Best method to obtain reference samples for Hydrogen Line work

On 2025-11-04 03:27, ‘Andrew Thornett’ via Society of Amateur Radio Astronomers wrote following in bold, then Marcus Leech responded with answers in italics.

Hi All,

Interesting discussion on the resistor in some barebones Nooelec’s.

So…..after all the tooing and froing can I ask:

1.Does providing a reference sample taken from the antenna side of the coaxial cable and LNA give an improvement over one taken next to SDR?

Yes, quite a bit–FOR CONTINUUM MEASUREMENTS.   Since there are many contributions to gain instability between the antenna and your final samples, you want the reference source as close to the antenna as possible.  You want everything “inside the reference loop”.

2.If yes to above, then does the resistor switch in some Nooelec Barebones devices also give improvement over the use of reference sample 3MHz away from the signal frequency as used in ezRA, in spite of the device’s limitations already discussed on this group?

For spectral observations, Dicke-Switching type scenarios are of limited utility.  Frequency switching is helpful to establish a spectral baseline (by determining the baseline passband shape), but it assumes that the pass-band of the receiving setup is the same over a 3Mhz shift.  It sometimes isn’t.

3.How about using a coaxial relay switch on the antenna side of the LNA to switch in a 50 ohm dummy load? In the H-Line group meeting yesterday it was suggested this could increase noise? How does it do that? How bad is that? Is it a major problem? Does that increase in noise offset all the benefit of getting a reference signal from next to dish?

Any loss IN FRONT of the LNA directly adds to the overall system noise.   If that relay inserts 0.5dB loss, then there’s an additional 0.5dB noise figure in the system.  See Friis for receiver noise-chain analysis.  The NooElec barebones with the switch was built with a GaAs switch with very very good specs from the manufacturer, but in practice the insertion loss was considerably higher, which is why the “barebones with a switch” doesn’t offer nearly as good a noise figure as the “retail version”.

4.One other question that came up yesterday is – the SDRplay devices can provide bandwidth 8MHz. If we use that whole bandwidth, does that mean that a reference signal collected from an offset frequency has to be over 8MHz away from 1420MHz? Is 8MHz enough or does it need to be further than that?

The idea behind frequency switching is to move away from the spectral lumpiness created by the sky signal–even in “quiet” regions of the sky. You only really need to move away far enough to avoid galactic emissions.  The extragalactic emissions (which are red/blue shifted by quite a bit), are quite weak, so a short-period integration a few Mhz away will provide a good estimate of the instrumental response, which is what you’re trying to account for.