Tuesday, July 25, 2017

A New Line of Transceivers ~ DifX

Dissing the Dishal Filter

(AKA Ditching the Dishal Filter)

 
 
07/27/2017 ~ Updated with some additional data.
 
Several months ago I started a project that would use a filter designed using the Dishal method. I have finally concluded that it is time to say Basta. For those who claim they have a working filter I suggest you go back a fully evaluate what you really have.
 
A short review of my reasons for Dissing (Ditching) the Dishal. After resolving the issue that every time I downloaded the software my virus checker put the Dishal.exe file in quarantine (not a good sign) I started off with a 6 crystal filter. That did not end well! Then I shifted to a 4 crystal filter and the curve below is the result of the second effort. This is a real plot of my data and from what I see, it has the classic Dishal shape and far different than my plot of the failed six crystal filter. For reference purpose I have now added what the theoretical 4 Pole. Dishal Plot should result based on my data

Mind you I did all of the steps relative to measurements using the G3UUR oscillator and employed my SDR transceiver to measure the loaded frequency shift to 1 Hz. The crystals that were selected out of the large batch I purchased all were identical as to normal and loaded frequencies. I purchased precision close tolerance MLCC COG (NPO) caps so we had many factors in place when the filter was measured and built. Essentially I used selected 120 PF and 150 PF caps that read higher but where each value was identical --ie two higher reading 120 PF caps to get you close to the nominal 121.6 PF computed value. The same applied with my 150 PF caps.
 
 
N6QW's Plot of the 4 Pole Dishal Filter
 

4 Pole Filter Plot from Dishal
 
 

Based on this plot I installed the filter (11.5 MHz Cf [nominal]) in a working 40 Meter transceiver and should mention that I used a second external oscillator to accurately develop the BFO (USB/LSB) frequencies. Once these were determined then that data was used in the sketch. It was important that my four pole filter was installed in a known working transceiver so that the only variable was the Dishal Filter. That installation is shown in the photo below.
 
 
 
The next phase of testing was to tune across the band and just listen to what I heard. The first notable item was that received signals (LSB) seemed to have an unusual fidelity like in what the "Enhanced SSB Crowd" likes. There was no evidence of a pinched audio or that there was sideband cutting. It was like hi-fi. Next I shifted to USB and tuned away 3 kHz and boom I could copy the signal in USB. So that was most telling in that two things were happening: 1) The opposite sideband rejection was terrible and 2) that the filter was too wide. Now the hi-fi sound made sense.
 
Next I tried the rig on transmit (into a dummy load) and while in LSB tuned an outboard receiver to USB and there I was on USB (not quite as strong but definitely not 50 dB down). So throwing caution to the wind I connected my rig to an antenna (the worst I could be was to have a DSB signal).

In short order I found a very loud station in San Diego (200 miles away) and gave him a call. Boom he came back to me and reported "you sound absolutely terrific". I then asked if he could check my other sideband --and got a shocking answer (he was an extra class ham, newly minted) and he said I don't know how to do that. I thanked him and quickly said 73's. I already knew I had a DSB signal.
 
About 15 minutes worth of work I installed a filter from a Yaesu FT-101 (3.180 MHz Cf) and after repeating listening to my signal into a dummy load --there was no USB component. True it did not have the fidelity of the Dishal; but it was a real sideband filter.
 
Oh the 15 minutes -- 5 of it was to install the filter and load the code in the Pro-Mini --the other 10 minutes was to install the new paint job -- Juliano Blue V.2. You can just see the 3.180 MHz filter where the Dishal was installed.
 
The New Juliano Blue Rig!
 
I must challenge those who built a Dishal Filter -- have you looked at the Opposite Sideband Rejection both on receive and transmit? You might be surprised. Even my good friend Bill, N2CQR commented that he had less than sterling results using Dishal and he uses the AADE software or the Program in the EMRFD DVD.  While the purists want to say I built all of this rig, even the crystal filter, you may actually have a marginal filter in the rig if you use the Dishal method. If you are getting a lot of negative 40M SDR Police reports --you surely are getting a message about your filter.
 
By the way my local Home Depot was having a sale on spray paint and so look for some future rigs in  "Hot Pink" and "Cranberry" colors. I am reminded of Henry Ford who initially produced only black colored cars and when asked about other colors smartly said -- "you can have any color you want so long as it is black". Why do so many of the commercial rigs come only in black? Now I do know that the Hiberling PT-800A, (a $17K rig) did come in a sorta Juliano Blue --but not sure they are still being produced.
 
Final word -- the Dishal software has been exorcized from my computer. Based on my experience I would not recommend the use of the Dishal method. The results simply don't match what you hear once installed in a rig. I truly believe those who say they have built a successful Dishal Filter may need to revisit their data and re-evaluate the filter performance in a real rig.
 
73's
Pete N6QW
 


Monday, July 10, 2017

A New Line of Transceivers ~ DifX

A New Transceiver on the Air!

* Some updates added on 7/11, 7/12, 7/13, 7/14, 7/15, 7/16, 7/22


A New "Juliano Blue" in the Wings?


I frequently have several transceivers in work at any one time and eventually some event occurs that causes me to actually finish off one of the rigs. It was really hot this past weekend in SoCal and so that was the triggering event -- there is just something cool about working in a garage that is running a hot 90 degrees F or for those who use the " other standard " we're talking 32.2 Degrees C.



Here are some highlights of this DifX [Different from a Bitx]:

  1. The size is 4X6 by 2 inches high. [For reference purposes the Bitx40 board is 25 square inches not including the display and controls.] I wouldn't call this micro-sized but Mini-Sized is probably OK. A bit of fudging here as I happened to have a 4X6X2 inch aluminum chassis that became the case enclosure. I bought this chassis years ago with the plan for a one "toob" CW transmitter --but who does CW anymore? To get more of a feel of the size of this power house look at the photo below and the relative size as compared to my venerable D-104 Microphone. This says a lot when your rig IS smaller than the Microphone.
  2. For the most part the circuits used are ones applied by me in other projects although this is only the third time I have used MMIC amplifiers and seeing how well they work will now become the standard for me.
  3. The rig operates on 20 Meters as that seems to offer plenty of activity (even DX) and there are less SDR police and old buzzard nets like on 40 Meters. Yes these are my contemporaries.
  4. The main IF uses  a 9.0 MHz Crystal Filter from INRAD [Model 351] and the Bi-directional amplifiers used for the Rx RF Amp and Tx Pre-driver are two, diode steered and relay switched, AG-303-86G MMIC amplifiers. Going into this stage I used a 2 dB pad and this has really added to the overall stability. Lets face it --these are cramped quarters and lots of potential for unwanted feedback.
  5. The Si5351 supplies the LO and BFO and the display is a 1 Inch Square Black and White OLED. Should mention OLED noise has not been a problem. The main Microcontroller is the Arduino Nano. A little detail here is that I provided clearance in the build so that a USB connector can be plugged into this board for any program changes without removing the Nano. A small but very important detail. The default step tuning rate is 100 Hz but additional selections via the encoder push button are 1 kHz, 10 kHz and 100 kHz. This approach gives plenty of flexibility to tune the entire 20 Meter Band,
  6. The Bilateral IF Amplifiers are the 2N3904/2N3906 Plessey circuit and the RxTx Mixer is an ADE-1L as is the Product Detector / Balanced Modulator. ADE1-L = 4 dBm devices so set the output on the Si5351 to 2 ma. Note the 2N3904/2N3906 are SMD. The immediate photo below shows the Model #351 4 Pole Crystal Filter which had been in the rig for about a week. On 7/16 I replaced that with the GQRP 6 pole filter. This involved a bit of noodling and modification to the lower board and to the very top board of the three board sandwich. The GQRP Filter being taller than the Model #351, meant I had to add more spacing between the first and second boards. The top board which is the audio and microphone amplifiers had to have two electrolytic capacitor swapped out for one shorter ones. In one case a 330 Ufd was replaced with a shorter 220 in parallel with a 100 Ufd. It was a small nightmare but it is all back together and is a testament to a well stocked junk box, fifty years of building experience and throwing caution to the wind. BTW can tell the difference with the GQRP Filter. Much tighter skirts and actually more output!
  7. The Audio amp is the LM386 driven by a 2N3904 and the Microphone Amp is a single 2N3904. The RC Shaping filter for the 988 Hz Square Wave Tone coming from the Arduino is also located on this board.
  8. The next photo shows how the board was reworked and move slightly so all would fit back in the case. Note the use of the solder lugs to "collect" the board so it doesn't move and the add of the two smaller electrolytic capacitors so as the assemblies moved upward all of the boards would fit in the case.
  9. The Power Output is 5 watts via a IRF510 and the driver is the 2N3904/2N3866 standard circuit from EMRFD.  This linear amp board is about 3.5 inches long and about 1.9 inches high and fits in the rear of the chassis box. The chassis is the heat sink. Note the shield between the driver and final stage and even the DC bias stage is behind the shield. Some additional bypass caps on the IRF510 have proven to add much to the stability.
  10. Other features include USB/LSB, MOX and Tune (via a 988 Hz tone generated by the Arduino). I wonder why the Bitx40 aficionados have not added the "Tone Tune Up" to their radios? In time grasshopper, in time. Note on the display the words SSB XCVR --when you place the rig in the Tune mode the wording changes to TUNE and when the timing cycle is over it reverts back to SSB XCVR. There was a bit of trickery to make that happen and not something found on other currently popular radios. Note the size in comparison to the 1/2 used pencil.
  11. The paint is standard Juliano Blue.
Some construction notes for this project. This is a really compact rig and even the install had to be sequenced so all circuits would fit in the enclosure and could be hooked up. Unless you have experience in building circuits in tight places, it is not recommend you replicate my approach. While the circuit pieces are modular, making them fit was a real challenge. One assembly consists of three stacked circuit boards with the bottom board being the IF, front end Mixer, PD/BM stages and the middle board is the MMIC bidirectional amplifiers and the top board is the Microphone and Audio Amplifier stages. The Driver and Final are on a separate board as is the Arduino.  From the photo below you can see it is pretty jam packed into the enclosure.

The OLED noise has not been an issue so check that box. BUT what has been a constant problem is the LEAD FREE SOLDER CRAP. That solder just does not stick and is a particular problem if you are using Manhattan construction. If you are doing ultrasonic wave soldering on 408 sized SMD parts --probably OK but not OK on an IRF510. Initially there was no output from the transmitter --yep a solder connection on the base of the 2N3866 just did not stick. You end up welding the parts and not flowing the solder. This new solder is real crap!!!!!!

The proof is in the pudding and my new rig just completed its first contact with WA7ND, Rick up in Bend, OR. The report was excellent and I was running the small outboard amp so the Pout was >100 watts. Forget that QRP crap and just move on. I have had a second contact on the rig with KD5CO, Chuck in downtown Texas. I couldn't resist an additional test with the second contact --yep two after burners and the Pout > 600 watts into my two element beam. Now this can be a DX machine. So this is proof that the rig is more than a one hit wonder. Today (7/11) I had a really great QSO with Jim, KA5BDO in New Mexico and he was very complimentary of how the rig sounded. He also suggested I use the Turner Dynamic Microphone for rag chews and the D-104 for DX chasing. You can see the new rig with the Turner Dynamic Microphone in the photo below. [That new rig IS small!] Just had a QSO with KH7XS, Bill out in Hawaii running 600 Watts -- another great report. This is a DX Machine in a mere 48 Cubic Inches.

This is one superb rig and lest you forget --it is a DifX! I can't fully express the joy in putting this rig on the air and to have real QSO's with this diminutive sized box. One asks could there be more features? With a bit of more effort on packaging and the add of one DPDT toggle switch, two band operation would be possible. The only adds would be a second Band Pass Filter and a second Low Pass Filter and of course the toggle switch. If I replaced the MOX switch with a band switch then no more panel space would be required. A 40/20 Meter rig would cover the bases or maybe the 20/15 Meter bands or even a 20/10 Meter rig when the sunspots kick back up. Maybe even a 40/17 Meter rig would be your cup of tea. It is merely some code changes to the sketch.

While it may not be possible for readers of this blog to build this project in a very tight space like this 48 cubic inch chassis box -- this is a worthy transceiver project that could be successfully built on an albeit larger scale (or box). That said the smallest SSB transceiver I have ever built was 33.3% the size of this rig --16 cubic inches. Virtually all of the circuits used in the rig have been previously published on this very blog (or articles in QRP Quarterly) so the module schematics are no mystery. If some one would like the Arduino sketch send me an email to the address on my QRZ.com page.
 
Block Diagram and Some Schematics


 
 







The Finished Rig~ Another Successful DifX!
 


As of 7/15 I have made over a dozen contacts with this DifX rig and am very pleased with the signal reports I have received. Only one negative SDR police report regarding opposite sideband suppression being only 40 dB down. Opposite sideband suppression could be improved with a different filter but  a commercial 9.0 MHz Filter I have might not fit in this build. On 7/16, I did replace the 4 pole filter with the 6 pole GQRP filter so we now have a better filter in the rig.  The signal quality is much better on receive and I am sure likewise on transmit.

73's
Pete N6QW

Wednesday, July 5, 2017

A New Line of Transceivers ~ DifX

Homebrewing a Chassis, Cases and Enclosures

Make your Rig Look as good as it Sounds!

 
There is just something about a rig that has that finished look when it has a painted case. Yes, it is something that came out of your garage or workshop but there is no reason why it can't look "finished". Long ago I adopted "Juliano Blue" and perhaps there was an unconscious bent to that decision based on my very early solid state experiments with the distinctively Blue CK722 transistor. Go Blue!
 
Of late I am not satisfied with a project unless it resides in a Blue Case --a Juliano Blue Case. So how do you get to the point of homebrewing a chassis, cases or enclosures. The first part is the right tools --no you don't need a $250 K CNC mill but for about $50 you can purchase a bench type metal break from Harbor Freight.

When I bought mine about 10 years ag they could be had for about $25. The type I have is 18 Inches long. The next thing you need is some stout metal bars in various lengths. The tool from Harbor Freight comes with an 18 inch bending bar. By using smaller bending bars in the 4, 6 and 8 inch range you now have a box break where you can easily make 6 X 8 inch chassis boxes -- thus the smaller bending bars.
 
For a chassis I make it in three pieces. Two pieces are "L" shaped and the final piece is the to cover. The two "L" shaped pieces are formed with ears so you can bolt/fasten to the pieces. Fastening also include "pop riveting". Two Chassis' are shown below one is a general purpose chassis and the other was one made for a transmitter power supply. Believe it or not most of the materials are galvanized metal available from Home Depot. The power supply chassis shows the "L" structure of the pieces and make for a very rigid chassis. The Sheet metal is very inexpensive and with a couple of homebrew chassis bulds you will recover the cost of the metal break.

 







 
 
 
In the photo above there is a trick to making these "L" section and it all has to do with the sequence of the bends. Excellent measuring skills are needed to pull this off and laying out the bends and cuts has to be done carefully. this is even extended to when the metal is in a flat state and that is when I actually drilled the holes. I have a stock of manila folders and so my first step is to make the part out of the manila folder stock. This in essence is a template. The first pieces bent are the top and bottom lips which are about 1/2 inch wide. At the location of the bend line I cut a 45 degree angle of material to the bend line on both the top and bottom sections. This when this is bent at a 90 degree ange the pieces meet up perfectly. Both sections are identical only in one of the pieces the end section is bent up at 90 degrees at each end. This is where the smaller bending bars come into play. Essentially an end of the bending bar is placed perpendicular to the bend line and the larger piece is bent upward. For normal operation the bending bar is "C Clamped" to the bed of the metal break and so with the smaller bars you can make boxes. the following two photos give a bit more detail of what is described here.
 








 
 
Now I would like to spend some time on how I build cases using angle stock readily available from Home Depot. The metal plate can be ordered from a company called "On Line Metals"
 
 

 
 
 
Yes this is an old project when I actually built analog VFO's as evidenced by the National Velvet Vernier. As soon as I started using the Si5351 I gave this rig away since I didn't want any evidence in my shack of an analog VFO. The aluminum angle stock makes for a very rigid structure and you can easily screw sheet metal to the sides and top.
 
 
 

 

 
 
The next critical issue is one of materials. Use what you have or think out of the box (enclosure). Sheet metal (versus aluminum) is readily available and inexpensive. Home Depot is a good source. I also have used single sided copper PC Board which often can be had a really good price. ABC Fab up in New Hampshire (eBay) sells various packages of boards. Recently I bought 18 pieces of 4X6 inch heavy duty board for about $1 a board --delivered. I also bought some 8.5 Inch by 12 inch board from the same source and two of my rigs have that for a base plate. One ham shared he found a source for large square tin cans which he cut open and flattened. Also check out computer shops as you can often find CD/DVD units which are defunct but its the case you want. Cookie tins are another great case source. So open your mind to the possibilities.
 
Oh --it is not a real case unless it is painted Juliano Blue!!!!!!
 
73's
Pete N6QW