a few degrees of phase error, reducing complexity in the Hilbert. If you aren't familiar with this topic look up "phase method" under SSB in a ham radio book (the ARRL Handbook is usually a really good deal).

b. Do the locking uC-assisted. Have it scoot up the LO a wee bit. If the beat becomes lower then the RF was above, else it was below. Then track. Have it do a sanity check once in a while if you expect wild changes.

--
Regards, Joerg

http://www.analogconsultants.com

input,

leads I,

a

than

Hmm maybe I need to explain it a bit better, perhaps I'd better do a block diagram,

______ ______________ [RF vco]---| |Q------>[IQ mod]-->| |--- ph err |lt5515| | | | bit of logic | [LO vco]---|______|I------>[IQ mod]-->|______________|--- f error | | | | [455khz*2]--[divide/2]--IFQ--' | '-not-[divide/2]--IFI-----'

The RF is sweep scanned so the LO needs to keep up. The idea is to try and do the LO tuning (almost) instantaneously, hopefully this will solve the problem with the vcos pulling each other into lock once they get within 1Mhz Its for my heterodyne Lidar.

Colin =^.^=

[snip]

I'd bet it won't work ;-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

into

maybe, but how much ?

Colin =^.^=

Draw up schematics for the "IQ Mod" and "bit of logic" blocks, then I'll decide how huge the amount ;-)

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC\'s and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

So you want to see if its gona work or not b4 you place your bet ahaha, thats making it too easy but oh well, ..

Well the [IQ mod] are just typical IQ modulators with 2 ordinary modulators summed together, ie IFQ.Q + IFI.Q and IFQ.I + IFI.I

When the LO=RF+IF, the I and Q signals are 455khz and so when mixed with

455khz the resultant inputs to the bit of logic should be steady value. when the phase/frequency relationship of the RF-LO changes they should reflect the error from the desired phase/frequency. Exactly how its reflected im not 100% sure, ive only imagined this in my head so far, wich is a little bit tricky for so many stages of mixing.

The [Ph err] just converts the two inputs as if they were quadrature related (wich I'm hoping they still are) into a phase. However its still very much all on the drawing board still.

To measure small phase error this is quite easy as its just the eqv of one input, for larger errors both inputs cld be combined, a.(b+dc) wich would give a wider range albeit crude at the extremes.

The [F err] just detects wich input is leading with a flip flop, obviously only makes sense when the phase error is changing quickly enough over > 360'

If it falls into a steady state of LO=RF-IF its liable to get confused, however this shouldnt be a stable steady state, as the slope of the [ph err] will be backwards.

Colin =^.^=