8
$\begingroup$

An answer on this question Suggested method for boosting 0.05 watt transmitter up to 10+ watts? suggests putting a filterless 19dB amplifier between the HackRF (or similar general-purpose SDR) and a larger amplifier — so only the final stage includes much in the way of filtering.

Is some caution warranted here for spurious emissions?

As far as I can tell, the HackRF's own filters are before (from a transmit perspective) its RF amplifier stage.

The project's block diagrams aren't very helpful in this regard, but http://www.wimo.com/hackrf-sdr_f.html has a detailed write up ("After the antenna input two MGA-81 GaAs MMIC amplifiers follow, one serves the input, the other the output path.") and provides this frontend block diagram:

HackRF frontend block diagram showing MGA81563 amplifier chips as the last step before antenna output

So after its filtering, the HackRF adds a +14dB gain of its own. After which, we might add something like the PennyWhistle, QRP-PA 2008 or PA-100 to boost the SDR's still-small signal closer to the 5W range — something like another 15–25dB further gain. Finally we might put this through something like the HF-Packer or HARDROCK-50 amplifier which provides another ~10dB gain, before the signal finally getting filtered again by that amp's LPF section.

Is all this gain going to add more spurious emissions than the power amplifier's LPF section is likely designed to handle? I've been told that one really should filter after every stage of amplification, but in this setup the signal goes through three stages before filtering. Perhaps the "pre-amp" stages are much easier to keep clean/linear, or is that a bit too optimistic?

How could I use the specifications for each of the pre-amp stages to determine all the expected types and levels of spurious emissions they are adding? Do these harmonics compound from one stage to another, i.e. if the input to the external pre-amp is already not clean how do I take that into account when calculating the distortion it will add?

Improve this question
$\endgroup$
2
  • 1
    $\begingroup$ The post-amplifier filters are there to strip off the harmonic distortion from the amplifiers, but I suspect the filters in this diagram are for another purpose: the signal coming off the mixer is essentially a square wave and so has a ton of harmonics. You want to filter those off ASAP so you don't waste power amplifying them. $\endgroup$
    – Phil Frost - W8II
    May 15, 2016 at 1:23
  • $\begingroup$ This will depend on your system, really. You may not need any additional filtering if the amplifier is sufficiently linear and the input signal is clean. It may also be that your amplifier will cause too many spurious signals to be generated (above, below, and harmonic to your desired signal) such as in Figure 6 of wp.ni4ce.org/wp-content/uploads/qst.pdf $\endgroup$
    – user2943160
    Jun 6, 2016 at 1:57

2 Answers 2

Reset to default
3
$\begingroup$

There are a couple reasons to filter between each stage.

Firstly, if you don't want to transmit those harmonics, then there's not much point in amplifying them. If you have an amplifier capable of a 10W output, but half the power is in undesired harmonics, then you are only getting a useful 5W out.

Secondly, you may want your amplifiers to be linear, but they never are exactly linear. In practice you'll get some degree of intermodulation distortion due to the different frequency components in the signal mixing with others. These intermodulation products have the potential to become spurious emissions if they aren't somehow removed before the final output.

If you don't filter between stages, then the distortion (both harmonic and intermodulation flavors) from one stage goes into the next. This increases the power at the input, and distortion gets worse with increasing power. It also means more frequency components to mix with others, compounding the issue.

How much filtering is enough? Well that just depends. By examining the signal at each stage with a spectrum analyzer (or my cheap substitute: another radio) you can see how much distortion is added by each stage and add the appropriate filtering.

Improve this answer
$\endgroup$
1
$\begingroup$

I don't think there is a clean answer to your question.

Cable plant designers wrestled with cascaded distortion for decades before fiber dramatically reduced the number of RF actives between the headend and the set-top box. Generally the rule - with some physical basis - was that second order distortion ("composite second order", or CSO) generally combined non-coherently, whereas third-order distortion ("composite triple beat", or CTB) combined coherently.

The only thing I can think of in your case would be to assume the worst case and assume that the amplified distortion products from each stage combine coherently at the end of the line.

I don't think this is a very good answer, but no one has taken a stab at your question for over a year - so I felt it couldn't hurt.

Improve this answer
$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .