Need to Measure Transmitter Spurs, w/o Spectrum Analyzer

Discussion in 'General Technical Questions and Answers' started by KK5JY, Oct 19, 2017.

ad: L-HROutlet
ad: l-rl
ad: L-MFJ
ad: Left-2
ad: abrind-2
ad: Subscribe
ad: Left-3
  1. KK5JY

    KK5JY Ham Member QRZ Page

    I'm assuming that using this method, the LPF filter cutoff needs to be between the fundamental and the 2nd order harmonic, correct?
  2. SM0AOM

    SM0AOM Ham Member QRZ Page

    This is also correct.
    I recall this method being used in the 60's for checking for harmonics from 21 and 28 MHz in TV band 1,
    and for VHF parasitics.

    A spectrum analyser in those days literally cost "an arm and a leg"...

    KK5JY likes this.
  3. AI3V

    AI3V Ham Member QRZ Page

    Well you start by not "double terminating" your tx.

    Loose the t and dummy load.

    Get a attenuator suitable for your tx power, and enough db's to bring your tx power down to somewhat below your sdr maximum level.

    You test this by adding some more attenuation, if you add 6 DB, ALL the signals should go down by 6 DB.

    Then you work on shielding the tx and sdr.

    Metal boxes and (temporary) battery power are your friend here.

    After all, if your tx is so poorly shielded it radiates from the power wires.......

    Good luck, you can do this. And you will find shielding and decoupling is not expensive or hard, just attention to detail.

    KK5JY likes this.
  4. AI3V

    AI3V Ham Member QRZ Page

    Purists will point and laugh, but 75 ohm attenuators go for a LOT less than 50 ohm ones on the hamfest circuit.

  5. KK5JY

    KK5JY Ham Member QRZ Page

    Well, it isn't shielded, but I hope it's not radiating from the power wires... the choke I have on the 13.8V wires is massive. :cool:
  6. G0HZU

    G0HZU QRZ Member

    If you have access to a reasonable (as in spectrally clean) sig gen you could make up a crude downconverter using the sig gen as the LO for a double balanced diode ring mixer and use a PC soundcard as the final IF at (say) 15kHz. The run some freeware spectrum analyser SW on the PC.

    You now have a crude spectrum analyser and the flatness response of the analyser will depend on the quality of the double balanced mixer. If you use a Minicircuits mixer here it should be fairly flat up to UHF.

    Note that there will be no image rejection with this method so every spur will appear twice as you tune the sig gen across the spectrum looking for spurious terms on the soundcard. This method will be quite slow, as it will take as long as it would take to tune a receiver across the range of interest. So it could take a few minutes to just do one sweep. But you did say it needed to be cheap...

    Even if you get access to a lab spectrum analyser you need to make sure the analyser is set up correctly for the measurement if you want to keep the overall measurement uncertainty of the system within a couple of dB. This means that the analyser's own internally generated harmonic distortion will typically need to be 15-20dB lower than the harmonic performance you wish to verify.
    In the case of the simple diode mixer and soundcard, the 2HI of the mixer might be +35dBm so to get reasonable measurement uncertainty you would have to make sure the mixer wasn't driven any harder than -30dBm. So for a 100W transmitter you would need 80dB attenuation between the transmitter output and the mixer input.

    This drive level of -30dBm would give 2nd harmonic distortion of about -65dBc in the mixer allowing reasonable uncertainty when trying to prove harmonic distortion of -50dBc in your radio. i.e. the mixer has to be 15dB better than the radio in order for the distortion in the mixer to have limited (< +/- 2dB?) influence on the measurement.
    You can also use this method to look at IMD or splatter on the main signal when it is modulated. Either use speech or a multi tone test signal for this?
    Last edited: Oct 19, 2017
    KK5JY likes this.
  7. KK5JY

    KK5JY Ham Member QRZ Page

    Well, the transmitter is actually pretty simple. It's just a carrier. I used the QRP Labs 5W PA circuit, so the only modulation is the envelope shaping at the edges of the carrier pulses. So hopefully the mixer products that are present in SSB radios won't be a problem here.
  8. KK5JY

    KK5JY Ham Member QRZ Page

    It took me a bit to think through what you are suggesting here, but this could be fairly easy to build, I think. If I understand correctly, you are suggesting that I burn nearly all of the transmitter power in the attenuator between the transmitter and the receiver, so that the path is a single shielded coaxial run with 50 ohms impedance. If that is correct, then what survives will be just enough of the carrier and any surviving spurs that I can just measure them directly in the receiver at that point. So it's not all that different from what I was trying, just wired differently.

    I'll have to go find the formulas for a PI attenuator, but making a 60dB or 80dB attenuator out of a few resistors should be straightforward. IIRC, the calculations are fairly simple for a resistive PI network, and I only need it to be able to withstand 5W for short bursts during measurements.

    Edit: I just put the numbers into an online calculator, and it suggested that an -80dB PI attenuator would use two 50-ohm resistors on the ends, and a 250-kohm series resistor. In hindsight, that seems pretty obvious, but either way it should be easy to build.
    Last edited: Oct 19, 2017
  9. G0HZU

    G0HZU QRZ Member

    With the soundcard method you could key the transmitter with a stream of dits at the equivalent of (say) 5wpm and then look at the OOK spectrum on the PC spectrum display. This will be able to display the OOK waveform very clearly and with a fast refresh rate. Much better than many oldschool HP swept analysers in this respect. The classic old HP analysers will be way too slow here and they won't have a low enough resolution bandwidth to display the 5pwm OOK waveform in full detail. They will be fairly hopeless for this test. But the mixer/PC/soundcard can do it very well indeed as it can exploit modern FFT techniques to achieve a very low detection bandwidth:)
    This would also give you a good indication if your transmitter has adequate filtering on the keying waveform. This will give an idea how well your transmitter is set up to prevent key clicks.
  10. G0HZU

    G0HZU QRZ Member

    Getting it to have a flat response up into the UHF region will be harder than you might think. Especially if you use old school leaded resistors rather than SMD. Best to go for chunky SMD parts here and make up a finely graded attenuator with lots of sections?

    Trying to do it all with just a few Pi sections (let alone just one!) will end in disappointment if you expect it to work up to UHF. You could easily end up with 8-10 Pi sections in a successful design?

    eg 0.5dB + 1dB + 1dB + 2.5dB + 5dB + 10dB + 20dB + 20dB = 60dB in 8 sections using SMD parts to make a graded 5W 60dB attenuator?
    Last edited: Oct 20, 2017
    WA7PRC likes this.

Share This Page