What is *the* best filter regardless of cost for RF purposes? I mean, which kind of filter most accurately approximates the *ideal* concept of a filter in terms of selectivity and whatnot? I'm looking for something that will in essence pass an unmodulated carrier only and 'eliminate' both sidebands due to phase noise from it. I want IRO of 140dB or better down from the (desired) carrier amplitude with as near vertical cut-off as possible (obviously). Take say 10Mhz as a ball park figure for the carrier. Any pointers? TIA.
--
This message may be freely reproduced without limit or charge only via
the Usenet protocol. Reproduction in whole or part through other
protocols, whether for profit or not, is conditional upon a charge of
GBP10.00 per reproduction. Publication in this manner via non-Usenet
protocols constitutes acceptance of this condition.
As a real filter, it sounds like you want a crystal filter, which can be a big deal.
Or as suggested, a PLL, preferably one that uses a very good VCXO.
The best PLL uses a VCOCXO and has distinct acquire and track modes or a very good phase-frequency detector, to minimize the active bandwidth once it finds lock.
I've mounted the OCXO on springs, to make it less vibration sensitive.
formatting link
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
Your layout and manufacturing quality has improved vastly over the years, John. Well done!
-- This message may be freely reproduced without limit or charge only via the Usenet protocol. Reproduction in whole or part through other protocols, whether for profit or not, is conditional upon a charge of GBP10.00 per reproduction. Publication in this manner via non-Usenet protocols constitutes acceptance of this condition.
Er, can you expand on that a bit? I mean, I know what a PLL does in its typical implementation, but how could it be used as something approaching an ideal filter?
-- This message may be freely reproduced without limit or charge only via the Usenet protocol. Reproduction in whole or part through other protocols, whether for profit or not, is conditional upon a charge of GBP10.00 per reproduction. Publication in this manner via non-Usenet protocols constitutes acceptance of this condition.
That is part of the NIF timing system, delivered in the year 2000.
formatting link
One problem was fan vibration, but a bigger problem was the SMB test connectors. When someone yanked a cable out of the module, the connector would SNAP! and we'd lose phase lock for a second or so. This is a bang-bang phase detector with a fast acquire mode and a slow track mode. The slow mode has super low phase noise and jitter, but is easily knocked out of lock.
--
John Larkin Highland Technology, Inc
lunatic fringe electronics
input - 50 Ohm to 400 Ohm transformer----- 400 Ohm line ----
400-to50-transformer - output.
From the 400 Ohm line to GND are 2 cheap 10 MHz crystals, one with a
35pF series trimmer,
the other with series trimmer and 4.7uH. The xtal resonances thus are 10 MHz +1.6 KHz and -1.6 KHz.
Transformers are 1:8CT Pulse electronics CX2049LNLT digikey
553-2287-1-nd. Autotransformer feature was not used.
The filter costs next to nothing. It won't turn my MV89A references into BVAs, but gives at least some keyhole insights . There could be easily 5 crystal pairs for strategically chosen frequencies.
For notches closer to the carrier the impedance level must be probably lower (1:4 or 1:2 xformers). That makes the notches less deep, but also less wide, so they cannot interact at the carrier frequency itself.
The filter withstands quite a lot of power because the crystals do not resonate on the carrier frequency. Just make sure you do not sweep them with +30 dBm accidentally.
If you would build a filter that is resonant on the carrier frequency, it would impair phase stability of the carrier and the crystal would blow up at @20 dBm. At -3 dBm (a level that the crystal can survive in resonance), your signal takes a hit from the thermal noise @ -173 dBm.
On the VNA, the notches looked that way:
<
formatting link
>
And, using John's array of ADCs and a virtual Win7 CPU, that was the phase noise result: Impact of the filter on the phase noise of an R&S SMIQ signal generator. SMIQ was chosen as an easy victim.
<
formatting link
>
Blue: without notch filter, red: with notch filter.
One ancient wave analyzer method is mixing the incoming spectrum with a VFO, then using a crystal filter at a convenient fixed frequency to do the required filtering at a fixed frequency. The filtered signal is restored to original frequency by mixing it with the _same_ VFO.
These days the crystal filter could be replaced with some DSP processing, say at 100 kHz. Some analog front end selectivity is needed ahead of the first mixer and after the second mixer, to avoid image responses.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.