600MHz BP filter. Help wanted

时间：04-08 整理：3721RD 点击：

I'm looking for a BP filter solution for a 600MHz radio with 6MHz bandwidth. The filter is to improve PA(1W) mask preformance, as well as LNA imtermods. Rejction requirement is not very high--min 10dB at 600MHz+/-10MHz and 20dB@600MHz+/-20MHz. Ideal insertion loss of 1.3dB.

Initial approach was ceramic filter but I gave up soon because of high price. Now I'm thinking of lumped filter. In ADS it looks pretty easy to realize the rejection requirements with a 2 poles or 3poles LC filter. But I'm not sure about insertion loss.

Questions:

What level of insertion loss should I expect, if I choose high Q inductor and caps and ceramic PCB substrates?

Would the temperature variation of lumped filter larger than ceramic filter?

ADS simulation gives some impractical LC values, like 0.17nH or fF. How can I make ADS to use standard values only?

Thanks for reply.

Dear LNA:

First I have doubts if you chose a correct frequency band; to my knowledge, 450-850 MHz is authorized by FCC and out of the USA by corresponding offices only for UHF TV.
Second, instead of making calculations, please try to use an old-style UHF TV tuner as a suitable band-pass filter. Such tuners had a tunable two-cavity coaxial band-pass filter, typically 5-8 MHz wide to -3 dB, then there was a local oscillator and a diode or transistor mixer.
More modern UHF TV tuners had a similar tunable band-pass filter controlled by varicap diodes, a FET preamplifier and mixer.
You can also make a similar filter tuned by trimmer capacitors, and tune it to a good response if you have an access to a sweep signal generator and a detector, both connected to an oscilloscope. In the past, TV repair shops had such equipment.

Now I would rather recommend to extract the UHF TV tuner from a discarded TV set, and use it for your radio. I am only afraid that your 1W transmitter at 600 MHz can cause interference and you may be prosecuted for it by FCC.

Thanks, jiripolivka.

This is licensed radio for remote villages so FCC is not a concern.

TV tuner is a good idea which I never thought of. The problem is that it is design for receiving only so the -3dB BW of 5~8MHz is lossy for transmitters.

Another obstacle of TV tuner is you have to tune every filter, which is not suitable for volume production.

Let's go back to lumped BP filter. I'm worrying about the impact of components tolerance. Assumming 1% L and C used, would the center frequency shift be 1%? . i.e. 600MHz+/-6MHz? That's terrible!

You should also worry about component Q factor. In my opinion, your specification with 1% bandwidth is VERY tough for a lumped element filter.

It is impossible to use LC design for such a narrow bandwidth filter at this frequency range. You may try ceramic filter or cavity filter.

Dear LNA:

The best solution of your filter is to copy the TV tuner BPF, with coaxial sections tuned by trimmers, and adjustable coupling e.g. by a loop.
You will certainly have to tune each filter quite carefully. Concerning the mass production, TV tuners are a good example that it can be done.
At 600 MHz, LC or lumped-element version will not work. I was always successful with coaxial cavities, this is why I recommend this technology.
It may be that a ceramic filter is available; the only advantage is the small size. There are companies offering to produce a quantity for a reasonable cost, mainly Chinese.
To my knowledge, temperature stability of high-permittivity ceramic material can be a problem. Many years ago I made a band-reject filter with coax cavities and tested it over temperature. At ~ 450 MHz, the max deviation was ~ 2-3 MHz.

Hi,
Another possibilty is for a higher volumen product the SAW filter, their from faar east producer arnt expensive, but other vendors as USA/Europa are not so bad (in prices)too...
K.

Are you sure that a TV tuner has such narrow band pass filters in the RF input section? I would expect that narrow filters are located in the IF, behind the first mixer, where the relative bandwith is much easier to handle.

If this is going to be a mass produced item I think you would be better off spending some effort on curing your problems at source. It would be a lot cheaper to incorporate a better designed PA that does not distort the modulation in the first place than to try to filter out the garbage after the effect. If you are relying on this filter to clean up the spectral mask, every time you change the TX frequency you will have to change the filter.

Peter

I did a quick search and found nothing detail of TV tuner BPF. Can you please give a circuit example or a current vendor of TV tuner filter?

What price level I should expect? With a resonable price I think the trimming process maybe acceptable.

What physical size?

Thanks again for sharing your ideas.

Added after 1 minutes:

I think SAW is very lossy. Not for final PA.

Added after 1 minutes:

Are you sure that a TV tuner has such narrow band pass filters in the RF input section? I would expect that narrow filters are located in the IF, behind the first mixer, where the relative bandwith is much easier to handle.

Dear LNA:
I have no direct design to offer, but for a 600 MHz BPF you can make coaxial cavities from metal sheet e.g. 0.2 mm thick, 15 x 15 mm cross section, and ~ 35-40 mm long.
The center conductor can be also made from metal sheet, approx. 5 mm wide. The resonators will be ~ quarter-wavelength , tuned by 5-15 pF trimmers at high-ends.
I would join two or three such resonators side by side, then use loops to match the input and output to coaxial connectors, and other loops to adjust coupling. Those loops will be made of ~1 mm wire, the ends connected to the low cavity end just between the center conductor (soldered to cavity center) and the cavity edge.
Many similar filters I built were usually adjusted for a flat pass-band response, 5...8 MHz wide; the complete adjusted filter had <1 dB insertion loss. The skirts depend upon the number of cavities; the best way is to try making and adjusting the filter.
As a good example to follow, please open an old UHF TV tuner and try to make a similar structure. UHF radio amateur manuals also offer good filter designs.
You can try to follow designs described in ARRL Radio Amateur's Handbook,. in my 2000 edition Chapter 16.22-24.
If you have a suitable UHF test equipment , making a good coaxial filter as described is just incredibly easy and low-cost. Your filter still has to be finally tuned up with your power amplifier and antenna for the best results.
If you prefer to buy, I noticed good UHF filters available from e.g."www.temwell.com.tw" but such filter may be costly. I prefer making my filters the easy way.

hi, jiripolivka. I just got time back to this topic. Is there any possibility you could share the artical you mentioned in ARRL Radio Amateur's Handbook? The 2000 edition CD may take weeks to ship to me. I can't wait that long. Thanks.

You said "... Your filter still has to be finally tuned up with your power amplifier and antenna for the best results...". Do you mean every filter has to be matched with specific PA and antenna? That is not good for field maintainance.

If you have the space, coaxial transmission line resonators would make a good filter.

If you don't have the space, I would probably go with coaxial High K ceramic resonators.

You will probably have pretty high loss, due to the limited resonator Q and the inevitable slight mismatching of the resonators. Any chance you can do the filtering before the final amplifier stage, where loss is not important?

You will need to be able to tune each resonator frequency, and probably the coupling. So that is a lot of manual tuning. Also, I would be a little worried about temperature drift.

Cell towers use giant copper resonators with air dielectric to achieve low loss and high Q. But they are massively huge filters!

[quote="biff44"]If you have the space, coaxial transmission line resonators would make a good filter.

Can you give more info on this coaxial transmission line resonator? Google returned nothing. I probably have no enough space for it.

Its high cost prevent me from ceramic filter.

All vendors I quoted give a insertion loss of 3dB min for 2 poles ceramic filter.
Filtering before PA can remove majority of unwanted emission, but the PA far out emission is still failing the mask. Plus this is a TDD system, so filtering between radio and antenna also provides better immunity to receiving.

What is the typical temp drift for such a filter, from -40C~+50C?

Hi,
I think you should search for "cavity resonators"...
Or these direction?http://www.scribd.com/doc/9487409/RS...sonator-System
K.

Hi LNA;
I think you can use some lump components and some distributed components (PCB Strip line or coil) mixed together.

I did find some helical filter on the market but insertion loss is too high. They are small sizes like 15mmX12mmX6mm. Why need big space?

Added after 24 minutes:

The problem with lumped discrete filter is wide BW, normally 3%, and freq variations over temp.

Some says ceramic filter is the best on temp variation but jiripolivka mentioned it's not stable. Not sure what knid of temp stability I should expect.

Also can some one tell me if RF mask filter at final stage is very common in 600~900MHz radio application?