No RF gear here?

I got my Red Pitaya to see the internet tonight - in a roundabout way ...

I got a cheap ($10 bucks - probably full of backdoors) generic wifi range extender off eBay, connected that to my network here at the lake, ran a cable from the extender LAN port to the Pitaya, and just like that the Pitaya was online to the internet ... Got the OS upgraded to the latest version, and downloaded a few apps from the bazaar.

I was just trying to set up a wireless connection to it, and didn't expect the RP to go online, as it has refused all other attempts to get it to see the internet.

The wifi extender conveniently has a 5 amp USB port, so it can power the Pitaya as well as providing a wireless connection.

Win W5JAG
 
The problem is caused by the stereo decoder. Ideally, a 38 KHz sine wave is generated from the 19 KHz pilot, and multiplied with the input signal. That results in AF from the 38 KHz side bands which with the LPF-ed input signal are fed to a matrix circuit to produce L and R. Such a decoder was immune to SCA and is immune to the HD side bands. It's cheaper to multiply with a square wave and thus arises the interference from the 15 KHz wide side bands of odd harmonics of 38 KHz. Various solutions have been used, best of all was by Pioneer. It used a digital demodulator (monostable) at 1.4 MHz and the pulse train was used to switch a (with the 19 KHz pilot locked) 38 KHz sine wave. Various articles at 88?108 MHz

One solution to IBOC self-noise on 50% duty cycle square wave MPX decoders that demodulate random noise with the 3rd and 5th harmonic of the 38kHz square wave is an IBOC filter. The IBOC filter removes most of the noise present in the 114kHz and 190kHz region.

I have designed and built a filter that has minimal impact on stereo separation.

Here's a plot of the early version.

https://sites.google.com/site/ej25awd/IBOC Filter Frequency Response 1024.GIF

My latest iteration was installed in a Tandberg 3001A and the measured stereo separation degradation was negligible as shown by the plot below.

Cyan and Green are without the IBOC filter and the Yellow and Red are with the IBOC filter installed.

attachment.php
 

Attachments

  • Tandberg 3001A stereo separation with proper IBOC filter board.GIF
    Tandberg 3001A stereo separation with proper IBOC filter board.GIF
    54.2 KB · Views: 264
Member
Joined 2018
Paid Member
One solution to IBOC self-noise on 50% duty cycle square wave MPX decoders that demodulate random noise with the 3rd and 5th harmonic of the 38kHz square wave is an IBOC filter. The IBOC filter removes most of the noise present in the 114kHz and 190kHz region.

I have designed and built a filter that has minimal impact on stereo separation.

Here's a plot of the early version.

https://sites.google.com/site/ej25awd/IBOC Filter Frequency Response 1024.GIF

My latest iteration was installed in a Tandberg 3001A and the measured stereo separation degradation was negligible as shown by the plot below.

Cyan and Green are without the IBOC filter and the Yellow and Red are with the IBOC filter installed.

760198d1559364895-rf-gear-tandberg-3001a-stereo-separation-proper-iboc-filter-board-gif

Separation graph looks good. It would be interesting to see the phase response of the filter. A band stop filter (99...129k and 175...205k might introduce less phase shift than a low pass filter. I'll investigate the issue as an almost ideal stereo decoder could be made with just 2 S/H circuits plus proper (but simple) timing derived from the 19 KHz pilot: no multipliers or matrix needed, nor a 15 KHz low pass filter. It's the issue that such a decoder is sensitive to all 38 KHz harmonics that requires attention.
 
What matters re base band MPX filters when multiplier or SH is used, is both amplitude and phase.

I know about how critical amplitude and group delay flatness are on the MPX baseband that is why my custom hand-built IBOC filter fulfills both. Check my earlier link and you'll clearly see that it is actually a dual notch and not just a plain low pass filter.

Phase shift is inevitable. As long the phase shift is consistent and results to a flat group delay, faithful processing of the filtered composite/baseband signal is ensured. The before and after stereo separation plot clearly shows that this has been achieved.
 
Today it was too warm for me me to go outside and work in the garden
(good excuse), and my RP was collecting dust much too long.
So I installed HDSDR in a windows7 virtual machine and also Gnuradio
and GnuradioCompanion on Linux. Both seem to work.
HDSDR has too much windows, waterfall diagrams of RF and demodulated etc.
I'd like sth. simpler, just enough to do QSOs. It also should be able to
decode / send CW. Have not done that since eons.

Charly25 and the RP transceiver seem to require extraneous hardware; that
is exactly what I do not want. I want nothing but a 432 MHz transverter after it
to do some motorcycle-portable EME QSOs with my pal with this huge antenna.

This is the 10.001 MHz output of my SMIQ generator in GNUradio companion at -11.3 dBm.

:) Gerhard
 

Attachments

  • 10.001NHzFrom SMIQ.png
    10.001NHzFrom SMIQ.png
    29.3 KB · Views: 134
Last edited:
Take a look at RadioBox by DF4IAH.

It is specific just to the RP and looks like it makes the 14 bit RP a standalone 0-60 MHz all mode rig, so a transverter and laptop just might be all you need for portable ops. It looks like the RP sensitivity is about 75 uV for SSB. Didn't see a CW number.

It is on the RP Bazaar or GitHub.

Win W5JAG
 
Member
Joined 2018
Paid Member
I know about how critical amplitude and group delay flatness are on the MPX baseband that is why my custom hand-built IBOC filter fulfills both. Check my earlier link and you'll clearly see that it is actually a dual notch and not just a plain low pass filter.

Phase shift is inevitable. As long the phase shift is consistent and results to a flat group delay, faithful processing of the filtered composite/baseband signal is ensured. The before and after stereo separation plot clearly shows that this has been achieved.

The graph on the google site denotes a low pass filter with damping poles, and is easy to reconstruct from the graph. Shown on the pic, only the amplitude response.
 

Attachments

  • IBOC fl.jpg
    IBOC fl.jpg
    91.5 KB · Views: 115
The graph on the google site denotes a low pass filter with damping poles, and is easy to reconstruct from the graph. Shown on the pic, only the amplitude response.

The group delay of your design from 10Hz to 53kHz is almost 3x worse and the amplitude flatness is 0.07dB worse. It is close to what I have but not as flat. The seemingly small differences do matter since MPX decoding rely on the flatness of these 2 parameters.
 
Member
Joined 2018
Paid Member
The group delay of your design from 10Hz to 53kHz is almost 3x worse and the amplitude flatness is 0.07dB worse. It is close to what I have but not as flat. The seemingly small differences do matter since MPX decoding rely on the flatness of these 2 parameters.

The reverse engineering from the graph was a 3 minute job for me and as I don't need such a filter, I didn't optimize it.
 
Take a look at RadioBox by DF4IAH.

It is specific just to the RP and looks like it makes the 14 bit RP a standalone 0-60 MHz all mode rig, so a transverter and laptop just might be all you need for portable ops. It looks like the RP sensitivity is about 75 uV for SSB. Didn't see a CW number.

It is on the RP Bazaar or GitHub.

Win W5JAG

It seems RadioBox is more experiment-oriented and not really for QSO-ing.
Interesting in the long run probably. Without CW, I could live with that for
receiving, but to implement the TX, that would be a lot of work; esp. since the whole
system is Verilog and I'm a VHDL guy. 15 years earlier, it would have been
impossible for me to say that -eeek- , but my design customers have turned me
around, a little bit, year after year.
I think, I'll stay with HDSDR for the next time.

I've been looking for a simplistic transverter for 32 MHz + 400 = 432 MHz.
I wanted to do 100*2*2 for the injection frequency, but there are things like
< SG3225EAN 400.000000M-KEGA3 EPSON | Kristalle, Oszillatoren, Resonatoren | DigiKey >
and it seems like a cheap 1st solution.

OK, impossible to lock against GPS but better than glorious but never finished at all.

For the receiver input I have chosen this one:
< SKY67150-396LF Skyworks Solutions Inc. | RF/IF und RFID | DigiKey >

and for the PA that:
< CGH40045F Cree/Wolfspeed | Diskrete Halbleiterprodukte | DigiKey >

40 Watts should be plenty with this partner station. There is still a hole between
400 mW from some AD SOT-89 amplifier and the 4? Watts into the PA.
I have some LDMOS from NXP, but it runs from 12V, not 28V like the PA and the
coax relay.

The whole thing will work from 8 18650 Lithium cells; maybe 2 or 3 extra for the
low-voltage stuff.
 
Last edited:
The reverse engineering from the graph was a 3 minute job for me and as I don't need such a filter, I didn't optimize it.

That is impressive :worship:

Here's the simulation of my design. The series R of the simulated inductors were taken from the actual prototyped values. Sorry no schematic.

attachment.php


The IBOC filter is still useful for minimizing splatter of an adjacent station when listening in stereo. I have a distant station that is 200kHz away from a local station that I'm receiving in stereo with minimal interference when using the IBOC filter.

A tuner even with a narrower IF filter will sound harsher without the IBOC filter. The IBOC filter will help in reducing adjacent interference. Stereo reception in general will also be quieter since we are not decoding most of the noise present above 53kHz.

The IBOC filter is actually an adjacent + RDS + IBOC filter.
 
RadioBox is definitely not a band cruiser, but looks like it would be OK for a single fixed frequency.

It looks like it will send RX audio out to a laptop, but TX audio needs to be supplied externally? So I guess one would need an audio board in addition to the transverter.

I got the wi fi dongle working on my Pitaya this afternoon, but I'm not really sure how I did it ... It seems to be really picky about what it wants to connect to through wi fi, and how wireless security is handled. Not sure if that is a feature, or a bug.

Win W5JAG
 
Worst thing is that you cannot trust the documentation. A different Linux
and everything is different.

My 432 MHz preselector filter was easy. I took a SAW filter from
EPCOS/TDK or whoever owns that company this week.
It is for remote controls and costs €1,50 @1 pcs.
Input and output are close to 50 Ohms, bandwidth is just enough.
loss is 2.7 dB, my mirror frequency is 70 dB down.
It can't get much simpler. 3*3 mm.
 

Attachments

  • Auswahl_001.png
    Auswahl_001.png
    532.6 KB · Views: 60
  • dsc_0514_gimp.JPG
    dsc_0514_gimp.JPG
    321.1 KB · Views: 55
Last edited:
... I've been looking for a simplistic transverter for 32 MHz + 400 = 432 MHz. I wanted to do 100*2*2 for the injection frequency, but there are things like < SG3225EAN 400.000000M-KEGA3 EPSON | Kristalle, Oszillatoren, Resonatoren | DigiKey >
and it seems like a cheap 1st solution.

So you are going to block convert down to and up from 32 MHz and use the RP as a tunable IF and RF source?

edit:

I infer from RadioBox that the RP can do up to 200 mv peak to peak out of its RF ports. This output level would work fine with the HF transverter I am still struggling with. I'm building for 100 mv TX input, but it will handle up to at least 300 mv without problems.

I am not sure I have enough RX gain if the RP sensitivity is 75 uV which is a number I saw in a RadioBox thread on the RP forum.

Just thinking, in case I want to use the RP instead of the hardware SG-9 ...

Win W5JAG
 
Last edited:
So you are going to block convert down to and up from 32 MHz and use the RP as a tunable IF and RF source?

Exactly. And if I used a 100 MHz *2*2 oscillator, it could also work
for 144 MHz without the f-multiplier and 44-48 MHz IF.
There is also a 400 MHz SAW filter for cleaning up after the multiplier.

A preamp should be easy for you. :)
 
Just a parenthetical -- but as the inventor of frequency skipping you might be interested in this film about Hedy Lamarr which is distributed by Zeitgeist Films and showing around the US at various locations. Bombshell: The Hedy Lamarr Story :: Zeitgeist Films

If Nikolai Tesla had her looks, he wouldn't have died broke!

This evening 8:00 pm on PBS (WNYC) -- a one hour documentary on Hedy Lamarr's inventions.
 
I have made a preamp for 432 this weekend.
Gain is 3 dB lower than promised for the demoboard and the
SAW filter is also not as good as measured alone.
I have not measured the noise. If that is OK, I won't complain.
I need 2 stages anyway and with 2 filters all will be nice.
The two stage version is etched already. it will have an additional
pin diode attenuator. And everything no-tune.

cheers, Gerhard
 

Attachments

  • singleSKY67150.png
    singleSKY67150.png
    592.9 KB · Views: 66
  • preampMitSAWfilter_einstufig.png
    preampMitSAWfilter_einstufig.png
    910.2 KB · Views: 60
Last edited:
Member
Joined 2018
Paid Member
That is impressive :worship:

Here's the simulation of my design. The series R of the simulated inductors were taken from the actual prototyped values. Sorry no schematic.

attachment.php


The IBOC filter is still useful for minimizing splatter of an adjacent station when listening in stereo. I have a distant station that is 200kHz away from a local station that I'm receiving in stereo with minimal interference when using the IBOC filter.

A tuner even with a narrower IF filter will sound harsher without the IBOC filter. The IBOC filter will help in reducing adjacent interference. Stereo reception in general will also be quieter since we are not decoding most of the noise present above 53kHz.

The IBOC filter is actually an adjacent + RDS + IBOC filter.

Filtering is vital for all RF gear and for most tuners an IBOC filter will improve reception of stereo transmissions. One alternative would be to replace the stereo decoder, preferably with one that only needs the lower side band (23...38 KHz) but it's fairly complicated. I'll have to decide which solution to take for my DIY FM tuner: a discrete version of the (famous) Pioneer digital system or linear multipliers, preferably with the LSB option as that will provide the best S/N ratio. DIY tuner under construction already...
 

Attachments

  • FM tuner.jpg
    FM tuner.jpg
    57.3 KB · Views: 60