The AM Forum

Received the following e mail from Craig W6DVC ... who is now producing the SoftRock Lite I-F version. I have no association with him - just passing along the information for those that might have been or are currently looking for SoftRock Lite I-F boards.

************************************************************

Hi Jay

I  am  handling  all  SoftRock Lite II kits for IF use and it is all
YOUR fault. Really it is. :-)

I  accidently  ran across your R-390A/SoftRock page and seeing how I
have  two restored R-390As I thought, fun easy project. Wrong!!! Tony
is  so busy kitting his newer SoftRock RX Ensemble and SoftRock RXTX
Ensemble kits, he couldn't  be persuaded to put together the 455 kHz
"IF  Special"  kits  I needed. Finding a kit on the secondary market
is nearly impossible. I did find a built 40 meter Lite II on ebay, but
the builder was  pretty  sloppy  and rebuilding it was more trouble
than what it was  worth.  After some negotiation, I made a deal with
Tony to take over  the "IF Special" part of the SoftRock stable of
kits. Not high volume,  but  enough  to  make  it  worth wile and
kinda fun helping people out when there is no other options in this
price range.

I  have  several  SoftRock  Lite  II  "IF  Special"  kits  in stock.
Depending  on  center  frequency  you  need,  I can ship as early as
tomorrow.

Price is $22 each plus shipping.

I need to know what center frequency you need and model of receiver or
transceiver you have when making inquiry.

Please contact me at: w6dvc@cableone.net

Thanks and best regards,

Craig W6DVC

Craig,

   You have done an admirable thing doing this. I wonder if you have any hope of growing this business, or are you content providing this niche product as a single product offering?

  question: What range of IF frequencies are you gearing up to provide?

I ask this because many radios have a dual IF. Hooking up to the first IF (often several Mhz) will allow one to see maybe 50 Khz of the band (or more depending on IF bandwidth and sound card chosen) whereas at 455 KHZ the IF bandwidth is narrow, such that looking up/down the band as a pan adapter is very restricted.

Jim
WD5JKO

You can easily change the frequency of operation, and even make it switchable if you like.

For 40mhz or higher IF, you need a different kit altogether.

I think that some have stripped the filter off the softrock completely for use as an IF reciever.
Which would mean all you need do is change the crystal.

I should contact this guy and show him some easier transformer designs. I would replace the input filter with a 6 db pad to force a 50 ohm input.
also it would be cool to have a FET source follower so it doesn't load down the IF stage at the mixer.

Frank said:
That's exactly what I had to do with the Drake R4A, Frank.
I used an MPF102 source follower with the SoftRock at the first I.F.. (5645 Khz)

Not only did it give me more signal to work with but the isolation is totally necessary ifn you want to use the receiver as intended. (for what ever reason??)

Post your design here.

The source follower is required when hanging the low impedance soft rock off a high Z tube circuit. A resistive divider will also work.   

I like the idea of a simple resistive pad. The coupling network for the low freq versions had a toroid with a TON of turns. Kind of a PIA to wind and not really needed.

Modified coil info (fewer turns) from an old thread. As good or better performance.

http://amfone.net/Amforum/index.php?topic=13686.msg105723#msg105723 (http://amfone.net/Amforum/index.php?topic=13686.msg105723#msg105723)

I think I mailed you one of my transformers Steve.
Also the softrock should be AC coupled to the radio to isolate any DC voltages.
So for universal applications a .1 uf on the input, FET source follower, then 6 dB pad into the transformer primary. I did a double balanced tayloe in my design but didn't need the FET. LO section was from  early softrock.

You did give me some info on how to wind a tranny that doesn't require a ton of turns, IIRC.

For 455 kHz work, KE1GF's downconverter works FB is is simpler than the Softrock.

Has anyone used a softrock with a Collins 51J-4 or 75A-4 yet?  If so what did you need to do to interface the 2?  Will I need a buffer amplifier?

Hi!

I have been using, for several years, a Soft Rock, or a Soft Rock-like mixer with my 75A-4.

I've used a Soft Rock-lite; a single-phase mixer designed by KE1GF: http://mysite.verizon.net/sdp2/id12.html ; and also the equivalent of a Soft Rock that is available here: http://zao.jp/radio/66RF/index_e.php.

They all work fine. [For the i.f. application, you get no benefit from the use of a quadrature mixer v. a single phase mixer].

If you use a single phase mixer you can run the application software on a laptop with a monophonic microphone input. If you use a single phase mixer with a desktop computer with a stereo line input, then use only one of the two inputs, and leave the other one disconnected.

I don't use a transformer-based tuned circuit at the input of the Soft Rock / mixer. I just capacitively couple the i.f. output I obtain from the 75A-4 (as described below) to a simple isolation transformer; and then through an appropriate voltage divider... to get the level right. The isolation transformer only serves the purpose of removing AC hum from the i.f. output of the 75A-4. I make the isolation transformer by wrapping a 6 turn primary and a 6 turn secondary (insulated hook-up wire) on a ferrite core that is the same type of ferrite core that some of us use to make transformers for Class E transmitters.

The attachment (below) shows the schematic of the circuit I used to pick up the i.f. output of the 75A-4. The 50 Ohm output of this circuit can feed a coaxial cable (e.g. RG-174U) of whatever length you need to reach the mixer. In my case, the cable is 10 feet long.

Summary: Interface circuit inside 75A-4 => RG174U cable (passing through one of the many vent slots in the 75A-4 cabinet) => r.f. isolation transformer => Soft Rock (or equivalent) with an appropriate voltage divider to obtain the proper level signal to drive the Soft Rock.

You might be able to use a much simpler i.f. interface circuit than I show below. For example, you could add a 470 Ohm resistor in series with R50, the 47k Ohm resistor that is already connected to the plate of V9. Then you could take the i.f. output from a 0.01uF capacitor connected to the point where the existing resistor and the new 470 Ohm resistor are joined. [Remember: that point is at B+, so use an appropriately-rated capacitor. On the output side of this capacitor...facing toward the coaxial cable... you should place a 470 Ohm resistor to ground. This is a safety measure to make sure that the capacitor is charged up to whatever voltage is on the plate of V9]. This would give you a 200:1 i.f frequency voltage divider (with respect to the signal on the plate of V9) ... which is more than enough to drive the Soft Rock (or the Soft Rock equivalent or the KE1GF single phase mixer).

Best regards
Stu

Stu,

Thanks!  Wow, that's great stuff.  I'm really looking forward to trying these out on the 75A4 and 51J4 and maybe others. 

I've never built up a circuit just from a schematic before, but I think I might be able to handle this one.  I guess one of those universal circuit boards should work or maybe a perfboard.  Or, did you make a circuit board for your interface?

thanks
Mike

Mike

At these frequencies (455kHz and 11kHz) things like ground planes are not very critical.

I use Radio Shack perf boards (the ones designed for DIP devices) to build things like this.

Stu

Does anyone know why a slight offset 9 or 10 KHz from the nominal IF frequency is preferable on the  SoftRock when using it as a spectrum display unit? I have heard that they are kind of deaf if they down convert right to 0Hz, no offset. Of course I realize that they are using low cost commonly available crystals so finding a crystal that is exactly at 4 X 455KHz is not likely without plunking down some cash but the offset requirement still has me curious.

I would think the ideal situation would be to have the softrock centered up exactly on the IF so you could get the maximum range of bandwidth either side of the IF center frequency, +/- 24KHz or more depending on the sound card sampling rate. I pick off my IF before the narrowband IF filtering where it is most useful for spectrum monitoring purposes so being able to see a wide bandwidth with minimal center frequency offset is important to me. This is also the senario where having both the I and Q channel available is  necessary.

Mark (WA1QHQ)

If you go through the mathematics of how the application software (that most of us are using) processes the digitized output of the Soft Rock... you will find that the input to the computer's sound card must be a passband (not baseband) signal... centered at a frequency such that all frequencies associated with the original receiver's i.f. output are within the passband of the sound card (just shifted down to the new center frequency). For example, if the i.f. output of the receiver is 455kHz +/- 4kHz (8 kHz total bandwidth), then you want the new i.f. frequency (after mixing within the Soft Rock) to be at least 4 kHz... so that 4kHz +/- 4kHz is within the passband of the sound card.

Since most plain vanilla sound cards will sample 48,000 samples per second, which implies a passband that is somewhat less than 24kHz, shifting the receiver's i.f output to be centered at 11kHz allows both sides of the i.f center frequency to fit comfortably within the passband (provided the i.f. passband of the receiver is less than 455kHz +/- 11kHz). Also, keep in mind that ordinary sound cards are not DC coupled... so "homodyning" the received signal down to baseband (0 frequency) is not going to work with an ordinary sound card.

That is why an i.f. of 11kHz is popular.

Of note, I can put the existing 50kHz output of my Drake 1A directly into a sound card I have that samples at 192 samples per second... and it works just fine with the PowerSDR software. [Because the passband of that sound card is a little less than 98kHz = 1/2 of the sampling frequency... so it can accomodate a passband signal centered at 50 kHz].

Separately, one can order crystal oscillators (not just the crystal, but a complete oscillator) from DigiKey... at a frequency you specify... for around $8.00. These oscillators come in a 4 pin metal package that fits into a 14 pin DIP socket. You use only three of the four pins: +5Volts, ground, and output. The output is a TTL level square wave at the frequency you have specified. This works fine as the i.f. source for any of the mixers used to convert 455kHz to 11kHZ.

Stu

Mark

As a post-script...

The way the Soft Rock (or Soft Rock -like) mixers work, you are actually mixing the incoming signal with a square ware (even if the Soft Rock local oscillator is a sine wave)... because the mixing device is a switch. Likewise, for almost any other type of mixer.

As a result, the output of the mixing process will contain a number of signals in addition to the desired signal (at the receiver's i.f. frequency - the mixer's local oscillator frequency). In fact, since the Soft Rock's on-off switching process includes a DC component, there will actually be a signal at the output of the mixing process that is at the same frequency as the input to the mixer.

Separately, the output of the receiver, prior to the receiver's i.f. filter, may contain a number of mixing products (from earlier receiver stages) or other signals (e.g. the local oscillator associated with a subsequent mixing stage) that the final i.f. filter would normally remove.

As a result, you may see an unexpected "mess" when you run the pre-filter i.f. output of the receiver into the Soft Rock.

I tried this with a Drake 2B a few months ago. I ran the 455kHz i.f. of the Drake 2B (conveniently available on the 4 pin accessory socket at the back of the 2B) into my Soft Rock. The Drake 2B's narrow i.f. filter is at the 50 kHz 3rd i.f. The 455kHz 2nd i.f. is somewhat bandlimited by the 455kHz i.f. transformers... but it has a lot of 405kHz 3rd local oscillator on it. What I saw was, the "mess" I am referring to, above. The fix this problem, I purchased a +/-10kHz wide 455kHz muRata crystal filter... that I put between the 455 kHz i.f. output of the Drake 2B and the input of my Soft Rock. After that, it worked great. The muRata filter removed the strong 405kHz 3rd local oscillator component, and probably some other undesired stuff as well.

Therefore, there may be some problems associated with running the pre-filter i.f. output of a receiver directly into a Soft Rock. [I also agree that, if you take the output of the receiver before the i.f. filter, then you will need to use an I/Q mixer to remove the "image band" components that will probably be present]

Stu

Hi Stu,

Understood and agree with everything you said and absolutely agree on the necessity to have a roofing filter ahead of the softrock.

I am new to the softrock but its pricipal of operation is easy enough to understand.

I am not sure what application you are referring to that requires the offset maybe they all do. Perhaps you can explain in a little more detail why this offset is necessary.

I don't see any warnings about this problem on the KB9YIG site but would expect when using any of the softrock receivers to find a deadband at the center of the coverage range where the signal would not demodulate properly.

Thanks for the input on the custom crytals from digikey, I wonder if you know any more about the technology employed to get these crystals on frequency in a quick and economic manner, are they actual xtals or are they programmable clock oscillators with their atendent relatively high phase noise.

Mark

Hi!

I don't know exactly how DigiKey can "program" these oscillators at the warehouse, or what the technology inside is. I don't know what the phase noise specifications are. However, they appear to work fine for the types of SDR receiver applications that most of us are using.

It is not easy to explain why the new i.f. frequency has to be sufficiently high to keep all of the frequencies coming out of the receiver's i.f. above zero frequency after they are down-shifted.

There are several mathematically equivalent ways to explain it. Let's try this one:

A. The i.f. signal, v(t), coming out of the receiver can be modelled as follows

v(t) = a(t) cos [(2pi x 455,000)t] + b(t) sin [(2pi x 455,000)t]; where a(t) and b(t) are the "in phase" and "quadrature" components of the i.f. signal.

In general, one needs both a(t) and b(t) in order to calculate and display the spectrum of v(t), and to demodulate any modulated signals contained within v(t).

B. If you multiply v(t) by the local oscillator, at frequency f, in simple (single phase) mixer... you get the following:  y(t) = v(t) x 2 x cos [(2pi x f)t]

There are 4 terms in y(t):  

y(t) = a(t) cos [(2pi) x (455,000-f)t] + b(t) sin [(2pi) x (455,000-f)t] + a(t) cos [(2pi) x (455,000+f)t] + b(t) sin [(2pi) x (455,000+f)t]

That is, you have 2 terms at the difference of the two frequencies (centered around 455,000-f); and 2 terms at the sum of the two frequencies (centered around 455,000+f).

It is easy to separate the two sum frequencies from the two difference frequencies... so let's assume that we have filtered out the two sum frequencies, and we have only the difference frequencies:

x(t) = a(t) cos [(2pi) x (455,000-f)t] + b(t) sin [(2pi) x (455,000-f)t]  

= a(t) cos [(2pi x h)t] + b(t) sin [(2pi x h)t]... where h= 455,000-f



To recover a(t) from x(t), we will (in the applicaton software, using digital techniques), we will multiply x(t) by 2cos[(2pi x h)t]; and then low pass filter the result.

But: x(t) x 2cos[(2pi x h)t] = a(t) + a(t) x cos[(2pi x 2h)t] + b(t) sin[(2pi x 2h)t]  

In order to obtain a(t) by low pass filtering the above result, it must be true that 2h is greater than the full bandwidth of a(t) (including positive and negative frequencies), and it also must be true that 2h is greater than the full bandwidth of b(t) (including positive and negative frequencies).

Likewise, if we wish to obtain b(t) by multiplying x(t) by 2sin[(2pi x h)t]; and then low pass filtering the result.

Therefore the new i.f. frequency, h= 455,000 - f, must be greater than half of the full bandwidth of a(t) (including positive and negative frequencies) and it must be greater than half of the full bandwidth of b(t).

Stu

Mark

I need to think about this some more (tonight)... but

As I think about the analysis I did (in the prior post) above... where I was assuming only a single phase mixer... I am beginning to think that using a quadrature mixer one could, in fact, mix the i.f. signal down to as low a frequency as you want... and still recover a(t) and b(t). This is a generalization of homodyning with two quadrature mixers to recover (separately) a(t) and b(t).

I'll post again after I think this through.

You still need to keep away from zero frequency, because sound cards are almost always not DC coupled... but maybe you could mix down to a frequency near zero... as you suggested

Stu

Mark

Please refer to the attachment (below).

If you use an application like Power SDR (as it is currently implemented), here is what will happen when the Soft Rock local oscillator frequency is too close to the center of the receiver's i.f. passband:

The PowerSDR software will treat the portion of the receiver's i.f. passband that is to the left (lower in frequency) of the Soft Rock's local oscillator frequency as part of the "image band". It will subtract it out (assuming the Power SDR applicaton has been calibrated properly).

Thus, the PowerSDR software will ignore (both in what it displays, and what it can demodulate) the portion of the receiver's i.f. passband that is below the frequency of the Soft Rock's local oscillator.

It is possible, in principle, to implement a different version of PowerSDR with a different algorithm, that will not "throw away" the portion of the receiver's i.f. passband that is below the frequency of the local oscillator. Thus, as you have suggested, you could pick the local oscillator frequency in the Soft Rock to be very close to the center of the receiver's i.f. passband. However, that would require a different version of PowerSDR, with a different processing algorithm... and it would require the quadrature mixing process to be very well balanced.

Stu

Stu,
I'm not sure, but it seems to me you're saying that you can't use frequencies below the LOSC set in powersdr?

You would, as you say need to use both I and Q, and set the image rejection (balance) well.

but I use the entire 96khz bandwidth, 48khz below center and the 48khz above center.

I think you would still want the 10khz offset from center as the LOSC. but if the IF stage of the receiver is wide enough, using the whole bandwidth available, above and below the LOSC works very well.

Maybe I've misunderstood...

There is still an issue with demodulating signals right at the LO (center) frequency, but with a good soundcard the width of that dead-band, or 'bad-spot' is minimal.

Clarification:

I am only talking about PowerSDR operating in Soft Rock mode. I am not talking about Power SDR operating with an SDR-1000, Flex 5000, or a similar software defined radio. When the PowerSDR application software works in conjunction with those radios, it implements the modified algroithm I described in my earlier post.

What I am saying is:

The physical local oscillator in the Soft Rock must be offset from the center of the receiver's i.f. passband by 48kHz if you want to see a (correct) spectrum display that is 48kHz above and 48 kHz below the center of the receiver's i.f. passband. This assumes that the receiver's i.f. filter passes frequencies 48kHz on either side of center (96kHz total i.f. passband width), and it assumes that your sound card is set to a 192kHz sampling rate.

Example, if you want to see 48kHz on either side of 455kHz, then the local oscillator frequency in the Soft Rock should be set to 407kHz.

Separately... with regard to the "carrier frequency" setting in the SDR software...

You can set that to whatever you want. It just affects the frequency that you have to tune to (and that gets displayed) on the SDR main display.

Stu

Not sure why that is Stu.

If I was to set the LO (crystal osc) to 455khz.   Set 455khz in PowerSDR. Use a 96khz sampling card.
I can see and demodulate from 407khz to 503khz.

Obviously it assumes that there is +/- 48khz available from the RX IF passband.

From the very early versions of PowerSDR nothing below the Lo frequency is "thrown away"
ONLY images from either side of LO are Nulled out, but valid signals remain on both sides.

This is where you're confusing me...


edit....

Now I see you said "running in softrock40 mode."

I don't know why there would be any difference in how the I/Q data is processed for the SR40 than for any other I/Q mixer.

I've just used Softrock 40 mode, and see no difference... my local osc frequency is still in the center of the display, and I am still able to see, and demodulate signal above and below the reciever's LO frequency.

Have tried a softrock, a Genesis G80 and a I/Q mixer that name escapes (DDS controller group)

What version of power SDR are you using Stu?

Stu et al

In going thru the archives on the softrock 40 group on yahoo I believe the question was answered by Tony Parks.

The Power SDR application or Rocky will take the entire audio spectrum below the maximum nyquist frequency set by the sound card +/- the center frequency set by the softrock crystal LO divided by four.

The reason for the offset is to avoid undesirable affects caused by
Ground loops between the softrock and the audio board.
DC offset caused by I/Q unbalance.
AC coupling that prevents DC response to the soundcard which would of course negate the first two items on the list unless the sound card is DC coupled.
LO leak-thru which I think can be nulled with I/Q balance, I could be wrong about this one.

At any rate there seem to be cures for most of these problems and they appear to be minimal in real world testing.

The phase noise issue of the digikey custom Xtal LO is of particular importance if the softrock is to be used for close in phase noise measurements as Frank WA1GFZ has already posted about. I have made critical phase noise comparisons of very good programmable clock generator chips vs Xtal reference oscillators where I multiplied both up to 10GHz to be able to actually see the phase noise above a very good spectrum analyzer's LO noise. The result was about a 20 to 30 db difference between the two. This is something that you would never notice for casual spectrum monitoring or SW receiving but would be a show stopper for narrowband phase noise measurements. I will investigate the digikey solution further.


Mark WA1QHQ

that makes the most sense...

If you don't want the "center spike" ( and the very small un-usable bandwidth at the lo freq) to show, you offset, and only use less than half of the bandwidth available to you.

Fact is, even in Softrock40 mode, power SDR does indeed display and demodulate signals above and below the LO freq of the softrock, or any other I/Q "mixer".

Ok

You guys have convinced me.

Stu

This sounds great, the only issue being the raggety PCs around here.

If I was going to the trouble, I'd want to be sure of a 96KHz capability on the sound card.

The people at the computer store do not know anything for example, if I pick up a used IBM NetVista (1GHz CPU, built in sound for $99), does it have the beans in the sound card?

So maybe I ought to try one and see how I like it. Just don't let the CRT based machinery see the heresy..

BTW I have been looking for the schematics of this board. Can someone direct me to them please?

I had a netvista....
The onboard audio device was not great for SDR.

BUT, the only way you'll know for sure is to try it.
Mine was likely a different board/chipset, as it was a 2ghz pentium 4



The schematics for SR40 lite WERE always available in the files section of the softrock40 yahoo group.

If you can't find them, I can email them to you.

I don't believe you're convinced....I bet you just don't want to argue the point Stu!  :-)

Me either, I just wanted to point it out that there's no reason if you have a soundcard capable of 96khz (or 192khz) that you can't use the whole bandwidth available.

sometimes there are circumstances where one only WANTS to use a small portion of the available bandwidth...

Anyhow, even at twice the price the softrock lite was a few years ago, it is a dandy reciever stand alone (40meters and down or up with a preamp) or using it as an IF adapter.

A lot of guys are using external USB sound cards that are of higher performance than what you would typically find in an off the shelf PC.

Schematics and build instructions including BOMs can be found here...

http://www.wb5rvz.com/sdr/

Mark WA1QHQ

Bruce
Et al.

Yes... I actually am convinced. After all of this discussion and throught, perhaps the easiest way to understand this is to look at the three (3) new attachments below.

You start out with the i.f. passband of the receiver (first attachment)... and this time I am showing positive and negative portions of the receiver's i.f. output spectrum.

You mix the receiver's i.f. output with a local oscillator (shown in the first attachment) to produce the spectrum shown on attachment 2 for each phase (I&Q)

However, through the magic of quadrature outputs and Hilbert transforms, you can (to some extent) null out the purple part of attachment 2... to produce what is shown in attachment 3 (mathematically, not as a real signal)

To the extent that you can null out the purple part, you end up with just the red part. Mathematically it occupies frequencies from -W to +W, where W is one half of the sampling frequency.

On the Softrock display, this appears as a spectrum whose full width is 2W... i.e. the sampling frequency.

Thanks for the mental exercise... and for straighting out my misunderstanding of what the Flex Radio algorithm is doing.

Stu