Help needed with LNB modification for external reference

[alibenpeng]

Hi,

like a few others, I am experimenting with modifying cheap a LNB to feed it an external reference clock for the local oscillator.

To do this, I removed the 25MHz crystal and fed in an external 25MHz with roughly the same amplitude via an LC bandpass filter, but I can't get it to work. At first, it looks like it works, i.e. the noise floor rises when I connect power, but I don't see any signals.

I found a PDF on the web which served as a template, although my LNB is a bit different and I used SMD components as you can see in the picture.

My problem is, I can't really tell which of the crystal pins is the right one to feed the signal into. As expected, with the original crystal, both have a 25MHz sine wave with slightly different amplitudes (but not as high as in the PDF, mine are only about 250mVpp).
If I feed my 25MHz into the pin which had the higher amplitude with the original crystal, the other pin has no signal on it, where as when I feed into that pin (tha one with the lower amplitude), the other pin has an attenuated signal.

Is is possible, that both pins have to be coupled in some way? Because in the PDF and several other examples I looked at, the unused trace is ignored if the crystal is removed, or cut, if the crystal has been left in place.

EDIT: Actually, now that I think of it, there was one other method I have seen. That one tied the signal in via a BalUn, with the balanced side on the XTAL pins. I couldn't test this method though, because I don't have a small enough toroid to fit in there.

I'm a bit stuck here, so I hope someone here can put me on the right track.

Cheers,

Ali

[alibenpeng]

Update:

I managed to create a test signal at about 10.489.750kHz which was clearly visible with an unmodified LNB, but with the modified one it just made the whole noise floor rise. Tuning the transmitter or receiver around brought no change.

I'm just not quite sure what that means, though, other than that it's not working properly.

[FredWDaniel]

Your crystal may be working in an "overtone mode" which is typically a 3rd, 5th or 7th harmonic frequency.  Therefore, look for a strong signal around 75 or 125 or 175 MHz and not 25 MHz.  Use a receiver or spectrum analyzer. 

[alibenpeng]

Your crystal may be working in an "overtone mode" which is typically a 3rd, 5th or 7th harmonic frequency.  Therefore, look for a strong signal around 75 or 125 or 175 MHz and not 25 MHz.  Use a receiver or spectrum analyzer. 

I am no expert on crystal oscillators, but from what little I've read on wikipedia, it seems like a crystal can only oscillate at one frequency at a time: either its fundamental frequency, or any one of the overtones. And I have verified that this one is oscillating at 25MHz. Also, these overtone crystals seem to be specifically designed for that purpose and are subject to tighter tolerances, which all sounds expensive and therefore unlikely in the absolutely cheapest LNB I could find on aliexpress. Kind of pointless, too, because that chip in there for sure has a PLL to get the LO up to 9.75GHz, so why bother with a special and more expensive crystal.

[PA0PBZ]

In the pdf you attached the crystal is replaced with an inductor, could that be the magic trick?

[alibenpeng]

In the pdf you attached the crystal is replaced with an inductor, could that be the magic trick?

Maybe. But note that one of the traces to the chip is cut, so it doesn't actually connect to both of the chip's XTAL pins.
Nevertheless there seems to be something to it, since I've seen another version of this mod that actually uses a BalUn to couple in the external signal, and the balanced side of that is connected to the chip in exactly the way that you probably thought the inductor on this one to be.
I don't think I'll be able to fit any kind of hand wound BalUn in there, so I'll actually try an inductor, maybe with a cap in series, between the XTAL pins. Given that the equivalent circuit of a crystal oscillator is nothing but a bunch of capacitors and inductors (ignoring the resistance), this might even work.