Happy New Year to all. May you make good progress on your homebrew projects, and may the radio gods act favorably on your behalf.
73 es HNY de N2CQR
Serving the worldwide community of radio-electronic homebrewers. Providing blog support to the SolderSmoke podcast: http://soldersmoke.com
Happy New Year to all. May you make good progress on your homebrew projects, and may the radio gods act favorably on your behalf.
73 es HNY de N2CQR
This is really an amazing project. Way back in 1967 (that's 55 years ago) John Aggers W5ETT of Ponca City, Oklahoma decided to homebrew a solid state version of our beloved Drake 2-B receiver. Triple conversion. No crystal filters. Twenty two discrete transistors and no ICs. Tuned circuits at 50 kHz to provide most of the selectivity. And he did it. Just look at the picture above. It even LOOKS like a Drake 2-B.
I sent this to our friend Dale Parfitt, W4OP who more recently built a receiver like this. He too was amazed by this project.
The article by John Aggers is very clear and provides a lot of good information on how he designed and built this receiver using the technology of 1967 and junk box parts. I was struck by the lack of diode ring mixers. And I was somewhat taken aback by his use of plug-in socketed transistors. The AF amplifier is our still-familiar transformer-less push-pull complementary pair design. John did a wonderful job on the mechanical tuning and slide rule mechanism.
Three cheers for John Aggers W5ETT. This article is a reminder of the great benefit to the hobby of writing up a project and putting out there in the world. Here we are, more than half a century later, reading John's article and learning from it. FB OM.
Page 8
This has been one of the major complaints about our beloved analog LC VFOs: The frequency tuning on these circuits is often not linear. For given amount of VFO frequency dial turn you can get vastly different changes in frequency. At one end of the tuning range the frequencies are nicely spaced and tuning is easy. But at the other end of the tuning range all of the frequencies are bunched together. This is one of the problems that leads some homebrewers to defect to the sad land of "digital VFOs."
But wait. It appears that the old designers found a solution to this problem. Just look at the tuning dial of my HT-37. The frequencies are all spaced out evenly. How did they do that?
I had been thinking that this success may have resulted from Hallicrafters' engineers using the series-tuned Clapp circuit. Here the main frequency determining element is a series-tuned LC circuit and not the parallel tuned LC circuit that we see in the more commonly used Colpitts circuit.
But hold on -- how could that be? The frequency bunching problem that we attributed to the Colpitts circuit must also exist in the Clapp, right? I went back to SSDRA where there was a good discussion of Colpitts and Clapp VFOs. The advantage of the Clapp was said to be in its use of a larger value coil which helped minimize the effects of stray inductances. But there was no mention of the Clapp offering improved linearity in tuning.
I have in front of me two transceivers: The Mythbuster uses a 9 MHz Clapp circuit (see below). The 17-12 rig uses a Colpitts Circuit. I checked the tuning linearity of both. Both appeared quite linear in tuning, with no real difference between the two.
Then I looked at the tuning capacitor in the Mythbuster 17-12 rig. It came out of an old Hallicrafters transmitter, probably the HT-44. I looked closely at the stator and the rotor plates. Both are curved. I'm guessing that this may yield a more constant change in capacitance for a given movement of the main tuning dial.
Next I opened up the VFO on the Mythbuster. (It is the VFO from an old Yaesu FT-101.) I couldn't see the stators very well but it appears that their shape is different from the square shape we often see in variable capacitors. Could it be that this variable capacitor was also made to provide linear tuning?
Back in 2013 Norm Johnson wrote about all this in the Antique Radios.com forum:
A capacitor that has uniform increase in capacitance with rotation will have the stations at the high end of the band squeezed together. Another type known as the straight-line frequency variable capacitor has, as you might guess, a characteristic that gives even spacing of frequencies with shaft rotation. These were popular in the 1920's but weren't very good for superhets where you needed to have a dual section capacitor that would tune both the RF and local oscillator, and have them track each other properly. The midline variable capacitor is more compatible with a superhet, and easier to make both sections track properly. This is the type that you see in most receivers from the late 1930's to the end of the tube era. They don't have quite the equal spacing between stations across the band that the old straight-line frequency caps had, but they're much better than the variables that change capacitance linearly with rotation.
I wrote an online calculator that helps in the design of the tuning. It shows what frequency range you'll get with a specific type of variable capacitor, including the effects of padder and trimmer capacitors. It also displays a dial scale that shows how the frequencies are lined up accross the dial.
http://electronbunker.ca/eb/BandspreadCalc.html
Steve W6SSP also provided some really good info back in 2013:
There are three types of open, variable plate caps;
SLC= straight line capacitance where the capacitance varies linearly,
these are the most common and have half-circle plates
SLF= straight line frequency where the plates are tapered to allow
for linear tuning of the frequency
SLW= straight line wavelength, you get the idea...
SLF and SLW caps have oblong plates.
The effect on tuning a receiver can be dramatic. One example is the
Hammarlund SP series of receivers where the ham bands are very
compressed at one end of the tuning range. They used SLC caps
in the VFO. On the other hand rigs like the Kenwood TS-520
and FT-101 series have linear tuning across each band. These use
SLF variable caps. Most old 1920's battery radios used SLW
where stations were identified by their wavelength.
Steve W6SSP
I looked at the tuning dials on my Hammarlund HQ-100 receiver. It is fairly linear in its tuning, but not as linear as the HT-37 or the Drake 2-B; on all of the tuning ranges the frequencies seem to spread out a bit at the lower end. My guess is that Hammarlund used the midline variable described above by Norm Johnson. The HQ-100 did use a ganged variable cap, with one section tuning the RF amplifier and the other tuning the local oscillator.
2) Try putting a series LC shunt circuit tuned to 41 MHz at the output of the carrier oscillator (between the oscillator and the buffer).
3) Reduce the voltage to the oscillator/buffer. I have this on a pot, so I can adjust it down to the point where the remnant of the harmonic is no longer audible, while keeping the main carrier osc signal sufficiently strong.
It seemed to work. I could now hear the desired frequency for spotting, without the confusing tone from the spur.
Why had I been able to do this back in 2002 in the Azores using a simple trimmer cap to ground? My guess is that I was using my Drake 2-B as the receiver. The trimmer cap to ground may have reduced harmonic output. And I was probably cranking back the RF gain on the 2-B to the point where I could hear the desired signal but not the remnants of the spur. I have no RF gain control on the Barebones Barbados receiver that I am using in this project.
So, what's the lesson from all this? Well, if you are faced with a serious technical problem, and you find yourself considering complicated and difficult solutions, go to the Dominican Republic for about a month (especially if it is January or February), and then take another look at the problem when you return. If you are unable to travel this far or for this long, taking a walk or taking a weekend break from a troublesome problem will likely have a similar mind-clearing effect.
The video above shows part of a February 1, 2022 QSO with Gar WA5FWC using the split TX/RX 17 meter rig. Gar is an amazing long-time SSB homebrewer who got his start with phasing rigs back in the day.
I was led to this magnificent receiver by the very humble 6U8 tube. Scott WA9WFA and I have been learning (mostly from Grayson KJ7UM) that the much used and sometimes loved 6U8s (three of them in our "Mates for the Mighty Midget") might be a bit long in the tooth, old even by Thermatron standards. I was worried when I remembered that my Drake 2-B has a 6U8 in it -- V2, the first mixer. So I Googled for more info and was led to this amazing receiver, a 2018 creation by KG7TR. How did we NOT see this for almost four years?
Here is more info and pictures:
http://www.kg7tr.com/75s-2b-receiver.html
Here is Mike's write-up of the project:
Mike KG7TR's web site:
As for the 6U8s, well Grayson says the tube has been getting something of a bum rap. And KG7TR has two of them in this receiver, so I will obviously have to give the 6U8 another chance.
--------------------
I didn't know that Lew McCoy had his own crystal and crystal filter company:
Bill,
Why do you guys make your Soldersmoke podcasts so darn intriguing such that I can’t listen to them in the background while I’m doing something else? Good grief! I start listening and before long you make me stop and chase down a rabbit hole to find something new that you mentioned that I had no clue was out there. Before long I’m doodling out a new sketch or playing with at a new design for something I really need to experiment with or build “next” or something I need to try. It is taking too much of my time!! J
I’ve been listening to your podcasts for years. Way back, before I knew you and before I knew you were doing these Soldersmoke blogs with Mike, KL7R, and just before he was so tragically killed, I was collaborating with him on a simple frequency counter project using a PIC microcontroller. We were making good progress on a neat design. I later completed the project but always kept his contributions noted as part of the source code.
I’ve been making PIC-based VFOs for years – dating back to about 2000 – aiming them at builders who were looking for something to go along with Rick Campbell’s (KK7B) receivers. Rick is a good friend now, after we met in the Kanga booth at Dayton where we both were demonstrating our stuff. (Bill Kelsey (N8ET) of Kanga, was the “marketer” for my kits as well as Rick’s for many years.) My original VFO kits used a DDS (high-end AD9854) that simultaneously produced I and Q signals which made it perfect for Rick’s phasing gear. Rick is a big supporter of my work but he still kids me about polluting his beautiful analog world with my “digital crap” (copyright KK7B term). When I came out with a newer version VFO using a Silicon Labs Si570 PPLL (I can hear already Pete Juliano groaning) it was a big improvement over the AD9854 in noise/spur reduction. I documented this all in a QEX article in about 2011 and Rick (and Wes Hayward) were very supportive/appreciative of my work.
I have used the Si5351 also and I understand Pete’s point of view. It’s “plenty good” for most amateur projects. However, it remains a fact that the Si570 is a better part and produces a cleaner signal. That’s the reason why the Elecraft KX3 uses a Si570. Granted, the newer Elecraft KX2 uses a Si5351 but it’s most likely because they wanted to preserve battery life (the Si570 uses more power but not nearly as much as the AD9854) and also to reduce the cost. I do understand! I also fully understand the ability of the Si5351 to produce I and Q signals via different channels. I’ve had extensive conversations about this with Hans Summers, at Dayton and online. I use a pair of Flip-Flops on the output of the Si570 instead. My PIC code driving the Si570 is ALL written in ASSEMBLER code. Yep! I’m an EE but have had a career mainly in software development and much of it was writing assembler code. I dare say there aren’t too many gluttons for punishment that do it this way. I do it because I want to understand every line of code don’t want to be dependent on anyone else’s libraries. Every line of code in my VFO’s and Signal Generators is MINE so I know I can debug it and it can’t get changed out from under me. (This problem bit Ashar Farhan hard on the Raduino of his BitX. Tuning clicks appeared because the Si5351 libraries he used changed between the time he tested it and released it. I was really appalled when I dug into this and resolved to NEVER use libraries that I didn’t write myself. Similarly, this also makes me have some distaste for Arduino sketches. I would rather see ALL of the code including the initialization code, the serial routines, etc, rather than having them hidden and get pulled in from Arduino libraries. That’s similar to the reason why Hans Summers didn’t use an Arduino in his QCX. He used the same Atmel microprocessor but developed/debugged it as “C” code with the full Atmel IDE/debugger.
By the way, Pete mentioned the Phaser FT8 transceiver by Dave, K1SWL, in a recent podcast. Dave is a very close friend, even though I haven’t met him in person since about 2000. We Email at least daily and some of it is even about radio. J I did the PIC code for the tiny PIC that controls the Si5351 in the Phaser. Yes, it’s written entirely in Assembler again! I do know how to do it for a Si5351. That Si5351 code is not nearly as much “fun”, though. I know, this will make very little difference to guys who write Arduino “C” code to control it but under the covers it’s a world of difference. It takes me about 15 serial, sequential, math operations to generate the parameters for the Si5351. None of them can be table driven and they all have to be performed sequentially. (This is all hidden in about 5 lines of complex, Arduino “C” code but the operations are all there in the compiled assembler code.) In contrast, my Si570 code is almost all table driven. I just have to do one large (48-bit) division operation at the end to generate the parameters. Yes, that’s a bit of trickery to do in ASM. There are no libraries do this.
I will point out one more advantage of the Si570 in comparison to the Si5351. It has the ability to self-calibrate via software instead of relying on an external frequency standard. In my Si570 app I can read up the exact parameters for the crystal embedded inside the Si570, run my frequency-generating algorithm “backwards” and determine the exact crystal frequency (within tolerances, of course) for that particular Si570. Then I update all the internal tables using that crystal frequency and from then on all generated frequencies are “exact”. I love this! Frequency often moves by about 6 kHz on 40M.
Oh yes, I must mention the difference of home solderability of the si570 vs the Si5351. Those little Si5351 buggers are terribly difficult to solder at home while the Si570 is a breeze. I know, many folks will just buy the AdaFruit Si5351 board and it’s already soldered on but, again, I like to do it all myself. No “magic Fruit” for me.
Now that I retired a couple of years ago and am getting out of the VFO kitting business I can finally build complete rigs instead of just making the next-generation VFO’s for everyone else to use. I recently build a tiny, Direct Conversion rig with a Si570 signal generator (of course) and a diode ring mixer (ADE-1). Look at my web page, www.aa0zz.com to see it, along with my VFO projects that I’ve been building in the past. As you well know, Direct Conversion is fun to build and the sound is astounding; however, they are rather a pain to use! Yes, I did make it qualify as a real rig by making several contacts all over the country. (Wes Hayward gave me the criteria: he told me that I must put any new rig on the air and make at least one contact before it qualifies as a real rig.)
The new rig that’s on my workbench is my own version of a phasing rig, experimenting with a Quadrature Sampling Detector (QSD, sometimes called a “Tayloe” mixer), using some ideas from Rick’s R2 and R2Pro receivers and many innovations of my own. At present my new higher-end Signal Generator works great, the QSD receiver works great (extremely quiet and MDS of -130 dB on 40 meters) and the transmitter is putting out about 16 watts with two RD16HHF1’s in push-pull. You can take away my “QRP-Only-Forever” badge too, not that I’ve ever subscribed to that concept! Still more tweaking to do with the TX but now I’m also working on the “glue” circuitry and the T/R switch. The SigGen, RX and TX are all on separate boards that plug into a base board which has the interconnections between boards and the jacks on the back. I’ve built DOZENS of variations of each of these boards. Fortunately they all fall within the size limit criteria to get them from China at the incredible price of $5 for 10 boards (plus $18 shipping) with about 1 week turnaround. Cost isn’t really an object at this point but it’s more of getting a hardware education that I sadly missed while I concentrated on software for so many years. it’s certainly nice to have willing mentors such as Rick, Wes, Dave (K1SWL), Don (W6JL) and many others to bounce my crazy ideas off. Yes, I’m having a ball!
I was licensed in 1964 but out of radio completely from 1975 to 1995. Do you like the picture of my DX-100 on my web page? My buddy in the 60’s had a Drake 2B and I drooled over it but couldn’t afford one.
Now I must finish this rig before you guys send me down another rabbit hole. Too many fascinating things to think about! I literally have a “priority list” on the my computer’s desktop screen. Every time I come up with a new project idea – something I really want to play with such as a Raspberry Pi, SDR, etc, I pull out the priority list and decide where it fits and what I want to slide down to accommodate it. That’s my reality check!
Take care, Bill. Thanks for taking the time to give us many inspiring thoughts and ideas.
73,
-Craig, AA0ZZ
After 46 years, finally a dial skirt |
N6QW Phase Shift Success -- It aint over 'till the fat lady sings |