Cancelling Ku LNB LO drift with its own comb, verified against a TV satellite beacon

[Ayushman Tripathi]

Hi,

I used the tracking beacon on a Ku TV satellite to kill the LO drift in an Inverto PLL LNB, and it turns out the LNB can correct itself. This is on a smaller Ku setup that I run my narrowband search pipeline on.

The issue at 11 GHz is that the LO in that cheap PLL LNB drifts by a few Hz per second, while Earth rotation moves a fixed sky source by only about 1.3 Hz/s at that frequency. So the instrument is sliding about three times faster than the thing I'm trying to measure, and a narrowband drift search becomes close to useless.

The LNB builds both its 9750 MHz LO and its 12.5 MHz internal reference comb from the same 25 MHz crystal. When that crystal moves, a sky signal and a comb line move in opposite directions, and the ratio between them is fixed at -LO/IF (that is the sky signal's apparent drift divided by the comb line's drift). So if you track one comb line, you have measured the drift of everything else in the band.

I verified this against the Ku tracking beacon on GSAT-30 at 83E, near 11449.5 MHz, following the beacon and one comb line together in a single capture. The ratio should have been -5.737 and I measured -5.745, with the two lines correlating point by point at r = -0.9999 where -1 would be an exact inverse match. That is what says one crystal is driving both, rather than two oscillators that happen to drift at similar rates.

Then I corrected the beacon using only the comb line, with nothing from the beacon itself going into the correction. Beacon's apparent drift went from -0.2960 Hz/s down to -0.00041 Hz/s. The satellite is geostationary, so zero is correct.

The nice part is that the comb lives inside the LNB, so it is there at any pointing, even with the dish looking at a wall. So I don't need a beacon or clear sky for routine running.

My setup is a small offset DTH dish (for current testing) with an Inverto PLL LNB, a HackRF One with OCXO and a Raspberry Pi.

Pictures attached:



My HackRF sweep of GSAT-30 at 83E across the LNB band, beacon at 11449.5 MHz : 

Sat beacon drift from my own capture, before and after the comb correction: (The satellite is geostationary, so zero is correct.)

Thanks

[Adrian]

Ayushman,

 Great work on calibrating your LNB for narrow band signal detection. Your corrections here should allow at least  Hz bin resolutions for narrow band potential detectable signals. Its great optimization fo making useable very inexpensive equipment.

Adrian

[Ayushman Tripathi]

Hi Adrian,

Thanks.

[Marko Cebokli]

I don't understand this. I have never seen a LNB with an built in comb generator. Which type/manufacturer is your LNB?

Marko Cebokli

02.08.2026 21:42, je Ayushman Tripathi napisal

--
--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sara-list/15104d70-3172-4488-b5ec-ad03439320ban%40googlegroups.com.

[Ayushman Tripathi]

Hi Marko,

It is an Inverto Single High-Gain Low-Noise 40mm PLL LNB, model 5928, the ordinary DTH type with 9750/10600 LO.
https://inverto.tv/lnb/445/single-high-gain-low-noise-40mm-pll-lnb

There is no comb generator in it. What I am using is the harmonics of its 25 MHz PLL reference crystal leaking into the IF output.

Correcting the beacon against one of those harmonics alone takes its apparent drift from -0.296 Hz/s to -0.0004 Hz/s, and the beacon is on a geostationary satellite so zero is the right answer. If it were instead arriving on the dish it would be mixed down by the LO exactly like the beacon and would have to drift the same way the beacon does, in which case the very same correction would have made the drift much worse rather than cancelling it to zero.

Ayushman

[A. C.]

I Marko,

I don't think this is an intentional design feature here, but rather a consequence of the minimalist PLL architecture in an inexpensive mass‑market satellite LNB design.

The PLL reference clock signal leaks into the phase detector and the VCO control path, and this modulation is carried through the LO chain and injected into the mixer,

where it appears as repeating multiples of the reference‑frequency clock spurs in the LNB IF output.

Ayushman has basically taken lemons and made lemonade by using these bleed‑through spurs as stable reference markers.

Because the spur comb is locked to the LNB’s internal reference oscillator, the spacing of the spurs remains extremely stable even when the LO itself drifts.

By comparing the spur positions to the known, very stable Ku‑band satellite beacon frequency he can measure the instantaneous LO frequency error and then correct for the LNB’s drift in post‑processing.

Very cool.

Adrian

You received this message because you are subscribed to a topic in the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sara-list/aMRp-euk4k4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to sara-list+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sara-list/c474dbee0f9274948adaaf49b44e2b13%40hamradio.si.

--

Adrian

"My God, it's full of stars!"
 Dave Bowman

[James Abshier]

Hi Ayushman,

Although I don't completely follow what you have done, it seems that you are assuming that the satellite beacon frequency does not drift. Geostationary satellites are not usually in perfectly circular orbits, so some Doppler shift could be expected. Have you considered this?

Jim Abshier

--

--
You received this message because you are subscribed to the Google
Groups "Society of Amateur Radio Astronomers" group.
To post to this group, send email to sara...@googlegroups.com
To unsubscribe from this group, send email to
sara-list-...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/sara-list?hl=en
---
You received this message because you are subscribed to the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sara-list+...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/sara-list/91c9d633-508c-4805-bd74-032cd78b7ba8n%40googlegroups.com.

[Ayushman Tripathi]

Hi Jim,

Yes, a satellite like this is kept very close to its fixed position, and even with the small movement that is left, the drift it can cause is about 0.02 Hz/s at the most. What I removed was 0.296 Hz/s, which is more than thirteen times bigger. So the satellite's own movement is far too small to explain what I was seeing.

https://www.satsig.net/pointing/satellite-station-keeping.htm

https://docs.fcc.gov/public/attachments/DA-05-1812A1.pdf

Also, the satellite cannot explain the correlation. The comb is generated inside the LNB from the same 25 MHz crystal, so it has no knowledge of the sat's motion. Measuring r = -0.9999 between the beacon and a comb line, at the ratio predicted from -LO/IF, is what identifies the crystal as the common source.

Thanks

[A. C.]

Hi,

 Although the so called, comb spacing, originates from the crystal reference—either the crystal’s fundamental frequency (e.g., 25 MHz) or a divided‑down value produced by the PLL’s reference‑divider—the LO (e.g., 9750 MHz) is not produced directly by the crystal. It is generated by a VCO whose frequency is continuously corrected by the PLL. The PLL divides the VCO frequency, compares it to the reference using a phase detector and control loop, and adjusts the VCO tuning voltage until the divided VCO matches the reference. This locking process makes the LO indirectly derived from the crystal, and any residual reference leakage or phase‑modulation coupling to the VCO produces the observed reference‑spur comb at integer multiples of the reference frequency.

 Also technically correct,  geostationary satellites do not remain perfectly fixed; they execute a small north–south and east–west motion that produces a slow “figure‑eight” ground track (the analemma). Because this motion is very small and the satellite’s velocity relative to an observer on Earth is only a few tens of meters per second, any resulting Doppler shift at microwave frequencies is negligible for practical purposes.

Adrian

You received this message because you are subscribed to a topic in the Google Groups "Society of Amateur Radio Astronomers" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/sara-list/aMRp-euk4k4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to sara-list+...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/sara-list/9a9d2a26-0869-43d9-bec9-f765d2965327n%40googlegroups.com.

[James Abshier]

Hmmm

Doppler Frequency = Carrier Frequency x (velocity/c)

For Ku band, say 12 GHz, and 10 meters/sec,

Doppler Frequency = 12e9 Hz x 10 m/s / 300e6 m/s = 400 Hz

This is negligible compared with 12 GHz but is much greater than .02 Hz

Jim Abshier

To view this discussion visit https://groups.google.com/d/msgid/sara-list/CACs6ujY%2Bp2SUrWAo3BtNS2u2H86OTuzitkq9nNzYRUH26i14fQ%40mail.gmail.com.

[Ayushman Tripathi]

Hi Jim,

Your 400 Hz is a frequency shift, not a drift rate. A steady offset produces no drift at all, only the change in it does. The satellite goes up and down once per day, which works out to about 0.03 Hz/s at the fastest.

What I removed was 0.296 Hz/s, ten times bigger, and that is using your 10 m/s which is more than the satellite actually moves.

Ayushman

To view this discussion visit https://groups.google.com/d/msgid/sara-list/4dc1f7a2-3c52-4fe2-8244-4f81e6c16ea5%40ameritech.net.

[A. C.]

Hi

Although the calculated Doppler values are mathematically correct for the stated satellite velocities, the underlying assumption about the satellite’s actual motion is not. A GEO satellite’s full orbital velocity should not be treated as if it were directed along the observer’s line of sight. In reality, its motion is almost entirely tangential, so only a very small radial component contributes to Doppler. Realistic radial velocities—typically a few‑tenths of a meter per second from station‑keeping and small inclination or longitude excursions—produce Doppler shifts of only about 10–15 Hz at 12 GHz, not hundreds.

By contrast, attributing a measured 0.02 Hz drift to Doppler would imply a radial velocity of roughly 0.5 mm/s, which is physically negligible and not representative of actual GEO orbital dynamics. Even natural occurring perturbations such as lunar gravitational forcing introduce very small radial accelerations (~10⁻⁶ m/s²), leading over many hours to radial velocity excursions of only a few centimeters per second—corresponding to sub‑hertz to few‑hertz Doppler at Ku‑band.

All of these Doppler contributions are extremely small compared with the satellite beacon’s own frequency tolerance. Unlike atomic clocks deployed in GPS constellation satellites, modern GEO communications satellites use high‑stability oven‑controlled OCXOs or even rubidium frequency standards as their master references; these provide stabilities in the 10⁻⁹ to 10⁻¹¹ range and result in ±1–10 kHz of allowable beacon frequency variation at Ku‑band. This level of stability is entirely appropriate for GEO communications payloads and modulation protocols, and the satellite’s internal frequency uncertainty is therefore hundreds to thousands of times larger than either realistic Doppler or the 0.02 Hz drift being discussed.

So even if some of the LNB’s corrected frequency variations are normalized against the satellite beacon—and the beacon itself can present small apparent drifts that reflect actual GEO dynamics—these effects remain negligible. The normalization method being used still stabilizes the LNB effectively, and the internal use of the comb as an accuracy indicator provides more than ample practical frequency stability for the narrowband precision required by the intended pipeline.

Adrian

To view this discussion visit https://groups.google.com/d/msgid/sara-list/CAAer-xPoR19Sjns-P7vjHz%2BjKNei6nPuW17pgBjtYu%2Bm%2BN_5HQ%40mail.gmail.com.

[James Abshier]

Hi Ayushman,

Thank you for your clarification. I had originally thought that you might be using the satellite signal for a frequency reference which is why I suggested the Doppler issue.

Jim Abshier

To view this discussion visit https://groups.google.com/d/msgid/sara-list/CAAer-xPoR19Sjns-P7vjHz%2BjKNei6nPuW17pgBjtYu%2Bm%2BN_5HQ%40mail.gmail.com.