No RF gear here?

So, it's mowing season, again, and in between rainy days, I had some time to think about this TX thing. I got about three lots done, before the new mower broke AGAIN ( same thing - tossed drive belt ) out in Oklahoma.

Anyway, since the rest of the rig is still in the process of being cased and sort of spread out, and the only way to get RF out if it will be to put an audio tone into it, meaning an audio signal generator has to be rigged up to it, while winching the mower which weighs about as much as a small car unto my trailer, it occurred to me a test jig would be handy for this.

So I took the surplus ( at the moment ) QRP Labs Si5351 frequency synthesizer, made a quick chassis in TinkerCad, and made one, nothing fancy, any and all thoughts of project creep were immediately banished so I could do this fast.

It's real simple - a 500 ma 5 volt supply for the synthesizer, and a 500 ma 12 volt supply for the TX strip. RF comes out of the Si5351 and is attenuated to the output level of the XVTR board ( 0 dBm ) and passed through the 2 pin XH connector to the TX board; the 3 pin XH connector feed 12 volts to the TX board, and takes RF back out of it to a 30 dB attenuator / dummy load, the opposite end of which is connected to an sma connector for direct feed to the RP, or left open as a dummy load.

The idea here is that with approximately identical RF input level and identical i/o connectors as used in the radio, any board made and debugged on the test jig can be directly plugged into the radio for service without any further work. Plug and pray, so to speak. Hopefully, lots of boards and circuits can be experimented with, without any need to take the radio out of service to do so.

I've not used the second output of this synthesizer, but I did include a coax line from it's output to a test point for an o scope probe, so at some point it can be looked at to see if it is as dreadful as the other Si5351. Each power supply ( they are independent ) also has a clip lead point if I need to grab a bit of 5 or 12 volt power for something else.

You can see a corner of the chassis lifted while printing, so this would be a throw away for a normal finished piece, but does not affect its usability for this purpose. This was printed at high speed to get it done quickly, I might make another one in a more entertaining color later. The xfmr is probably overkill for this purpose, but it was laying out on my test / build bench back in my ham shack, so I just grabbed it instead of looking around for a smaller one.

Hopefully TX strip building will start this evening or tomorrow.
 

Attachments

  • IMG_20200501_080451310.jpg
    IMG_20200501_080451310.jpg
    129.4 KB · Views: 149
  • IMG_20200501_080528401.jpg
    IMG_20200501_080528401.jpg
    148.3 KB · Views: 153
I worked a little on a starter TX strip over the weekend. With the RP, I measured the RF out of the attenuator at -2 dBm, so I missed my target of 0 dBm a little bit. I measured the T attenuator / dummy load at 31 dB, so that was pretty much spot on.

I tried a few RF circuits ranging from a 40673, to bipolar parts. The 40673 had the best harmonic structure, but was a bit complicated with variable bias looking forward to a future ALC power control, and it had the lowest power output. All of the RP screen shots are taken through the 30dB attenuator, out the SMA port on the test jig, and directly into an SMA input port on the RP.

The bipolar devices were much simpler and had better power output, but with worse harmonic structure. I tried Fairchild PN2222A, and PN3563, and a China "331" marked as 2N3904, which is shown in the picture mounted up on the test jig. This is as simple as I could make it - the base has a 4.7K / 1K voltage divider and the emitter is at 47 ohms, unbypassed. At 47 ohms, the 2N3904 gets really hot, but can probably survive without a heatsink. At 22 ohms, power output begins to sag almost immediately without a heat sink. The heatsink I have on it is overkill at 47 ohms ( It's all I had in my travel kit ), but would probaby be about right at 22 ohms - it gets that hot. Output tank is a 2.87 uH toroid, resonated with a 60 pf trim cap, and RF is taken from a three or four ( can't remember ) turn link, wound midway on the primary for convenience.

All of the "331" semiconductor parts that I have used so far, both SMD and leaded, have performed satisfactorily.

I may make this a double tuned bandpass, or may not. Right now my objective is just to generate enough legally clean RF at the antenna connector, in the simplest manner possible, to have a reasonable chance of making QSO's and to drive an outboard amplifier, then go for a higher degree of sophistication in later TX strips.

The test jig seems to work well. I broke down and added an on / off switch. I'm satisfied with it and can't see anything that needs changing, other than a more colorful case.
 

Attachments

  • 40673.jpg
    40673.jpg
    752.9 KB · Views: 143
  • 40673 fft.jpg
    40673 fft.jpg
    95 KB · Views: 161
  • 331_2n3904.jpg
    331_2n3904.jpg
    993.1 KB · Views: 147
  • china 2n3904 fft.jpg
    china 2n3904 fft.jpg
    96.9 KB · Views: 61
Member
Joined 2016
Paid Member
Excellent.... I'm diy-ing a small loop antenna for 40-20-17. Using some 3/4" heliax for the loop. Got to either make or buy a big HV air spaced variable capacitor, then motorise it. Should be fun!
Just ordered a nanovna to tune it. Looks good for the money! I have Anritsu sitemasters for the day job but they don't go low enough!
 
... Just ordered a nanovna to tune it. Looks good for the money! ...

That does look nice. I might have to pick one of those up.

I just got through ordering a box of parts that when assembled should make another 3d printer, a small 200mm x 200mm Prusa I3 clone. That 500mm x 500mm monster is just overkill for most jobs. It takes a long time to get that big bed heated and stabilized.

I am cautiously hopeful I can have a simple 1 or 2 watts by this time next week. I have another case already printed out, except for the front panel, so I can start doing some trial assembly, doesn't really matter whether or not I have a TX strip to install.
 
Really ugly, but works.

This is the first low level amp converted bit by bit to the W7ZOI / W1FB "Beaverton" feedback circuit topology.

Compared to the simple tuned circuit amp, changing the tuned circuit to the ferrite bifilar load, lost about 5 dB. Adding the emitter degeneration added about fifteen dB. Adding the negative feedback from the collector to the base lost about 1 dB. I really can't tell that feedback does much of anything with my simple test equipment.

This is the same "331" brand 2N3904 transistor that are $0.81 USD / 100 on AliExpress, often with free shipping. The RP says it's down only a few dB at 49 MHz. At some point I may put in a better transistor, but for now this 2N3904, or whatever it really is, seems adequate. The screenshots are the same as before, through the 30 dB attenuator and then direct into the RP. This broadband stage is doing about 22 - 23 dBm and the xvtr board is 2 dB hotter than the test jig, so there should be plenty of RF to drive the final(s) to a couple of watts, maybe a bit more.
 

Attachments

  • broadband 49 mhz.jpg
    broadband 49 mhz.jpg
    94.6 KB · Views: 54
  • broadband 29 mhz.jpg
    broadband 29 mhz.jpg
    93.8 KB · Views: 47
  • broadband 14 mhz.jpg
    broadband 14 mhz.jpg
    97.2 KB · Views: 48
  • broadband 7 mhz.jpg
    broadband 7 mhz.jpg
    97.8 KB · Views: 59
  • broadband 2n3904.jpg
    broadband 2n3904.jpg
    739.8 KB · Views: 66
Last edited:
Okay, the transmitter test jig is working very well, and the first version of a xmtr strip will be complete and ready to put in service in a few hours.

The first RP screen shot is the Si5351 spectrum output at 14 MHz as shown by the RP, at - 2dBm as it come out of the test jig, and then direct via coax to the RP input.

Next, is the cheap China 2N3904 and cheap China "2N3553" (?). This actually does not look too bad, but the reality is when you look at it on the scope, it is clearly clipping that part of the wave below the zero crossing point, but that does not seem to be apparent in all of the FFT plots that I looked at. This was flat out to about 40 MHz, and down about - 4dB in the 6 meter part of the spectrum.

I do not have any real 2N3553 here with me, to compare these transistors against. It is possible that these are some type of 2N3553 or something pretty close to it. I have here known genuine Motorola 2N3866 and 2N5109, and compared to these transistors, the cheap China 2N3553 worked about as I would expect. The case finish is very different from the real TO-39 Motorola transistors that I have, and, you can't tell it from the picture, but the cheap China case dimensions are very subtly ( smaller diameter ) different. Whatever these transistors are, at first look they appear to be completely satisfactory for HF and 6 meters.

I wanted to get rid of the clipping, and it turned out to be the final part of the amp. Regardless, I changed the "2N3904" driver to a Fairchild JAN 2N2222A, and turned the current down in that stage. I replaced the 2N3553(?) with a genuine 2N5109 and also turned the current down in that stage. This resulted in a very nice broadband amp that with the -2 dBm drive, is ruler flat at +30dBm output over the entire range the RP can see. The o scope shot is not pretty, but that's what the input from the Si5351 looks like - just amplified, a lot.

As far as I can tell, for radio amateurs subject to FCC jurisdiction, and for xmtrs operating in amateur radio bands below 30 MHz and below 5 watts, spurious have to be down - 30 dB from the peak output. For transmitters below 30 MHz, and between 5 but less than 400 watts, the standard goes to - 40 dBm.

So with this small PA, it should be easy to meet the - 30 dB legal requirement. Maybe with just a 3 pole low pass. I'm going to clean up the wiring on the board a bit, while I decide whether to put the low pass on this board, or put it on a piece of perf at the antenna input on the chassis.

And maybe drive down to the Radio Shack to see if they have any 2 or 3 watt resistors and look around the store. They are a dealer, and have a lot of maker stuff. The 1 watt resistors in my dummy load / attenuator are not big enough ...
 

Attachments

  • Si5351 14 mhz.jpg
    Si5351 14 mhz.jpg
    92.2 KB · Views: 51
  • china PA.jpg
    china PA.jpg
    94.1 KB · Views: 41
  • china 2n3553 14 mhz zoom.jpg
    china 2n3553 14 mhz zoom.jpg
    96.3 KB · Views: 42
  • IMG_20200506_140011150.jpg
    IMG_20200506_140011150.jpg
    711.8 KB · Views: 59
  • IMG_20200507_072359945.jpg
    IMG_20200507_072359945.jpg
    704.9 KB · Views: 61
  • 2n5109 14 mhz.jpg
    2n5109 14 mhz.jpg
    103 KB · Views: 59
  • 2n5109 o scope.jpg
    2n5109 o scope.jpg
    77.2 KB · Views: 55
Last edited:
The low pass filtering turned out to be tedious. I could not get the Si5351 VFO to -30 dB with a three pole low pass in any configuration. The XVTR board has a lot lower spurious output than the VFO, so it probably would have been okay with the best three pole low pass I could come up with, but the insertion loss was really high, more than 3 dB.

A five pole low pass worked pretty well. Eventually I was able to get a -3 dB cutoff at 18.35 MHZ, 23 dB second harmonic attenuation, and 43 dB third harmonic attenuation with just a little more than 1 dB insertion loss. The o scope also looks a lot better with the square wave harmonics starting to be stripped off.

I used the calculator at wa4dsy.net for the starting point, and of course all of these calculators always generate odd ball inductance values that can only be obtained with a hand wound toroid by sheer luck, so there was a lot of tedious trial and error with capacitors to dial the thing in.

The ARRL website has a good paper on seven pole filters, especially designed for single ended output stages to maximize second harmonic attenuation.

Version 1 of the TX strip is done; it's attached to the test jig for safe transport back to home base where the rest of the radio is. The second pic is the three pole low pass; third and fourth pics are the five pole.
 

Attachments

  • IMG_20200507_195147553.jpg
    IMG_20200507_195147553.jpg
    767.4 KB · Views: 75
  • 5 pole o scope.jpg
    5 pole o scope.jpg
    75.4 KB · Views: 65
  • 5 pole low pass.jpg
    5 pole low pass.jpg
    92.3 KB · Views: 64
  • 3 pole low pass.jpg
    3 pole low pass.jpg
    93.1 KB · Views: 71
Last edited:
... the transformers are from Pulse Engineering CX2031, CX2041, CX2039, CX2047, 2074, 2147 etc or MACOM MABAES0060, MABAES0061 etc. ...

Sadly, I think they're too small for these proto boards I'm using, even if I go to the boards with smaller pads. I'll have to look for some bigger parts from a surplus house, and make the xfmrs by hand for now.

I have Nigel's PCB design program, but I am a long way from being able to make my own PCB's. I'm not very good at getting it right the first time, so PCB's may not be a good solution for me.
 
Sadly, I think they're too small for these proto boards I'm using

Mini-Circuits has baluns and other TF transformers in breadboard friendly sizes, but there are so many different types on their web site that it's hard to find what you want. I just bring up the RF transformers and baluns page input the desired frequency and impedance ranges, then mouse over the package column and look for the big ones.
 
Some progress, but not enough to mark anything off the to do list.

Completely de - Dentron'ed the SG-9, and wired it up ( I think ) to switch modes between SSB and CW, but I haven't tested any of this yet. The SG-9 does not have a sidetone function, so I'll have to build that. I don't want to hear relays clacking, so I'll also have to build a hang circuit. The need for a sidetone explains why the SG-9 did not mute its audio amp on transmit, and has an open, unused, input to that amp. Since my little audio amp functioned exactly opposite, with near zero chance of opening up the other input without major rework of the board, I went back to using the SG-9 audio, and just swamped the input from the product detector, so the high gain is less of a problem. Partly because of this snafu, I just rebuilt that entire control board, again, also getting rid of all the complicated transistor switching, and just made it as simple as possible, hopefully freeing up enough room on the board for a sidetone oscillator and relay hang circuit in the process.

I didn't add anything to switch or turn on / off an external PA, since I don't yet know what form one will take, and so far my track record of predicting things like this has been poor. I'll worry about that when I get to it. A generous contribution by Tubelab, of some big watt Motorola RF power transistors, has opened up a lot of new possibilities for PA's. The new relay on the board applies power to the TX strip when the SG-9 switches to transmit. I thought I could avoid this, and just leave the board powered up all the time, but the transistors just get too hot idling and eating all the power. The TX strip runs off regulated 12 volts for now, but I can move a jumper and let the board get the full voltage of whatever the radio is hooked up to, for higher voltage / current transistors.

That yellow LCD has a lot of contrast and is probably a good choice in a bright environment, but for now I swapped it out for a blue LCD to tone it down some while I finish this thing out. I've ordered some other colors / backgrounds to see what looks / works best.

The breadboard test jig for TX strips was pretty useful, so I made one for the whole radio. I put some clips down the middle for a wiring channel, but grossly underestimated the amount and bulk of wiring that would be in there, and ran out of channel well before I ran out of wiring, and there is still some wiring to be added to fully implement CW. There is also a lot of wiring around the front panel, and it is pushed as far forward as possible. The leads to the encoder are very long so I could conveniently use the VFO on the bench, and, to avoid rewiring it, the extra lead length is just stuffed into the void between the DDS board and the display. There is a lot of wasted space in these devices that use the 1602 type LCD.

I am now completely out of RG-174/U and need to order some more. The stuff I have used was I.S.C. wire and cable, part #2750; probably got it at Dayton a decade ago, and it is reasonably flexible, but if there is something smaller and softer that anyone knows of, I would be interested in that.

I have all of the needed VFO functions implemented through the momentary contact switches on the panel and the encoder. The plastic chassis makes it possible to size mounting holes such that if the opening is correctly done, parts like switches, jacks, etc., can simply be threaded into it, so you don't have to look at a bunch of lock nuts on the panel. The LCD displays are so cheap, about as cheap as four of those metric hex nuts at the local hardware store, that I think the best approach to finally mounting the LCD is to just glue it in place, with a trim bezel glued in place around it.

These printed chassis might look flimsy, but they are actually quite sturdy. The overall size of the jig is 216mm x 216mm x 53mm, IIRC.

I have not yet sourced all the jacks for the external accessories. The shafts on the encoder and volume control are a little long, but I think rather than replace those, I'll just print up some compatible knobs. I was concerned that all the bulky wiring would make it a formidable task to put the complete radio in a smallish package, but, looking at the breadboard, if I were to slice the thing in half along the wiring channel, and then fold it back on itself along that axis, it looks like it could work. I'll need to add more mm of vertical height; right now the breadboard is at the absolute minimum height to mount the VFO. I think front panel space to mount all the stuff that needs to go on the panel is the biggest issue to be solved. It may not be solvable.
 

Attachments

  • test panel, rear view.jpg
    test panel, rear view.jpg
    150.5 KB · Views: 159
  • test kig, top view.jpg
    test kig, top view.jpg
    235.1 KB · Views: 160
  • test jigs, side by side.jpg
    test jigs, side by side.jpg
    153.2 KB · Views: 158
Tubelab_com: I read some older posts were you mentioned that you sometimes used ADS but not always, how can that be??? What old school trick did you use?

When I was doing my Master's I would never dare to build an LNA or a power amplifier before simulating it in ADS or Genesys, for LNA I used S-Parameter files and use the optimizer to achieve a certain NF, Gain over a specified bandwidth, in the case of power amplifiers I used the Harmonic Balance analysis and would do simulated Load Pulls to find the optimum load impedance, after that I exported the design to Momentum to perform EM analysis using the method of moments, everything had to be taken into account, the substrate thickness, its permitivity, the copper thickness, etc... the optimizer was my daily bread!

How did you guys do this before all these tools?
 
you mentioned that you sometimes used ADS but not always

I worked in RF design starting in 1984 before there was MDS (forerunner of ADS). The computer processing power to run any of this was generally not available either. We did have access to Super Compact that ran on a large networked computer, but access was through a slow terminal, or worse, an old Teletype! I never used it, but my cube mate did.

One day we had a little wager. We both set out to design a 4 pole narrow band crystal filter for the IF circuit in a hand held two way radio. We analyzed the crystals individually to extract their S parameters, and agreed upon the parameter files.

He set down in front of the terminal, and I set out with a pile of Smith charts, and some reference books. Once I had a tentative schematic, I made a breadboard, then spent about a week with a pile of parts and a network analyzer.

After a couple weeks of work, we both had a "design." Mine was on copper with real physical parts, his was on paper......oddly enough they were very similar with only a few minor component value differences.

I used MDS and later ADS to design and optimize many RF circuits once it became powerful and capable for doing these tasks.

the optimizer was my daily bread!

I agree that it was often the "magic bullet" to cut weeks off of a complex design, once you find the right set of constraints and weighting factors to keep it on the right course.

I had to design a triplexer. a complex low pass filter network to connect three RF power amplifiers, each on a different frequency band, to one antenna. Only one transmitter would be active at a time, but all would be connected, for zero switching time. I started with three separate LPF's and three S parameter files containing the complex impedances looking back into their outputs.

I would set up an ADS optimizer run each night before leaving, and check it when I got in the next morning. Often it had stopped due to convergence failure, or it had gone into an infinite loop trying the same stuff over and over.

It took a couple of weeks of playing with the inputs to ADS, but a successful design eventually emerged. Another useful but temperamental feature was the Monte Carlo sim. It can find the weak spots in the design when real world parts with tolerances are used.

The last few years of my career were spent building high linearity RF power amps for use in LTE transmitters in the 10 to 100 watt range. Here, ADS will only get you close. Finding the balance between efficiency and linearity while meeting a strict emission mask takes some bench time. Again it's a pile of parts, some fancy test equipment, and an in house PC board lab. I could get a new two layer board in 4 hours, and a 4 layer board in a day. Want two layers of Rogers Duroid stacked on two layers of FR4, yes I could get that too.
 
I worked in RF design starting in 1984 before there was MDS (forerunner of ADS). The computer processing power to run any of this was generally not available either. We did have access to Super Compact that ran on a large networked computer, but access was through a slow terminal, or worse, an old Teletype! I never used it, but my cube mate did.

One day we had a little wager. We both set out to design a 4 pole narrow band crystal filter for the IF circuit in a hand held two way radio. We analyzed the crystals individually to extract their S parameters, and agreed upon the parameter files.

He set down in front of the terminal, and I set out with a pile of Smith charts, and some reference books. Once I had a tentative schematic, I made a breadboard, then spent about a week with a pile of parts and a network analyzer.

After a couple weeks of work, we both had a "design." Mine was on copper with real physical parts, his was on paper......oddly enough they were very similar with only a few minor component value differences.

I used MDS and later ADS to design and optimize many RF circuits once it became powerful and capable for doing these tasks.



I agree that it was often the "magic bullet" to cut weeks off of a complex design, once you find the right set of constraints and weighting factors to keep it on the right course.

I had to design a triplexer. a complex low pass filter network to connect three RF power amplifiers, each on a different frequency band, to one antenna. Only one transmitter would be active at a time, but all would be connected, for zero switching time. I started with three separate LPF's and three S parameter files containing the complex impedances looking back into their outputs.

I would set up an ADS optimizer run each night before leaving, and check it when I got in the next morning. Often it had stopped due to convergence failure, or it had gone into an infinite loop trying the same stuff over and over.

It took a couple of weeks of playing with the inputs to ADS, but a successful design eventually emerged. Another useful but temperamental feature was the Monte Carlo sim. It can find the weak spots in the design when real world parts with tolerances are used.

The last few years of my career were spent building high linearity RF power amps for use in LTE transmitters in the 10 to 100 watt range. Here, ADS will only get you close. Finding the balance between efficiency and linearity while meeting a strict emission mask takes some bench time. Again it's a pile of parts, some fancy test equipment, and an in house PC board lab. I could get a new two layer board in 4 hours, and a 4 layer board in a day. Want two layers of Rogers Duroid stacked on two layers of FR4, yes I could get that too.

Sounds great! I would love to read more stories like that, I guess that in the end there is no substitute for bench time.

I have some equipment that will work in the low microwave range, like 2 GHz, and I still have some Rogers substrates that I treasure like solid gold, so I can mess around with amplifiers and such, however my goal nowadays is to do more stuff in the lower RF rather than in the microwave range, a range in which microstrip is still not needed, you can still use plain old FR4, and handmade toroidal transformers or air core inductors rule.
 
A box of 3d printer parts made it in, so I assembled them into a cheap printer ( trickier than it looks, and I actually now know a bit about these printers ) and, to test it, printed up a new "chassis" for the test jig.

Then, a situation came up that I couldn't adequately deal with remotely from the lake, so I've had to come back in to the office for a few days, and, on a whim, tossed the newly remade test jig in the front seat of the car to bring it back to home base.

A couple of days left out in the front seat of the car, and the pictures tell the tale. This PLA plastic looks to be completely unsatisfactory for the local climate conditions. It's not even hot yet - probably has not been more than 90 degrees the last couple of days, which folks around here would consider mild for mid June. ABS plastic is supposed to be more heat resistant, but is not as user friendly to print ( high temps and noxious out gassing ), and I don't have any, but I guess I will have to get some and see if it can stand up to the local conditions.

The carcass is sitting on a G-500 Trans Oceanic for those interested in such things. I haven't done any more work on the radio, but I have been using it as a receiver every day. The receiver performance has, so far, been very satisfactory.

edit: btw, what do you guys think about the grey versus black ? I like the grey better, I think.
 

Attachments

  • IMG_20200617_091350412.jpg
    IMG_20200617_091350412.jpg
    128.7 KB · Views: 111
  • IMG_20200617_091405284.jpg
    IMG_20200617_091405284.jpg
    110 KB · Views: 59
Last edited: