Topics

Crystal CW filter with Version 4 uBITX boards? #ubitx

Jim Sheldon
 

Has anyone successfully added a narrow crystal filter to the uBITX version 4 boards using Ashhar's included header connections? 

So far I've not been able to get a signal through filters at 4.6,  6.2 or 8 MHz.  The filters use closely matched crystals, They are Chebychev filters and have been designed with "ELSIE" for a 200 ohme input/output impedance with 4:1 transformers to match them to Ashhar's specified 50 ohms. Software has also been modified to allow moving the BFO and IF frequencies to match the filter's frequencies.  Can't seem to get enough signal through them to be useful though the software DOES put the IF where it should be.  I can hear a signal generator on whatever frequency the uBITX is tuned to, but it takes better than 10 microvolts input at the antenna for what I'd consider to be a usable CW signal in the headphones.

I built the filter using a 4 crystal - 200 ohm 8 MHz circuit with crystals and other components purchased from INRAD (Vibroplex now owns them) but either I haven't figured out the proper 4:1 transformers or there's just too much loss through the filter.  I have tried it with 4 and 6 MHz filters I made from scratch with only slightly better results (6 microvolts at the antenna for a usable signal)

Sure would be nice if someone was able to get something working to the point it was usable and repeatable, even on perf-board. 

The idea of switching the IF frequency to switch filters appears to be sound - just need to find the magic crystal/capacitor/transformer combination and for that I need someone with better knowledge of crystal filters and better test equipment that I have.

Thanks

Jim Sheldon, W0EB

ajparent1/KB1GMX
 

Did you change the 2nd lo to convert the 45mhz signal to  the new IF??

The Existing IF is 11.995mhz (more or less).  Why are you messing with 4.6, 6.2, 8 and other frequencies?

Likely the filter part has a chance but everything else is plain wrong.

I have limited use for CW but making a filter using matched (within less than 50HZ) crystals at 12mhz
isn't that magic and narrow filters are easier.  However the end termination will NOT be 200ohms
and likely lower maybe 50 ohms which is also OK.  Are you switching the existing filter out or what?

Allison

Bill Cromwell <wrcromwell@...>
 

Hi Jim,

My first question is in line with Alison's reply. Why not use 12 MHz xtals and keep it consistent? I have dome something similarin a couple of radios. But I got xtals to put the narrower filter on the same nominal frequency as the wider existing filter. Works a charm.

I am using narrow filters on the audio with my uBitx. The NEScaf works best for me. I really don't to make a $2500 radio out of it.

73,

Bill KU8H

On 07/06/2018 01:47 PM, Jim Sheldon wrote:
Has anyone successfully added a narrow crystal filter to the uBITX
version 4 boards using Ashhar's included header connections?

So far I've not been able to get a signal through filters at 4.6, 6.2
or 8 MHz. The filters use closely matched crystals, They are Chebychev
filters and have been designed with "ELSIE" for a 200 ohme input/output
impedance with 4:1 transformers to match them to Ashhar's specified 50
ohms. Software has also been modified to allow moving the BFO and IF
frequencies to match the filter's frequencies. Can't seem to get enough
signal through them to be useful though the software DOES put the IF
where it should be. I can hear a signal generator on whatever frequency
the uBITX is tuned to, but it takes better than 10 microvolts input at
the antenna for what I'd consider to be a usable CW signal in the
headphones.

I built the filter using a 4 crystal - 200 ohm 8 MHz circuit with
crystals and other components purchased from INRAD (Vibroplex now owns
them) but either I haven't figured out the proper 4:1 transformers or
there's just too much loss through the filter. I have tried it with 4
and 6 MHz filters I made from scratch with only slightly better results
(6 microvolts at the antenna for a usable signal)

Sure would be nice if someone was able to get something working to the
point it was usable and repeatable, even on perf-board.

The idea of switching the IF frequency to switch filters appears to be
sound - just need to find the magic crystal/capacitor/transformer
combination and for that I need someone with better knowledge of crystal
filters and better test equipment that I have.

Thanks

Jim Sheldon, W0EB
--
bark less - wag more

ajparent1/KB1GMX
 

The filter is not lossy its shunted by a lossy termination, the other filter.

The other filter is NOT an open circuit at some other frequency its a highly reactive load.
as such it may be possible to build two diplexers to route Rf through based on frequency but
at HF a suitable diplexer with at least 30db isolation is going to have a few DB loss and
you need two the losses will be high and you will need to tune them.

Simple solution, relays.  Sophisticated solution, diode switches.  You remove the
existing filter and build it next to a second more selective one with switching to
use A or B. Dead bugging them on a strip of copper clad for a good ground
path and putting a SPDT relay at each end (or diode equivilent) will work
very well.  You will need to coordinate the switching with software to assure
the right BFO/LO2 combo.

Allison

Jim Sheldon
 

Then why did Ashhar Farhan even bother to add those headers and say it would work?  How about letting him explain his reasoning.  I'd rather not chop up the board to add relays or diode switches.  

Jim

On Jul 6, 2018, at 9:55 PM, ajparent1/KB1GMX <kb1gmx@...> wrote:

The filter is not lossy its shunted by a lossy termination, the other filter.

The other filter is NOT an open circuit at some other frequency its a highly reactive load.
as such it may be possible to build two diplexers to route Rf through based on frequency but
at HF a suitable diplexer with at least 30db isolation is going to have a few DB loss and
you need two the losses will be high and you will need to tune them.

Simple solution, relays.  Sophisticated solution, diode switches.  You remove the
existing filter and build it next to a second more selective one with switching to
use A or B. Dead bugging them on a strip of copper clad for a good ground
path and putting a SPDT relay at each end (or diode equivilent) will work
very well.  You will need to coordinate the switching with software to assure
the right BFO/LO2 combo.

Allison

ajparent1/KB1GMX
 

The third way is to put a higher selectivity filter in series with the wide one.
The added filter must be inside the pass band of the first. and you need
switching to get back to wide (SSB).  The problem with this scheme is
you incur the loss of two filters cascaded plus interface loss between them.
Most folks that do that insert a gain stage to isolate the filters from each other
and offset the losses of two filters.

The fourth way is two identical wide filters and a pair of mixers and two LOs.
The two filters are then used as a variable bandpass (analog) filter lots of
hardware and gain matching needed due to the mixers.  You get wide filter
and the offset overlap is a narrow filter depending on offset.  This also allows 
varying the IF center (Passband tuning) frequency without changing the BFO.
The bad part is the hardware to do that is nearly as complex as the radio.

Allison

ajparent1/KB1GMX
 

So you take off the parts and use the header as the connection points.
What is the big deal?  

There is not enough room on the board for a second filters or relays.

I can't find his saying it would work anywhere groups.io.  

Allison

Jerry Gaffke
 


At one time, Farhan was thinking two filters in parallel, a CW filter at 1.8432mhz.
https://groups.io/g/BITX20/message/36041

And if that doesn't work, headers would allow us to remove the current filter,
use pluggable filters.


On Fri, Jul 6, 2018 at 08:03 pm, Jim Sheldon wrote:
Then why did Ashhar Farhan even bother to add those headers and say it would work?  How about letting him explain his reasoning.  I'd rather not chop up the board to add relays or diode switches.  
 

ajparent1/KB1GMX
 

>>The idea of switching the IF frequency to switch filters appears to be sound<< 

Yes, it can IF you have a way to isolate the other filter that will
look like a RF short or reactive load at the off frequency.  That device
is called a diplexer.  In short 4 filters to isolate two crystal filters
based on frequency.

Its a bunch or caps and coils one path is high pass and the other is low pass.
Minimum isolation needs to be greater than 20db and 30db might be enough.
The problem is the loss will be about 1 to 3 db each diplexer and you need
one at each end (matched pair).  Each diplexer will need to be tuned and
the diplexers reactance is not correctly tuned may upset the filters shape
and loss.

I've done this for splitting one coax path for monoband antennas antennas
and multiple radios using one multiband antenna.  I've done this for HF
through microwave.  For an example I  have a few 2m/438 ones I've made
up for satellite work.

Have you ever tested a filter at the off frequency to see what its presenting?
EMRFD does discuss this and its why most high performance designs put
some form of isolation amp between the mixer and filter.  Reason the mixer
wants a 50 ohm broad band load and most filters are anything but that.

The assumption that the filter if not passing the signal is a open circuit is false.
Filters work by reflecting the signal or absorbing it.  After all the off frequency
signals have to go somewhere.  IF its reflecting it then presents a composite
impedance that is unfriendly.  If its absorbing then you've lost that signal anyway.
This is why simply paralleling filters is a busted idea.

Google High pass, low pass diplexer. 

You asked for knowledge and experience. I have that, if you don't like
the answer I can't help you.  Likely no one has had a V4 board long
enough to try it and based on a few decades of experience you are
doing what fails and hopefully understand why.  You can save them
pain.

Allison

ajparent1/KB1GMX
 

Jerry, I finally found that.

Doesn't say its tried.  The headers can make hookup easier if it doesn't.

I've answered why his losses are so bad.  I'm surprised its not worse.
I'll bet he hasn't checked for additional birdies as the intact 12mhz
filter can still pass stuff to the product detector.

Adding diplexers is awkward as the board does not allow. and they can
be lossy.  I'd not be surprised if the blow by was also degrading the filter.

Me I'd go with striping that part of the board and reusing them on a mezzanine
board with the needed switching for two filters. 

I keep thinking is wishes were ponies...

Allison

Jerry Gaffke
 

I can believe there's all sorts of potential for trouble with running crystal filters in parallel.

Though here's a report that it works:
    https://groups.io/g/BITX20/message/48530

Would be interesting to do a full evaluation.
 
Jerry


On Fri, Jul 6, 2018 at 09:00 pm, ajparent1/KB1GMX wrote:
Jerry, I finally found that.

Doesn't say its tried.  The headers can make hookup easier if it doesn't.

I've answered why his losses are so bad.  I'm surprised its not worse.
I'll bet he hasn't checked for additional birdies as the intact 12mhz
filter can still pass stuff to the product detector.

Adding diplexers is awkward as the board does not allow. and they can
be lossy.  I'd not be surprised if the blow by was also degrading the filter.

Me I'd go with striping that part of the board and reusing them on a mezzanine
board with the needed switching for two filters. 

I keep thinking is wishes were ponies...

Allison

ajparent1/KB1GMX
 

Jerry that pointer is to bitx 40/20 tuning clicks.  The Sotabeams audio filter was
mentioned and I know that would work well.

Allison

Jim Sheldon
 

Guys,
I know I stirred up a can of worms with this topic, but I already have the NESCAF audio filters in all of the uBITX rigs (5) that I've built.  However, Ashhar DID say (I have an email from him telling me the I/O impedance for those header connections was 50 ohms and Alex, PA1FOX apparently has made it work like Ashhar intended.  All I was trying to do is get information on anyone having successfully accomplished this, not a bunch of engineering theory on why it won't work.  

If Allison had read Alex's (PA1FOX) a bit more closely she'd see his reference to switching IF frequencies to switch filters.  His click problem was created by the way he originally handled the firmware for shifting the IF to the new filter frequency - he was doing it far more often than necessary and that was creating the clicks.

That part of the problem has been solved.  

I'm not an engineer but I am a good technician and taught electronics for the US Army Security Agency at Fort Devens, MA for a number of years between 1968 and 1981 when I retired.

My major problem is I'm not afraid to test something that isn't supposed to work and see if there is a solution to why it won't work or prove the nay sayers wrong.  You guys are so busy picking apart the idea and using theory to "prove" it can't work.  I'm trying to prove you wrong - LOL.

Let's just agree to disagree on this particular subject and see what Alex, PA1FOX has to say if he answers my private email to him requesting further info.

5 MHz isn't a good selection though as the 9th harmonic of 5 falls in the 45 MHz first IF range and someone also posted on huge birdies caused by that selection (Ashhar had suggested 5 MHz as a possibility in one of his very early posts on the V4 boards).

Jim Sheldon

------ Original Message ------
From: "ajparent1/KB1GMX" <kb1gmx@...>
Sent: 7/7/2018 6:17:06 AM
Subject: Re: [BITX20] Crystal CW filter with Version 4 uBITX boards? #ubitx

Jerry that pointer is to bitx 40/20 tuning clicks.  The Sotabeams audio filter was
mentioned and I know that would work well.

Allison

Jerry Gaffke
 

Post 48530 is with regard to a uBitx.
No mention of sotabeams anywhere in the thread.
It includes the following statement
"During the tuning process, at every frequency change, the function was called
and set the correct  IF oscillator frequency to 4Mhz or 12Mhz (dependant on wide or narrow)."

That was the cause of his clicks.
And that sounds to me like two crystal filters at 4 and 12 mhz,
he would have to move the BFO when selecting the other filter (also move clk1).
I followed up with a post for clarification as it is not very clear,
but there was no further response.

I can believe it does not work terribly well.
But could be worth trying, or at least worth modeling the impedances of
the unused filter.
Farhan's original selection of 12.0 and 1.8432 mhz would place the filters
very far apart, that might help.

Jerry


On Sat, Jul 7, 2018 at 04:17 am, ajparent1/KB1GMX wrote:
Jerry that pointer is to bitx 40/20 tuning clicks.  The Sotabeams audio filter was
mentioned and I know that would work well.

Dave Dixon
 

I am using a sota beams audio filter on my version 3 u bitx and it works brilliantly. Dave G0AYD.

On 7 July 2018 at 12:17, ajparent1/KB1GMX <kb1gmx@...> wrote:
Jerry that pointer is to bitx 40/20 tuning clicks.  The Sotabeams audio filter was
mentioned and I know that would work well.

Allison


F1BFU - Fr - 79
 

Hi Allison

Use sota beams audio filter with an additionnal amplifier (LM386 for example° ?
I have a uBitx V3 and a Sota Beams audio filter on wait. 
How you have connect the audio filter ? On the HP of uBitx or on the headphones connector ?

73 Gilles F1BFU / FR

 

2018-07-07 18:23 GMT+02:00 Dave Dixon <wylyeguy2@...>:

I am using a sota beams audio filter on my version 3 u bitx and it works brilliantly. Dave G0AYD.

On 7 July 2018 at 12:17, ajparent1/KB1GMX <kb1gmx@...> wrote:
Jerry that pointer is to bitx 40/20 tuning clicks.  The Sotabeams audio filter was
mentioned and I know that would work well.

Allison



ajparent1/KB1GMX
 

When and if you get it to work please post your results and what it took to make it work.

Yes the ports are 50 ohms, that's never been secret but if you have two 50 ohm devices
(source and load) and then parallel a third 50 ohms device what is that, not 50 ohms.
Only the IF and the Balanced mod are 50 ohms, the filter is 50 ohms in the band pass and
whatever outside it.  

I never said the idea to switching IF frequency will not work. ONly paralleling filter will not work well
due to one more or less shorting out the other.  You have already written that part up.  I did expllain
those of us that can design would do it and it would work well.  Also every stinking cheap am/fm
radio has a dual frequency if both 455 and 10.7 and thats been done for over 50 years.
Been in engineering or over 50 years, my specialty is making the won't work things,  work.
But that unusually means some fancy design or understand how it works to start with.

Two things.   I told you how to do it and it will work if you do that.   I said your hack
of paralleling them relies on things that will be guaranteed to work poorly unless lucky.
Also Farhan said 1.842mhz filter and you didn't try that.  

Now your own words:
>>>The idea of switching the IF frequency to switch filters appears to be sound - just need to find the magic crystal/capacitor/transformer combination and for that I need someone with better knowledge of crystal filters and better test equipment that I have.<<<
I have both and your proceed to disagree.  
Until other post their work it is vaporware.  I have your posting it didn't work.

>>>>So far I've not been able to get a signal through filters at 4.6,  6.2 or 8 MHz.<<<<

You didn't try 1.8432, and you did get some signal but the attenuation is high.  
Silly question did you try the filter without the 12mhz filter shorting it?  That
would prove your filter works.  You could also do that using a bench jig.
You only need signal generator and a scope maybe a few resistors to simulate
a 50 ohm load.

Me I'd mod the 12mz and make it a jones filter (varicap tuned variable bandwidth)
as used in some older Tentec and the Elecraft K2.   I like that as it means the board
only needs parts added (varicap diodes and resistors to a control point) and no
hacking it up.  It is an old idea and discussed before.


Allison

iz oos
 

Adding another IF narrower filter might be the best solution. However some may be almost equally satisfied adding a good audio filter. Good receivers by Lowe had 200hz wide audio filters for CW. I have tried the hi-per-mite filter by SQRP and is a great filter one could connect at the speaker output or I believe before the audio final as the filter can provide audio gain depending on the settings.


Il 07/lug/2018 18:45, "Gilles Delpech" <gilles.f1bfu@...> ha scritto:
Hi Allison

Use sota beams audio filter with an additionnal amplifier (LM386 for example° ?
I have a uBitx V3 and a Sota Beams audio filter on wait. 
How you have connect the audio filter ? On the HP of uBitx or on the headphones connector ?

73 Gilles F1BFU / FR

 

2018-07-07 18:23 GMT+02:00 Dave Dixon <wylyeguy2@...>:
I am using a sota beams audio filter on my version 3 u bitx and it works brilliantly. Dave G0AYD.

On 7 July 2018 at 12:17, ajparent1/KB1GMX <kb1gmx@...> wrote:
Jerry that pointer is to bitx 40/20 tuning clicks.  The Sotabeams audio filter was
mentioned and I know that would work well.

Allison



ajparent1/KB1GMX
 

Giles,

Before every hijacks the topic which is focused on using crystal filters
in a dual frequency IF.

I don't have that SOTA filter.  I build CW radios with narrow crystal filters or in the case of
analog I/Q phasing use analog filters of the desired bandwidth.   Though I usually
add a stage of active filter in the design to reduce high frequency hiss even to
SSB radios.

Any of the switched capacitor filters are good and there is software suitable for a faster micro
like Rpi or the STFx series to do it digitally.  Just create another topic for that and not disrupt
this one.

Allison

Jerry Gaffke
 

Allison,

You mentioned a number of ways to have switchable/variable crystal filters
that have been proven to work in other rigs.
I'm not aware of any commercial rigs just slapping two crystal filters in parallel.
There's likely a reason.

Though perhaps the reason is that they didn't have
such an easily programmable oscillator at the time.

If I ever get time, and get everything else done I'm expected to do,
I'd like to build two pluggable filters for a uBitx, say 12 and 1.8432 mhz.
Simple, and should just work.
Scan each with some sweeperino like thing, verify they are working properly.
I'm fine with pluggable filters on a $120 rig, I don't often have to switch between
CW and SSB mode, and generally have the cover off anyway.
Would rather have pluggable than more of those d**n relays.

Could be worth trying just running them both in parallel, sweep the combination
and see how much worse it is.
Images should not be a problem, since we also have the 45mhz filter in there.
Since everything is unshielded, will likely be more birdies.
There may well be weird reactances present off frequency, though I'd expect the filter
that is on frequency with it's 200 ohm load to be of much lower impedance.
Having never swept a crystal filter, I could easily be wrong.
But could work well enough to be a keeper.

There are lots of corners cut on the uBitx that you never see in commercial gear.
That's why it's $120, not $600. 

Jerry


On Sat, Jul 7, 2018 at 10:23 am, ajparent1/KB1GMX wrote:
When and if you get it to work please post your results and what it took to make it work.

Yes the ports are 50 ohms, that's never been secret but if you have two 50 ohm devices
(source and load) and then parallel a third 50 ohms device what is that, not 50 ohms.
Only the IF and the Balanced mod are 50 ohms, the filter is 50 ohms in the band pass and
whatever outside it.  

I never said the idea to switching IF frequency will not work. ONly paralleling filter will not work well
due to one more or less shorting out the other.  You have already written that part up.  I did expllain
those of us that can design would do it and it would work well.  Also every stinking cheap am/fm
radio has a dual frequency if both 455 and 10.7 and thats been done for over 50 years.
Been in engineering or over 50 years, my specialty is making the won't work things,  work.
But that unusually means some fancy design or understand how it works to start with.

Two things.   I told you how to do it and it will work if you do that.   I said your hack
of paralleling them relies on things that will be guaranteed to work poorly unless lucky.
Also Farhan said 1.842mhz filter and you didn't try that.  

Now your own words:
>>>The idea of switching the IF frequency to switch filters appears to be sound - just need to find the magic crystal/capacitor/transformer combination and for that I need someone with better knowledge of crystal filters and better test equipment that I have.<<<
I have both and your proceed to disagree.  
Until other post their work it is vaporware.  I have your posting it didn't work.

>>>>So far I've not been able to get a signal through filters at 4.6,  6.2 or 8 MHz.<<<<

You didn't try 1.8432, and you did get some signal but the attenuation is high.  
Silly question did you try the filter without the 12mhz filter shorting it?  That
would prove your filter works.  You could also do that using a bench jig.
You only need signal generator and a scope maybe a few resistors to simulate
a 50 ohm load.

Me I'd mod the 12mz and make it a jones filter (varicap tuned variable bandwidth)
as used in some older Tentec and the Elecraft K2.   I like that as it means the board
only needs parts added (varicap diodes and resistors to a control point) and no
hacking it up.  It is an old idea and discussed before.


Allison