Harmonic operation

I have done this with My UHFSDR to push its UHF operation up to 1Ghz (limited by Mixers) and i thought i would do that same test with the LimeSDR

i has asked some time ago what kind of signal is sent to the mixers in the Limesdr and was assured that it was Square wave …—… that was important to me —
UHFSDR also uses a square wave Lo.

UHFSDR running Quisk (Odroid-X2/Ubuntu16.04-32bit)
LimeSDR runnung GQRX (Odroid-C2/Ubuntu16.04-64bit)

Jim Veatch sent me these Band pass filter for 2M band and has a 3db point at 180Mhz and drops fast from there …

The idea was to test the possibility of operating with reduced performance above the 3.8Ghz Lime limit.

What is 54.166 X3? 162.498
What is 3.456Ghz X3 ?

The Large spike next to the weather transmitter is the leaking Lo of my UHFSDR.

Unknown yet is max frequency range of LNA and the Mixers … TIA and the PGA (what ever that last amp’s called – i forget) does not matter since they are IF range.

No, i don’t believe that the I/Q transformers can go to 10Ghz with out alot of attenuation … there would need to be some modifications …

Also in the video the weather transmitter is on the left of center – i think it should have been on the right side but i need to recheck what i did …

What if you were to use an external circuit to do the TR switching and trippling amplification.

Larry in El Paso

Im sure that you could …

T/R switching might be needed for some applications …

Its still unknown what the unfiltered , unmatched output of the LimeSDR is …

The Mixer’s Lo is a square wave and should be rich and powerful ODD harmonics.

My Avatar shows a spectrum analyser depiction of my UHFSDR output. Fundamental is set at 28Mhz and 3rd harmonic is 9db below the fundamental. Each odd harmonic out is less and less reduced. The 17th harmonic is 30db down (UHFSDR) …

Im not sure it a trippling amplification is needed … It depends on what your after – are you looking to achieve a specific freq?

7th harmonic of 3.456Ghz (24.192Ghz) is getting up there … I wonder when the board material becomes a factor in signal loss … Theres no way that the existing board components – especially the I/Q transformer – will pass that kind of a signal, i don’t think.

Oh … here is my Avatar !!!

That reminds me that I meant to enquire about driving the LimeSDR GPIO. @Zack, @joshblum, do you know if this is accessible via an API? And if not, where would be the best place to add this?

T/R switching, like transmit receive switching for a TDD use case? This is something we are going to need FPGA control over because this needs to be automatic. Typically for this sort of use case I would design a set of registers in the FPGA that drive the frontend switch IOs based on a Tx condition.

  • Theres two registers,
    • one that represents the IO state when transmitting,
    • and one that represents the IO state when transmitting is off (receiving).
  • The FPGA uses the the transmit state (driven by tx fifo activity) to switch between these registers
  • Then the driver needs to setup these registers in advance given some configuration by the user.

From the software side, this register configuration is usually based on an antenna selection. But in the case of the limsdr, there is no frontend switching, so there isnt yet a way to convey this software wise. But with some fpga support and target use case, I’m sure we can imagine a special antenna selection that engages this configuration.

Also with a lot of radio amateur use there will be a single antenna, with TX and RX on the same frequency and the need for PTT switching. I believe that a popular API for apps with this on Windows is ExtIO.

Sounds like we need to schedule the work required to support both these use cases, plus perhaps other non-TX/RX switching uses. Given there are multiple GPIOs, some could be used for switching in/out external filters and amplifiers etc. also. So some simple mechanism for just toggling GPIOs would be nice.

We definitely want GPIO!! We’re going to not only want it for PTT in and TX/RX, but filter switching as well. From my testing, we’ll want bandpass on the input and harmonic filtering on the output. The odd order TX harmonics can be quite high depending on the band.

I did some harmonic testing at 10/24/47/78 GHz. The LimeSDR makes a good signal source with a few add-ons…

This is the LimeSDR set up on the bench. I GPS locked it for Freq stability, That is the pigtail on the top right. You can see he Lime is locked if you look at the LED’s. The fourth LED (below the red one) that is usually off is on Green. This indicates Lock. The antenna is WA5VJB 2-11 GHz LPA.

On 10/24 GHz, I used 3456 out of the LimeSDR. With the 10 GHz rig still sitting on a shelf, signal was S9+. I didn’t bother to point it.

Next I set up 24/47 GHz. I left the LPA on for the first test. LimeSDR still set at 3456. Here you can see the S meter at S9 on 24,192.

I changed over to a WR-42 horn and signal was S9+10. A bit hard to see exactly on my trusty FT-790 analog S meter…

Next, 47 GHz. Signal was a good S9. Both 24 and 47 are mounted on the same frame and they were about 6 feet away.

For fun, I tried 78 GHz. Here’s the rig set up.

Signal was very good! Here’s a shot of the signal in HDSDR.

The movement in the trace is because the rig had not fully warmed up yet. I should qualify the measurement. My 78 GHz rig uses a WA1MBA Amp/LNA and is highly optimized. I have earned VUCC #2 with it. Longest DX so far is 205 km in SSB.

I’m not sure I would see the signal without the LNA…

Fun Stuff!

How about receiving?

Hi Mike,

Are you inquiring about maybe 2M and up through 3456 for receive?


No, I already use the Lime on 2M, 1.25 and 70cm Rx…

Im wondering about Rx with existing hardware (minus the MN19 was it?) up to 10Ghz …
Looks like your equipped to test farther above …

I have a HB100 and a 10.4Ghz GunnPlexer … both i have yet to power up …

Also — some day — i plan on using a HughesNet internet dish and hardware (minus the modem) and its starting to sound that the LimeSDR would work wonderfully with a 12Ghz HughesNet i think

Hope you get some time to try the LimeSDR for Rx this weekend in the microwave bands

Just to see what can get through the 3.8Ghz transformer …

Just do some quick test:

  • Using input RX_H , signal generator set at -60dbm , set signal generator at 3GHz and tune LimeSDR to 3GHZ by SDRconsole V3.0 , LNA max , TIA Max , signal measure on SDRconsole -24dbm, change signal to 4.5GHZ and tune SDRConsole to 1.5GHZ signal measure on SDRConsole -55dbm, change signal to 6GHz and tune SDRconsole to 1200MHZ signal measure -73dbm, at 10GHz signal measure -74dbm , SDRconsole tune at 3334MHz

So i infer that the base line gain @ 3ghz is 36db … (also fundamental Lo)
then @ 10ghz (input is still -60dbm) that the chain has a loss of 13db

One of the assumptions ill haft to make is the lower signal level FROM the 3rd of the Lo. Its never the same level of the fundamental. Im going to go with 12db less than fundamental.

Im not sure what it means right now … I think most people would use an LNA for the noise figure and not the gain.

Tried something for 10Ghz Rx too …

I dont have much gear for 10Ghz … Im open to donations : filters, waveguide, coax transitions …

I don’t think you’re seeing 10 GHz.

Powered up my 10 GHz rig. It uses a 27 dBd dish driven by a 4.5 watt amp. LimeSDR had a WA5VJB 2-11 GHz LPA on it. Tuned to 3456. 10 GHz transmit at 10368.

Seeing Nada…

Darn … my reply, question from email while i was cleaning in the garage didn’t post …

If that is so … are you sure that the Lime was Tx’ing above 3.8Ghz or was there a signal mixing in the Rx chain of your various microwave radios ?

The Rx1_H parts layout isnt that different from Tx1_2 (EDIT reason**its listed a W not H) save a 12nH(L13) … hope i have the V1.4 schematic.

Im wondering if you goofed with the antenna selection … I have a microwave associates 86503 gunnplexer to try(ied) and i did read in an old article that they are fairly spectrally pure. I’m not sure 100% on that – one man’s old reference might not hold water anymore …

I tried the Gunnplexer and looks about the same as the HB100 …
Ill have another video for N1JEZ and WA5VJB to see in a bit …

Added video now …

and had the averaging setting too high … set down to 0 seemed to help

Completely forget to calc what frequency the MA-COM 86503 was running at …

Really need to get the LimeSDR in a shielded case … this signals might just be from Rf getting into a Clock gen or who knows … I just don’t have anything yet …

I accidentally left the LimeSDR/Odroid-C2 running overnite … and it dawned on me that one way to check if the signal was coming from the antenna or coupled into the LimeSDr is some other, way was to simply switch gqrx to one of the other antenna ports.

With the MA-86503 running again … i switched to Rx1_L and Rx1_W and indeed the signal diminished significantly …

I could get the gunnplexer signal to show in Rx1_L and Rx1_W when the gunnplexer was brought less than a foot from the LimeSDR.

Next thing i should do is wire in a waveguide to coax transition (or try to make a 10Ghz dipole with the end of the Wifi coax) and eliminate the Laptop Wifi antenna to see if it helps reduce the phase noise / FM looking signal that’s with the main spike.

Hello Mike,

Just looking what he is doing and there is a possibility that he is receiving the 10GHz signal.
What I can see, his LO is set on 1/3rd of the receiving frequency and the receiver is performing the sub-harmonic mixer with zero IF. The losses a big as he is using the gunplexer at 10ghz, so at least 10dBm into a small horn antenna and he can barely see the signal.

I am using the same approach when some signal on 76 Ghz or 122 GHz is required on my bench. As my HP SA is going up to 22Ghz (max 23 GHz) I am using a harmonic mixers to extend the operation on 38, 47, 76 and up to 140GHz.

Of course, the loses in such LimeSDR harmonic mixers are so big that this is almost unusable.