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-BCInc
ad: abrind-2
ad: l-gcopper
ad: Subscribe
ad: L-MFJ
ad: Left-2
ad: Left-3
  1. KK5JY

    KK5JY Ham Member QRZ Page

    Yes, it says that. In its original configuration, I believe it is class D, since that configuration uses a switching FET, and is directly driven by a square wave DDS and driver.

    There's no way it can be class C, no matter what they say, because it doesn't depend on a tank circuit to develop power, and also because the wave shaping survives the PA stage. As you pointed out, if it was really Class C, that wouldn't be the case.

    The configuration I am using is driven by a sinusoid, and the switching FET has been replaced with an RF MOSFET that is sufficiently biased that it should be Class A.
  2. G0HZU

    G0HZU QRZ Member

    They would have to be quite close to you, probably within a km or so. The spurs from a DDS like this depend on the frequency word sent to the chip. So for some frequencies the DDS will generate spurs that are a bit more obvious and they may fall in band at -65dBc to -75dBc. But on a typical ham band they will probably be -75dBc most of the time. Further away they can easily hit -65dBc. At 29.7MHz output with a 125MHz clock I think the (Fclk -3RF) will be at 125-(29.7*3) = 35.9MHz and this spur could be as bad as -55dBc. But it will probably be closer to -60dBc. Across the rest of the HF band the spurs should be typically -65dBc.
  3. KK5JY

    KK5JY Ham Member QRZ Page

    Oh, now I see what you are talking about. I believe one of the Analog publications talks about that. I can't remember if it was in that data sheet or some other document. But yes, I would expect the impurities from the DDS process to be more pronounced as one approached the clock frequency. For this project, I'm just focusing on HF bands, which keeps the carrier rather far from the clock.

    Now that I'm thinking about that, I never thought of the DDS as being a source for decreasing gain as the frequency increased. I have been blaming my driver circuit, but I wonder how flat the DDS is from 80m to 10m... I should be able to measure that with an oscilloscope, and that is something I do have access to... I'll have to check the probes to see how frequency-flat they are.
  4. G0HZU

    G0HZU QRZ Member

    The raw output from the DDS chip pin should follow a sinx/x response so by 30MHz (with a 125MHz clock) the level at 30MHz should drop by just under 1dB wrt a low frequency like 1MHz. However, a lot depends on the reconstruction LPF after the DDS as this will cause more droop and ripple in the DDS response. But a typical 9850 will follow the sinx/x profile quite closely at the raw output pin of the chip. So you can predict the droop at 30MHz with a simple equation. If your measured results differ then it could be due to loss/ripple in the LPF after the DDS or maybe an issue with the flatness of the test gear.
  5. KK5JY

    KK5JY Ham Member QRZ Page

    That might also explain a lot. This is just one of the cheap AD9850 boards that have been made in large quantities. In fact, they are cheaper to buy in the board form than it is to get the IC itself in singles. The schematic I found for that board shows that it uses three inductors in series, mixed with seven capacitors (three in series, four shunt). So I can see where the roll-off might be more pronounced than a simple sinc function.
  6. G0HZU

    G0HZU QRZ Member

    I've seen the ebay AD9850 DDS boards and they look to be amazing value for the money. I've not actually played with one of the ebay boards myself but one concern I have is that they appear to be based on the 'clockgen' evaluation circuit for the AD9850. This uses a 200R elliptic filter with high passband ripple. So the LPF on the ebay module may not be a 50R filter and it may be designed to have a fair bit of passband ripple. Is it easy to spot if the termination resistors are 200R or 50R on your DDS PCB?

    I think there are still things that could go wrong because the PA could develop its own spurious signals through various forms of instability. Usually when this happens it is fairly obvious something is wrong.

    If you have access to a fairly modern digital scope with a high sample rate (i.e. >1Gsa/s) you could try playing with the FFT feature on it to use it as a crude spectrum analyser for the HF bands. However, much care will be needed to avoid confusing alias terms that may appear on the FFT display. Also most scopes are only 8 bit so this will limit the spur free dynamic range to very close to your own spec requirements. Also, you will be limited in frequency range by the bandwidth of the scope and the rolloff in the FFT response as well. But a reasonably modern scope with a fairly decent FFT would at least give you some idea as to the spectral purity (to spot obvious instability problems) and it can do it quite quickly. The problems start if you try and lower the sample rate because the scope FFT display can quickly fill with confusing alias or sampling artefacts that will look like spurious terms.
    Last edited: Oct 22, 2017
  7. N1OOQ

    N1OOQ Ham Member QRZ Page

    That's what I've found with the digital-scope FFT functions... they're often so full of sampling artifacts that you can't make heads or tails of them.
    Usually, if there's any serious departure from a clean sine wave, you can see it in a simple scope trace. For instance, something 30 dB down is at an amplitude of about 3% of the fundamental. This is noticable from peak to peak on the scope.
  8. KA9JLM

    KA9JLM Ham Member QRZ Page

    I can hear in now.

    You: Honey can I build a radio kit, It only cost 20 dollars.

    Spouse: OK that is fine.

    You: Honey my new radio is built. I need to get a spectrum analyzer to adjust it. It cost 1500 dollars.

    Spouse: I want half of everything. :mad:

    K0OKS and AI3V like this.
  9. KK5JY

    KK5JY Ham Member QRZ Page

    I have heard that story from other guys. I haven't lived it myself, but I have heard it from others. ;)
  10. WA7PRC

    WA7PRC Ham Member QRZ Page

    Some of these look good enough for the purpose:

Share This Page