Transverters/Microwave filters

[VK3XDK]

Gday everyone,
Due to the problems with PCB manufacturing consistencies Im developing two new transverter boards covering a range of 1.2 to 5.7 (L) and 5 to 12Ghz (H).
Built on Ceramic substrate with 2 stage RX and 2 stage TX MMIC devices.

Instead of relying on a pcb filter these boards are designed to have a offboard filter (and Ive layed the board out to allow the use of a pipecap filter on the back of the board as an option)
Im thinking a "multiband" Transverter using a YIG filter may even be possible!!! (thanks to Dave VK3HZ for the Idea)

These fellas are very early in the prototyping stage and will not be available for some time.( All of my other transverters are still available)

Obviously "commercial" filters will offer VERY good Selectivity, LO rejection and low losses would be my first choice with these new boards BUT
as we all know, these filters can be expensive and/or hard to obtain so Ive been pondering other solutions.

My initial experiments with pipecap filters have been VERY disappointing: only 30-40db of out of band isolation, large losses (6-10db)and terribly gentle rolloffs.
Has anybody had success with these filter?? Maybe someone could put me onto some designs that work??
Any other filter Ideas??

Cheers, Graham VK3XDK



cheers, Graham VK3XDK

[VK3XDK]

Thanks Mat,

Ive been playing with 1/2inch pipecaps mainly.
The screw sizes ive tried are 5/32 (seem best) and 2mm
Ive tried probe lengths of between 4mm and 8mm using brass sewing pins through the board.
It seems the filters work the best at lower frequencies where the screw is nearly at the full length of the pipecap. Maybe the pipecaps would be better if shortened at 10ghz??

I have a larger pipecap filter (3/4?) that I made some time ago by just soldering two sma connectors to the bottom of the pipecap with the pins sticking in (I should submit a photo!!)
This fella has reasonable performance at 3.4ghz but still a bit lossy (at around -6db)
It does made a very handy test bench filter.

[VK3XDK]

Here they are.

[VK4OE]

Hello Graham and other readers.

Whilst my experience with pipe cap filters is a little like your description of larger bandwidth than can be achieved with other technologies, my general impression is that they should be better than you report.

That led me to question the "open bottom" type of construction that is visible in the couple of photographed examples that you attached. Intuitively speaking, that may be the cause. And as Matt has sugggested, the length and location of the filter probes is very significant for optimum functioning of a pipe cap filter.

In a simple 5.7 GHz transverter that I made a couple of years ago, I needed to 'go' to an assembly of two pipecaps to achieve the out-of-band rejection that I needed. I did this with the two pipe caps soldered onto brass plate, SMAs dooubling for the input and output connections as well as probes, and a short length of .085 semi-rigid for the 'between filter probes' plus the connection between them. It worked very well.....

Just thought that I'd describe a positive experience.

Do note also that Kuhne like to use silver plated filter and screw hardware - I've never got to that, but it may make a difference!

Cheers and 73,

--Doug Friend, VK4OE.

[VK2EM]

G'day Graham

I have made two transverters for 10GHz, and various other oscillator/ multipliers for microwaves, and now a transverter for 5760MHz.

I did a lot of experimenting with pipecap filters, including reading Paul Wade W1GHZ "Pipecap Filters" and "Pipecap Filters Revisited"... and Kent Britain WA5VJB "Cheap Microwave Filters", "Pipecaps Explained", and "Copper Pipe Endcap 10GHz Bandpass Filter".

I did some experimenting with probe lengths/sizes, and using 1/8th brass screws for tuning. I was never happy with the results until I read somewhere that someone was using the centre conductor of UT-141 coax for the probes, so I used that. This improved the response somewhat.

I then thought that if the response improved with larger diameter probes, then maybe a larger tuning screw may improve things, after all a screw is an uneven device, and the 'average' diameter of a 1/8" screw is much smaller than 1/8"(3.18mm). I tapped out the caps to 4mm, and used M4 brass screws. The tuning was then much improved, much smoother and repeatable.

I mounted a cap on a piece of 1/32" FR4, and played with probe lengths (solder and test method) and screw depth, measuring loss and bandwidth.

Copper & Brass Caps and 5328 FR4 Multiplier

Multipliers

I didn't measure the exact loss, but as Paul W1GHZ says "Gain is cheap". So with some ERA-3's and ERA-1's, sufficient levels can be attained for multiplying and mixing (generally +7 to +10dBm).

Also, one pipecap filter is not enough for post-mixer filtering, so two two pipecaps with a Fet interstage amplifier is suitable.

73 Bruce VK2EM

[VK5ZD]

Hi

There's a lot of useful information in this article: http://www.w1ghz.org/new/Pipe-cap_Filters_Revisited.pdf

73
Iain

[VK3XDK]

collection of todays playing.
no real improvements on the standard pcb type!!

I can get some good sharp peaks with reasonable insertion losses BUT im finding "out of bandwidth" isolation often very poor (only about -40db), wonder if shielding between the probes would help??

[VK3XDK]

More playing!!
I have found that soldering sma connectors on the side of the pipe cap (about half way) works just as well as the pcb/pin method, It has the advantages of being easier to make (etched pcb not needed) and can be remote mounted

Im still not 100% happy with the band passes that i have been achieving (using either method) but combined with an image rejection mixer may get good results??
Obviously the pin length has the biggest effect on the response and a compromise between insertion loss and sharpness needs to be made. In fact the biggest issue that I have been trying to resolve is not the peak of the filter but the outer skirt of the shape. (looks like a VERY bad VCO). Seems like there is leakage between probes that is very difficult to stop. I tried a bigger screw (for more shielding) but it dint help.

I also tried using a brass pcb standoff as the main element, the tuning screw goes right down through the standoff for tuning. I figured that getting rid of the thread (bad RF surface) at the high current point of the element would vastly improve the response, unfortunately It made little difference except maybe being easier to tune (larger thread contact area).

Also tried the sma at various places and about halfway up the pipecap seemed to be the best for ease of build vs performance.

Most of my testing has been with 3/4" pipecaps (as Ive run out of the others), 1/2" would be better for 10ghz (and maybe 5.7??)

results of the filter in photo 3
3400Mhz -3.8 to 4db of insertion loss
@3256Mhz -30db
@2968Mhz -42db (Fairly happy with this one)

5760Mhz -1.8 to 2db of insertion loss
@5616Mhz -16db
@5328 -28db (I wonder if it will be enough with an image rejection mixer??)
@5000Mhz -36db

10368Mhz (Interesting results)
Less than 2db insertion loss

@10224 -30db !!
There is a pronounced dip bottoming at -48db @ 10187Mhz (some waveguide mode??)
@9936 -24db

I could of spent lots of time on my mill and lathe to create a much better filter but that is not what this is about, my main aim is ease of construction and most of the filters took less than 30min to build.

Back to the testing, Graham VK3XDK

[ZL1SWW]

Hi,
Yes I have played with Pipecap filters at 2.4 GHz and was a little disillusioned as well with performance. Well you could get good attenuation at the skirts with wide spacing and short probes but bandwidth / insertion loss suffered. Longer probes gave better coupling / less insertion loss but outer skirts were less steep (broader) and hence LO rejection was poor.. Can't win! Maybe stagger tuning two in series is good but takes more space.
Often wondered about those Toko ceramic filters. Trouble is they seem to not go into the higher bands (highest I have seen is 3.4 thru 3.5 GHz for WiMax use). Sad they seem to just slightly off the edge for Ham stuff in general...

Was honoured to meet up with Paul W1GHZ in ZL and had a good chat about microwave stuff, very knowledgeable guy. Liked his take on microwave on the cheap. As mentioned, gain is cheap..

Stiil in search of a good filter....

Cheers,

Simon,
ZL1SWW

[GM3SBC]

Hello All,
I always look in on you lads down there. All interesting stuff, keep up the good work.

Just two comments on the pipe filters which may be of interest.

1) I have always found that it was difficult to tune and set the screw accurately using the two lock nut method because of backlash etc. I often wanted to keep the height of the filter assembly at a minimum so I decided to use only ONE nut which is soldered to the top of the pipe cap as per normal. Having soldered the nut I now use a mini hacksaw and cut a slot through the nut to approximately 60 percent across the thread. I now apply a down force on the face of the nut above the slot area until the slot partially closes. Now insert the screw and test the level of resistance to turning the screw. I set the slot distortion at a point were the screw is very reluctant to be turned. Once set on frequency the screw never moves. There are no nuts to tighten and destroy your setting position !

2) I made some 2.3 Ghz 28mm I.D. pipe caps with a double sided PCB base soldered to the pipe cap base. I have been been frequently observing a detuning of the filter from one day to the next. Up to 6 dB additional loss from the tuning peak has been common. I now believe that this has been the result of the use of a THIN PCB board being used ( 1/32 inch ) and small distortions of this board are occurring due to strain being exerted from the externally base mounted SMA connectors / cable assemblies. This results in changing the position of the probes inside the cavity and bingo it is no longer on frequency. It does not require much movement !

So its just a point to consider, that mechanical rigidity of the entire pipe cap assembly is vitally important at microwaves.

Best Regards
Ed Murphy GM3SBC

[VK3XDK]

Interestin Ed,
I wondered how they would go with temperature variations, and ive noted that board flex creates big variations. (perhaps another benefit of the side mounted sma??)

I would like to try some silver based solder and possibly silver coating (if there is an easy option, ie paste or chemical)

[VK4GHZ]

More playing!!
I have found that soldering sma connectors on the side of the pipe cap (about half way) works just as well as the pcb/pin method, It has the advantages of being easier to make (etched pcb not needed) and can be remote mounted
.
.
.
5760Mhz -1.8 to 2db of insertion loss
@5616Mhz -16db
@5328 -28db (I wonder if it will be enough with an image rejection mixer??)
@5000Mhz -36db
.
.
.
Back to the testing, Graham VK3XDK

Hi Graham, and others.

Three years on, and any updates?
I have some 20mm end caps here, some SMAs and about to try your mid-mounted SMA idea.

My application is a bit different to conventional ham requirements!
My non-ham flying buddies operate their video downlinks on 5740 MHz (Fatshark Ch1) and 5800 MHz (Fatshark Ch5) as LIPD, whilst I'm down on 5665 MHz (remembering the 6cm ham band goes down to 5650).
I want to knock back their signals at my RX, because, frankly, these broad-as-a-barn-door video receivers suffer when other in-band transmitters are closer to your own receiver than what your own TX is.

I've assembled an inline 5/4 WL stub notch filter/T arrangement, tuned for 5800 MHz.
It's 5/4 WL long because a 1/4 stub length @ 5.8 GHz (8.9mm long) would finish somewhere inside the SMA connector!
And if I accidentally trimmed too much off, then I'd cut on down to a 3/4 stub.

@ 5665 I have 4.33dB insertion loss due to the notch curve not being as sharp as what I would like.
@ 5740 there is 11.17dB atennuation
@ 5800 there is 16.5dB attenuation.

Although happy with the 10dB+ cut at my mates freqs, the 4dB insertion loss on my own freq is a bit more than what I'd like.

Hence, now looking at trying a single-stage pipe cap bandpass filter tuned @ my operating frequency of 5665 MHz.
I need a filter that's a bit sharper than what the stub notch is offering.

So, any developments in pipe cap filter construction considerations?

[VK3MAT]

Hi Adam

how about a wavegide filter? i have been reading this document i came across the other day. http://www.w1ghz.org/filter/Waveguide_F ... _build.pdf
these mightn't be sharp enough but still offer plenty of rejection?

i guess it will depend on your design criteria size / weight?

would love to know your thoughts,

Matt