Filter problem

Do you know the nature of the ferrite in the toroids?

Manganese-zinc ferrites tend to be lossy at frequencies above a few hundred kHz. Nickel-zinc ferrites have a higher resistance but get lossy above a few MHz.

If there are specialised ferrites for the 150MHz range I've yet to hear of them.

They are FT-37-43, which are claimed to be good from 25 to 300 Mc/s.

Type 43 is okay. Type 62 is better for chokes and baluns, where you don’t care about loss.

For a one-off, I’d get some 50-cm micro coax jumpers with U.FL connectors and make open-circuit shunt stubs. 154 MHz is just under 2 metres, so at a velocity factor of 0.67, a quarter wave is 32.4 cm.

You’ll need a bit of series resistance between each stub and the next, because otherwise you’ll get a parallel resonance between stubs of slightly different lengths.

It’ll add some loss, and may need a transformer on each end to keep the notches narrow enough. ( You might want to be hanging 50Ω stubs off a 5-Ω point, for instance.)

This is easy to dork in LTspice. It isn’t the lowest-loss thing in the world, but you can tune it with dikes.

If you don’t mind using half-wave shorted stubs, you can tune them by sticking a sewing needle through the jacket into the center conductor, which lets you adjust in both directions. I use thumbtacks in RG-58 like that fairly often. Good Medicine.

Cheers

Phil Hobbs

What's the power level?

You might make a bandpass filter out of commercial inductors and caps and a few padders or trimmer caps. Build it on dremel'ed FR4 and bolt it into the box for good grounding. One narrow deep movable notch might help a lot.

Or do one of those SSB phasing tricks to kill the image sideband. Yes, I know it's FM.

Wiliams has the stuff you need to design a passive bandpass filter. The arithmetic is tedious. My NORMA program might help.

Milliwatts at high impedance.

That's only one section, I am going to need a lot more than that because the wanted and unwanted frequencies are only 5.3% apart.

I have found some calculations in the RSGB Handbook but my big worry at these frequencies is unwanted 'components' in the form of wires (or box sides) with inductance and stray capacitance everywhere. If I make a simple tuned circuit with an air-cored coil (or helix) across the width fo a small die-cast box, the return currents from the grounding of the capacitor at one end will flow back through the box wall to the grounding point of the coil at the other end.. These currents will spread out and may interact with a similar circuit layout at the other end of the box to give unwanted coupling.

I had this happen many years ago with something that worked at 100 Kc/s, so the problem could be much worse at 150 Mc/s. (It also caught out the designers of the RA17, hence the hacksaw slot most of the way through the chassis.)

I am avoiding the use of printed circuits, it is all being built on tagstrips and standoff pillars - and a lot of the circuit can be supported off the valveholder tags (but not the filters).

That was my estimate - it would fit in the overall housing (though not in a die-cast box of the type I am using to make the modules). It might even work if it were simply threaded into the loom with all the other wires.

Would it be better to use inductive or capacitive coupling, to reduce the losses?

The impedances are high because it is in the anode circuit of a valve. I could use a tapped resonant circuit or a ferrite transformer to bring down the impedance to 50-ohms, which would make it easy to test with a VNA.

Perhaps a quarter-wave line could be used as part of the impedance transformation? Audio screened cable has a characteristic impedance of about 120 ohms, so a termination of 50 ohms at the outpute end of a quarter wavelength would appear as 288 ohms at the input end.

I wonder how tightly you could roll it up without destroying the effect?

Some of the circuits I have already built for this project are at:

The stuff I’m talking about is 0.050” diameter or thereabouts. Widely available for cheap on AliExpress. You can coil it as tight as you like.

I don’t know what your plate Z is at 2 metres, but it isn’t super high. I’d say a 1:10 transformer on each end (500Ω->5Ω), plus three OC stubs with two 2Ω resistors to make a 3-section pi network, would be a good place to start spicing.

You don’t want much conductance at the open end. Just the featureless cut end.

Cheers

Phil Hobbs

Heck, 150 MHz is almost DC.

Retro look.

Not with valves, it isn't

Not appearance but practicality. The plan is to avoid semiconductors altogether; with valves it is much easier to make one-offs on tagstrips and try out different components and layouts. At 150 Mc/s there are unspecified hidden capacitances and inductances waiting to catch you out. Some valves for those frequencies were designed with a specific layout in mind ( ECC91, QQVO 2-6, QQVO 3-10, QQVO 3-20, QQVO 6-40).

I have known laminated printed circuit boards to track across between the layers. With transistors this would hardly be noticed but with the higher voltages and much higher impedances of valves, it can cause all sorts of strange intermittent faults.

I wasn't thinking of using that line as part of the filter. The equipment is modular, so there have to be connecting leads between the various parts. My idea was that by using 120-ohm cable and cutting it to the right length, the transformation could be done in two steps; line and transformer. This would alow a large impedance ratio to be obtained whilst keeping the ferrite transformer ratio to a sensible value.

We were discussing building a bandpass fiter. At low power, the parts can be one per cent of a wavelength long.

And at low power, tubes should be small, so plate capacitance will be small, and can be rolled into the first bp filter capacitance.

What's the plate capacitance of the tubes you plan to use? Is your output differential?

I breadboard with surface mount transistors and passives all the time. These parts are small and planar, which tubes aren't. And a hunk of copperclad has a beautiful ground plane on the back side.

A transistor, especially a GaN fet, has transconductance measured in Siemens. Tubes are mS. Don't need a socket or a heater supply. Really hard to break.

MMICs are really (and literally) cool.

Come on, try it. It's 2025.

I was intending to make the characteristic impedance of the filter 50 ohms so I could set it up easily with a VNR.

ECC91= 2.5pf max with anodes paralleled EF91 = 2.1pf (3.1pf if shielded)

No, see:

I can go out and buy a ready-made transceiver but I want to make one using valves. It's a hobby but I want to approach it in a professional manner.

Am 13.06.25 um 17:46 schrieb Liz Tuddenham:

You can get my FT-505, 560 W PA input. It even has a 7360 beam deflection tube in its SSB modulator for added exoticism. I bought it when I was still at school / from my 1st internship. Only handover in S/W Germany near Luxemburg. Currently unable to carry it.

cheers, Gerhard DK4XP

Xc is over 400 ohms. Basically a pure current source.

There are filter forms that are 50 ohms on one end and Hiz on the other.

There is an inductor in the anode circuit, which resonates with the valve capacitance (and a trimmer) at the output frequency.

At lower frequencies, pentodes are generally considered to be current sources and triodes are nearer to voltage sources. As they approach their highest operating frequencies, electron transit time messes that up. The triode is a mixer, so Miller capacitance doesn't apply because the two input frequencies are not the same as the output frequency.

What form do they take? Are they band-pass, band-stop or low-pass?

But then you want a bandpsss filter, I think.

There's no Miller effect only if the grids are zero impedance to ground at 150 MHz.

Generally one starts with a lowpass filter and translates that into a bandpass or bandstop.

This is the classic reference:

Some lowpass filters are terminated on both ends and some just one. You can also use a transformer on one end.

Is there any chance you can make enough room in your setup for something of the following sort?

That's exactly the sort of thing I had in mind. The author's dimensions of 4" diameter and 6" high might be a bit of a problem but I could try scaling it down and adding more turns to the helix. Perhaps a piece of

A length of thinner pipe would be easier to accommodate, so perhaps I could experiment with something intermediate between between a coaxial stub and a helical resonator.

Amazon won't let me see it because I am not in the USA. In any case, won't buy anything from Amazon Do you have another reference to the book (I presume it is a book)?