Thursday, June 22, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.24

Back to the Dual Conversion DifX Build

6/23/2017 ~ See video at the bottom for more MMIC Applications 

Now that I was able to resolve the Crystal Filter that will be used in the second IF (it is not a Dishal nor any other form of homebrew filter) I have turned my attention to the issue of various amplifier stages that are used elsewhere in the rig. One such circuit is the bidirectional amplifier board I developed where there are two amplifier circuits comprised of a single 2N3904 in each leg. One leg is the receiver RF amplifier stage and the second leg is the Transmit Pre-driver stage. Relay switching is used to direct the signals in/out of the board -- a total of three relays.
While the 2N3904's are cheap, the relays are not! While the stage was broad band there were issues of making the gain constant over a range of 30 MHz (160 to 10 Meters). Since most of my rigs are single band units or perhaps two bands like 20/40 Meters this was not so much of a problem.
So in noodling the problem to have broad band amplifiers stages that are constant gain, are 50 Ohms, involve minimum switching (no more than one relay) and are termination insensitive. While you could use the Hayward/Kopski Termination Insensitive Amplifiers -- I wanted this to be a Pete design and not use the work of others.
My noodling took me back to 2010 and my very first article in QRP Quarterly where I used MMIC (Microwave Monolithic Integrated Circuit) amplifier blocks. The project was a 20 Meter MMIC based QRP SSB transceiver. A pair of MMIC amplifiers are diode steered so that they are bilateral (operating in two directions depending on which amp is powered on and steered for the proper mode). Yes Alice (or Virginia) it has a Digital VFO,

(Interesting note in that same issue we have G3UUR Analyzing Crystal Filters)

The MMIC device selected was the TriQuint (Watkins Johnson) AG-303-86G, This device is good to 6 GHz and has a fixed gain of 20 dB with a Z in/out of 50 Ohms and operates from 5 VDC. It also is termination insensitive, Did I also mention they work quite well.
Thus it was an easy decision to once again employ the MMIC amp and we have a board operating with a test transceiver. The alternative for those who don't want to enjoy using MMIC's in your next rig then you can always use the Hayward/Kopski TIA amps. Shown below is a board that was just installed in the transceiver that was recently used to evaluate the 128X32 OLED noise issues.

BTW I did have several conversations with TriQuint and they provided information about the use of the 1N3070 diodes as the "best choice" over the 1N914 or 1N4148. The actual building of this bilateral stage will present construction problems in that the MMIC's are surface mount and really small. It will be difficult to build this amp using Manhattan techniques. So while the benefits are superb, the construction may well beyond the skill/capability of those who have never done this type of homebrew construction. In the last 7 years the price of the MMIC's used has increased significantly so that may be a factor in your decision but my having a stock of the devices makes it a non-issue for me.
As you are perhaps gathering,  the Dual Conversion DifX will have features and functionality not found in other currently popular designs.
For those sitting on the edge of your chairs regarding the detail of second IF filter --patience grasshopper (for those who remember the Kung FU TV series). Oh I will tell you it is not 9.0 MHz.

Below is a video of MMIC's (four of them) used in a 40M CW transceiver built about the same time as the 20M MMIC SSB Transceiver. I do not know of many other transceivers that employed MMIC's in various transceiver stages. Perhaps another tip of the spear from N6QW.

A video from 2017
A video from 2010
Pete N6QW

Friday, June 16, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.23

Commercial and Surplus Crystal Filters

6/20/2017 ~ Short diversion -- OLED noise with a 128X32 OLED Blue Display
Believe it or not this test is important to an upcoming project that will soon be revealed.
Pete N6QW

6/18/2017 ~ In the original post I mentioned that I saw three German Filters on a board for sale from Israel but did not know much about them. Thanks to one of the regular blog readers (Thanks Jim), we now have info on these filters and from what I can see they appear to be symmetrical which is important if they are installed in a bilateral circuit such as the Termination Insensitive Amps and the Z in/out is 1 K which is a 20:1 match to 50 Ohms. A 9 turn to 2 turn matching transformer gets you close 9^2 = 81 and 2^2 = 4. 81/4 = 20.25.
The good news is that they are 8 pole filters which will really tighten up the skirts and resolve the "feathering" issue complained about by one of the 40 Meter SDR police and cited in the post. Thank again Jim for the info and the plot of one of the filters taken "bare" without any matching which  would affect the ripple content in the pass band. If that ripple is smoothed your are looking at 60 dB
rejection and pretty narrow skirts.
Don't rely on my analysis but  for some this might be a good buy and you get an AM filter too! This is definitely not a Dishal homebrew filter and would be hard to replicate in a cold, dimly lit garage with the only tools being a cheap VOM and an 80 Watt RS soldering iron.

I just bought a board with the three filters -- $34 shipped and that gives me three filters.

Pete N6QW



Having sworn off Dishal and other homebrew filters there has to be some alternatives for those who would like to homebrew a rig. The reasons are many  for not homebrewing a filter and chief among these is that it is difficult to achieve consistent results from these somewhat arcane processes without the benefit of an extensive amount of test equipment. Anyone who tells you they did it with just a VOM is Fake News. Now if you just want to try your hand at it --by all means build a filter.
But with the advent of the SDR radios and those who lurk around 40 Meters looking for aberrant signals, there is a strong possibility you will get a report that some one has spotted energy above 2800 Hz (especially on 60M) or that your filter is feathering. (OK try to figure that one out.) Yes a homebrew filter will let you enjoy the experience of making a rig where you literally built everything by hand. But I would bet that in most cases the homebrew filter that results unless you are extremely lucky will not match a commercial filter.
So OK where can one buy a reasonably priced SSB filter? One place is the GQRP Club. Join GQRP, become a SPRAT subscriber and buy a high quality filter for about 12 Pound plus shipping. At a $2  to 1 Pound exchange rate --about $30 with shipping puts one of those jewels in your hands. Many of my rigs have this filter. In fact two W7ZOI designed rigs have two such filters in each rig. The IF is 9.0 MHz and works very well for most applications save 17 Meters. Below is that filter in the Zia Transceiver built in 2014 which uses the Hayward/Kopski Termination Insensitive Amplifiers. The photo right below shows another GQRP filter installed in the LM373 Rig.



For the same amount of money INRAD sells a 4 pole Filter Kit --also on 9.0 MHz. I have two of those filters and they work very well. It is their model # 351. The Crystals are color coded and come with both SMD and Leaded Caps so you can pick your method of building the kit. Below is the #351 in a 2017 project implemented with the SMD caps.
Just yesterday I toured eBay looking at commercial/surplus filters. Wow some one from downtown Serbia is selling Elecraft SSB filters for an amazing price. There are also filters from Kenwood, Yaesu and Icom radios. Some are pricey but some are really at a very good price.
There were several Heathkit filters and some listed starting at $10. I have used a Heathkit filter in a transceiver and it works very well. The only problem --it is rather large. The filter IF is a 3.395 MHz and so this lends itself to a dual conversion approach. In the photo below the signals ahead of this board were converted to frequencies in the 8.8 MHz range. The PTO (lifted from a Ten Tec Triton IV) operating a 5 MHz converted the signals to the IF at 3.395 MHz using the TUF-1 on that board. The relay on the board enabled adding AGC to the IF or ALC. The device is a DGM 3N209 in a circuit developed by G4GXO in Sprat 128 --it is bilateral! In the upper right hand corner is a diode ring modulator/demodulator that had both resistive and capacitive balance. That is not something you find in many homebrew rigs. I built this in 2009.
But just as I saw some really good filters there were some questionable ones. There were several boards from Israel showing three German filters on one board. The price was OK --it is just that I didn't immediately recognize the filters. More research needed here.
The 8.8 MHz (or so) Kenwood Filters look like a good fit with a 45 MHz 1st IF. Many of the commercial filters are 2.6 KHz wide and so would give better results on received audio as well has having a Hi Fi sounding signal on transmit with all of the benefits or presence, brightness, color, with great lows and highs.
Frequently with a homebrew filter you will get tons of complaints about having pinched or restricted audio --typically form those operators using SDR radios with 72 Inch LCD screens. So they can really see your signal!!!!
So expand your horizon's and think beyond homebrew filters -- it avoids a lot of work that frequently results in a marginal filter that unless you are extremely lucky will have a difficult time competing with a $30 or less commercial filter. Oh -- the Heathkit Z in/out = 2K Ohms, the GQRP is 500 Ohms and the INRAD #351 is 200 Ohms. Pretty easy to match and a known quantity!

BTW I bought this filter after finding out more of the specifications and  this now lends more weight to my argument -- I just bought three filters for $30 and $4 shipping. These filters are better (being 8 pole) than any 4 or 6 pole Dishal filters. After purchasing the board I got a prod from eBay essentially saying buyers who purchased this board also bought Si5351's --so there are individuals out there who are already on this path. (6/18/2017 N6QW)

That is the other bit of good news. In the past having the filter without the matching BFO crystals was a huge problem. No more fellow homebrewer's. With the Si5351 --you simply change the frequency in the sketch and you are off and running with a new filter. Hooray --that is a problem no longer.
Pete N6QW

Tuesday, June 13, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.22

Have You Ever Wondered?

Arduinos and Si5351 PLL Clock Generators are inexpensive (well depends where you live); but for around $20 (US) you can purchase a ProMini and a si5351 Clock Generator. I have not seen many designs where two Arduino's and two PLL clock generators are employed. So why not? Yes you will hear from the lurking illuminati that you don't need to use two but has any one really explored the possibilities opened up by such an approach.
Let us think "Out of the Box" for a minute. Suppose we take a device like the Pro-Mini and loaded that with code so that CLK2 would provide the injection frequency for a second mixer in a dual conversion transceiver (the 1st conversion would be an up conversion to say 45 MHz).Thus the second conversion would be to the IF frequency and CLK0 would provide the BFO injection frequencies. You got that OK?
Now think for a minute of a two position switch mounted on the front panel so that one position would be to select USB, the second position would be for LSB. You might even have a second encoder with a PB so that by depressing the encoder push button you could have pass band tuning limited to the range of separation between the USB/LSB frequencies. Pretty cool.
Now about the second Arduino and PLL Clock Generator. Suppose that this time for the up-conversion we use the Si570 controlled by an Arduino Mega. The use of the Si570 would cure the gaff from the EMRFD Illuminati about phase noise and would use a true encoder and not a pot like in the Minima. I am thinking with the Mega you would have more digital and analog IO so that controlling band pass and low pass filters would be unencumbered by pin limitations. A bigger display (and more programming space in the Mega) would let you display to your hearts content. The two PLL Clock generators will also address the naysayers who complain about the spill over onto the third clock. The upper frequency limit of the Si570 is about 5X the Si5351 so other possibilities here.
So maybe it is time to add a few more Arduino's and PLL Clock generators in our rigs.
Pete N6QW

Monday, June 12, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.21

We must look to the past often times to see the clarity of the way forward. I am reminded of something that Thomas Edison experienced. (Don't know if this is factual but may be more of an urban legend thrown in for good measure.)
It seems there was a major fire at his Menlo Park Laboratories and with that fire many of Edison's projects were burnt to a crisp. One of his assistants was lamenting to Edison about how terrible the situation was in that many of his projects were no more. Edison, ever the visionary, said this is actually good! The assistant could not believe what he heard --Edison chimed in -- "most of what burnt up didn't work and I just didn't have the heart to take it to the dump". The fire actually solved a problem for Edison --with the clutter removed he could focus on  more productive things.
So it is with all things Dishal! I had a ceremonial fire today and gone is all that wretched crap that didn't work and probably never would work. But with that fire, I have a new plan and I think a much better solution for the second filter. I am somewhat mad that I didn't noodle this before.
While I work the new approach the blog will be silent. But I am really jazzed that where I am headed will in effect be a super DifX.
June 12, 2017
Pete, N6QW

Friday, June 9, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.20

6/10/2017~ More on Dishal Dyslexia Part 2

There always has to be a reason why something as fundamental as 1 + 1 does not equal 2 and so it is with the results arising from building a Dishal Crystal Filter. So lets us take a few steps back and examine why 1 + 1 may not equal 2.
The Dishal software in large part requires the user to make some initial measurements outside of the software itself. Starting with the G3URR oscillator and my SDR receiver (capable of measuring to 1 Hz and verified by first locking on to WWV) I measured the crystals in the loaded and unloaded condition. Again the receiver is calibrated to be accurate to 1 Hz and the G3URR component values are 470 PF (NPO COG) in the Colpitts circuit and the load cap was 30 PF NPO COG. So the data was taken and crystals were found that were very close in frequency and thus were the basis for entering the data into the Dishal software. It is pretty automatic once you enter the data.
The Dishal software can be tweaked by factors such as the number of crystals and the ripple factor. Once these are selected then you get a plot of the filter. But the software also spits out the Filter Center Frequency, the values of the coupling capacitors and the impedance that must be matched to 50 Ohms. So far so good.
Now the problem is one of what BFO frequencies do you select. Starting with the predicted Center Frequency I chose BFO frequencies about 1.5 KHz above and below that value. This choice was one based on the stock of 9 MHz crystal Filters I have and the matching BFO crystals are just that -- 1.5 KHz above and below. That isn't gospel but merely a place to start. Those values were input to the Arduino Sketch. It became pretty obvious that those were not the right values.
My next step was to disconnect CLK2 from the SBL-1 and in its place use my FeelTech signal generator to supply the BFO signals. Two cautions here with the first being that you crank down the output to 1.4 Volts PTP before connecting to the SBL-1 and second that you have a 10 NF cap between the FeelTech and the SBL-1. What I found was that the values that seemed to work were only  about 1 KHz away from the Cf. There is something wrong here.
A 1 KHz BFO would place the signal on the top of the pass band and not on the slope of the curve! Thus  1 + 1 is not 2. This bothered me and that is when the light bulb went on. Dishal is lying to me!!!! Well not entirely true but close to the truth (I must sound like Donald Trump.) Is this being taped?
This is when I tried something and this may provide an explanation for some of the crappy results I am seeing using Dishal. The first thing I did is to connect the radio to a dummy load so that no signals were incoming. Then I listened to the rig as I tuned the BFO (FeelTech) from about 2.5 kHz below the supposed Cf to about 2.5 kHz above and just listened -- I noted the change in the background noise and at one point heard a "null". Was this in fact the Center Frequency? I think it may be --tell me I am wrong. This value was close but different that the Dishal value. So I think my evaluation may have some validity. Using this value I added/subtracted 1300 Hz to set the frequencies. In looking at the location of +/- 1300 Hz this puts the frequencies about 3 dB down on the curve and so that may add some additional validity to my selection.
I have changed the Arduino sketch with the 1300 Hz values and will now await the 40M band to perk up so I can run some listening tests and get some reports on the transmitted signal. I very likely missed this part in the Dishal Tutorial (who reads that stuff anyway) but the setting of the BFO frequencies does indeed affect the filter performance.
Lurking here is the theoretical (Dishal Predictive Curve) versus what really results and that may be a shift of the Center Frequency because of imperfect components, stray capacitance, lead length and whether you have dandruff or AFF (Athlete's Foot Fungus). So many things could impact if indeed 1 + 1 = 2.

My Dishal experience has consumed way too much time and the results so far do not justify the effort. Despite what you may have read in all of the testimonials and the touted results --did those users actually install the filter in a radio and "air test it" and match their real world experience with what the curves may be telling them? I can only share what I have seen and so far it has not been pretty.

Pete N6QW

Dishal Dyslexia 

I am disappointed with the results from using the Dishal method where quite often what you get does not match the predictions or expectations. There are those reading this blog who are now quickly preparing a flaming response. There will be the usual comments: you didn't do it properly; or you did not make the measurements correctly or you built it in a haphazard manner or on and on.
My 6 pole filter sucks and is real crap. I then decided to build a four pole filter using four different crystals but ones from the same batch. I was very careful (again) to follow all of the steps and to double check the build, the test setup and the measurement process. The outcome was not unlike the six pole but  somewhat better. I did find that the skirt rejection was better than the six pole and closer to the predicted curve but installed in a radio sounded "not so good" when tuning across a signal. That to me is telling --how does it work in a rig.
So below are plots of the four pole filter responses both from the predicted value and the actual measured value. Using the FeelTech signal generator for the signal source and my Rigol scope for the detector. There was a 50 Ohm series resistor in the source end and a 50 Ohm termination resistor on the output side. The readings were taken as RMS readings.
Here is how I converted the Vrms data. First I multiplied the readings by 2.828 as this converts the readings to V Peak to Peak. I thought this would be useful to know and have as a matter of record the PTP readings. I then squared that PTP reading and divided that by 400. The 400 factor accounts for 50 Ohms and does in effect take the value back into RMS. The answer is in watts. If you multiply that by 1000 you now have milliwatts. Or you can simply take the squared value and multiply by 2.5 (1000/400). This answer is now in milliwatts. If you take 10 X LOG10 of this number you will have dBm. That is what I did. So find fault as you will; but the curve will represent the relative impact of the pass band and give you a prediction.
Here is my curve based on the measurement process. I took readings every 200 Hz. There might be some advantage at this point to repeat the test using 100 Hz data points. But this curve does look like the Dishal curve for the four pole filter. Installed in a radio I can say that on receive it does not sound that good.
The Dishal Predicted Curve!


So what do you do now? You have the filter, it looks like the prediction but does not sound too good. There are opportunities for changing things. One change: make the bandwidth less. I set it at 2.5 kHz so it would have that "hi fi" or enhanced SSB sound. A filter at 2.3 KHz night be a better choice, and doing so will change the cap values and the matching transformer. So before I do that I want to do some more testing with the current configuration.
This now  begs the questions: Why did the 6 pole shoot craps and Why the 4 Pole while looking like the prediction not sound so good? Maybe the sounding not so good is that the bandwidth has been opened up and I am used to hearing like about 2.1 kHz. Stay tuned.
Pete N6QW

Monday, June 5, 2017

The LM373 Transceiver ~ Three weeks on the air.

The LM373 Transceiver - What a Rig!

The LM373 is a Real Gem!

6/7/2017 ~ New Video of the LM373 Transceiver!
6/6/2017 ~ Added >10 dB Attenuator to handle super strong signals. It is a "T Type" controlled by a small relay with the Series arms being 33 Ohms and the Parallel arm being 22 Ohms. A switch on the front panel controls the In/Out. Kind of amazing --with the pad Out of the circuit you can hear the noise on the signal and with it In --the noise drops like a rock. Good affirmation about RF amplifier stages and their contribution to noise. The Pad/Relay was mounted on a small circuit board which was soldered vertically to the main board.

 The panel mounted toggle switch that selects Pad In/Out is located below the volume control and above the earphone jack. This is a solution to the strong signal overload.
Pete N6QW
Every so often the planets align and you end up with a really nice rig. No nice is not the right word but the correct terminology is that you end up with a Superb Rig. I have fully documented the building of the LM373 Rig at
Frequently I joke that I have two boxes of electronic projects with a giant box containing rigs/projects that worked once, or ones that don't work or ones that will never work (the 6 Pole Dishal Filter is in this group). Then I have a smaller box of rigs/projects that work very well. The LM373 Rig is definitely in that box.
Rigs can look good --or can even be called pretty; but how do they work is the real test. Typically the unofficial criteria is how do they sound on the other end, can they work DX , how do they receive, and for those who lurk the bands with their SDR store bought radios with 72 inch screens is there any energy below 200 Hz or above 2800 Hz. This latter criteria seems to be the one that is the most controversial with those of us who homebrew their rigs.
So now I would like to share some LM373 Performance information. With IRF510 for a final it will put out in excess of 5 watts (6-8 can be easily had). With my intermediate amp it will do >110 watts and with the SB200 in line following the intermediate amp it will do 700 watts (easily). Couple that with a 2 element beam (which I will talk more about a bit later) and you can break pile ups --which I have. Stations have been worked running QRP and most of the stations worked at 110 watts and of course stations worked at high power. Interestingly about 1/3 of the DX stations were worked running the 110 watts -- the beam antenna is a great field leveler.
Now how does it sound -- I get frequent rave comments about the audio (subjective from non SDR monitors). Punchy and articulate are frequent descriptors --kind of amazing since the audio stage is a single 2N3904 that was modeled in LT Spice. There have been no reports about spurs, lack of sideband suppression or pinched sounding. All this from a homebrew rig!
On the receive side it hears really well and that has been a frequent question from the other end after hearing the transmit side. Well I have found the hearing part an interesting aspect of the rig. The LM373 has built in AGC Circuitry and the data sheets include suggestions for making this adjustable versus fixed values. That will be a further experiment.
In W5BAA's design he used a second LM373 as the receiver mixer stage and riding along with that is that you do get some gain from that stage since it is an active (versus passive) mixer. In my rig I used an SBL-1 as the mixer stage which as a passive mixer (non-gain) does have a conversion loss. Ahead of the SBL-1, I added a single manually adjustable gain stage. Mind you the manually adjustable is a trim pot on the circuit board. My original intent was to set it for about 10 dB and forget it. Well I have found that on really strong signals and with the beam down the throat of the station that the rig will overload. Crank back that trim pot ( 2 to 3 dB of gain) and everything clears right up without any other adjustments. This does not happen all of the time --just on the rock crushing signals and sometimes I do see those kinds of signals. With the beam moving off the signal heading this clears up the signal too. So I smile as I now know the beam is doing its job!
The solution to the infrequent overload can take many forms and two of those would be to make the AGC a panel control as shown in the LM373 Data Sheets or add a 10 dB attenuator (a simple  relay with a 10 dB pad on one side of the contacts) as I did with the KWM-4. The attenuator would be easy to implement right on the 2N3904 Rx RF Amp Board. This has perhaps a better approach as the engagement is a simple toggle switch versus a panel mounted pot.  So this whole discussion boils down to that the receiver is sensitive and can be too sensitive if upstream there is not some signal limiting for extremely strong signals. A more elaborate AGC would also 'save the day'.
Now how does it work with DX -- well my DX standard is the European / VK-ZL path. I have worked the following pre-fixes S51 (Slovenia), EA3 (Spain), E51(Cook Island), LU1 (Argentina), XE1(Mexico), ZL1(New Zealand) and 9Y4 (Caribbean). Mind you that so far I have made 50 contacts total with the rig and 15% have been DX.
Sides have been added to the case (painted black) and now I am working on a top cover. One possible top cover might be a sheet of 1/8 thick Lexan plastic so you can the "innards" another is a metal cover with a speaker built into the cover. Still noodling that piece.


There is just something about being able to work DX stations (some running Flex 6700's) and to have them say nice signal --all from a homebrew rig using 1970's technology.
Pete N6QW

Friday, June 2, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.19

6/4/2017 ~ The Dishal Saga continues.

Bob, N7SUR shared his Dishal experience (he was very successful) and in his configuration he has a TIA amp ahead and following the filter. He is also using a AD9850 signal source. I added a front end amplifier and the results improved somewhat. This added about 15 dB of improvement. The real test would be to add an amplifier on the back end and make another run. But NOT THERE yet but better than earlier. The theory about amplification is that the scope may not be detecting the lower signal levels (masked by noise) and with a boost this enable better readings.

Pete N6QW

6/3/2017 ~ More Listening Test with New BFO Frequencies

(BTW only 1 kHz below/above the apparent Center Frequency --what does that tell you?)

The Dishal Filter must be on par with Fake News!
Pete N6QW

Four weeks of work and this is what you get --not very satisfied!!!!  Several blog posts ago I mentioned about Method #1 which simply is finding 6 crystals that are no more than 50 Hertz spread among any of the crystals, use 150 PF NPO caps and build a ladder filter. Assume the Zin/out to be 150 Ohms and get on with life. I would bet that approach would have netted a much better filter. If you are considering the Dishal approach opt for Method #1.

Pete N6QW

Sunday, May 28, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.18

The Final Filter

6/3/2017 ~ A definition of Covfefe: Running the same exact test on the same piece of crap will change the outcome!
My question is why is there only about 25 dB difference is the peak response and the rejection? The bandwidth looks about 2.2 kHz but the ultimate rejection looks poor. I would feel better about the results if there was say a 45 dB difference in the peak response. The shape looks fairly flat but not within 0.2 dB as predicted by Dishal for a 127 Ohm termination. I can only reflect that what we have going here with the Dishal approach is much like the touted Gas mileage is indeed YMMV!
Pete N6QW

I have received a private email from someone who knows and they have shared that my disappointing results with the Dishal Filter possibly is linked to Covfefe. I surely hope not but this malady may be affecting a lot of electronic failures from now on and for the foreseeable future.
Shown below is a second data run which is not inconsistent with the 1st run. The results are disappointing. I wouldn't call this a very good filter.

Pete N6QW

The filter is built and this may be the last one I ever build. Talk about wasted man hours. I will now begin some characterizations tests. Getting to the precision caps was a lot of effort and without the AADE LC meter this would have been impossible. I have added a plot of the response and now need to go back and do some fine tuning  of the caps.

To use a crass common vernacular term --this was a real "ball buster"! I am not satisfied with the result. This was a lot of effort for a curve that looks no better than picking 6 close crystals and using 150 PF caps in a ladder configuration. One has to ask why go through the agony?
Pete N6QW

The above plot is with the data converted to dBm. No resemblance to the Dishal curve but we now have a baseline for "Twizzling" the caps. I am not all that hopeful that this will dramatically improve.
The BW looks like about 2.4 kHz but not enough data points to look at the 60 dB points. I only scanned about 5 kHz at every 100 Hz point. More points and maybe at 50 Hz increments would have further defined the curve. But this is a starting point.

I am so glad that my brain works while I sleep as that often keeps me out of trouble or at least may minimize some downstream trouble.
When I awoke the other morning I got a message from my brain about the effect of stray capacitance and the design of my compact board. The pads as cut are 0.5 inches by 0.5 inches and there is a pad to pad capacitance as well as a pad to ground plane capacitance. So I took my handy AADE LC meter and measured away. Boom the pad to pad capacitance is in the range of 3 PF and fairly consistent from pad to pad and the pad to ground is higher in the range of 4 PF. So those values are in parallel with what I am soldering to the board. Thus I need to re-look at the values I have chosen for the caps and this may negate the need to have trimmers installed.
It may take a day or two but this is a precision crystal filter and thus will need to rethink the install. There has also been traffic on the illuminati reflectors about capacitor Q and how this impacts the use of caps in various tuned circuits. I purchased precision MLCC 1% caps so I am not starting in a deep hole for those who will be quick to email me about the Q of capacitors. I have been advised accordingly so am aware of this concern. Thanks Greg.
Pete N6QW

Tuesday, May 23, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture ~ 2.17

[For more info on the LM373 SSB Transceiver see

Reworked Color Display for the Dual Conversion DifX

The Crystal Filter Build Out

5/25/2017 ~ Where "Noodling" pays big dividends.
Yesterday I made a "linear layout" of the six pole crystal filter and the result was a series of 11 pads where the crystals and caps would be installed and that assembly was about 6 inches long. This is an unwieldy dimension to handle inside of a compact radio. Then I spotted a photo of an 8 pole filter as used in the Elecraft rigs and saw how they had compacted the build. So that led me to think a smaller footprint. I now have come up with a compact version that get the footprint down to something like 2 (or 3)" X 4" inches --much more accommodating. I also had concerns about signal leakage between the sections of the compact layout and think that can be fixed by the addition of a shield between the sections. See my "noodling" sketch below. (I are not an artist and barely an engineer!)

I have already created the design for the compact board in G Simple and have the dxf file ready to cut a board on my CNC Mill. I will add a photo of the cut board when I am done.

Noteworthy I have received an email about my capacitor selection process and a link to a video about polarity of capacitors and the stability of capacitors. Hopefully the extra $$$ I spent for the 1% caps and the NPO criteria will abate some of these well founded concerns. I will also exercise caution in NOT applying too much heat to the caps during the soldering process.
Pete N6QW
The caps arrived today and I also measured the 150 PF NPO caps I had in stock. One of values needed (4 places) was 151.2 PF. Using my AADE LC meter, I measured 13 caps and here are the results.

The results of this test shows that of the 9 caps we have nailed down 5 of them. The two 115.4 PF I think will be asy to achieve with the 100 PF NPO I bought and the 15 PF trimmer. The 430 PF that were to be the basis of the 437.1 PF are so small I can hardly see them so I may have to invoke a work around.
I also designed the PC board and it is about 5 inches long and 3 inches wide. There may be a way to compact this but for this 1st run it will be linear. Hope to cut the board tomorrow and start soldering in crystals and caps.
Pete N6QW
I have been awaiting the delivery of some 1% caps and they will be arriving in the next couple of days so in the meantime I will flesh out the circuit board for the crystal filter. So let us examine what is required. My Dishal filter for the 6 Crystal Filter has a total of nine capacitors and they are as follows: 
  • 2 X 151.2 PF
  • 2 X 437.1 PF
  • 2 X 115.4 PF
  • 2 X 151.2 PF
  • 1 X 156.8 PF
Noteworthy is that we really only have 4 values of capacitors including 4 caps of the same value. The precision caps I ordered are 100 PF, and 430 PF and these are 1% and NPO AND Surface Mount. I have a bag of 150 PF NPO caps which I will measure to find values all below 150 PF. (Hopefully finding four that are all below 150 PF but close in value. I have an AADE LC meter and that is my measurement tool.
The PC Board will require a total of 11 Pads to accommodate the crystals and the caps. To get me exactly on the values needed I will make the pads large so that there room for the 15PF  air variable trimmer caps in addition to the fixed value ones and the crystals. Playing with these parts before I cut any pads is critical to the final assembly. Thus 437.1PF = 430 PF + 0-15 PF Trimmer
I will add to this posting with the layout of the crystal board after I get it laid out.
Pete N6QW

Tuesday, May 16, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.16

I am now documenting the LM373 build on my website so see the
following LM373 Link So for those tired of seeing me gush over my vintage 1970's designed, new rig you will have some relief. Have made about 2 dozen contacts including some at 600 watts. You got to have power to make lots of contacts so get over that QRP stuff! Soon we will be back to the crystal filter build on the DifX.
Pete N6QW

5/19/2017 ~ QSL Card from the 1st contact with the LM373 Transceiver. Here is proof! Also noteworthy I was running only 15 Watts and the Mosley 2 Element Beam. What fun! I am now up to 10 Contacts with the LM373 Rig.

Pete N6QW

5/17/2016 Some added photos of the innards of the LM373 rig. We have now made six contacts with the new rig. I still need to tidy up the wiring. Yes, I succumbed to an IRF510 in the output which is not my favorite RF device --but it was an expedient. I am redeploying the K5BCQ 5 Watt final to the DifX dual conversion rig. With the outboard amp this will now do in excess of 100 Watts -- more than QRP!!!!! In the last photo you can see the LM373 IC square in the middle of the photo. All that work being done with one IC. To the aide and aft of the 9.0 MHz crystal filter is the 11 VDC power supply and the N6QW solid state power switch. The relay on the back panel only switches the antenna between T and R and one extra set of contacts switches in the outboard amp. In addition to the final wiring clean up I will add some aluminum support structure so that the front/back panels are rigid. This rig has taken about six weeks to build with some of that time awaiting the new Rigol scope.
Pete N6QW


While I gather up some high quality capacitors to build the six pole 11.5 MHz Crystal Filter, I wanted to share more exciting news about the new LM373 20 Meter SSB Transceiver. Yes that is a photo on the masthead --Pretty Cool Blue isn't it?
This transceiver is more than a one hit wonder as I now have made three contacts with the new rig and I am please at how it works. It is also important to note that it is not a bilateral approach and thus truly not a Bitx! By that I mean that signals pass through the IF stage (which is a single IC -- the LM373) and steering through the single direction stage is done with 1N914 signal diodes. On receive the receiver mixer stage (SBL-1) is connected to the front end and the BFO is injected into one port of the LM373. On transmit the output from another SBL-1 acting as the balanced modulator is fed into the front end and now the LO is introduced into the port where formerly we injected the BFO. The port where the audio is recovered on receive is now the transmit output stage. Thus three ports on the LM373 are switched in going from Receive to Transmit --all in a single 14 Pin Dip package!
As I often saw with analog VFO's and BFO's the transition from R to T frequently caused a shift in frequency because of the loading effect as the LO and BFO were shifted to other parts of the circuits. That is not the case with the Si5351 --there is no shift. This single pass through an IF stage in older designs employing Analog LO/BFO's was problematic in this aspect. I wrestled with this issue and had some elaborate schemes to mitigate the problem. The bilateral approach where the LO and BFO were not shifted was a solution. Now with the Si5351 this is a solution with a single pass approach.
I have challenged myself to use as much stuff as I have in the junk box without resorting to placing orders for parts and that has been very rewarding in that "you use what you have!"
Beyond the LM373 I am using my standard Audio Amp stage as well as the bidirectional 2N3904 amp stage and the driver with the 2N2222 and BD139 produces a whopping 400 MW. I now will add the 5 watt K5BCQ RF Power MOSFET board that uses the Mitsubishi RDD series of transistors. Right now the 400 MW is driving an outboard SS amp all the way to about 15 Watts output.
Touring the knobs, the left most smaller knob is the audio gain and the right one the microphone gain. The three switches on the bottom center are for USB/LSB, Tune (Red PB) and the right switch is the MOX. The right hand large knob (ala Drake R4B) is the main tuning knob. There was some careful thought (about 1 minute) to the panel layout relative to ergonomics. Since I am right handed the main tuning and step rate change is the most frequently used so it is to the right not unlike the Drake rigs!

The new Rigol DSO has been very useful as I worked on peaking a tweaking this new rig.

Pete N6QW

Thursday, May 11, 2017

A New Line Of Transceivers ~ DifX

Transceiver Architecture 2.15

News Flash --- The LM373 SSB transceiver is on the air and the 1st contact was with W4DNQ, Ron who is in Florida. The contact took place at 1525 PDST on May 11, 2017 on 14.220 MHz. Here it is "al fresco"

Lets Build the 11.5 MHz Crystal Filter ~ Part III

I re-measured the 9 candidate crystals using the G3URR oscillator only this time using the 1 Hz position for fine tuning on my SDR transceiver. The field is now down to 6 crystals with five of those being exactly the same frequency in the loaded state and the 6th is only off by 5 Hz. As we used to say when I worked in Aerospace --Close Enough for Government Work!
So now I was ready to go back into the Dishal Software and using the new data, I  re-plotted the curve for 6 crystals. Again the setting of the Band Pass ripple materially affects the Z in/out and the capacitor values. Less of a ripple translates into a Lower Z in/out. So there has to be some reasonability (trade-offs) between Ripple and Impedance. Below is the new plot and we will now highlight some of the factors.

First and foremost is that it is a 6 pole filter that is characterized by steeper slopes and a 3dB bandwidth of 2.35 KHz. The band pass ripple is 0.13 dB (pretty flat) and the Impedance is 127.4 Ohms. 127.4/50 = 2.55 and a 8 turn to 5 turn broad band transformer is 64/25 = 2.56 ==so really close.

There are 9 capacitors needed for the 6 pole filter and in looking at the values these are easily achieved using high quality (NPO) caps in parallel with very small air trimmers (0-15 PF which were provided to me by a kind ham located on the east coast --Thanks Bob!)

An old time filter measure was to compare the 6 to 60 dB "shape factor". The BW at 60 dB is only 5.43 KHz --so that should do well with those California Kilowatts located just down the street from me. BTW the math shows 1: 2.225.

My next steps will involve acquiring some capacitors and building the actual filter. Stay tuned.

Pete N6QW

Wednesday, May 10, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.14

Back to the LM373 SSB Transceiver.

While I take the time to re-measure the 9 crystals for the Dishal Filter, I thought I would update the LM373 SSB Transceiver project. This rig qualifies as a DifX as it is not a Bitx and the subject matter that will be covered has direct applicability to the DifX Dual Conversion project.
W5BAA's LM373 SSB Transceiver (Ham Radio November 1973) used 11.0 VDC as the supply voltage for his rig. If you look a the photo of his rig in the article you will see it connected to a 12 VDC Gel Cell. So connecting the dots, this gave him some lee way as the Gel Cell discharged. My evaluations of his approach actually demonstrated that his design did indeed "love" 11 VDC and so I followed his direction.
But now you have to have a source for 11 VDC (at least for the lower level stages (which is most of the rig) and you also needed a means of efficiently switching 11 VDC to the steering diodes to change the application of the circuit elements.
So task #1  was to come up with a 11 VDC regulator circuit. For this I went to an LM317 TO-220 style that is good for 1.5 amps. My circuit has the element of being adjustable which you will shortly see the why. The Internet abounds with regulator calculators and favorite circuits. I simply lifted one of those circuits and used values from one of the voltage calculators.
Essentially the LM317 samples the output voltage and using a voltage divider feeds that back into the V adjust pin. The ratio of the feedback resistor and the resistor from V adjust to ground sets the final output level. The LM317 can handle input voltage up to 40 VDC in and will regulate voltages up to 37 VDC out. The delta difference is needed for the regulator to work properly.
In my design the two resistors were 220 Ohms and 1715 Ohms. The 220 Ohm is a stock value but 1715 is not! So I homebrewed a 1715 resistor by placing in series a 1K connected to a 200 Ohm pot wired as a variable  resistor and that in turn is connected to a 680 Ohm resistor to ground. Thus with the pot a "0" we have 1680 Ohms and with the pot at 200 Ohms we have 1880 ohms. The pot also compensates for the tolerances of the 1K and 680 Ohm. The circuit works very well and this is shown below.
So now we have a 11 VDC source and as was mentioned when I first covered the LM373 Transceiver, W5BAA used steering diodes to route signals through the single direction IF block and it is through the application of the steering voltages that the LM373 rig goes from Receive to Transmit. Long ago I designed a circuit to do this. It was out of necessity as I was using an analog VFO (Yes Bill one of those old technology devices) and a power relay to switch voltages. Well the back EMF was so large (even with a snubber) that the VFO would jump frequency. This resulted in my solid  state switch design shown below to replace that pesky relay and has been updated for the 11 VDC. This same switch will be used in the DifX Dual Conversion rig although at +12 VDC.
Stop! I can see all of those who are into power FETs that will quickly make a posting about doing this with power FETs --you know will supply the arguments  that no current is drawn and how when you take this to the field battery life is precious and how your design is so much better. Well good for you. I am not taking this to the field, I have a stock of parts and it works just fine.
So how does this work. First I thought it would be cool to use an optoisolator like the 4N35. That device is also important as the next element in the circuit is quad NAND Gate and a bouncing input signal will give a lot of false triggering --the 4N35 acts as a buffer between your mechanical PTT switch and the 7400 IC. Two of the NAND Gates are wired as inverters with one Gate triggering the second Gate. With the PTT in an OFF state the upper 2N3904 has a signal applied to the base which powers on the TIP32C and we have juice for the receive condition (R). When the PTT is engaged the 7400 is toggled so that the HIGH output is not on Pin 6 but Pins 3,4,5 and thus the upper 2N3904 is now OFF and the lower 2N3904 is ON and we have juice for the Transmit stages.
You can build your self a truth table to verify what I said. Since there is a pull up resistor on Pins 1 and 2 that means it is HIGH (with the PTT disengaged) which results in Pin 3 being Low. But Pin 3 is bussed with Pins 4 &5 which make them low and now Pin 6 is High --driving the upper 2N3904 to ON. Now when you trigger the 4N35 with the PTT the output goes low which means the 10K is to ground and that drives Pins 1 & 2 to a Low state. Now Pins.3.4.and 5 are high and Pin 6 is Low. But since we are tapping off of Pins 3,4,and 5 and in a  HIGH condition we now see a signal on the lower 2N3904 and we have a transition from R voltage to T voltage. No relays are involved for this part of the circuit. Below is the build.
Yes there is a combo of leaded and SMD parts and the layout is close to the schematic. This is a keeper and no power FETs were involved in the build! When I connected the regulator and solid state switch to my LM373 I measured the voltage at the R and T ports and then adjusted the 200 Ohm pot so that they read 11 VDC at the R and T pins (collectors).
After some additional tweaking I now find that with my single 2N3904 pre-driver stage it will produce 10 Millwatts into a 50 Ohm load - (2 Volts Peak to Peak) . Now we will proceed with the driver and final stages. This is really moving along.

Tuesday, May 9, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.13

Lets Build that 11.5 MHz Filter Part II

In Part 1, (2.12) we shared how we measured all 25 crystals and found a grouping of 9 that were essentially "carbon copies" of each other. I never play the lottery because of that exact situation. I suspect all lotteries are rigged and the likelihood of having nine exactly the same crystals is somehow rigged. So the first thing I will do is to re-measure all of those nine in hopes of finding 6 out of that batch that are exactly the same. That may happen yet today.
So once you have the crystals and the two frequencies that were measured on the G3URR oscillator proceed to the WA5BDU tutorial on how to calculate Lm and Cm. I made an assumption that the crystal holder capacitance was 3 PF --probably not good or bad. Regardless, using the equations provided by Nick Kennedy I found the Cm to be 16.356 Femptofarads (10 to the -15) and using that value you can calculate the Lm which is 11.7 MHy.
Now for the actual crystal filter calculator go to the ARRL QEX download page and for the year 2009 find the Zip file 11/09_Steder_Hardcastle and download to your computer. Do NOT place it in the Program Files or Application Files!!!!!!!! I have a directory called N6QW and in that directory I have a folder called Dishal --extract it to that folder. Milton Dishal wrote a piece of software to simulate crystal filters. Steder and Hardcastle take that software and turn it into "ham speak" so it can be used by people like me.
Now I had a problem with the QEX download --my computer kept telling me it contained a virus and promptly removed it from my computer no less than 3 times. The fourth time on the Zip file I used a tool called compatibility --it seems when I now use the Dishal software my computer thinks it is a windows XP machine and all works good.
Here is where the dishal software can save you lots of effort --if you enter the shifted frequencies it will actually calculate the Lm and Cm --and they match my hand calculation. After playing with it a bit I can see that it would be ideal to have six crystals in the filter and here is why. See below.
The first plus is the 3 dB bandwidth which is about 2.2 KHz and at 30 dB down is probably less than 3.0 KHz. The 20 dB numbers and 40 dB numbers look OK. It appears to have a more symmetrical shape ( a factor that improves by having more crystal in the filter), thus getting away from the idea of strictly an LSB filter.
I found out the pass band ripple is linked to the input impedance. In this case it is 102 ohms. 102/50 = 2.04 and if we did a 7 turn and 10 turn winding on a FT37-43 core we would have 100/49 = 2.04 --so a perfect match. But if you drop the ripple to 0.2 dB then the input impedance raises to 127 ohms and this is a ratio of 2.54 when transformed to 50 Ohms. So if we use an 8 turn and 5 turn winding on a FT37-43 core that transforms to 64/25 = 2.56 ---really close.
The software has a built in  LC matching network calculator so that you can essentially design "L Networks" for the matching. I prefer the broadband match.
The 0.2 dB pass band ripple also "twizzles" the caps so that the values are much nearer to close tolerance standard value caps. I have some low value air trimmer caps (0-15 PF) that could be paralleled with the standard values so that the filter will be dead on.
I also believe this filter will have to be built in such a way where the crystal cases are flat to a ground plane and the crystal can soldered to that plane. I will use SMA connectors to the filter and it will be totally enclosed in a metal box. This will be bullet proof.
My plan is to update this posting with the re-measuring of the 9 crystals --to see if I should play the lottery today.
Pete N6QW

Friday, May 5, 2017

A New Line of Transceivers ~ DifX

Transceiver Architecture 2.12

Let's Build That 11.5 MHz Crystal Filter!


Sunday May 7, 2017

 I have completed the first round of crystal measurements and have found several groupings of crystals that have potential for a filter. One group has nine crystals and my next step is to re-measure all 9 to verify that indeed that all have the same specifications. I used different colors to highlight the crystals with the same specs. See the table below:
Stay tuned for the next round of measurements.
Pete N6QW

See video at end added today on measuring the loaded crystal frequency --yes being done with a Raspberry Pi3 and a SoftRock V6.3 SDR radio.
My bag of 25 crystals arrived (May 4th) and in case you are wondering what I bought here is the Mouser Part Number (See below) The one thing nice about the manufacturers specifications is that the ESR (Equivalent Series Resistance) is provided and that number is 35 Ohms.(That could be good or bad but at least I know what they say it is,)




It is amazing how crystal filters are suddenly at the top of everyone's list. The man himself, W7ZOI, Wes Hayward on May 1, 2017 just released an exciting document that details the "innards" of our beloved crystals as used in filters. You can find that document here  Essentially Wes describes the crystal model and how today's crystals present some issues when we try to build crystal filters. Download this and keep it under your pillow at night.
In Transceiver Architecture 2.10, I mentioned that there are some excellent you tube videos on the subject of homebrewing crystal filters and w0qe, Larry Benko, in his part 13 and 14 has taken up that task. Part 13 deals with some of the front end analytical work and Part 14 gets to the nuts and bolts.

One tool used in Part 13 was the use of Elsie (free Student Version from Tonne Software) to simulate a crystal filter. The "free version" will only let you simulate a 4 crystal filter but that is good enough for our purposes. The professional version lets you add many more crystals.  Elsie will not design a filter for you --BUT it will let you "twizzle" the parameters so that you can visualize how changing parameters impact the shape of the filter curve and the SWR. Interestingly the Elsie parameters match closely what you will finally need for a good filter. The Zin/out looks like less than 200 Ohms and the coupling caps are in the range of around 100 PF.
Using the Elsie Software (with help from Jim Tonne, the developer) I was able to simulate what a crystal filter might look like using my 11.5 MHz filter frequency. If I could match this model in practice, we would have a pretty good second filter for our DifX Dual Conversion Transceiver. It is really interesting to see how changing the bandwidth and Zin/out materially affects the shape of the curve. My latest run at this suggests that the Zin/out should be about 100 Ohms and the bandwidth 2.8 KHz which of course will place me in the league of the enhanced SSB guys. But those numbers are not the final values --just data.The combined plot below shows the filter shape and the SWR. Don't overlook the SWR as that impacts the filter capability to pass signals without distortion!

In Part 14 Larry, presents a plan on how to go about actually building a filter and it starts out with "get a bag of lots of crystals" and then "measure them and record the results in an Excel Spreadsheet" -- hmmm sound familiar?

The Elsie Software presents an idealized filter with crystals that are exactly alike (you'd like to see to 10 decimal places).  But that is not the real world as the Elsie software only captures just a few decimal places-- like out of the box maybe two or three places. As filter builders we in essence TUNE each crystal so that it will match the Elsie model. This is not a five minute job! But it can be done. In Part 14 Larry, show how he kept changing parameters and how that fine tuned the filter. I suspect the first time through the process will be like being 7 years old and waiting for Christmas and it is only June.

My first step is to identify and then  measure each crystal and finally record the results. Fortunately long ago I built a G3URR oscillator and will use my Raspberry Pi SDR radio (with Quisk) as the measurement tool or I might use my Omnia with HDSDR --so I have some choice here. Or my new Rigol Scope that has a built in frequency counter.

But having just run some tests using the Rigol Scope  "counter function option" has shown that will not work. The reason is that the scope (or my lack of knowledge on how to have this do something different) only takes you to 4 decimal places. I ran a few crystal into to the G3URRR test oscillator and the scope counter read 11.4998. This is not enough resolution! They tutorials all sing the same song. If you use an external oscillator method to excite the crystal for measuring Lm and Cm then you need to have a generator with a 1 Hz resolution. So you will need to have something as good to measure the crystal response.

My friend KV4QB, DuWayne was so kind to send me an SNA Jr. board and it will produce the 1 Hz resolution for the DDS generator. I have an Omnia SDR and with HRSDR can read the signals to 1 Hz. Keep in mind we want no more than a 50 Hz spread for a SSB filter and as I found out no more than 10 Hz for a CW filter -- the Rigol scope will work to get you into the ball park but will not tell you the seat number.

w0qe suggested several actual crystal filter design programs (including the AADE one I mentioned). But did indicate several that were more user friendly and so I will look to using one of the ones suggested.

For now I will have to take a short respite while I measure the crystals and record the results. Stay tuned. Yesterday using my handy Brother tape label machine I punched out 25 numbers and will start first by affixing the numbers to each of the crystals. Then I will need to clear a spot off of my work bench so I can do the measuring. I may get a chance to video the crystal measuring process as that may be helpful t those wanting to build the 11.5 MHz filter.

Too many projects and not enough time --but I will continue to work on the LM373 transceiver and keep you posted on its progress.

Pete N6QW