A question for the signal theory experts (Preamplifier gain)

How much above the kiwis receiver noise floor should the preamplifier noise floor be at least?

I understand that the system noise figure drops further, the more gain the preamplifier has. But the maximum possible gain is limited by the kiwis dynamic range. So there has to be a sweet spot somewhere?


  • What type of antenna will you use?
  • I'm working on a magnetic loop preamp, but the question was more of general nature.
    Obviously the noise floor from the preamp has to be higher than the receiver noise floor, in order to improve the overall noise figure, but by how much?
  • If you have enough low noise preamp gain to have raised the noise floor by 10 dB the Kiwi noise will then be degrading the result by less than a half dB, IF IMD or other factors haven't made things worse.
    But it does depend upon signal and noise levels from the antenna system which in turn depend upon antenna size, frequency etc. It is more difficult to approach propagated and galactic noise floors with a small antenna, whether loop or dipole. It may not even be possible at mid-upper HF given typical site signal levels and available active devices.
  • I haven't approached this topic from your question point of view. Generally I try to adjust the preamp gain using input voltage to achieve the highest signal gain without raising the noise floor much. That is when the noise floor starts rising, it is time to stop increasing the input voltage. I have found that when a preamp is adjusted for maximum gain the IMD products increase as well as the general noise floor. Adjusting for the best S/N ratio across the 30MHz bandwidth is what I have been trying to achieve. My approach is probably to simple to be effective, so take this for what it is worth. Hi Hi
  • Anecdotal answer only
    From other SDR forums/threads I picked up the "raise the the gain until the noise floor lifts by about 5db".
    That might have been more for VHF but I started to use that as a baseline for HF tests.
    To test the affect on on digital modes I powered my active loop well under voltage (9.5V at the PSU when minimum 12V required at antenna) and watched WSPR decodes.
    Despite the noise graphs seeming to be dead with almost no lift during dawn/dusk I was getting good decodes.
    From that I increased the voltage until I was seeing no more improvement in decodes.
    In that use case, broadband magnetic loop, digital mode, there was no obvious advantage beyond about 3db (lowest) increase in noise floor.
    The floor lifted fairly evenly over the HF range so that wasn't like getting 3db at 15MHz forced 18db at 500kHz.
  • >
    >broadband magnetic loop, digital mode, there was no obvious advantage beyond about 3db (lowest) increase in noise floor

    That's highly likely.

    Glenn was spot on with his previous statement.

    > It is more difficult to approach propagated and galactic noise floors with a small antenna, whether loop or dipole. It may not even be possible at mid-upper HF given typical >site signal levels and available active devices

    With your loops and LZ1AQ amplifier, I suspect that the antenna performance is likely to be about 10dB worse than the sensitivity required to hear the galactic noise floor on the upper HF bands, so as soon as you see the noise from the amplifier raising the KiWi noise floor, you won't really notice any further improvement.

    Very few (if any) 1m diameter active loops are capable of hearing down to the galactic noise floor on these frequencies, so it's not that you are doing anything wrong, it's just the laws of physics constraining things. There have been many previous threads on this subject if you wish to trawl back through them.

    This is one of my gripes about Youtube videos where folks compare different receivers, but are using something like a Wellbrook loop as the antenna. In most cases the antenna is the limiting factor and not the actual receiver. In fact I despair about the trend with amateur transceivers where folks seem to gravitate towards equipment with very large dynamic range figures, whereas in all likelihood they will never be able to take advantage of it, because they are using compromise antennas in noisy urban environments. If you take a look at a lot of the KiWi's that are on line, many of them are struggling to achieve 20 or 30 dB dynamic range (max signal level to noise floor) for these reasons, so having a receiver capable of >120dB dynamic range is pointless.

    I got close to achieving maximum sensitivity by using a combination of my TC2M broadband vertical, followed by a DXE-RPA1 copy 13dB low noise pre-amp and then a passive amplitude slope equaliser and selective BC band notches. By doing this I could have much greater RF gain on the HF bands, but roll it off on the LF bands where the natural noise floor was already higher than the KiWi baseline noise figure of somewhere around 14dB. However even this was not good enough to hear emissions from Jupiter at around 20MHz, which is what I was hoping for.


    Martin - G8JNJ
  • edited September 2019

    Ah, I see that's what you're really after, Martin....

    The SKA, LWA, LOFAR radio astronomy arrays have produced a lot of interesting designs which were created with as one object noise temperatures below the galactic background noise when using uncooled amplifiers in the 20-80 MHz range. The strong signal and inter-modulation behaviour of the used LNA's is often not as good as needed on HF below 20 Mhz but then interference is likely less of factor. The antenna elements are often crossed dipoles at about 45 degrees from zenith and mounted above a ground screen mesh. Since a lot of the research for these mega projects are tax payer funded quite a bit of information on the design and testing of the individual elements of these arrays is publicly available.

    Here are some examples from one of those design documents:

    Best regards, Ben

  • Hi Ben,

    Yes I had seen that design, unfortunately I think it would fall over in the presence of the European short wave broadcasters. But you may get away with it if it's only required to operate over a small (normally quiet) frequency range, in a remote site with the antennas pointing up at the sky.

    However there are quite a lot of documents like this on the web regarding low frequency arrays, that are particularly relevant to the design of effective active antennas, and I know that Glenn has got a library full of them :-)


    Martin - G8JNJ
  • Well with my present loop, I'm often in the borderline clipping range (>-30 dBm) particularly in the 31m and 25m broadcast bands, while noise levels above 18 MHz don't exceed the kiwis noise floor.
    It's a trade-off between occasional ADC saturation and loss of sensitivity in some bands.
  • Yes for various reasons most loops tend to favour frequencies below about 10MHz.

    A simple amplitude equaliser can fix this problem be reducing the level on the lower frequencies without unduly affecting the overall S/N ratio, as the natural noise floor is much higher on the LF bands and active loops generally have adequate sensitivity on these frequencies.


    Martin - G8JNJ
Sign In or Register to comment.