 |
| View from rocket. I'm in front of the swing-set, with hat on. Billy to my left. |
SolderSmoke Podcast #196 is available.
5 May 2017
http://soldersmoke.com/soldersmoke196.mp3
Shenandoah Rocket Launch -- With video!
5 May 2017
http://soldersmoke.com/soldersmoke196.mp3
Shenandoah Rocket Launch -- With video!
Pete and the QRPARCI HOF: We thank you for your support!
Note the strange seasonality of these kinds of events... Always in the Springtime...
April is the cruelest month. Especially the FIRST DAY!
Feel Tech Follow-up -- the DC blocking cap issue.
HBTV? Roku Challenge?
A tube of Desitin? Why?
SDR Misadventures.
SDR on a tablet Just like WA7HRG
Pi Day More RaspPi Linux observations TOO HARD
The problem of Si5351 and updated libraries -- old sketches don't work!
My SDR question: If we go with direct sampling, no need for I and Q at front end No image problem, right? But why is there often a digital I and Q in the direct sampling receivers?
BITX MODULE
Getting RF in through the DDS jack.
Electric Radio: Reading from the history of SSB. And from the T.O.M. article
Joe Carr K4IPV : Homebrew Hero from Falls Church Va. Any more info on him?
Pete's newest DifX
Pete's new digital scope.
Bill's NE602 RIG, Epiphyte History.
He put AADE crystal filter impedance match circuits in backward That's why he needed RF amp. Duh.
The perils of a crowded box
Building RF amp board (BITX Module design) for the NE602 rig.
NE602s do not put out a lot of power. uW
BITX Module Amp circuits very stable.
Chassis as the heat sink.
Mic amp? We don't need no stinkin' mic amp! But yes, we do. 741 on the way.
Parts on e-bay very expensive! Back to the hamfests. On to Manassas!
Congrats to Tony Fishpool G4WIF, who received the Don Cameron, G4STT, Award for an outstanding contribution to low power amateur radio at the RSGB AGM meeting in Cardiff.
MAILBAG
Posts
Lexicon? Twizzle? A well-worn path: try Jabberwocky, " 'Twas Brillig, and the slithy toves ...". My birthplace was Hokitika on NZ's West Coast, South Island, and a compatriot described the weather there as "mizzle", miserable drizzle. :)
ReplyDeleteRegarding the Digital IQ, I believe Pete was on the right track. I and Q Data is an universal and complete description of the information content of the RF signal. By that it not only allows to demodulate all "classical" modulation (like SSB), it gives a foundation to demodulate any modulation. This is why all modern modulation schemes (e.g. QPSK,QAM) are fundamentally built on the I and Q representation of RF information. One can not "reinvent" this, all you would end up doing is describing the same thing differently (and possibly not as generic). To demodulate a known modulation differently, one would need to reformulate the underlying principles in a specific way only applicable to the one scheme... which may make sense when dealing with analog minimal hardware(i came to think of http://hackaday.com/2014/12/29/demodulating-bpsk31-with-opamps-and-555s/), but digitally not. So the answer is, it makes no sense to do it differently for a universal receiver, as there is no known better way.
ReplyDeleteI&Q is pretty much unavoidable. But you can do it different ways...a phase shift of 90 degrees can translate to a time shift. So take a sample, (call it I), then wait a bit, and take another sample (call it Q). If you time it right, they're separated by 90 degrees. So one A-to-d can spit out both I & Q streams: even #'d samples are I-stream, odd #'d samples are Q-stream.
ReplyDeleteSince you fellows are not Xenia-Dayton bound, have there ever been thoughts of a Soldersmoke gathering in N. VA or maybe sunny CA ? Bring the world to one of the soldersmoke hubs - just don't sked for April.
ReplyDeleteChuck, WB9KZY
From Allison KB1GMX Part 1:
ReplyDeleteFYI: I've made good filters using the short 49 type crystals with good result.
Remember that the mesh (crystal plus adjacent caps) will load the crystal
and lower its frequency.
For crystal testing I still use the freq counter from 1973 (Yaesu YC355d). However
Pacific antenna has a good module that does the job well. I'm sure there are others.
Rigol scopes, good. Rigol DSA815T spec analyser with tracking gen is nice too.
For RF hunting rather than a scope I use a diode probe.
I've built many radios using the LM173 back 40 years ago. Did the 9mhz with
20/80 coverage and many others using IFs like 9.6, 6 and .455mhz. Noisy chip,
uses the same structure as MC1350 for AVC. FYI those where never cheap
typically 3$ in 1975 (about an hours pay then!). I have a few left. the LM172
was handy for RF AGC.
Dual conversion is for those that are willing to shield well, if not you will hear bleeps and burps.
If you shield well and take good care its not so hard is you pay attention gain distribution.
I've done it a few times, lots of small boxes in a box works well.
For the relay switched 602 radio see what I wrote back in 2006 (QQ) and the KNQ7A as
example/sample of work done. The QQ article used the switched scheme and worked well
as the drive to both stages is the same. Its very common scheme and works fairly well.
FYI the 603 only needs maybe a 100-300millivolts before you hit overload (re: mic amp).
The original topology was the Atlas 180 that used DBMs for mixers, 2n3866 post mixer
amp a 1350 in the IF driving a DBM. THe LO and Carrier osc are switched using FETs
but a relay (DPDT) would do it. I just finished building a design using two DBMs a
2n3866 filter and a IF amp like HyCas (3904 and J310) driving a dbm. Very low parts
count and low RX power needs. No RF for RX on 40M only input band pass then the
TX uses high pass low pass from the product det (RC, LC) to get he TX RF run that
through a band pass filter to get about -15dbm signal into amp lineup for using the
a pair of RD16HHF to get about 20W at 13.8V. LO is Ad9850 and Arduino for LO
and crystal osc with buffer for carrier. I call it HL-1 (hardware Lite). It was tested
with VFO and run a while that way, properly buffered it should not shift with loading
change.
The CA3020 audio amp is a oldie but good. Problem it's extinct and only good to about
10mhz give or take. Also the IMD is not so great at full power audio about .1% was doing
well (thd in DBM about -30db).
My choice is a handful of transistors. Typical for HF/6M is two stages of 2N2222A (not PN2222!),
2N3866 or 2219, to a final. Often now the driver is push-pull driving push-pull final. Finals are
IRF510, RD06HHF, or RD16HHF though 6M.
Allison KB1GMX
From Allison KB1GMX Part 2:
ReplyDeleteMy common Terms:
Scratching and daubing.... tuning stripline circuits by cutting away conductors or
adding bits of copper. Typically done at high uhf and up.
Drillin and blastin, chassis construction involves much drilling and cutting (blasting).
Regarding the direct sample question... for Bill.
Direct sample is still I and Q and there are two techniques. One is two A to D running in quadrature the other is to take
samples at high enough rate that you can create a I and Q sample by selecting the samples that are delays in time by
90 degrees at the selected frequency. Its a matter of not thinking in the frequency domain but in the time domain
recognizing they are connected. For example we can say 1mhz or periodic waveform of 1 microsecond.
A rehash... Analog IQ (aka KK7B R2 designs) we use a 90 degree phase shifter at RF and also for the audio.
What is the 90 degree phase shifter doing? A time delay that corresponds to time at frequency such that
the occurrence of one is 90 degrees (or 1/4 the waveform time) earlier than the other. you can see that on
a dual trace scope same waveform shifted in time.
Sound cards are just fancy A to D and also D to A with some added audio hardware that limits it to audio. But direct
sample uses the same basic hardware but way faster so that we can sample at far higher than the frequency of
interest. So we get more data faster and in far smaller samples in time. Keep the last item TIME front of you.
Now I and Q are effectively samples in time one being either 90 degrees earlier in time or later. So going back
if we are at 1mhz or 1 microsecond for a cycle then 90 degrees is 90/360 or 1/4... or 90 degrees earlier or later
in time by .25 microseconds. So from a single stream of data from a fast A to D taking samples at say 40mhz we
get enough samples like one every .0025 microseconds. That means we pick one sample call it I and 100 samples
later (or previous in the stream) we have a Q sample we can do the math at that instant and iterate.
Obviously we need a continuous stream of samples and with enough memory we can store
hours of it and we can also pick groups of data and process them into audio.
Time vs frequency, phase, they are related.
73's
Allison/kb1GMX
Hii Bill & Pete...
ReplyDeleteYes, Pete you were kind of on the right track, as was Bill. The pahse relationship bewteen I & Q allows for easy demodulation of signals through very simple arithmetic operations.
Basically, I and Q are streams of samples... numeric values in a buffer, so as the buffers fill on the input side, demodulation is a simple case of performing one of the following operations on the I&Q value and passing the result to the audio codec.
USB = I + Q
LSB = I - Q
AM = SQRT (I * I + Q * Q)
FM = atan(Q * I_old - I * Q_old, I * I_old + Q * Q_old
You can read more here:
https://github.com/df8oe/mchf-github/wiki/How-does-your-mcHF-software-DSP-work
Regards
Gerry