I recently got myself a new toy...

[BriansBrain]

I recently got myself a new toy...

Siglent SSA3021X with Tracking Generator

Siglent SSA3021X.jpg

Now...
I got one of these Chinese 0-1KW 88-108MHz Low pass filter a while ago and have never used it yet

ebay - Chinese - 0-1KW 88-108MHz Low Pass Filter.jpg

So I put it to the test, tracking generator to the board input and the output to the RF input
Both set to 50 Ohm.

Chinese - 0-1KW 88-108MHz Low Pass Filter- Under Test.jpg

This was the result, it does not look correct for a 88-108MHz Low Pass Filter

Chinese - 0-1KW 88-108MHz Low Pass Filter - Siglent.jpg

Am I doing something wrong or is the board not within spec ?

[Bton-FM]

Big boy toys

[BriansBrain]

Big boy toys

[BriansBrain]

The next test with the Siglent is the LPF in the 150W Chinese Linear that uses the MRF9180

For the test I broke the print from the MRF9180's output so the test is just the LPF as this picture shows.

Chinese - 150W Linear LPF Test.JPG

The Siglent I did a Normalize on the two cables first

Chinese - 150W Linear LPF Test - Siglent.jpg

Looks good to me

[ronald001]

Hey Brian, is this a prebuilt kit?

I have a DIY kit, and it seems to be a bit different then yours.

https://nl.aliexpress.com/item/32874446 ... hweb201603_

How long are your pieces of coax? ( White 25 Ohm and Blue 50 Ohm )

Thanks

[ronald001]

Am I doing something wrong or is the board not within spec ?

These chinese boards are not always accurate - According to the schematics of my 170W kit i have to use 2WDG in the output filter.

I measured this filter with my NanoVNA and it looks really bad like yours

[BriansBrain]

Hey Brian, is this a prebuilt kit?

The 150W Linear = YES

I have a DIY kit, and it seems to be a bit different then yours.

Yes.... my LPF has 4 coils yours looks like they have only 3
+ Each of my coils has 4 turns - yours seem to have 3

How long are your pieces of coax? ( White 25 Ohm and Blue 50 Ohm )

?
The amp is now ''ON AIR'' so I can't have a look now

I measured this filter with my NanoVNA and it looks really bad like yours

I thought the 150W Linear LPF was reasonable

It's the first LPF in this thread I checked that is BAD

[nrgkits.nz]

When I test filters, I always use a directional coupler and measure the return loss while going into a 50 ohm load on one end of the filter, this will give you much more accurate results. The way you’re measuring it above will give you the insertion loss which will give you a relatively straight line across most of the pass band from DC to the roll off point, however you’ll find that the return loss will be quite bad in certain parts of the pass band - this will affect your PA causing unnecessary heat and a reduction in output power.

[ronald001]

This was the test result measured with my nanoVNA



Not sure how it relates with your test method..

[BriansBrain]

When I test filters, I always use a directional coupler and measure the return loss while going into a 50 ohm load on one end of the filter, this will give you much more accurate results.

Useful info... thanks

The way you’re measuring it above will give you the insertion loss which will give you a relatively straight line across most of the pass band from DC to the roll off point

Yeah...
I just thought I would give basic roll off point test

[BriansBrain]

Not sure how it relates with your test method..

My test was a simple... TG out to RF in

[BriansBrain]

When I test filters, I always use a directional coupler and measure the return loss while going into a 50 ohm load on one end of the filter.

This is the only Directional Coupler I have at the moment

Bi Directional Coupler 80MHz to 1000MHz 60dB 200W Connecticut Microwave 441270-S.jpg

So when you say ''measure the return loss'' you mean by using the Reflected output of the coupler ?

I'm looking for a 30db at a reasonable price ... any suggestions

[Bton-FM]

When I test filters, I always use a directional coupler and measure the return loss while going into a 50 ohm load on one end of the filter.

This is the only Directional Coupler I have at the moment

Bi Directional Coupler 80MHz to 1000MHz 60dB 200W Connecticut Microwave 441270-S.jpg

So when you say ''measure the return loss'' you mean by using the Reflected output of the coupler ?

I'm looking for a 30db at a reasonable price ... any suggestions

You could build your own but probably won't have a good frequency response at UHF?

[thewisepranker]

Bton - you can build a directional coupler for any band you like - you're limited by the number of decades of operation, not upper frequency.

Building your own doesn't really help as you'd need to be able to characterise it - which you should do with off-the-shelf ones anyway. It's not a very good idea to just go by the numbers written on directional couplers - I've got a 30 dB coupler here rated from 30 MHz to 3 GHz. It doesn't achieve 30 dB coupling factor until 500 MHz. It's pretty flat up to 3 GHz, give or take a dB or two at most.
It's a very good coupler, but you just need to know how things are in the real world. This is what I've got: https://audessence.com/broadband-rf-dir ... l-coupler/
I've swept it using a network analyser and built an interpolating spreadsheet so I can type in any carrier frequency I like, type in the SA reading and get a power output value within a decent tolerance in seconds.
I've got another, lower powered one, 25 W and 20-30 dB coupling factor. I think it's Narda, I got it cheap on eBay.

Brian, the best way to tune a filter is to look at the return loss rather than the insertion loss, which seems backwards. If you can look at both i.e. with a network analyser, even better, but if I only had a SA and TG, I'd get it lined up with the return loss then check it for sanity by measuring insertion loss, tweak as necessary and repeat.
Yes, you could use the reflected port on the directional coupler but at 60 dB coupling factor (as on the one you posted a picture of) you're going to run out of dynamic range using a spectrum analyser and tracking generator. In your first post, fourth picture I can see you running out of dynamic range at around 45 dB. You can reduce the RBW to improve the dynamic range but it's not really the tool for the job as you'd have to wait a few seconds to see what (if anything) your adjustments have done. A 30 dB directional coupler would probably suit you best.
Alternatively, you could get a high power, high value DC for the high power work, and a low power, low value one for aligning filters if your filters are separate to the amplifier and you can get a 50 Ohm connection on either end (like in your third picture in this thread).

[thewisepranker]

So I put it to the test, tracking generator to the board input and the output to the RF input
Both set to 50 Ohm.
Chinese - 0-1KW 88-108MHz Low Pass Filter- Under Test.jpg

This was the result, it does not look correct for a 88-108MHz Low Pass Filter
Chinese - 0-1KW 88-108MHz Low Pass Filter - Siglent.jpg

Am I doing something wrong or is the board not within spec ?

It's quite funny, I see low pass filters on here quite a lot with perfectly neat coils like that, but they won't ever be tuned and will probably perform worse than your example.

It's not correct - you need to tune the coils to align the filter. Try spreading them out with a ceramic screwdriver if you've got one whilst having it connected to the spectrum analyser and tracking generator. If you do it with your fingers you'll find that the trace moves a certain way when you touch it, potentially more than the adjustment you're making, so when you let go it has gone even worse even thought it looked better when you were touching it and moving it. A potentiometer "twiddler" can also work.
It takes a bit of practice. As I said in my previous post, being able to see return loss will help you but only after a lot of practice, it's not very intuitive, especially using the return loss method. It can be frustrating if you're starting out. I doubt many will admit it.

I've seen people on here ridicule filters with the coils spread out, but that's the way they should be. Not too much - if they're spread out too much you should rewind for slightly less inductance to keep the Q up, but a bit spread is perfectly acceptable as it's the only way of tuning the filter unless you use those World War 1 Arco trimmer capacitors in your shunt elements.

You want to bring those peaks at markers 3 and 4 down as low as possible, and have as little loss as possible in the pass band (around your carrier frequency). There's no reason why you shouldn't be able to get 0.2 dB in the pass band.

On the second one you tested, with the blue loop of coax on the output of the transistor, be aware that the input to that filter might not be 50 Ohms. This is a completely different can of worms. I don't really like integrated filters for that reason - you can't tune them properly.

[Albert H]

I don't mind stretched coils - it usually suggests that someone has actually tuned the damn filter! One common cock-up is to put the coil against the board (like in the picture above) destroying the Q of the coil, and causing high induced currents in the ground plane. I saw a (whisper it) Broadcast Warehouse PA that had suffered from a big current being induced into the ground plane, with the copper foil rippled and burnt and one of the filter caps had desoldered itself!

My Band II PAs usually have the output filters on separate boards, and the coils are around 8mm above their board, with the groundplane etched away under the coil. The improvement in Q is huge. All the coils are offset to prevent mutual coupling and the high voltage caps are flat to the PCB, but clear of the coil windings. When I next have one nearby, I'll put up a picture. The design was published in the 1960s in the ARRL handbook - they gave a really easily reproduced PCB layout that could be used on most of the VHF bands without modification.

[nrgkits.nz]

Another important thing to bare in mind is to normalize the analyzer so you can as close to an accurate reading as possible. If you are measuring return loss you can normalize with an open load on the directional coupler - since no coupler is ever going to be perfect, and there will be differences in your cable, connectors and analyzer also. Normalizing attempts to "zero" out all of this so your DUT (device under test) is the only thing which will influence the end result you're looking for - not adding the losses in your cables, parasitic inductance's etc... and so on.

[nrgkits.nz]

I don't mind stretched coils - it usually suggests that someone has actually tuned the damn filter! One common cock-up is to put the coil against the board (like in the picture above) destroying the Q of the coil, and causing high induced currents in the ground plane. I saw a (whisper it) Broadcast Warehouse PA that had suffered from a big current being induced into the ground plane, with the copper foil rippled and burnt and one of the filter caps had desoldered itself!

My Band II PAs usually have the output filters on separate boards, and the coils are around 8mm above their board, with the groundplane etched away under the coil. The improvement in Q is huge. All the coils are offset to prevent mutual coupling and the high voltage caps are flat to the PCB, but clear of the coil windings. When I next have one nearby, I'll put up a picture. The design was published in the 1960s in the ARRL handbook - they gave a really easily reproduced PCB layout that could be used on most of the VHF bands without modification.

Albert what sort of configuration are your filters? The ones I have constructed are usually a 9 pole chebyshev with 5x capacitors and 4x inductors. I usually use 16SWG tinned copper wire and pairs of 27p ATC100B capacitors for the inner poles (54p total), and single 27p caps for the 2x outer poles.

I also have another filter here which I bought second hand locally. It was used in a mobile RT installation in the A band below 88MHz for taxi's etc... its in a metal box and configured with 3x high voltage adjustable capacitors and 4x wire wound inductors using 18 SWG tinned copper wire. The rolloff was originally set at around 90MHz of thereabouts, but I widened the coils as much as I could and re-adusted the capacitors, now the roll off is a little above 108MHz, enough for a usable passband upto around 103MHz before there's too much attenuation.

lpf.jpg

Edit: Here's a filter I've spent this evening constructing and aligning, the analyzer trace is the insertion loss. I discovered my directional coupler had issues (Mini-Circuits ZFDC-20-4L) so had to resort to using my Diamond SX-200 SWR/Power Meter and a transmitter on 8 Watts into the filter to adjust for best return loss. In the end I had no measurable SWR/Reflected power on the meter and an insertion loss no worse than -15dB on the analyzer.

analyzer.jpeg

filter.jpeg

[BriansBrain]

I'm looking for a 30db at a reasonable price ... any suggestions

You could build your own but probably won't have a good frequency response at UHF?

I had a go at making one of these RF-Power-Sampler which is the same thing ( or is it ? )
http://www.n4ga.com/50db-hf-rf-sampler/

RF-Power-Sampler Diag.JPG

My Attempt

RF-Power-Sampler - Mine.JPG

On Test =

RF-Power-Sampler - Mine - Siglent.jpg

As you can see it's OK over the FM band

[Albert H]

Your sampler looks OK. I still think that SO239s are best used for CB! I stopped using them way back - as soon as I tried a TDR (Time Domain Reflectometer for those who don't know the term). I was amazed by the discontinuities in the cable from the PL259 / SO239 combination. N-types are very much better. I had cheap SO239s melt at just 90 Watts back in the 80s!