LPF Chebychev Q re: input/output capacitors

I found a 13 element 150 Mhz to 174 Mhz adjustable (pinch/stretch the coils?) filter in my junk box, It has seven capacitors marked as below.

10 E---30pf---36pf---40pf---36pf---30pf---10 E All caps are silver mica, labeled 850 volts. I wonder about the values of the first and last capacitor, also why are the special?

I have found a calculator for multi element chebychevs, but can't match these values, any clue.

formatting link
If I could match the cap values I would have an idea what value the inductors are.

Possible hint: The input/output connectors are RCA phono style. The literature says it's 100 Watt filter.

I assume it's a 50 ohm filter, but here to be corrected.

So, I wonder about the values of the first and last capacitor, also why are they specially marked?

Why can't I match up the cap values with the calculator?

Mikek

Reply to
amdx
Loading thread data ...

Thanks for that URL!

Did you notice using 161.5MHz cutoff, 100 ohm and 0.1 dB ripple you get close [11pF] with all the coils near 150nH?

but what bothered me is EVERY calculation came up with the middle three caps being almost identical in value ?!

I have to tell you getting 'even' ripple with 13 elements is NOT easy! Perhaps this filter is a hybrid, lead in, 7 element highly tuned, then lead out? might be easier to control.

Anyway, if you get a chance it's a bit fun to go through all these classic filter designs using the BASICS, all those derivations etc. They were done just to get closed solutions using equations. With today's PC's you can pretty much do anything you want.

Reply to
RobertMacy

Yes the site has a lot of interesting programs. Took me a lot of searching to find a calculator that would handle a

13 element chebychev, it will calculate out to 19 elements.

I tried many combinations, I'll check your solution.

Several years ago I used the radio amateur handbook tables to make Chebychevs, I don't have it handy to see if that is common to have close values on the caps.. I have an intuitive feeling that the more elements you have, the more likely that would be.

I hope to rewire this as a filter for an FM transmitter, maybe tune it down to about 90Mhz. I don't care about ripple, although, it would be interesting to be able to put one of the low loss bumps at my transmit frequency. It is overkill, but a project. It is a great enclosure, 6 shielded compartments.

That's not going to happen, I'm lucky solve simple algebra equations.

Thanks, Mikek

Reply to
amdx

post the structure of the coil(s), ID, or OD. wire gauge, number of turns, and length of coil(s)

Using free femm 4.2 I [or anybody who wants to jump in here] can give you some pretty accurate values for the inductance(s)

measuring ID and OD yields pretty accurate wire gauge. is that silver plated copper?

or just one coil as a reference.

Reply to
RobertMacy

My thinking was, I'd need to wind new coils, and pad the existing capacitors to move the cutoff frequency down to 90 Mhz. Here are couple of pictures. Turned out good. Very clean for being 40 years old.

formatting link

formatting link

Wire diameter = 0.050"--- # 16 wire, silver plated copper.

All coils have 1/4" ID.

Coil 2, 3-1/4 Turns L= 0.350" Coil 4, 4 Turns L= 0.475" Coil 6, 4 Turns L= 0.300 Coil 8, 4 Turns L= 0.450" Coil 10, 3-1/2 Turns L= 0.390 Coil 12, 3-1/4 Turns L= 0.315"

C1= 10 E Unknown Value C3= 30pf C5= 36pf C7= 40pf C9= 36pf C11= 30pf C13= 10 E Unknown Value

There's the info as best I can provide.

Mikek

Reply to
amdx

Can't see those pictures, photobucket, dropbox, often tinyurls, just hang my systsm(s)

Meanwhile, if you don't have it, download free femm 42.

Does your email on these postings accept attachments?

Reply to
RobertMacy

I'm not sure I understand how to answer that question. If you send me an email, yes, I can receive attachments. (can't everybody?) I can send attachments.

If it helps, I posted them in abse.

Mikek

Reply to
amdx

No, sometimes google gmail, along with some email servicers, strips a zip attachment, throws away the email, and doesn't tell anyone.

I have NO idea what abse is.

Did you download femm 4.2? so I can send you the model for coil 6?

Reply to
RobertMacy

It is a newsgroup. alt.binaries.schematics.electronic

I haven't yet, go ahead and send it, my address is good. Did you get a value for coil 6? Need to go to work soon, probably won't get much time until tomorrow, dinner with the relatives. :-\ I'll check this group after I get to work. Mikek

Reply to
amdx

will do, but for me also by tomorrow

Reply to
RobertMacy

This combo 161.5Mhz, 100 ohm, 0.01db does not have capacitance values of; 10 E---30pf---36pf---40pf---36pf---30pf---10 E

I was hoping to hit capacitor values that match the ones in the filter. Then the inductor values would pop out. Mikek

Reply to
amdx

perhaps the first caps are either 'selected' or unusual value.

Anyway, the inductance I got for coil 6 seems to be a bit anemic anyway, at around 56nH, but I did have a fairly high Q. Have you ever noticed how the values of the filter components 'look' like the coefficients of a polynomial? Like take (x+1)^5

1,1 1,2,1 1,3,3,1 1,4,6,4,1 1,5,10,10,5,1 1,6,15,20,15,6,1 except for the ends, tracks your cap values a bit, and as expected for a Chebychev, squashed in the middle.

I sent you the models and screen shots of results. NONE of the parasitics are taken into account in any of these. But, it's an interesting starting place.

Note the mesh in the finite element analysis program, femm 4.2, is so small that it resolves the skin effect with very good detail. You can plot the eddy currents inside the wire, inside TURN 2, and the plot is smooth. Not very important since you get the same answer for inductance whether fine mesh or coarse mesh.

You should get these free PC tools, they're great for the professional AND for the hobbyist who wants to learn. femm 4.2 LTspice octave

Reply to
RobertMacy

When I get the time I'll pull out a couple of the coils and measure them. When I rewind it for 90Mhz, I'll stick with the approx. 1 wire spacing between turns. If the holes permit, I'll go with 18 gauge wire just to help gain on the lack of silver plating. I still don't know why I can't use that simulator and match the values in the filter. I have compared the program to another and got the same answers. If I can get them to fit, I might replace the RCA connectors with BNC connectors. I got all the simulations you sent. Did you every find a way to see the pictures? If not and you are curious, let me know and I'll send them to you. Thanks, Mikek

Reply to
amdx

haven't gotten them go ahead and send

The caps are 'symmetrical values, but from the description, the coils are NOT. Plus, with 13 elements, tuning becomes extremely important, so parasitics get in there and move your 'expected' values around. A good example, the largest coil is 56nh [based upon number of turns and shortest length] inductance of wires is in the range of 20nH/inch, so the lead length to each cap shifts its value around a lot. You mentioned the whole thing is in a single box? and there's not a lot of room? That's a few pF distributed around to each coil shifting their values, too. Since I haven't seen the picture, I couldn't estimate and put into the LTspice model for you.

Reply to
RobertMacy

got the images, thanks.

Whatis the EXACT part number, mfger of those caps?

better yet, data sheet and either Spice models, or S-parameters

Reply to
RobertMacy

No Part number, Manufacturer is UNELCO 30pf 850 VDC. As you recall the first and last caps don't have values, they are built slightly different with a lead on one end only that solders to the connector. It is labeled 10 E, I think that might be the part number. I'm speculating that if I could see the backside of the other caps it would have a part number similar to 10 E. That's all I got. If it turns out to be really important to you, I could get one unsoldered from the compartment. The filter is a TFD6102A, lots of them available made in 1977, I believe it came from a Motorola radio.

Mikek

Reply to
amdx

I sent you the LTspice plot when you insert all the inductors you measured, looks like 0.5dB ripple and cutoff at 165MHz, without any resistance in the coils.

Reply to
RobertMacy

"catalog" values for n=13, 0.5dB, & r=1 cheby are:

Cheby filter request: n = 13 r = 1 ripdB = 0.5 dB

R = [1.0000000000000000 1.0000000000000000]

1 1.7609970153470449 2 1.2772630942432961 3 2.6877820746492245 4 1.3802216881020906 5 2.7595619419312416 6 1.3955024245638863 7 2.7713887548761735 8 1.3955024245638865 9 2.7595619419312416 10 1.3802216881020908 11 2.6877820746492249 12 1.2772630942432939 13 1.7609970153470504
Reply to
Simon S Aysdie

What value did you use for the first and last capacitor? Mikek

Reply to
amdx

In that "normalized" (catalog) filter, they are the first and last values.

Try this:

Chebyshev filter Low-pass filter Equal-ripple pass band Bandedge loss = 0.500000 dB. Upper passband edge frequency = 165.000000 MHz Specified stopband type Multiplicity of zero at infinity = 13 Overall filter degree = 13 Input termination = 50.000000 ohm Output termination = 50.000000 ohm

1 +---R---+ 50.000000 ohm 3 +---C---+ 33.972288 pF 4 | L 61.600829 nH 5 +---C---+ 51.851370 pF 6 | L 66.566395 nH 7 +---C---+ 53.236112 pF 8 | L 67.303366 nH 9 +---C---+ 53.464269 pF 10 | L 67.303366 nH 11 +---C---+ 53.236112 pF 12 | L 66.566395 nH 13 +---C---+ 51.851370 pF 14 | L 61.600829 nH 15 +---C---+ 33.972288 pF 17 +---R---+ 50.000000 ohm
Reply to
Simon S Aysdie

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.