Radio 4 heard on 396kHz

[G0HZU_JMR]

I looked for other harmonics on the analyser last night. I couldn't see any on the analyser that were exact multiples of R4. I could see faint carriers from MW stations on the higher harmonic frequencies but these were several Hz off the expected harmonic frequency of R4. I assume all these stations will be on multiples of 9kHz.

[Pfraser]

Third harmonic (presumed)* on 594kHz on the Wrexham SDR

*Still too early for MW propagation to bring in signals from abroad?

Just detectable via CW mode in Stourbridge.

[Hartley118]

I recall in my younger years half a century ago visiting relatives in the Midlands who lived within a few miles of Droitwich. It was normal to be able to receive the station on 400kHz on any radio as well as on the expected 200 kHz. I attributed this at the time to non-linearity in the receiver front end, but subsequent experience supplying mobile phone base stations makes me wonder about non-linearity in transmitter elements such as cable and connectors (encountered as the Passive Intermodulation - PIM - problem in base station design).

I wonder what proportion of the spurious signal that’s under discussion here is contributed by the receiver front end and how much actually comes from the transmitter.

Martin

[Radio Wrangler]

attenuating the receiver input from the antenna will show any harmonic creation in the receiver. If the harmonic creation is external, both the fundamental and harmonics should drop by the same amount. If the harmonic creation is inside the set, the harmonics will drop faster than the fundamental, the higher order the harmonic the faster.

David

[turretslug]

Not a hint of a murmur of R4 on 396kHz tonight amidst the typical dog's dinner of everyone else's noisy things. Plenty of signal from Poland on 225 as usual, so the radio's probably OK, accompanied by thunderstorm-style ticks and crashes- hopefully someone en route is getting a bit of rain!

[G0HZU_JMR]

I just checked with the spectrum analyser and 198kHz is visible at -45dBm again and the second harmonic is still there at about -117dBm tonight. That's a difference of 72dB. Normally, the rule of thumb for a spectrum analyser is that the internal front end distortion produced by the analyser has to be 20dB lower than the distortion being measured if the measurement uncertainty needs to be less than +/- 1dB. If the analyser distortion is only 15dB lower than the DUT then the uncertainty grows to about +/- 3dB.

So to keep the uncertainty below +/-1dB my analyser would have to be capable of being distortion free down to -92dBc with a fundamental fed in at -45dBm. So the 2HI (second harmonic intercept) of the analyser would have to be -45 - (-92) = +47dBm. The datasheet suggest the analyser's mixer has a typical 2HI of +52dBm but this degrades to +42dBm at low frequencies. I had 10dB internal attenuation inline so this boosts the input 2Hi from +42dBm to +52dBm at low frequencies. However, the effective 2HI of the analyser might degrade still further down at just 198kHz.

Therefore, I tried measuring the 2HI of my analyser with a clean (filtered) 198kHz test signal at -20dBm and I saw a -85dBc harmonic with 10dB internal attenuation. This improved to -95dBc with the input drive level from the sig gen reduced to -30dBm. So I think my analyser can make this measurement with low uncertainty for a drive level of -45dBm at 198kHz. However, for belt and braces I then added the 5th order 370kHz HPF at the input to reduce the level at 198kHz to make certain there was no distortion from the analyser caused by the 198kHz drive signal. With (or without) the HPF inline the result was the same in that the 396kHz signal was still at -117dBm this evening.

Not all spectrum analysers are as strong as this one. Some R&S analysers at work only manage a 2HI of +25dBm at 10MHz for example. So great care needs to be taken when using a spectrum analyser or any receiver for stuff like this. It really is best to find out the effective 2HI (second harmonic intercept point) of the analyser and then ideally make sure the internal distortion introduced by the analyser is 20dB lower than the distortion of the device under test.

[kalee20]

The OP hasn't mentioned whether the 396kHz is evident on more than one receiver - if not present on other types, it would point towards a receiver non-linearity (in addition to the tests Radio Wrangler suggests).

However, other contributors have detected a signal, which indicates that 'something' is real.

Designing a LW AM transmitter isn't an easy task, even with the current 5kHz audio cut-off the carrier is only about 1.5 decades above the highest audio frequency, which gives the output filters, (pi networks, whatever) quite a hard job. And an efficient RF output PA is going to generate considerable harmonics which need to be filtered before reaching the transmitting aerial.

Given that the 3rd harmonic falls in the MW broadcast band, so not exactly, if there is any detectable output at these frequencies, it wouldn't surprise me if the matter has been swept under the carpet. 396kHz though, is radionavigation, which I don't know much about - are spurious emissions a nuisance in practice or has GPS rendered usage redundant?

[saddlestone-man]

Many thanks for the comments, a very interesting discussion. I would say that the jury is still out on whether this signal is real, or an artifact of the front end of my (and others) radio.

There are less NDBs on this band than when I last listened a few years ago. So any second harmonic (if it exists) of R4 would be less noticeable and probably wouldn't do any harm.

I've just tried two more radios on the same aerial as the Sailor 66T. Results are:
Target hf3 receiver: nothing heard on 396kHz.
Trio R-600 (my favourite radio from that period): R4 heard on 396kHz, but much weaker than on the Sailor.

best regards ... Stef

[G0HZU_JMR]

I'm only about 20 miles from the Droitwich transmitter and I seem to be getting a consistent signal level of -45dBm at 198kHz for both day or night. I also see about -118dBm at 396kHz day or night. I'm using a passive coaxial loop antenna.

I dug out my old Trio TS430S HF transceiver and this has a general coverage receiver covering LF through to 30MHz. It uses diode switched filters at the front end so it can be prone to a certain amount of 2nd order distortion in the pre-selector switching diodes.

However, when I measured the 2HI of the TS430S at 198kHz I got a result of +48dBm at a -20dBm drive level and the same result of +48dBm at -30dBm. This isn't a great distortion result for a comms receiver but it should still be good enough when receiving 198kHz at a signal level of -45dBm. With the loop antenna connected the TS430S received the 2nd harmonic of R4 at 396kHz at about -118dBm which matches the level seen on two decent spectrum analysers.

The result was also -118dBm when I put the 360kHz HPF inline between the loop antenna and the TS430S. By connecting the AF output to a PC soundcard I could look at the AF spectrum in USB mode and this let me see the PSK modulation on the 396kHz carrier to prove it was a harmonic of R4. The level of the PSK sidebands was about 6dB higher compared to 198kHz and this is to be expected when the frequency is doubled.

I also had a look at 594kHz using the TS430S and I could just see the PSK modulation on a very weak carrier using the PC soundcard. So there is also a tiny amount of 3rd harmonic distortion here as well. I'm not how significant this is because I'm only 20 miles from the transmitter and the 396kHz harmonic signal is down in the noise when receiving it using the TS430S in AM mode.

Providing the 198kHz fundamental signal isn't huge, a really good comms receiver should be able to make a reliable measurement of signals at 396kHz and this assumes the comms receiver has relay switched preselectors and a decent first mixer. The mixer in the TS430S uses a pair of balanced JFETs. I don't think the preselector in the TS430S front end provides any rejection of 198kHz when the receiver is tuned to 396kHz so any harmonic distortion could be in the diode switches or the mixer. However, my 2HI tests show that the receiver is still good enough. Adding the 360kHz HPF made no difference and this proves the receiver is OK in my case.

[G0HZU_JMR]

See below for an old Agilent app note for my spectrum analyser. App note 5980-3079EN

https://www.testunlimited.com/pdf/an/5980-3079EN.pdf

If you scroll down to page 21 it explains the impact of the internal distortion of the analyser and provides an equation and a graph. The graph is Fig 2-17 and this is shown as an attachment below.

This shows why the analyser's (or any receiver's) own internal distortion should ideally be 20dB lower than the distortion device being tested if the measurement result needs to have less than 1dB uncertainty. If the difference is only 10dB the error grows to about +/- 3dB.

If the analyser's internal distortion level is the same as the distortion of the device under test then fig 2-17 below shows that the result could appear 6dB worse than it really is if the phases sum. If they cancel in phase then the analyser will show no distortion term at all and this can cause a lot of confusion. This effect can happen with a real test using a real analyser (or receiver?) so great care must be taken to make sure the test gear doesn't mask the real result.

From Agilent App Note:
The mixer level (mixer level is defined as the power at the RF input port minus the nominal input attenuation value) of the analyzer is set such that internally generated harmonic distortion products are at least 18 dB below the harmonic distortion of the DUT. This guarantees that the distortion measurement uncertainty due to internal distortion combining with DUT distortion is less than 1 dB.

At work the usual goal is 20dB rather than 18dB.

I'm not sure how much distortion can occur in a ferrite loop antenna but hopefully it will be low if the loop is tuned to 396kHz such that it can also reject 198kHz. My loop antenna is made from coax with no ferrites anywhere. I've got some tuned ferrite loop antennas here so I could repeat the tests to see if the loop antenna can introduce its own distortion from the 198kHz R4 signal.

[Radio Wrangler]

Those were quite nice spectrum analysers, but the early ones in the family had a bug somewhere in the Bullwinkle-Natasha chipset (custom ADC and DSP) that could cause patterns of samples to vanish. So narrow spiky things in spectrum pictures might not be shown. It took a while for this to be noticed, it was spotted by customers first, and then a while before custom chips could be rolled.

David

[HMV 1120]

In a similar fashion to the above comments, I see that the second and third harmonics of Radio 5 have become audible this afternoon on the Weston SDR on 1818/2727 khz and are still audible at intervals across the 80 meter band. 2727 in particular is very strong but exhibits QSB not present on 909.

I think Clevedon is nearest transmitter, maybe it needs some attention?