EMRFD Message Archive 7317

Message Date From Subject
7317 2012-03-10 07:08:25 va3vfo HB Filter for 6 KHz
Hi- new to group, love the EMIRD book. Fills a void in info and not too "heady"
for beginers.


I hv no formal RF / Electronics training but have
been self learning for many years. Please forgive if some questions are too rudimentary. Group probably too advanced so will mostly lurk, however:

Re: building a home brewed crystal filter for use at centre frq of 8215.kHz and BW of 6000 KHz . Is this a worthwhile endeavor or is the expense/ difficulty of buying the required 8215 crystals not worth it? I am looking for an experimental option for an AM
Filter for an FT102 Yaesu xcvr. Inrad has them premade for 120$ but I was looking for other possible alternatives such as using ceramics etc.

Is the operating centre frequency the main stumbling block to any realalistic experimental substitute?

Thanks
G
7318 2012-03-10 07:31:26 Tim Hills Re: HB Filter for 6 KHz
I'm building a homebrew dual conversion receiver mostly for MW and LW
DXing but I went with 10.7MHz and 455KHz for the IFs because the filters
are more available and cheaper.

I'm sure someone on the list has a link to a custom crystal source.

Tim Hills
Sioux Falls, SD


7319 2012-03-10 08:43:58 William Carver Re: HB Filter for 6 KHz
>
> > Re: building a home brewed crystal filter for use at centre frq of
> > 8215.kHz and BW of 6000 KHz.

Homebrew "lower sideband" ladder filters (xtals in series, shunt
coupling capacitors to ground) are fun and inexpensive using crystals at
frequencies that are used for microprocessor clocks, TV color burst
frequencies, etc.

Narrower filters, where coupling capacitors are much larger than the
parallel capacitance of the crystals, are quite successful. Wider
filters are more difficult: without tricks the width can't exceed the
spacing between the series resonant and parallel resonant frequency of
the crystals. As you approach the pole-zero spacing of the crystals it
gets increasingly difficult to synthesize a flat top and symmetrical
skirt selectivity. 6 KHz is probably going to be a bit difficult. This
is one reason why wide commercial filters are usually lattice topology
rather than ladders.

The xtals and coupling capacitors determine the centre frequency you get
and for a particular series resonant crystal frequency you have quite
limited ability to position the passband to an exact frequency.

8215 KHz is NOT one a commodity frequency so as far as I know the
crystals would have to be made. And if someone makes crystals series
resonant at 8215 KHz the center frequency of the resulting lower
sideband filter 6 KHz wide will be higher, perhaps as high as 8218 KHz,
a problem if the centre must be 8215 for a commercial radio.

I would suggest getting some inexpensive crystals at a commodity
frequency near 8215 KHz, say 8 MHz series resonant. Try your hand at
making a 6 KHz wide filter with those as an experiment. You can see
where the passband falls with respect to the series resonant frequency
of the crystal and you can evaluate the shape of the filter with simple
lower sideband topology. Then if it meets your needs you can pursue
having crystals made for 8215 with more confidence it will be
successful.

Bill
7320 2012-03-10 08:48:09 Alex P Re: HB Filter for 6 KHz
International Crystal Manufacturing ICM



7322 2012-03-10 22:21:15 va3vfo Re: HB Filter for 6 KHz
TNX to everyone for their replies. Bill has really provided the exact explaination I was looking for, vis-a-vis AM vs. SSB topology requirements etc.

INRAD does have this filter for sale, as it is comm
7324 2012-03-11 04:54:39 Ashhar Farhan Re: HB Filter for 6 KHz
What happens if we brew a lattice filter using the same xtals? Second
q, can't we pull the crystals with a trimmer like we do in a vxo?
- farhan

On 3/11/12, va3vfo <va3vfo@yahoo.com> wrote:
> TNX to everyone for their replies. Bill has really provided the exact
> explaination I was looking for, vis-a-vis AM vs. SSB topology requirements
> etc.
>
> INRAD does have this filter for sale, as it is comm
7325 2012-03-11 08:27:25 kb1gmx Re: HB Filter for 6 KHz
7327 2012-03-11 10:15:24 William Carver Re: HB Filter for 6 KHz
Think of a zero-bandwidth ladder filter: all the coupling capacitors
would be infinite and all the crystals would be on the same frequency.
The filter zero-width passband would be at the series resonant frequency
of the crystals.

As the coupling capacitors are reduced to finite values the width
increases. The resonant frequency of the "mesh" formed by each xtal and
the the coupling capacitors on either side is going to increase because
in that mesh the xtal is "in series" with the capacitor values.

This thinking indicates that if the coupling capacitors are reduced in
value to produce a bandwidth of "B", the center frequency will be the
xtal series resonant frequency of the crystals plus B/2. So to a first
approximation a 6 KHz wide filter at 8215 KHz will require crystals that
are series resonant at about 8212 KHz.

At that frequency the pole-zero spacing will be >6 KHz, so a 6 KHz wide
filter will be theoretically possible. But as you approach a significant
percentage of the xtal's spacing the parallel resonance of the crystals,
above the passband, makes the upper edge of the filter passband droop,
and the symmetry of the skirts goes to hell. I'd certainly want to
experiment, either with some 9 MHz xtals or better yet, with PSPICE,
before I laid out $15+ per crystal.

International Crystal made some crystals at 8215 for a prototype SSB
filter. I'll ask if they can retrieve an average number for Q, Lm and
the parallel capacitance. From those numbers you could do a paper filter
design and put it into PSPICE to see what it will look like. But it may
not still be in their records.

A Cohn filter has the same xtal frequencies and the same capacitors. You
must accept whatever impedance those capacitors, chosen to get a
specific bandwidth, produces. But no big deal: at HF a transformer wound
on a binocular core is darned near "ideal".

Unfortunately Cohn filters of more than about 4-5 poles start getting
weird aberrations at the upper and lower edges of the passband (Wes
explained that to me about..oh..25 years ago?). But if you have Wes'
software you can compute the specific series resonant frequencies of the
xtals needed to produce any of the polynomial filters, Butterworth,
Chebychev, or even the oddball filter I like, equiripple phase.

Bill
7328 2012-03-11 10:22:43 William Carver Re: HB Filter for 6 KHz
On Sun, 2012-03-11 at 17:24 +0530, Ashhar Farhan wrote:
>
> What happens if we brew a lattice filter using the same xtals? Second
> q, can't we pull the crystals with a trimmer like we do in a vxo?
> - farhan
>
> On 3/11/12, va3vfo <va3vfo@yahoo.com> wrote:
> > TNX to everyone for their replies. Bill has really provided the
> exact
> > explaination I was looking for, vis-a-vis AM vs. SSB topology
> requirements
> > etc.
> >
> > INRAD does have this filter for sale, as it is comm
7329 2012-03-11 10:51:05 William Carver Re: HB Filter for 6 KHz
On Sun, 2012-03-11 at 17:24 +0530, Ashhar Farhan wrote:
>
> What happens if we brew a lattice filter using the same xtals? Second
> q, can't we pull the crystals with a trimmer like we do in a vxo?
> - farhan

You can pull the effective series resonant frequency of a crystal with a
series capacitor. Note the word "effective". An xtal is not a simple
series resonant RLC, it also has a 3-6 pF capacitor in parallel with it,
so it has both series and parallel resonance.

When you put a capacitor in series with the xtal it moves the series
resonance of the mesh, just as the coupling capacitors do, but the
effective Q, and L also are changed. As long as your trimmer is large,
say 50 pF or so, the effect might be tolerable....it depends on how
picky you are: if you're trying to obtain a flat Chebychev bandwidth or
no ringing with one of the flat group delay filter types things go to
pot pretty quickly.

In a batch of crystals there will be some scatter in the series resonant
frequencies. I try very hard to match up that batch to the spread that
the (non-Cohn) filter design calls for to minimize the amount of tuning
that has to be done on individual crystals and maximize the tuning
capacitor size.

For any of the "polynominal" filter types....Chebychev, Butterworth,
Cohn, Bessel, equiripple phase, Gaussian-to-12 dB, ETC, every "mesh" in
the filter should be on the same frequency. That frequency is the center
frequency of the resulting filter. This "mesh" is formed by the crystal
itself, in series with the coupling capacitors on each side.

Once you've picked an xtal to go into each place in the filter, and have
a set of coupling capacitors picked out, you can measure the series
resonant frequency of the crystal IN SERIES WITH THE COUPLING CAPACITORS
ON EACH SIDE, and add a series tuning capacitor to tune this mesh to the
center frequency of the filter. And as I said above, those capacitors
are hopefully pretty darned large. Hundreds of pF tuning capacitors
won't destroy the design.

----------------------------

I have never designed a lattice filter. Wish I knew how. I have books
that tell you, but I've never slogged through the math enough to learn
the process. "Crystal Filters" by Robert G. Kinsman (ISBN 0-471-88478-2)

But you can see that if all the crystals have the same parallel holder
capacitance their effect will be cancelled by the bridge-nature of the
lattice.

Bill
7330 2012-03-11 16:03:43 kb1gmx Re: HB Filter for 6 KHz
>
> I have never designed a lattice filter. Wish I knew how. I have books
> that tell you, but I've never slogged through the math enough to learn
> the process. "Crystal Filters" by Robert G. Kinsman (ISBN 0-471-88478-2)
>
> But you can see that if all the crystals have the same parallel holder
> capacitance their effect will be cancelled by the bridge-nature of the
> lattice.
>
> Bill

I have and it depends on getting crystals with pole/zero with
the right spacing and two (or more frequency pairs). The OLD SSB handbooks had the methods for filters in the LF (455khz) and HF
(5-8mhz) range using FT243 holder where you could find a bunch
at the same frequency and then etch or grind up the other for the higher. That's is rather old school. While FT243 rocks are findable
pairs and active rocks are not so common. To day I look for rocks
in the series or parallel realm and then find pairs that are 1.5 to 2khz apart and shoot for 2 pair (2 low and two high) that form pairs
for a back to back half lattice. I also built a fair filter using
two about 1khz apart @9mhz and padding the upper with series C (47pf)
that produced a decent 2khz filter when all the cap and intermedite
coil twiddling was complete.

Asymmetric filters (USB/LSB) can be done that way as well.

It's an alternate and tends to produce a filter that is fairly symmetrical and if the right rocks are chosen clean outside the band pass. But a good filter still needs at least 6 crystals and 8 being better. Using 6- 8 rocks in a transistional ladder filter will be pretty good as well with the difference that finding 8 with close
frequency in a bunch being more likely.


Allison
7331 2012-03-11 17:12:07 William Carver Re: HB Filter for 6 KHz
Allison, I've BUILT lattice filters, just never designed one, hi. I made
a filter at 455 KHz for my Hallicrafters S-85 when I was a kid. That was
a total bust. Didn't have ANY knowledge then so didn't recognize the
colorful surface of the plated FT-241 crystals was CORROSION, hi.

The last one is a 10 pole, 8 zero filter 250 Hz wide at 3 dB at 5.2 MHz.
It's a computer-design filter called a "TBT", for "transitional
Butterworth Thompson" because it's partway between a Butterworth and
Thompson (like a Bessel). It does NOT ring, sounds incredible, and it
has a 4:1 shape factor....but at 3/100dB rather than the usual 6/60 dB.

Great, but I couldn't DESIGN it if my life depended upon it. Just
happened to meet a sorta-retired professional filter designer who took
pity on me.

One way to find SSB-spaced pairs, useful in the high end of the HF range
(say 8-16 MHz) is to buy two batches of xtals. Find someone who makes
series resonant and "20 pF load capacitance" (maybe 18 pF for some
manufacturers). The only difference is their center frequency, and
that's pretty close to the width of an SSB filter.

In the last ten years I've used filters from microwave baseband
electronics: frequency division multiplex boards retired from telep[hone
servce. They're a little wider than ham filters, and that's fine with
me: voice quality is better. Then I build corresponding narrow filters
for CW/data use.

Bill
7332 2012-03-12 14:04:18 kb1gmx Re: HB Filter for 6 KHz
7333 2012-03-12 15:16:21 William Carver Re: HB Filter for 6 KHz
Roger on all Allison.

The guy who designed the 250 Hz CW filter wrote an early IEEE paper on
the TBT math (and corresponding Chebychev design). He contacted me and
said my ladder filter was OK, but "why ladder?". When I said "Cause
that's all I have tools to design with" he said "Maybe I can help", and
collaborated for six months, gratis.

He wrote his own software and naturally considers it proprietary. And
wrapped up within that software is the math, which would probably be
nearly overwhelming even with a calculator or spreadsheet.

Yes, the idea of homebrewing an SSB lattice was to buy some 8.0000000
MHz series, and 8.000000000 18 pF. The 18 pF would be series resonant at
perhaps 3 KHz lower and you'd have the material for an 8 MHz LSB
lattice.

The 5.2 MHz SSB filters were from Motorola FDM boards. They are ladder
filters using two pole resonators and I found enough in Kinsman's book
to believe that HE designed them! A bit odd, but I took some apart to
make matched four pole roofing filters with 0.5 dB insertion loss.

Wes blew the lid off decades ago, when he explained how to make simple
ladder filters. And if you put your Big Boy pants on you can go from
simple to some really spectacular filters.

Bill
7335 2012-03-13 07:59:25 Donald Re: HB Filter for 6 KHz
I enjoyed this thread. I kinda went the opposite direction.

I built a 40M tube receiver a few years back with a 4MHz IF. My first filter was a lattice type using 2 uP crystals - one series type and the other 20pF.

It was much simpler than the filter configuration Allison and Bill describe. The idea for using a lattice came from the old 60's Handbook receivers of which I had built at least two. The receiver project was in part to see if I could use all toroids in a tube rig including the LO.

The lattice filter worked OK but worked poorly for SSB and too wide for CW. I think there was nothing I wanted to spend time on that would have produced anything but disappointing results.

I eventually change to a 4 pole ladder filter using the same process I used in SS rigs following W7ZOI and others projects and books. I am certain I could improve this filter but it works very well.

I sort of proved to myself that you could build a vintage single conversion receiver using many modern component and techniques. I didn't want to use expensive transformers or crystals.

http://i732.photobucket.com/albums/ww328/Don_K5UOS/RXladderfilter.jpg

http://files.qrz.com/s/k5uos/k5uos.jpg


Sorry if I changed the thread somewhat. I know it is a bit unusual to mention tubes in the context of EMRFD.

Don


> I have and it depends on getting crystals with pole/zero with
> the right spacing and two (or more frequency pairs). The OLD SSB handbooks had the methods for filters in the LF (455khz) and HF
> (5-8mhz) range using FT243 holder where you could find a bunch
> at the same frequency and then etch or grind up the other for the higher. That's is rather old school. While FT243 rocks are findable
> pairs and active rocks are not so common. To day I look for rocks
> in the series or parallel realm and then find pairs that are 1.5 to 2khz apart and shoot for 2 pair (2 low and two high) that form pairs
> for a back to back half lattice. I also built a fair filter using
> two about 1khz apart @9mhz and padding the upper with series C (47pf)
> that produced a decent 2khz filter when all the cap and intermedite
> coil twiddling was complete.
>
>
>
> Allison
>
7336 2012-03-13 11:32:02 William Carver Re: HB Filter for 6 KHz
Don, I have no problem with trying things on the bench vs modeling. Or
tubes for that matter (I'd LOVE to have a couple of new 7360's but at
$30-$50 each for 40 year old tubes I have to pass).

As the saying goes "there is more than one way to skin a cat". It's more
than a cute saying, it's a belief in diversity, a denial of
one-solution-fits-all.

Bill
7337 2012-03-13 11:32:35 Juanjo Pastor Re: HB Filter for 6 KHz
Hello everybody,

A trick used by Ricardo Llaurado, EA3PD, for lattice filter some 30 years
ago is to use third harmonic crystals in their fundamental for the filter.
There is quite a wide dispersion of fundamental resonance in third harmonic
crystals. He used CB crystals, but now there are other widespread crystals
designed for third harmonic resonance, and they are dirt cheap. A 6 KHz wide
filter should be easy to implement then...

73, 72 de Juanjo, EC5ACA. EA-QRP #104, G-QRP #9742, QRP-L #1662,
FP #899.

Juanjo Pastor
C/San Roque, 4-1º
46460 Silla
SPAIN

----- Original Message -----
7338 2012-03-13 15:40:19 kb1gmx Re: HB Filter for 6 KHz
7341 2012-03-17 05:47:58 Tim Re: HB Filter for 6 KHz
My suggestion below doesn't help the OP because he has to have a center frequency of 8215kHz but maybe homebrewers would take it to heart:

If one wanted to build an AM-bandwidth LATTICE filter using cheap microprocessor crystals there is one pair of widely available microprocessor crystals that might fit the bill: 5.9904 and 6 MHz, with a delta of 9.6 kHz between them. I think every major lineup of microprocessor crystals includes both these frequencies.

As others have pointed out the difference between a series-cut and say an 18-pF cut crystal for the same stamped frequency, is going to be several hundred hz (just a few Mhz) to 2 or 3 kHz (10 or 15 MHz crystals) but I don't know how to build a AM-bandwidth lattice filter from those until the center frequency approaches 20+ MHz. K8ZOA has a very nice web page
7342 2012-03-17 09:28:16 Juanjo Pastor Re: HB Filter for 6 KHz
As I said in my previous post, there are cheap 30 cents rocks that are third
harmonic mode crystals and they have some 5 KHz dispersion in their
fundamental frequency. I have tried several of them in a 74HC04 simple
oscillator, I found that they are maked for their third harmonic: 24, 32, 40
MHz... oscillating at about three times lower than marked frequency: 7.99,
11.66, 13.32 MHz and so. The op wanting to build a 8215KHz centered filter
will have to order at least 4 crystals custom made. Guessing about 20$ per
rock, that is 80$ plus developing and adjusting the filter. Anyone is
willing to accept the challenge? I guess the 120$ Inrad filter is not so
expensive in the end...

73, 72 de Juanjo, EC5ACA. EA-QRP #104, G-QRP #9742, QRP-L #1662,
FP #899.

Juanjo Pastor
C/San Roque, 4-1º
46460 Silla
SPAIN

Tel.: +034 96 120 17 67
Movil: 651 35 35 11

----- Original Message -----