DC that varies with frequency. Lowpass and slice. It's just a>classic discriminator with a high-Q resonator.
Yes, John Larkin, SHOW us ;-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
make DC that varies with frequency. Lowpass and slice. It's just a >classic= discriminator with a high-Q resonator.
I will kill everything on this subject except doing getting to an IF frequency and then building a proper digitized data stream to process it. The data rate is 1 MHz, I don't think a PLL is going to work. I don't think a 978 Ghz tuned ceramic resonator that can theoretically cause a phase blip in conjunction with a discriminator that needs probably on the order of 0.5 us of delay will work either. Jeesh
DC that varies with frequency. Lowpass and slice. It's just a>classic discriminator with a high-Q resonator.
One-trick pony alert.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics Electro-optics Photonics Analog Electronics
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
The guy needs to build a real receiver, not a science project. I suppose there are other tricks he could try, but it would be more like a direct conversion receiver with a real I/Q data stream and not some kind of dumb ass analog PLL circuitry or better yet, a discriminator built at 978MHz that provides a whopping 45 degrees of phase difference providing that you can come up with 0.5 usec of delay at
978 MHz to pull it off. Oh -I know - why not suggest he put 300 feet of coax in the box in order to get the delay?
will>make DC that varies with frequency. Lowpass and slice. It's just a>classic discriminator with a high-Q resonator.
You will NEVER see a schematic of a TRULY FUNCTIONAL circuit posted by Larkin.
And Hobbs is on Larkin's payroll.
So you have a paid mutual admiration society in play here ;-) ...Jim Thompson
--
| James E.Thompson, CTO | mens |
| Analog Innovations, Inc. | et |
| Analog/Mixed-Signal ASIC's and Discrete Systems | manus |
| Phoenix, Arizona 85048 Skype: Contacts Only | |
| Voice:(480)460-2350 Fax: Available upon request | Brass Rat |
| E-mail Icon at http://www.analog-innovations.com | 1962 |
I love to cook with wine. Sometimes I even put it in the food.
FYI, I appreciate all the replies. I realize that opinions vary and that some may be considered esoteric or otherwise. In any case I am still pursuing this and, most importantly, I am learning a great deal.
will>make DC that varies with frequency. Lowpass and slice. It's just a>classic discriminator with a high-Q resonator.
Nice try, Jim, but you won't catch this trout with that mouldy bait.
Re admiration: I admire you too, when you're not in pseudo-Allison pissing-contest mode. For that matter, I admire Phil A.--he knows his field cold, and can be pretty helpful when he's on his meds. I also admire his courage. It can't be easy, living with that disorder of his, whatever it is. It's bad enough having a bad back or a trick knee, without having your mind act up on you.
Brent seems to think that you have to have a delay line to get a phase slope, so his scorn falls completely flat. A one-trick pony, as I said.
(To Brent: FSK has been in use since long before there was packaged logic, and probably since before there were transistors. World War II IFF systems used it, unless I'm mistaken, and so did second-generation radars.
Resonances make nice steep phase slopes with frequency--that's how slow light and Pound-Drever-Hall laser locking work, for a couple of very modern examples. Atomic resonances in rubidium vapour can slow light down by at least six orders of magnitude. (I'm not sure what the record is at present--probably further than that.)
Crystal discriminators have been used for narrow FSK and NBFM since forever. Any old Radio Amateur's Handbook tells all about it.
And then there's the classical Foster-Seeley discriminator, which has been in very wide use since the 1920s. In that one, the zero degree input signal is summed with the +-pi/2 phase shifted signals from the two ends of a center-tapped tank circuit, and then the two combinations are rectified with diodes. Detuning changes the phase shift in the tank, which makes one of the rectified outputs go up and the other one go down, so that their difference is a zero-centered frequency deviation signal. You can also demodulate FM with an AM radio, just by mistuning it a little. No delay lines. Not rocket science.
Using an I/Q demodulator on a signal with a modulation index of 0.3 is a waste of good circuitry. The I signal contains no information to speak of; it varies from 1.0 all the way down to cos(0.3)=0.955.)
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics Electro-optics Photonics Analog Electronics
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
That's one thing to do -- then you find that your IF is high enough that your demodulation is still inconvenient, so you have to put in _another_ IF that's lower yet.
That's radio design.
--
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
You are right that I haven't got my copy of Gardner in Sydney, but wrong in assertig that I've misunderstood him.
For me, the OP's PLL is trying to track the frequency of a signal that runs either at 1.0003GHz (a one) or 9.9997GHz (a zero), including the periods when the frequency ramps up or down between these two states. A second order phase locked loop won't track perfectly through the ramp periods.
But a second order loop won't do it perfectly.
Certainly true - we've all been telling the OP to forget the pure PLL approach anyway.
Perhaps.
Make the control loop slow enough, and insert a sufficiently long string of zeros into the data stream and the phase error can mount up to more than 20 degrees - this figure is implementation dependent, and you haven't actally specified the implementation for which your claim is true.
Only if the data being sent is guaranteed free of long strings of ones or zeros - if the tracking LO gets close to 1.0003GHz or 9.9997GHz any noise on the transmitter's frequency can turn into data errors.
The ICCC modem protocols always specified that the data stream was multplied by a longish pseudo-random binary sequence to minimise this kind if problem.
Why is that important? He just needs to tell a zero from a one, with a good BER. During a burst, the phase error will be far from zero.
Doesn't need to. One or two start bits would be fine, or Manchester encoding, or.... If you have a sufficiently long string of zeros or ones, any PLL above first order is going to zero out the static phase error.
In the OP's case, since his phase deviation is so small, he could just AC couple the output of the PD.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics Electro-optics Photonics Analog Electronics
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
Huh? The I/Q demodulator that I have modeled for a UAT receiver contains a lot more information than you suggest. when the IF frequency gets sampled/converted to an I Q data stream the IQ data stream will contain either +300 KHz data or -300 KHz data. Without the I/Q data stream you cannot tell if the demodulated frequency is + or - 300 KHz. So I think that the I stream is much more significant than you state above.
You and Larkin have not explained how well your direct FM discriminator holds up to a frequency that is 1 MHz off and 30 dB higher in amplitude.
The poster wants to get practical ideas for building a receiver. You two are suggesting stuff that is not practical.
will>make DC that varies with frequency. Lowpass and slice. It's just a>classic discriminator with a high-Q resonator.
end_of_the_skype_highlighting
Complete gobbledegook. You obviously have no idea how an FM discriminator works. Frequency below centre --> negative voltage; frequency above centre --> positive voltage. Capiche?
We haven't got anywhere near amplitude or SNR yet. AFAIK the OP and I are the only ones who have even tried to attach meaningful numbers to the problem.
Your lack of experience makes your opinion less than persuasive.
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics Electro-optics Photonics Analog Electronics
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
What is the consequence of momentary signal losses e.g. due to receiver front end blocking due to very strong signals at nearby frequencies ?
How fast does the demodulator and frame synchronizer relock after such events ?
Since you appear to work inside the aviation frequency band, there can momentarily be quite strong signals at nearby frequencies.
A plane passing over you at an altitude of a few hundred meters and as a direct line of sight propagation condition exists, so quite strong out of band signals can be received, even if the transmitter on the plane is quite low power.
Fixed primary radars generate huge pulses of radiated power. Even if your receiver is not directly subjected to these signals, a plane flying over you can reflect the radar signals towards your receiver for a few seconds.
You plan to use SAW filters in the front end. What is the combined bandwidth of these filters as measured at say -3 dB, -26 dB and -60 dB points ?
If it is something in the order of 30 MHz as some have speculated, I do not understand how a PLL could operate reliably directly at 1 GHz in such hostile RF environment. Then there are other issues with the PLL as such frequencies as others have pointed out.
I would suggest mixing the signal down to 50-100 MHz, filter it to 1 MHz bandwidth and then do the demodulation with whatever means, if continuous reliable operation is required.
Excuse me,but i made a resonator tunable in the FM band and it was maybe 3" diameter and 5" tall; estimated Q 300 or so. The technique used gives a Q roughly related to the frequency. At 1Ghz, the resonator would be rather small. The IEE paper was old in them thar daze (30 or so years ago when i did the "cavity"). The structure can be described and analyzed in two different ways: (1) shielded coil, and (2) waveguide with slow-wave structure. Take a hollow cylinder of conductive material (i used copper); you may put conductive ends on it to decrease losses some. Take a coil (nominally self-supporting) that is about 2/3 diameter WRT the inside diameter of the cylinder; align axis of both and length maybe 3/4 of the cylinder (all guess from inexact memory), one end soldered to inside of cylinder. Moving one of the end / closing caps will allow tuning. I had I think 3.5 turns of 1/4 inch copper tubing for the coil / slow wave structure.
Not if he sets himself up a phase-locked loop that is fast enough, and of high enough order, to alow the controlled oscillator to track the signalling source - which is roughly what he seemed to be proposing in his original post.
If you've designed the system right in the first place, and the OP clearly hadn't thought through the system demands when he made his original post.
or.
Sure, but you are assuming - without having made it explicit - that you can get away with tracking the long term average frequency of the singal source.
or
ke
ou
n
a
This isn't actually a good idea if the data contains arbitrarily long streams of ones or zero, for the reason I've already spelled out. It doesn't take much hysterisis in the bit slicer to make AC coupling tolerably safe, but you do need some.
-- Bill Sloman, Nijmegen (but in Sydney at the moment, just enjoying a first sip of Scarborough's 2007 Chardonnay while admiring our view if Sydney Haarbour at night).
And you have no idea how I Q demodulatin works. Pretty much every single modern radio receiver uses it these days. Even in older technologies in the aviation world such as VOR and Com radios theoverwhelming trend is to use it. Can you actually imagine a coaxial resonator cut to 978 MHz with no selectivity filter anywhere in the system as being a viable solution to a UAT receiver? Did you know that the UAT receiver needs to process bursts of data from multiple aircraft (translate this to each burst has a different carrier frequency) and each burst of data lasts a couple of hundred usec? How is a PLL going to lock to each carrier frequency? By the time it locks, the data burst is gone. The UAT specification is written relatively recently, so the assumption made by the specification is that the receiver would be implemented using modern processing techniques. These modern techniques allow for more efficient use of data transmission.
With the I Q data stream as long as each carrier frequency from the various aircraft is close to its assigned carrier the receiver can instanly determine whether the modulated signal is above or below the carrier frequeny by 300 KHz. (And the I data stream is very important for this purpose).
You do not know my experience level. You might argue that my arguments are not persuasive, but my experience level is adequate.
This sounds just like an ordinary helical resonator that you are describing. Based on decades old design nomograms, with those dimensions, you should be able to build a helical resonator for 100 MHz, with an unloaded-Q of 1500. With a loaded-Q of 300, the insertion loss would be about 1.5 dB.
With Ql of 300 at 100 MHz the -3 dB bandwidth would be 330 kHz. This will quite comfortably pass the FM broadcast signal, but adjacent channels, the attenuation is not that great.
Using the same nomograms at 1000 MHz and targeting for Qu of 1500, the shield diameter would have to be 25 mm, quite a beast compared to current SMDs :-). If Ql is again set to 300, this would give 3.3 MHz bandwidth, quite a lot more than the required 1 MHz and the attenuation at say +/- 30 MHz would not be too spectacular.
If the diameter would have to be reduced to 10 mm, the Qu would drop to 600 and with Ql=300, the insertion loss would be 6 dB (acceptable after the first LNA stage).
Without some regenerative tricks, a single helical resonator would not be sufficient in front of the demodulator at 1000 MHz.
Using a PLL makes NRZ codes with no start bits somewhat difficult, just like an AC-coupled baseband amplifier, and for the same reason. The solution is the same too. It's just a DC restore problem, and has has nothing to do with hysteresis--you just need enough linear range of the PD that it doesn't matter whether you're going from 0 to +20 degrees, 0 to -20 degrees, +10 to -10, or anywhere in between. It won't get outside that range unless the loop unlocks.
Given that you have to have a carrier in order for the PLL to work at all, none of this is very hard.
This is all pretty recherche stuff for a guy who thought that the 4046 PD2 wasn't a phase detector, and blamed it on Gardner. ;)
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics Electro-optics Photonics Analog Electronics
55 Orchard Rd
Briarcliff Manor NY 10510
845-480-2058
email: hobbs (atsign) electrooptical (period) net
http://electrooptical.net
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.