Thoughts on locking a Si570???
KD6RF
Hi All,
I have recently become interested in SDR, as well as weak signal modes.
Along with my work in antenna design (I will be offering an ultra wideband, TSA based 144MHz – 3.5 GHz antenna this year, once I complete mechanical design and test) ,SDR and weak signal have really renewed my interest in ham radio.
I have been reading with interest the progress being made toward accurizing existing radios here in this forum..
So, a few questions to those with more practical experience in PLL implementation –
I am working with the Si570 chip, for SDR, with an eye toward stable/accurate operation. As you may know, the Si570 is a cool chip, but lacks any straightforward method for accurizing.
One possible method:
Use the temperature/frequency curve as the control input. Assuming that it is monotonic, one can imagine setting the chip temp, and therefore it’s ref freq, by changing it’s temperature.
Naturally, like many of my ideas, this has been done before, for example in laser frequency control like this -à http://connection.ebscohost.com/c/articles/9780618/thermal-phase-locked-loop-frequency-stabilization-internal-mirror-he-ne-lasers-0-633-m
I can imagine this working well for fixed frequency control of the Si570, where relatively long lock-in-times, dictated by long thermal time constants, are not an issue.
Does this sound like an idea worth pursuing?
Assuming fixed frequency operation works well, what are your thoughts on a frequency agile system, i.e. tuning up and down the band in a normal fashion. Would you think that the phase discontinuities that would almost certainly crop up when the freq is changed, and resultant loss of lock , would be a show stopper for normal non-WSPR OPERATION?
This would be quite a project, and hope you will share your thoughts on it’s practicality before I start blindly trudging through the design process!
Glenn Elmore
Hello and welcome to the 2 meter wspr group! I'm very glad to have you interested and involved. We are clearly interested in many of the same things.
I'm not an expert on the SiLabs chipsets, though I use some of them in SDRs and there is one in the OEM GPS we're using for the external PLL project.
Early on, I did consider trying to use on of the DDS approaches to generate accurate, referenced frequencies for WSPR and other uses. At that time, my conclusion was that the digital word length in almost all of these available chip sets was too short. I decided that even with the recognition that the Icom radios I was considering, themselves have a resolution granularity and thus don't guarantee precise output. I measured my IC706 at about 20 milli-Hz steps, if I remember right.
The other issue, particular to the SI57x chips, is that they use a screwy reference 114+ MHz or something and, worse, don't make it externally accessible. This means that one would have to lock the main output rather than the chips reference. One of the models, I forget which, does have a "fine tuning" input that can be used to generate fine frequency control but without a means of accessing the reference it controls, making the whole chip accurate and useful in the normal way becomes a problem.
Also, the step size on these is too course. They have really nice performance and rather good phase noise at most offsets for such an inexpensive solution but, as I remember, steps at HF are on the order of Hertz rather than a thousand times smaller than that.
For these reasons, I moved to different approaches in locking up my Icom. As I show on the pages, linked to from another thread, I used both direct replacement/injection of the radio's master oscillator and also the phaselock of it. Phaselock has the advantage of being limited by the PLL loop bandwidth and thus is less "invasive" and less likely to compromise the radio's spectral purity. For those modifying radios not having suitable test equipment, it seems the safer approach.
Another advantage of locking a system to a more standard reference, e.g. 10 MHz, is that is useful for locking not only all LOs in the entire system, which could be several if one has a microwave station (see my very old and hard to read Ham Radio magazine article at http://www.sonic.net/~n6gn/hr88/articles.html ).�
I don't want to discourage experimentation but I did consider various approaches, including the temperature control, and decided it didn't have the return to merit the effort.� You may think of a way to avoid some of these shortcomings and make it viable. They certainly are nice little devices!
Glenn n6gn
� Hi All,�
I have recently become interested in SDR, as well as weak signal modes.
�
Along with my work in antenna design (I will be offering an ultra wideband, TSA based 144MHz � 3.5 GHz antenna this year, once I complete mechanical design and test) ,SDR and weak signal have really renewed my interest in ham radio.
�
I have been reading with interest the progress being made toward accurizing existing radios here in this forum..
�
So, a few questions to those with more practical experience in PLL implementation �
�
I am working with the Si570 chip, for SDR, with an eye toward stable/accurate operation.� As you may know, the Si570 is a cool chip, but lacks any straightforward method� for accurizing.
�
One possible� method:
Use the temperature/frequency curve as the control input.� Assuming that it is monotonic, one can imagine setting the chip temp, and therefore it�s ref freq, by changing it�s temperature.
�
Naturally, like many of my ideas, this has been done before, for example in laser frequency control like this -�� http://connection.ebscohost.com/c/articles/9780618/thermal-phase-locked-loop-frequency-stabilization-internal-mirror-he-ne-lasers-0-633-m
�
I can imagine this working well for fixed frequency control of the Si570, where relatively long lock-in-times, dictated by long thermal time constants, are not an issue.
�
Does this sound like an idea worth pursuing?
�
Assuming fixed frequency operation works well, what are your thoughts on a frequency agile system, i.e. tuning up and down the band in a normal fashion.� Would you think that the phase discontinuities that would almost certainly crop up when the freq is changed, and resultant loss of lock , would be a show stopper for normal non-WSPR OPERATION?
�
This would be quite a project, and hope you will share your thoughts on it�s practicality before I start blindly trudging through the design process! --
You received this message because you are subscribed to the Google Groups "2 Meter WSPR" group.
To unsubscribe from this group and stop receiving emails from it, send an email to 2-meter-wspr...@googlegroups.com.
To post to this group, send email to 2-mete...@googlegroups.com.
Visit this group at http://groups.google.com/group/2-meter-wspr?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
�
�
K6PZB
I'm confident you have seen this: http://www.hanssummers.com/ddssi570.html
relating to the SI570.
Glad to have you on 2m WSPR!
John, K6PZB
------------------------
KD6RF
I am using the old R7000 as the 2M receiver. I can tell the temperature out in the lab (where the radio is) by what frequency it is reporting :( It moves a few hundred Hz over the 40 to 65 degree lab temp...
Using a 2 Meter reference dipole up about 8 ft for now - I may have to cook up a tappered line and stack 2 dipoles, EZNec says around 12 dBi. Or maybe just build Carol's nice stacked halo design.
Q - what accuracy is considered acceptable? I have a couple of Rubidium standards coming in next week, which look like they have about 5E-11 accuracy, with RMS drift over a few minutes in the 1E-13 range. Good enough for the measurements y'all have in mind?
Another Q - Is 20 to 50 watts the standard on 2M? Is more power discouraged?
Glenn Elmore
You're not the only one that is reporting temperature. I saw Joe Taylor
at a previous Pacificon and told him I could tell something about the
temperature in his ham shack. Evidently his is out on a porch with less
temperature control than the rest of the house. It is good for ~ 4 Hz of
frequency variation even at HF :-)
WRT accuracy, it probably isn't nearly as important as stability since
even stock WSPR has A/B frequency adjustments to correct for errors that
are constant with time. Stability is a much bigger issue and the answer
depends upon which WSPR and what band. There is roughly an order of
magnitude less tolerance to drift with WSPR-15 than WSPR-2 (conventional
WSPR). Additionally 1296 MHz is roughly an order of magnitude less
tolerant than 2m.
WSPR used differential modulation so considerable drift can be tolerated
by the decode algorithm. Something on the order of 4 Hz can still get
decoded, provided the shape of the curve is reasonable. For HF, on paths
which almost always involve the ionosphere, there is almost always a
significant fraction of one Hertz of smearing due to ionosphere
characteristics and I when the band is opening or closing, there can be
1 or more Hz of Doppler due to the changing path length. VHF and UHF are
actually better candidates for being able to see improved as-transmitted
stability and accuracy. At 430 MHz we have been able to compare
different OCXO or GPS referenced systems to better than .1 ppb, even
over very non-LOS paths. The HF smearing makes WSPR-15 unusable there
but that is not so for VHF and above.
Whether or not there will be additional information gleaned from systems
that are .1 ppb or better is still TBD but so far, we have been able to
discover several new effects because of the higher accuracies and
stabilities.
Certainly an Rb standard is plenty good, as is a GPS disciplined high
quality OCXO. Even the eBay OCXOs I found for the external PLL project
will do better than .1 ppb, maybe quite a lot better in a 2 minute
interval after a little care with additional thermal insulation and
digital FLL parameters.
I would say the 1 ppb is probably fine for most normal 2m WSPR while one
might want one to two orders of magnitude better than that for WSPR-15
on 1296.
Regarding power level, there may be multiple opinions. Mine is that on
VHF and above, we are likely using antenna directivity and we have a
very different QRM situation than those on HF where low power is the
norm. Some on here have tested with as high as 500 watts and I've seen
no ill effects nor any real compromise of anyone else's spots because of
it. Combined with WSPR 15 and a lot of antenna directivity, who knows
what previously-unrecognized propagation mechanism we may discover? We
have already pretty much confirmed ACS and wing-tip-vortices as real
and, perhaps, useful mechanisms. There may be others. I have long
wondered if mid to upper microwave might not be far more useful than
most suspect. 1 megawatt ERP and WSPR-15 might be just the thing to
discover mechanisms that could provide unheard of DX. The late K6UQH,
who frequented 10 GHz in the SF Bay area for several decades, was of
this opinion. WSPR and high ERP might be just the tool to look for this.
So, my opinion is that unless someone discovers an actual QRM problem
with a given transmit power or ERP, try whatever you want. It is a very
different situation compared to HF!
best
Glenn n6gn
KP4MD
KD6RF
Thanks for the listing!
I changed my TX freq to 144.490555 as of this morning (in honor of the venerable old 555 timer chip).
Let me know if this is a problem for freq coordination, and I will be happy to change it elsewhere.
Antenna has changed to my design - VTenn Vivaldi, 6.5 dBi. at 22 ft. Will run EZNec plots if interested. Antenna is good up through 4 GHz with gain that increases up to around 13 dBi as freq increases.
Polarization is 45 degs, with 30 degree elevation, still pointed due North.
Rig is Icom IC-706MKiiG, set to 20 watts.
Locked to Efratom FRS-A.
Efratom is coupled to Icom's crystal ref osc tuning coil with 6 turn link. It turns out that the Icom is perfectly happy with sub-harmonic locking. Shove the Efratom's 10 MHz signal in, and the Icom's 30 MHz ref osc locks right up.
KD6RF
Anyway, freq set back too 144.490540.
KP4MD
KD6RF
EZNec can't really handle it all that well. It works ok for the low freq end (with 150 wires or so). If we use a more dense mesh, it can't handle the small "loops" well. But it's not enough wires for proper modeling of the high freq end.
So, I only use EZNec for modeling little enhanements on the low freq end.
Not sure what the capability of other NEC based apps is, but it would be quite the chore to mesh up the plates if you don't have some sort of general shape mesh generator.
Line loss, 10 meters LDF4-50, plus the bits of LMR400 flex (from VK1OD calc) is:
0.45 dB @ 146MHz, and rises to 2.5 dB @ 3456 MHz
So, the rise in gain beats out the line loss by several dB.
It's nice to have a single feed line for all freqs 2M and up, although I have yet to try using a duplexer for satellite work.