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: Left-3
ad: L-MFJ
ad: abrind-2
ad: HRDLLC-2
ad: Left-2
ad: MessiPaoloni-1
ad: L-Geochron
  1. KK5JY

    KK5JY Ham Member QRZ Page

    Do you have a recommended DBM chip or design to use? My RF fabrication skills are a little lacking, so it would have to be something that is easy to build with. In fact, my whole transmitter project is basically made from well-defined blocks that are chained together. I'm trying to figure out if this is something that can be done as a kit that others could build and use easily.
  2. SM0AOM

    SM0AOM Ham Member QRZ Page

    Any useable home-brew attenuator needs to be sectioned, otherwise parasitic capacitances will ruin the high-frequency response.
    It is not advisable to use larger attenuation values than 20 dB in each section.

    To measure "reasonable" harmonic suppressions, what you are needing is a "selective microvoltmeter" or "measuring receiver",
    which can easily be improvised by a DBM such as the SBL-1, a signal generator and a sound-card.

    Assuming that a dynamic range of about 60 dB is sufficient, 70 dB of attenuation from +37 dBm gives -33 dBm into the mixer which should not stress
    its distorsion properties. 6 dB conversion loss means -40 dBm into the sound-card for the fundamental.

    A 30 dB noise figure in a sound-card is reasonable, so in a 1 kHz bandwidth, there will be -114 dBm of noise.
    Any harmonic component stronger than about 70 dB down from the fundamental will then be detectable with such a set-up,
    which is simple but time-consuming way of measuring harmonic suppression.

    I would suggest that you post the schematic of the output stage here, so it may be determined by inspection if the
    design permits a sufficient harmonic suppression. If the output circuit is a tuned pi-network or a multipole low-pass filter,
    the built-in suppression usually suffices.

    WA7PRC likes this.
  3. KK5JY

    KK5JY Ham Member QRZ Page

    The final PA is this kit: ... I did the recommended mods, including upgrading to a real RF MOSFET.

    The output LPF is this kit: ... one for each band, in a relay-switched array:

    There are schematics in the PDFs at each of those links.

    If you use their entire kit, they drive the 5W PA with a square wave from a Silicon Labs DDS and BS170 switching MOSFET driver stage, and they still seem to get (barely) FCC-compliant output from the far side of the LPFs. The design I'm using is an AD9850 DDS, generating a sinusoid, running through an LM7171 op-amp driver stage. So in theory, it should be at least somewhat cleaner input to the 5W kit PA stage, which should mean the "big spurs" should be even lower than their measurements.
    WA7PRC likes this.
  4. KK5JY

    KK5JY Ham Member QRZ Page

    This is also doable. I'll have to run the calculations to see what resistors I'll need, but I can probably do a "cheap stripline" attenuator made of 20dB sections. It might be helpful to build it that way, because then I don't need gigantic 50-ohm resistors, since each section is only dissipating a fraction of the input power. I guess the first section will burn the most power if the sections are of equal attenuation, but I could make the stages anything I want if I'm dividing it up anyway.
  5. KK5JY

    KK5JY Ham Member QRZ Page

    By the way, I noticed that the MFJ-762 step attenuator...


    ...places the small attenuation values close to the input. Is this so that the dissipation (in mW) per stage is similar throughout the chain? Or is just the way they decided to build it? As I'm tinkering with the calculator, it seems like a multi-stage attenuator would have less stress on the first stage if the power is stepped down a few dB per stage until the power level is small (maybe down 10dB or 13dB from the input level), and then run the big attenuation stages after that. Is that a reasonable assumption?
  6. SM0AOM

    SM0AOM Ham Member QRZ Page

    I would say that your practical harmonic suppression will be limited by wiring parasitics and spurious coupling
    between LP filter input and output. Narrowband antennas with tuned matching circuits will also increase suppression.

    It is reasonable to expect a suppression in the order of 60 dB of the 2nd and 3rd components,which is entirely
    sufficient at this power level. Even when a square wave is fed to the filter, the 3rd harmonic starts at -9dB relative to the fundamental.

    WA7PRC and KK5JY like this.
  7. K7JEM

    K7JEM Ham Member QRZ Page

    No. That's just the way it is configured, since you can run any attenuation or none at all. In any event, it's only good for 125mW of power. But if you proceed it by a 20dB attenuator, you could run 5 watts in and only deliver 50mW to the input of this device. A 20 dB attenuator for HF is pretty easy to build, especially if you only need to have 5 watts on the input.
    WA7PRC and KK5JY like this.
  8. WA7PRC

    WA7PRC Ham Member QRZ Page

    I see LOTS of used legacy SAs listed on a certain auction site for less than $500.

    I also see lots of RF power attenuators from ebay seller "Henry Radio" such as this one for $16.95 shipped (link):
    It's good for up to 50W, up to 2GHz. You just select the attenuator value (1, 2, 3, 4, 6, 10, or 30 dB), and add a heatsink. My homebrew step attenuator uses 1/2W resistors:
    IIRC, on a network analyzer, it looks OK to about 100 MHz... and starts to look wonky above there. But, the price was right. That, or the MFJ unit would be OK for testing a piece of gear on lower HF. Above HF, you'll likely want something better. ;)
    Last edited: Oct 20, 2017
    KK5JY and K7JEM like this.
  9. WB2WIK

    WB2WIK Platinum Subscriber Platinum Subscriber QRZ Page

    That actually won't work, at least not at radio frequencies.

    Reason: The series resistor has capacitance and coupling between input and output will greatly reduce the amount of available attenuation -- you'll never hit anything close to 80 dB.

    That's why an "80 dB attenuator" is usually made of eight 10 dB attenuators in series or at least four 20 dB attenuators in series, with shielding between the sections.
    KK5JY likes this.
  10. KA9JLM

    KA9JLM Ham Member QRZ Page

    I am trying to figure why a variable attenuator is needed for this.

    Is it for guessing when you have no clue ?

    A fixed attenuator should work just fine.

Share This Page