An idea for a low cost stable 10GHz receiver

The satellite Es’Hail-2 is expected to be launched by the end of 2016. This will be the first geostationary satellite carrying an amateur radio transponder. As the launch date comes nearer, it becomes interesting to find a low cost solution to receive its 10GHz downlink.

Several amateurs have been experimenting with low cost LNBFs designed to receive satellite TV. These operate in the Ku band and usually cover the frequencies 10.7GHz-12.75GHz. However, many of these LNBFs have also good performance in the X band, and particularly in the amateur 10GHz band (10GHz-10.5GHz). In fact, the ASTRA-type LNBFs have a local oscillator which can be setted to either 9.75GHz or 10.6GHz. The 9.75GHz local oscillator mixes 10.386GHz (the narrowband terrestrial subband) to 618MHz, which is a frequency covered by most SDRs and conventional scanners. The satellite subband, which is 10.45GHz-10.5GHz gets mixed down to 700MHz-750MHz, a frequency which is also easy to deal with.

The receive performance of these LNBFs seems to be very good in terms of noise figure, but their frequency stability is not so adequate for narrowband applications. The local oscillator is derived either from a DRO or a PLL controlled with a crystal oscillator. The PLL is more stable than the DRO, but since the crystal is not ovenized, a stability of 0.25ppm is the most one can expect from it. Of course, this is 2.5kHz at 10GHz, which is way too much. Some amateurs have been modifying these PLL based LNBFs to use an external frequency reference instead of the crystal oscillator. See for instance the tests done by G4JNT and F1CHF.

Many of the PLLs are driven by a 27MHz crystal oscillator. The idea is to replace this with an external 27MHz signal generated by a PLL which is driven by a very stable 10MHz oscillator (this frequency being the most common for a frequency reference). Ideally, the 10MHz oscillator should be GPS disciplined. Dieter DF9NP has designed a 27MHz PLL to be used for this purpose. Luis EA5DOM has a nice document describing how he modified an LNBF to use the PLL. This is probably the best solution in terms of performance, but it can get a bit expensive. The PLL alone is about 30€, which is more than the LNBF.

It is possible to get OCXOs in the surplus market for low prices, but 27MHz is not a usual frequency for an OCXO. However, QRPLabs provides a kit for an OCXO/Si5351A synthesizer. The Si5351 is a PLL/MultiSynth-based frequency synthesizer and the intended application of this kit is to be used with other kits such as the Ultimate3S QRSS/WSPR transmitter. The nice thing about this kit is that the Si5351 has to be driven from a 25MHz or 27MHz crystal, and so this kit contains a 27MHz OCXO. The 27MHz signal is accessible on one of pins in the output header of the kit. With a price tag of 15$, this is probably the cheapest way to obtain a stable and clean 27MHz signal.

The oven is built by hand using PCB pieces, so it will be interesting to see how stable it is. The assembly manual says that, in practise, the frequency stability for a 10MHz signal is better than a 0.01Hz drift in 2 minutes. This is equivalent to 10Hz at 10GHz, which more than enough stability for most narrowband applications. Another interesting aspect of this kit is the possibility of using the Si5351 instead of the crystal oscillator to generate and feed the LNBF a frequency different from 27MHz. For instance, a 27.515MHz signal will make the LNBF mix 10.368GHz down to 432MHz instead of 618MHz, permitting the use of a 70cm receiver. However, the phase noise of the Si5351 might not be so good. The Si5351A can generate 3 different frequencies from 2.5kHz to 200MHz simultaneously (with some restrictions), so it can also be a very useful tool to have around in the lab or shack for other applications.

Another possibility that some amateurs have experimented with is to use a 3MHz surplus OCXO and a simple multiply by 9 circuit to obtain a 27MHz reference.

I have already ordered an OCXO/Si5351A kit and started modifying my LNBF. I will test how well this solution works in a future post.


  1. normally you can receive astra and eutelsat (hotbird) with one dish with lnbf mounted with 6 degrees offset (13 and 19 degrees east) …

    now if you mount another lnbf directly on the other side it should be possible to receive eshail with your astra dish (6 to 7 degrees to other side from astra lnbf) … eshail is at 25.5 to 26 degrees east … right??

    so that would be a solution for any ham (who already has a sat dish for astra) … without the need for another dish ….

    1. Yes. Es’Hail-2 will be located at 25.5ºE, so your idea may be feasible. I have no experience with receiving different satellites with an offset LNBF as you describe. Performance will be lower for the LNBFs not in the focal point of the dish. Are there any ballpark figures for what decrease in performance to expect?

      The current link budget for the SSB transponder gives you an estimate of 15.4dB SNR at areas in the End Of Coverage (i.e., Brazil or India). The receiving dish is supposed to have a G/T of 13.98dB/K, and presumably this corresponds to an 89cm dish. In Europe, a 60cm-75cm dish should be usable. My remark is that actually 15.4dB SNR is more than enough to copy SSB, so it should be possible to use a slightly worse receive chain than this.

      However, given the price of dishes, I think that unless space is a problem, it is worth to get a dedicated dish for EsHail’2. I’m using a 95cm dish that I got for 23€ at a Spanish online store for satellite TV products.

  2. Daniel.

    An even better solution is to use QRP-LABs PROGROCK with a cheap GPS receiver.
    This gives exceeding good stability via GPS locking and you can set the frequency, to the nearest Hz, via BCD switches and a push-button.
    Best of all is that it’s cheap. Dirt cheap!
    I have built 4 PROGROCKs to date and they are very good indeed. If you are ultra-fussy, you could replace the standard synth with their ovened model for not much extra cost. The outputs are 3V P/P squarewaves, so should be followed by an LPF.
    I thought the QRP-LABS GPS receiver was a little bit expensive, so instead I used a £7(UK) UBLOX-NEO-6M from EBay and just tapped off the 1PPS signal from pin 3 of the GPS chip. It works VERY well indeed and I plan to use it as the reference for a 10Ghz PLL brick. You’ll find more info on my BLOG.

    1. Yes. Some other people have pointed me to the QRP-Labs Progrock. The problem with the Progrock is that it works roughly as follows: The Si5351 generates whatever frequency you tell it. The frequency is slightly off because of crystal variations. A frequency counter locked to the GPS 1pps reference measures the Si5351 generated frequency. The frequency setting of the Si5351 is updated to correct for any errors.

      So this is not the same as a GPSDO. It’s a (GPS referenced) NCO. The problem is that the Si5351 can only set the frequency in steps. I don’t know how large are these steps (it actually depends on how you set up the multisynths and so on), but when multiplied up to 10GHz they may convert into too big steps. Thus, I’m worried that you will get a noticeable frequency jump each time the Progrock updates its frequency. With a real GPSDO you should get continuous frequency correction. If you’re setting this up, I would love to know how big this effect is in reality, to judge if it’s much of a problem or not.

      In any case, I’m now going to install a 10MHz GPSDO and 27MHz PLL by DF9NP. I’ll post in the blog how well this works.

Leave a comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.