Premium Rcvrs, FlexRadio, HPSDR, SoftRock, SDR, DSP, PCs, OSs, etc.!

[flintstone mop]

It's nice that this thread was resurrected. I guess some folks are reporting here with their plus and minus views for the softrock lite boards they built, or asked someone with toroid winding patience to get their board up and running.

I finally have a SoftRock Lite v6.2 running (thanks Stu) in a Dell with the Flex software and at times I see 30% CPU. No hiccups in the audio. Sound card is a SoundBlaster Live card @48khz sample. One of those 'unsupported cards'

There's a lot of buttons to push and many flavors to choose from.
I'm very impressed with SAM (Synch detector mode) This $15 project saved me $600.00. I do not have to buy an expensive SE-3. It virtually eliminates all of the selective fading distortion from music programing on short-wave.

It has turned my R390A into a plastic ($8000) radio............ha!

Fred

[KF1Z]

Ok Fred... good deal...

I wonder if you're using the Power-SDR software right from Flex-radio site?

Or, are you using the Power-SDR that was re-written with the softrock in mind?
(available at the softrock group files section)


The other one to try is the WinRad software... that Sync-AM is much better than Power-SDR

[flintstone mop]

The software I got was from the Flex web site. They offered the SoftRock board in their setup screen.
I can try Winrad and listen to any differences. This seems to be the nice thing about these "new radios". The SAM (in Flex) has a slightly different audio characteristic, restricting the high freqs a little more than straight through AM in the 12khz bandwidth mode.
Good stuff

Fred

[AB2EZ]

Out of curiousity, I tried unplugging the I output or the Q output, of my SoftRock-like mixer, from the stereo input to my sound card. In theory, all of the information needed for the PowerSDR application is in either one... so it should work with just one.

I tried this on AM, synchronous AM, and also listening to a sideband signal on LSB.

In all cases, I could not hear any affect of unplugging either I or Q... nor could I see any affect on the display. Everything continued to work fine with just I or Q

I'm using my mixer board with the i.f. output of my receiver.

Question: does anyone know what functions of the SoftRock + Power SDR actually require both I and Q?

Stu

P.S. As I think about this, I believe that the only purpose of using both I and Q is to reject "image band" signals. For example, if you mix a 444 kHz local oscillator with a 455 kHz i.f you get a mixing product at 11 kHz. However, if your receiver were putting out anything at around 433 kHz, then that would also appear as a mixing product at the output of the SoftRock (or equivalent). For i.f. applications, your receiver already has a sharp filter that removes the image band (i.e. signals that are 22kHz away from the signal you are trying to listen to. So... for i.f. applications, you do not need I and Q.

A simple mixer that down-converts your i.f. 11 kHz is cheap and easy to build:

http://mysite.verizon.net/sdp2/id12.html

[KF1Z]

Stu,

As I understand it....

The two signals are In phase, and Quadrature phase.

In one complete cycle the "I"  is 0 to 90 degrees, and 180 to 270 degrees,
And the "Q" would be 90 to 180, and 270 to 360 degrees.


But, what's actually happening is,  instead of the signal being chopped into 4 segments, as that would lead you to believe...

Each line ( I and Q) carries ALL the RF information... but the Q is simply being shifted by 90 degrees.

However,  if you have a "closed" system, as in and IF detector, you do not need both signals, for the reasons you gave.


BUT,
To look at say 190khz of band on your display, you need a soundcard that can sample at 96khz.

And THAT"s where the I/Q signals are used to seperate (or cancell out) the image signals, that appear on either side of the center frequency.
Because all the RF information is in each channel (I/Q or L/R) but one appears inverted.

[Tom WA3KLR]

The latest issue of electronic design has the theme of "Wireless Everywhere".  It has an article (19439) by Louis Frenzel W5LEF, a regular author, entitled "SDR Transforms Amateur Radio.  The printed magazine article is 6 pages and features the FlexRadio FLEX-5000A.  Other makes of SDR are included also.

There is a side article (19438) also, 2/3 of a page, where Lou interviewed Dave Sumner at the Dayton Hamvention.

The online version of the article can be found at:

http://electronicdesign.com/Articles/ArticleID/19439/19439.html

[flintstone mop]

I had a nice road-test time with soft rock lite board and Winrad and the Flex software. And the winner  (for me) is the flex software. Lot's of nice features for variable bandwidths and the SAM (synch detector) is so nice in the Flex. I was not impressed with the slow weird lock-in time of the Winrad in the synch detector mode. And the audio quality was not as good as the Flex.
The Flex could be used for AM QSO's as it will lock in faster than Winrad.
I'm a happy camper....don't have to spend no $600 for an SE-3...........ha!!

Fred

[WA1GFZ]

slower the lock, (lower the corner) the cleaner the audio.

[Steve - WB3HUZ]

I tried this on AM, synchronous AM, and also listening to a sideband signal on LSB.

In all cases, I could not hear any affect of unplugging either I or Q... nor could I see any affect on the display. Everything continued to work fine with just I or Q

Very interesting. The SW must not be using a synchronous detection mode requiring I&Q (possibly squaring or a narrow filter)?

[WA1GFZ]

I think you can download the source code if you want to see how it is done.

[Steve - WB3HUZ]

You're assuming I would understand it!!!

I think you can download the source code if you want to see how it is done.

[AB2EZ]

Steve

If the rf (or if) signal had been "homodyned"... i.e. heterodyned to baseband...by the SoftRock... then you would need both I & Q for demodulation. Actually, some would argue that, to avoid conmfusion, the use of the terminology "I" and "Q" should be reserved for homodyning.

Since the "I" and "Q" output signals from these mixers are not at baseband... they both contain all of the information required for such things as synchronous AM demodulation.

As a way to see this... consider a typical double conversion receiver:

You start out at some r.f. frequency. You mix that using a crystal local oscillator to some i.f. frequency (e.g., 10.7 MHz). Then you mix that using a variable frequency local oscillator to a second i.f. frequency (e.g., 455 kHz). Finally, you demodulate using whatever methodology you wish (including coherent demodulation methods,if you wish). 

In all of these mixing operations, in a conventional double conversion receiver, you don't attempt to obtain what Flex and others would call "I" and "Q" i.f. signals. This is because shifting the center frequency with a mixer doesn't remove any information provided you use a filter in front of the mixer, in each stage (if necessary), to remove any "image band" signals that are present. For example, if you are going to mix a 28.8 MHz signal down to 10.7 MHz in the first mixer... by using an 18.1 MHz local oscillator... you have to take some steps to ensure that there are no signals centered at 7.4 MHz (i.e. 10.7 MHz below the local oscillator frequency) entering the mixer along with the desired signals centered at 28.8 MHz (i.e., 10.7 MHz above the local oscillator frequency).

The way the SoftRock design removes signals that might be in the image band is by creating the needed filtering function using the phasing method... the same way a phasing type SSB transmitter removes one of the sidebands. If you already have sufficient filtering in front of the SoftRock to remove any image band signals... then you don't need the "I" & "Q" signals.

Stu

[Steve - WB3HUZ]

then you don't need the "I" & "Q" signals.

But you do for many forms of synchronous detection. My question was how does the PowerSDR process the I only channel to accomplish synchronous detection, since they are obviously not using the already existing I&Q from the QSD?

[AB2EZ]

Steve

I assume that they just use DSP to, in effect, extract I and Q from the single output of the i.f. => 11 kHz mixer.

Mathematically

If you multiply (using DSP to do it) the single output of the i.f. => 11kHz mixer by cos (2pi x11,000 x t) you get the baseband "I" signal.

If you multiply (using DSP to do it) the same single output of the i.f. => 11kHz mixer by sin (2pi x 10,000 x t) you get the baseband "Q" signal.

Stu

[Steve - WB3HUZ]

Probably. Seems silly to waste processing power to do this when you already have incoming I&Q signals. I'm guessing very little processing power is required. Highly optimized coding for the Hilbert Transform has been around for a long time.

[WA1GFZ]

Don't you need the I/Q to cancel the low frequency image? My TCI 8174 takes the final IF and generates I/Q in the DSP before it demodulates.
I think the Cubic receivers like the 3150 and 3280 do the same thing.
I would think if the DSP has enough horsepower you don't need I/Q but remember this all started using the stereo sound card as the interface.

[AB2EZ]

If you don't have a filter in place to remove any image band signals prior to the mixer (SoftRock or equivalent) then it will be mathematically impossible to remove image band signals after the mixing process unless you have both I and Q outputs from the mixer.

On the other hand, if you do have a filter in place to remove any image band signals prior to the mixer (SoftRock or equivalent) than only 1 output from the SoftRock is needed.

Deriving the I and Q outputs from either output of the Softrock (for the case where there are no image band signals to worry about) is a lot less processing intensive than the other tasks that the DSP has to perform (e.g. as you mentioned Steve: the Hilbert transform)

Electrons don't get tired... so the issue of whether something seems like a waste of processing power is only important if it pushes the required processing beyond the capabilities of a particular processor.

As an aside:

Each of what we call the "I" and "Q" outputs of a SoftRock (or equivalent) board contains both of the baseband I and Q components:

The input to the Softrock (or equivalent) board (not including image band signals) is:

s(t) = { I(t) cos ( 2 x pi x f x t) } + { Q(t) sin (2 x pi x f x t) }

where I(t) and Q(t) are the baseband I and Q components.

The pair of outputs from the SoftRock (or equivalent) board (not including image band signals) is:

x(t) = what we call the I output of the SoftRock board = { I(t) cos ( 2 x pi x 11000 x t) } + { Q(t) sin (2 x pi x 11000 x t) }

y(t) = what we call the Q output of the SoftRock board = { I(t) sin ( 2 x pi x 11000 x t) } + { Q(t) cos (2 x pi x 11000 x t) }

where I(t) and Q(t) are the baseband I and Q components.

Thus, you still have all of the "work" of extracting the baseband I and Q components from the SoftRock (or equivalent) board's output(s), x(t) and y(t)... whether you have both of them, or just one of them.

I believe.... based on the fact that unplugging one of the two inputs to the sound card has no affect (not even a click or a pop)... other then to reduce the reading on the S-meter by 6dB and the level of the displayed spectrum by 6dB... that the first thing that the PowerSDR algorithm does with the I and Q inputs is as follows (using DSP):

1. Apply the Hilbert transform to the Q input
2. Add the inputs together (after step 1) to cancel the image band... and to yield the equivalent of what you would get if there were no image band signals to start with, and you only had one of the SoftRock outputs to work with.
3. Proceed to perform the demodulation by extracting the baseband I and Q components from the result of step 2.

Best regards
Stu

[KF1Z]

What about the image created when you mix your IF with the local osc on the softrock?

Is that showing up on your display Stu.?

I assume you'd still be creating an image anyway...

[WA1GFZ]

stu,
If I remember the TCI/BR 8174 DSP block diagram this is exactly what they do with the single ended low frequency IF. I think they use a TI DSP running at 40 MHz clock.  The S Meter display is as accurate as my HP8640B step attenuator.

[AB2EZ]

What about the image created when you mix your IF with the local osc on the softrock?

Is that showing up on your display Stu.?

I assume you'd still be creating an image anyway...

I'm not sure if I understand your question.... but perhaps this will help:

Incoming (to the SoftRock board) r.f or i.f. signals at frequencies that are in the band of frequencies that is 11 kHz +/- 11 kHz above the SoftRock local oscillator frequency will produce the desired outputs (display, demodulated audio, etc.) from the Power SDR software... regardless of whether you use both outputs from the SoftRock board, or just one of the outputs from the SoftRock board.

Incoming (to the SoftRock board) r.f. or i.f. signals at frequencies that are in the band of frequencies that is 11 kHz +/- 11 kHz below the SoftRock local oscillator frequency (usually called the "image band" of frequencies) will produce, in effect, interference that simply adds to (in the case of CW, SSB, AM, or SAM reception) the desired outputs (display, audio, etc.) from the PowerSDR software... unless you provide both outputs from the SoftRock board to your sound card.

If you use both outputs from the SoftRock board, then the normal/desired portions of both outputs will add (making the desired signals look 6dB larger); and the image band/undesired portions of both outputs will subtract (cancel)... making the image band signals go away. This is just a consequence of the way the mathematics works out.

If there are no incoming (to the SoftRock board) r.f. or i.f. signals within the image band of frequencies (i.e., because they have already been filtered out), then you don't need to use both outputs from the SoftRock board.

Stu

[W3RSW]

Interesting quadrature and detection discussion.

For a little eye candy and brief respite from technical burn-out, here's a screen shot of some of Cathy's latest work from Phil's QS1R board showing two receivers, oscilloscope time & phase functions, various waterfalls, etc. 

Note that this is Cathy's test GUI/server combo display; not the "pretty print" SDRMaxII GUI normally used in a more polished version.

Also see the Yahoo group for further clarification and updates to Phil's SDR board and applications.
http://groups.yahoo.com/group/qs1r/

This is rapidly becoming the SDR to have.  Buy one; get another for list ! You'll not be sorry and will be fully diversified !   

[WA1GFZ]

HPSDR Mercury just made the last board turn and will have the option to use the 170 MHz. A/D Phil's is a very nice RX but I would love to do some testing to see how it stacks up. Pretty display. I would think the only limit to receivers displayed is computer horsepower and DDC bandwidth.

[W3RSW]

Yeah Frank, this is great stuff.  Wot say... "bleeding edge?"
- great marriages of high speed ADC's and FPGA's.

As some have mentioned before I don't think Phil's ready for a/b comps. until he gets the pre-amp/bandpass filter/attn. board finished. 
But even so, the bare bones board is plenty sensitive enough on HF and with discrete or inf. variable very sharp filters.  It's so neat to see the whole HF spectrum up through 60Mhz in one glance... just pick a signal.

No where to go but up in SDR.  Fun times to be a' livin.'

After seeing spectrum, having instant picks, etc. I don't see how we can resist putting at least SoftRock equiv. detection on everything from an S-38 up through R-390's on the "low" end and going the full SDR route on the high end.  It's very addictive.  I'll bet YaeComWood won't hold off incorporating SDR for long or at least upgrading their TFT displays to match some of these fantastic GUI's.

....er, or at least until Russia, et.al. start up a hot war
I still unplug 'em during storms.

[WA1GFZ]

Rick,
It is very hard to get a preselector tight enough to get the performance. I remember working on one that would do the job. It was used by the Navy and almost as large as a small fridge with a wip antenna. Most of the inductors were wound with 1/4 inch copper tubing.
This is also the reason TRF went to superhet about 75 years ago.
Look at your noise floor when there is static or lightning. The whole floor jumps up across the spectrum. Kinda hard to sort things out when the A/D is saturated for many samples.
An analog system seems to deal with it better...so far

[W3RSW]

Well I can see that reasoning. 
but if the noise floor jumps up and down noticably is that an artifact of the video display at say 30fps (*well ok, 15fps makes a movie*), or something else slowing the response time? - or is it real ADC's response or saturation?  There are already millivolt (h, maybe even volt) signals with wild mod. swings, overmods, harmonics, etc. from myrads of BC statons, SW beaters, 200kc wideband VLF,etc. operating simultaneously. All communications through real world equipment, even the 'data' itself takes a finite time for turn-around from/to the on/off state unless you have perfect square wave response 'to infinity.' 

Even analog filters no matter how wide "ring."  AVC takes time to recover on an analog rig if your using it in any real world application say long constant SSB.  ..or want to listen to a wide variety of AM signal strengths.  So yeah on VHF line of sight, perfectly low noise, unmolested amp/detection figures are nice.   ....

...well, sort of heading to an argument of some sort here in my muddy thinking land but maybe you can see the trend.  Galatic background is horrendous to any noise figure in HF even if everything else is perfectly quiet, no storms of anykind anywhere.  Phil's board for most unmodified applications has a cut-off just above the 6 mtr. band.  Intermod now, etc. that's another story.

So day-to-day usage is kind of what I was refering to for the QS1R, not absolute performance approaching Racals'...  (but it's getting close.)

Besides if I had a perfect, quantum receiver I wouldn't even need to post this.  You'd have it before I sent it.
 whoooo