After the building frenzy that resulted in the completion of the 40M WBR regen, I was left with another entirely functional WBR on Board #2. This one consisted of the beautiful AF Stage v2, hooked up to the funky but functional RF Stage v1. Like the packaged WBR, the input tank was set up for 40M, using the same tapped coil and caps specified in the original schematic.
Having a backup is great, but it was more appealing to modify the second board for use on 80M, the other band that works well with my tube transmitter. I knew the input tank had to change, and set about figuring out what the new values should be. I’m familiar with the concept of LC circuit resonance, but oddly it didn’t immediately occur to me that the resonant frequency of the tank determines the frequency you are listening to. I can be a bit thick that way sometimes. After some email conversations, light dawned, and just as an academic exercise I computed the resonant frequency of the tank using the original component values. Yup, 40 meters.
The inductance of the input tank is determined by the total number of turns on toroid holding the tapped coil. And, the capacitance is determined by the three paralleled caps: the fixed capacitance, the trim cap, and the capacitance provided by the varactor diode. There are a number of degrees of freedom here, multiple combinations that will result in a given solution – a true engineering problem. The best way to solve those is to make a model, and I used a spreadsheet.
The formula for determining the resonant frequency of an LC tank is:
and the commonly used empirical formula for inductance of toroid wound on T68-2 material is
(this is from a web calculator you can find here).
My model is designed around the following inputs:
- the value of the fixed capacitor, in pF
- the min and max capacitance of the trim cap, in pF
- the min and max capacitance of the varactor diode, in pF
(I looked up the capacitance range possible with the MV109 varactor from an online data sheet.)
The model calculates the inductance in uH of the coil, in increments of 2 turns. (To center tap the coil we need to have an even number of turns.) For each inductance value, I computed the minimum frequency possible (with maximum capacitance on the trim cap and the varactor), the maximum frequency possible (minimum values on both caps), and the predicted tuning range (difference in frequency with the trim cap at a minimum, computed for the minimum and maximum capacitance on the varactor.) Frequencies are reported in kHz.
After playing with the model for a bit, I realized that I could probably increase the tuning range a bit by reducing the value of the fixed capacitor. That would make the variable range of the varactor account for a larger portion of the total capacitance in the tank. Of course, lowering the overall capacitance in the tank will raise the frequency of the tuning range a bit, but that could be compensated for by choosing a different inductance value.
I had hoped to get a bit more tuning range out of the WBR. Although others seem to report covering all of the 40 meter band, with the original component values I was only getting about 2.5 kHz, I probably could have extended the range a bit below 7 MHz, but I wanted the band edge to be the bottom of the tuning range, to make navigation a little easier. As a result of playing with the model, I reduced the fixed capacitance to 75pF from the 82pF originally specified, and this increased the tuning range to about 3.5kHz, much better.
So how well did the model work? Okay…
The only part of the model that did not really work well was the predicted tuning range, which was much larger than what I actually observed. I couldn’t explain this. My best hypothesis was that the range of capacitance from the varactor is actually much less than the specified values from the data sheet. I don’t know how to measure the varactor in the circuit, so this discrepancy remained a mystery — until now.
As I prepared the materials for this blog post, I noticed to my horror that the constant I used for computing the toroid inductance was incorrect! I used 4.7; as you can see in the image above, the correct factor for T68-2 toroids is 5.7. As a consequence my computation for the inductance was too low, and I wound coils with too many turns on them. This has the effect of reducing the tunable range, as well as shifting it. Dang…
In the images below I am showing the corrected values in the spread sheet. The rows highlighted in green are what I should have chosen, those in pink are what I did use.
On 40, the predicted solution with the smaller fixed capacitance was:
and on 80 meters
You can download the spreadsheet from here:
WBR Tank circuit spreadsheet model
It was done using LibreOffice, an open source productivity package, and is saved in xslx format, which is compatible with MS Excel and many other spreadsheets. There are no macros in sheet, just formula calculations.
So this mistake cost me a little bandwidth – I won’t be tuning in CHU at 7850. On the other hand, the CW portion of 40 and 80 are a little more spread out, making tuning a bit less touchy. I don’t think I will go back and rip out the tanks to change them.
I wound 56 turns on the T68-2 for 80 meters, dropping down to 26 gauge wire to fit them all on. My wife patiently helped me count the turns while I wrapped the coil – she likes the aspects of ham radio that remind her of the fiber arts she enjoys. With her help winding the coil went smoothly; working on my own I would have been stopping to count again every three turns and might well still be at it at this point.
With the changes to the tank circuit installed, I tried out the 80m WBR before putting it into an enclosure. Hooking up my doublet antenna, and applying power, I instantly got CW pouring out of the speaker. Yes, I know that I had already tested this board on 40, but it was still a thrill to hear it operating on another band. I adjusted tuning range, 3500 – 3870 in this case, and got ready to put it in a case.
In the previous post I was complaining about the cost of a commercial enclosure big enough to include a speaker, but I had actually located one and ordered it back before Christmas. It is an LMB Heeger enclosure, made of painted aluminum, and it was back-ordered at the time. I did not receive it until mid-January, and by that time I had decided to skip the speaker for the 40 meter rig.
The case was not ideal. When I got it I realized that the way it clam-shelled together precluded mounting the speaker on the top as I had planned. This would force the front panel to be a little cluttered, all of the controls and the speaker would have to be mounted on the front.
Mounting the speaker itself posed mechanical challenges. I have a set of large Greenlee punches (a shrewd yard sale purchase a few summers ago) which have a diameter just right for the small speaker I used (excessed from a K1 after the battery holder option was installed.) I tried this punch out on a piece of single side PCB material. It worked fine dimensionally, but how to protect the speaker cone? I tried a bit of window screening, but was not satisfied with the way it looked:
And, the PCB board was warping under the stress of the mounting screws. Not acceptable.
I happened to glance at an Heathkit speaker sitting on a shelf in my shack went it struck me – I have a big metal shear, perhaps I could find some perforated metal sheet to cover the speaker. eBay to the rescue. There are dozens of patterns and sizes available, relatively reasonably. I chose a stainless steel sheet, in a gauge my shear could handle, and a hole pattern that results in about 51% open area. There were patterns with smaller holes that looked better, but at only 23% open. I figured that little LM386 needed all the help it could get, so I went with big holes.
There was one more small issue that needed addressing. Trying to spot the transmitter in the regen was challenging, because of the huge sensitivity of the regen detector. Experimenting with the 40m rig, I found that the grounding of the antenna provided by my T-R switch was not sufficient. Keying the transmitter simply blew away the WBR, resulting in a horrible groan emitting from the phones. The only way I could spot was to physically disconnect the antenna from WBR, and turn the RF gain to minimum. Then I could get a usable heterodyne to tune with.
To facilitate this, I added a toggle switch to the back of the case of the 40m rig, that disconnects the antenna jack from the circuit. This worked well; using the switch in combination with turning the RF gain down, I could get the receiver to the neighborhood of my transmitter. (It’s not exact, as the RF gain control also slews the frequency a bit. But close enough.)
Since I had discovered this before boxing the 80m rig, I decided to put the switch on the 80m front panel, right next to the RF gain control, and interrupt the circuit there. This turned out to be a bit of regen naivete on my part: on the front panel the switch has little effect. I guess there is enough signal leaking from the coax coming from the jack at the back panel to render the switch useless. I may try to rework this somehow. In the mean time, to spot the 80 meter rig I have to physically disconnect the antenna and move it at least 6 inches away from the jack. These regen detectors are truly amazing.
That’s it for regens for now. I am collecting parts and circuits to attempt a low voltage tube receiver, and will report my progress here as that project evolves. In the mean time, I’ve got a bunch of bench maintenance to do; if you build it yourself, you have to fix it when required, and I have some kit built equipment that needs some TLC. I am also working on an accessory for my tube transmitter which will be will be described here soon.
We still have a lot of winter to go.
So the WBR lives on 80M too. Excellent!
Yes, spotting can be a bit of a pain with a regen. It’s not so much that regen detectors are sensitive (which they are), but also because they crumble in the presence of strong signals. I think the fact that all that gain is concentrated into one stage doesn’t help either. With a superhet, the gain is distributed between several stages, so if signal leaks into one of the stages, it doesn’t get amplified so much. With a regen, there is only one stage, and if (or when?) signal leaks into that stage, it gets the full benefit of all the receiver’s available amplification in one fell swoop!
Now you’re entering the sacred chamber of regen folklore by getting into tubes. I’ve got your blog on my RSS reader and will be following your progress with interest!
I’m finding it workable with the antenna disconnect, plus turning down the RF gain and AF gain to a minimum. It’s a rather approximate business anyway, since the RF gain control moves the frequency a bit when you turn it back up (but not as much as I expected.) I have yet to make a 100% rx/tx homebrew QSO yet, but I will keep trying.
I’m really excited about experimenting with a tube regen. As much as possible I am going to try to build and test in stages, so I will report my progress (or lack of it) as I go along. It’s great fun.
Pingback: Another radio, another VXO | 73, de N2HTT