Table of Contents:

Background - Research

The Build

December 29

First QSO

The Flea Turns 80

BACKGROUND - RESEARCH

I was recently introduced to the Flea or Pulga, (Puça en Català ).  I was in the midst of my Pixie Project and didn't want to appear unfaithful, but the attraction was powerful. It is the creation of Joan and Eduado (EA3FXF & EA3GHS), two Catalonian hams who saw possibilities of expanding the Cubic Incher into a transceiver.  http://ea3fxf.googlepages.com/flea

Currently, I am awaiting parts I have ordered to build this little rig.  In the mean time, you can get a preview of whats to come by checking out the links below.

The Spice model for the transmitter portion of the Flea rig, and one for modeling the filtering of the receive section can be found here.  I think its great when people publish Spice models for their creations.  If you don't have LTSpice installed yet, you can go to pulga waveforms and see some of their Spice results.  Amazing selectivity in the "filter" model.

I believe that the values in Figure 5 for C1 & C2 in the SPRAT-FLEA.doc, below, are reversed.  I am still puzzling over the bandpass(?) filter they use in Figure 5. I have posted some questions on this to the Yahoo EMRFD group.

THE BUILD - "First we take Manhattan, then we take Berlin" - Leonard Cohen

Above is the circuit provided Joan Morros, EA3FXF, in the SRAT-FLEA.doc.  I have built the transmitter portion which is basically anything that is not directly connected to the LM386. Assume R8, R3 and D2 are removed.   I have installed R2 and V1, as this allows the oscillator to function without keying the transmitter.  The oscillator worked on power up, but the transmitter would not key.

This is Joan's Spice model circuit for the transmitter only.  This circuit runs fine in Spice and provides the expected outputs.

Note that this 'circuit', compared to the circuit in Figura 5, above, is simplified in some sections and expanded in others.  E.g., everything between the "A" and the "B", in the upper left corner, represents the crystal.  And I noted a few 'discrepancies' when comparing with Figura 5.  The first is that the values for C1 & C2 appear to be reversed.  Or, are they?

Here's where I started to get lost.  In an email on the EMRFD Yahoo site, Eduardo and Joan explained some things about the transformer (L1, L2 & L3) and the capacitive transformer created by C1 & C2.  I was pointed to an online calculator

Looking at the calculator, it appeared that C1 & C2 in the 'results' section had their labels reversed.  Checking the Help file confirmed that.  In Spanish, abajo is lower and arriba is upper.

In their email to me they gave the following data for using the online Capacitive Transformer Calculator:

>The R_loss of the resonant circuit is 14kohm.  L3:L1 transforms the 50ohm of the antenna to 14kohm.  L2 and L1 has 1:1 turns ratio.  The collector sees a 50ohm load impedance.  You can reduce L2 to get more power output (but the sensibility of the receiver could be lower).   C1 C2 is a capacitive transformer, from 14kohm to base-emitter resistance rpi=220ohm

If you open the Calculator above, you will see the values that I entered.  If you press calculate you should see "rp" has a value of 14001.9 ohms which I am assuming is the R_loss in the info from Eduardo/Joan.  The value for C1 (lower) = 505.458 pF, and C2 = 82.819 pF.  The first thing that bothered me was that these were different values than those found in C1 C2 in the circuits above.  So, was I doing something wrong when using the Calculator?  I don't know the answer to that yet.

Intuitively, it makes sense that the larger capacitor will be the lower one, given that it has the lower X_C (44.79 ohms) vs the upper with an X_C of 273.37 ohms.  Its like a voltage divider?  And, this is consistent with the example provided in the Calculator's Help file.

Questions: Why isn't this the case in Figura 5, above?  And, why does the oscillator in Figura 5 work with C1 at 270 pF, but won't work with C1 C2 reversed, as it seems to me they should be?

Shown above is my current build, with C1 & C2 not in place.

This my first Manhattan project, and I have enjoyed it.  I'm using a combo of 3/16strips of PCB that I snip into chunks, and ISA card edge connection strips as per this article. I used a cheap wood gouge to cut the V_CC strip up the right side of the left picture.  I used the ISA pieces, with sections chopped from IC socket connectors, to make the crystal socket and transistor socket.  I used a single ISA 'bar' for the junction for the diode in the foreground of the right picture.

The circuit uses a ferrite bead as an RF choke in the base circuit of Q1.  Above, its just to the right of the yellow trimmer cap, in the left picture, and to the cap's left in the right shot.  I have no experience with using ferrite beads as RF chokes.  I am wondering if my application above could be problematic?  I may replace the green magnet wire that runs through it with some 26 AWG hook up wire.  I might also try 'insulating' the bead from the PCB?

December 29, 2008

I continue to have problems getting the transmitter to work properly.  I have received some outside help and I will write that up soon.

I wound one coil for the transmitter, based on my thinking which I get into below the graphic.  This transformer did oscillate and did transmit.  But, the output into a 50 ohm dummy load was about 8 mW.  A little low for a 1 W rig.  Next, I wound a coil that mimicked the one in the graphic below, which made sense because the transmitter in the Flea is the One Incher, more or less.  This oscillated and when keyed, Q1 got real hot, but it was only putting out about 300 mW with a 10.25 Vdc supply.

Right now I am trying to make sense of the 'sense' of the windings on a toroid transformer.  Since the Flea's transmitter is lifted directly from the One Incher, I looked at that transmitter to get clues on getting the transformer right.

If I were winding the toroid for the circuit above (assume the toroid picture is not there), my thinking would be this:

• The "top" of each winding could be the start or finish of a winding.  I pick start.
• All of the windings will be in the same direction.  E.g., each new turn passes "down" through the toroid and proceeds in a CCW direction around the toroid (as does the primary above).

But, applying this to the schematic above and looking at picture of the toroid, I find this:

• The Antenna winding is wound in the same direction as the primary, but is connected in reverse to the primary.
• The Collector winding is wound in the reverse direction to the primary and is connected in the reverse direction.

So, my task is to sort out my confusion on this issue.  The Spice model circuit, back up the page, shows 'sense' marks on each winding.  Unfortunately, the Spice circuit is a 'representation' of the actual circuit, and even though I have had some professional counselling on the topic, I get very confused when I try to map the 'directions' from Spice model to the Figura 5 schematic.

QUESTION: Given a schematic, such as Figura 5, how does one proceed in winding a "correct" transformer, L1, L2 & L3, such that the "sense" of the windings is correct?  All helpful suggestions will be appreciated.

December 30, 2008

I posted a query regarding my question on QRP-L and qrp-l.org and got a few responses, off-list.  These responses helped me to realize that I had not framed my question very well, and that maybe the Cubic Incher graphic was more confusing than helpful.

I am now going to re-post to these two groups as follows:

I am coming to realize that I did not frame my question well, nor did my graphic help.  The graphic with the circuit and a view of its toroid is for the Cubic Incher (CI) transmitter.  I posted it because the Flea, which I am building, uses the same general circuit of the CI.  But the CI uses a T50-2 toroid while the Flea uses a T37-2.  The toroid drawing for the CI does not try and show the correct number of turns, but rather, it to shows how the turns are to be arranged and how they are connected.  I.e., is shows the transformer's geometry.

I have never built the CI, but I am presuming that the toroid for it should be wound and connected exactly as is shown.  [I have found two representations of the CI circuit and toroid and they are the same.

My question could be framed like this: If I was given a schematic of the CI transmitter, but it did not have the drawing of the toroid, by looking at the schematic, how would I know how to wind that toroid?  In other words, I am looking for the convention, if one exists, that would direct me to building the correct toroid.  And of course my REAL question is related to the Flea schematic which does not have a toroid drawing.

Now, I am aware of the "dot-convention", and you will note that in the Spice model circuit for the Flea transmitter, further up on this page, dots have been placed on the coil windings.  But, what I am having trouble finding is how the dot-convention relates to the geometry of the particular coil.

I am attaching the only graphic I found relating to dot-convention and the geometry of the coil structure.  I found this late last night and I am still studying it.  I am assuming that one can 'slide' the coils around on the magnetic lines of coupling to get a view of how it might be structured in a toroid?  Looking down on the top of the left coil, we see that its wound CCW from the top down, and looking down on the right coil we see that it is wound CW from the top down.  If we slide the right coil around on the lines of coupling, so that the two dots are adjacent, then we see that in this arrangement the coils are now both wound in the same direction, but their "dotted" ends are adjacent to each other.  If this observation is relevant, then I am still not sure how it extrapolates to becoming an answer to my question above.

FIRST QSO - January 4, 2009

Here is the complete Flea transceiver.  Its been a journey, and today I fired it up, called one CQ and N6MUK came back.  Paul had just completed building his Ten Tec kit he got for Xmas.  So, we were the first QSO for each other!  QRP to QRP.  He was running about 3 watts and the Flea was putting out just under 900 mW.  I got a 529 and gave him the same.

There is a bit of a learning curve in getting the Flea operating, and it seems to require having another receiver to monitor the CW tone to determine any chirp.  This adjustment is done with C3 and hopefully it won't need too much adjustment over time.  As you adjust C3 for the best keying, the transmit frequency changes.  While I still have to confirm this, I think that the Receive/Transmit offset changes as well.  The net result is that when I sent my CQ I thought it was way below 7040 Khz, but when Paul came back he was right on my frequency and had a tone of 500-700 hz.  This was comforting to say the least.

THE FLEA TURNS 80

I was pretty happy with the 40M Flea, but I was never sure that "I had it right".  Yes, I could hear signals and yes, people heard me, but things didn't seem just right.  For one thing, the adjustment instructions say: "increase C3 until you can hear band noise and signals".  That didn't happen, or at least I don't think it did.  The Flea adjustment is somewhat complex, with one control adjusting multiple things simultaneously.   More about that later.

I will say more about the MAS 'Minimal Art Session' QRP contest on another section of my pages, but basically it's an 80M QRP contest in Europe that sets your handicap by the component count of your rig.  Michael Rainey, AA1TJ, turned some of us on to the event, and suggested that we have some hams from this hemisphere partake.  And it was Michael that got me interested in the Flea.  But, for the MAS, it had to operate on 80M.

My initial effort to convert to 80M was sort of a "by guess and by gosh" type of thing.  Other than the crystal, the only circuit components involved were the 3 winding transformer and C1/C2.  How hard could it be?

I started with doubling the inductance of the T1 primary and then did a few mental calculations involving ratios, got the iron hot and got on with it.  Hmmm?  I couldn't adjust C3 to get oscillation.  Maybe I have the inductance too high?  So, off come a turn or two.  And then a couple more.  Eventually I get oscillation, of sorts.  So, I start playing with the secondary windings.  More turns for the collector winding?  Nope.  Maybe less?  No.  Reverse the collector winding?  

It went on and on like this until I heard a signal.  It was a net starting up on 3578.5 so I gave net control a call and he came back, had my call correct and gave me a TU.  But, the rig was very unstable, to the point that I couldn't duplicate yesterday's setup on the following morning.  So, I ripped out the xfmr and caps and started over.

I'd like to say that at this point, I got very scientific and did everything by the book, but I can't.  It's my nature to take intuitive leaps and so I took another, wrestled with it and was finally beaten in the end.  Only then did I get down to "designing" the rig.

PLAYING DESIGNER

While Eduardo ea3ghs provides an online Capacitive Transformer Calculator http://www.rootshell.be/~ea3ghs/ctrafo.php?f=3700&L=13.5&Qu=45&Rin=220&R... that will calculate C1 and C2, I had not been able to get it to reproduce their C1 & C2 results for their 40M design, so I had no confidence in using it to create an 80M version.  I did painstakingly reproduce the calculator into Excel which allowed me to study what the design task was, and also to use functions in Excel such as Goal Seek.

The task was this: working from the chosen inductance, which I chose to have the same inductive reactance as the primary in the 40M version, you calculate the capacitance necessary to resonate it at the crystal frequency.  This capacitance consists of C1/C2 in series, with C3, a 7-40pF variable, in parallel with them.  I assumed that I should allow about 25pF for C3.  After studying the ratio of their C1/C2, I found it to be 21%.  So, I built a spreadsheet that would calculate the capacitance to parallel a C3 of 25pF that had a ratio of 21%-79%.  And it worked!

SPICING IT UP

The next morning, I realized that I had another design tool already prepared for my use.  Their Spice model.  All I had to do was plug in my calculated values and run the model!  This gives you a typical example of how focused (over-focused?) and fixated I get with a single idea, to the extent that all other ideas cease to exist.

But having the realization was only a start.  For example, when I want to change their crystal model to represent the colour burst crystal I am using, I couldn't figure out how they were letting Spice know the frequency.  One of the labels in their crystal model was 40m760.  I knew theirs was a 40M circuit, but couldn't figure out the 760.  After reading a bunch, including the stuff on crystal models in EMRFD, I got it.   40m760 is the European method of labeling the Lm as 40.760mH.  Combining that with the Cm of 0.012pF gave a frequency of 7196 Khz, which explained why the Spice model had a frequency way above the 7040 Khz crystal in the regular schematic.

Another issue was that their Spice model had an extensive set of Spice directives which were gibberish to me.  Even when I had changed the values in the schematic to reflect my 80M design, it wouldn't work.  I finally copied the circuit (used Duplicate) onto a New Schematic sheet, leaving all of the directives behind, and after giving the new schematic the parameters of the transistor, I had it working.

You can probably imagine how much easier it is to make changes, lots of changes, to various windings and capacitors in Spice, by clicking and entering, than it had been with removing a little T37 transformer and changing windings and then soldering it back in, lots of times.  And then there were the things I learned about the circuit.

For example, Eduardo and Joan (Juan) had informed me on the EMRFD site that I could increase transmit output by taking turns off of the collector winding, but it might degrade the rigs ability to receive.  A few on the lists and off list, said no, that was wrong.  They told me I needed to increase the feedback (more turns) to the increase the positive feedback.  Well, when I was trying various combinations of windings for the two secondary windings, I noticed that when I increased the inductance of the collector or "tickler" winding, the output decreased.

It just so happened that on that morning I had been reading a little book called "Designing Analog Chips" that Michael Rainey had sent me, with the suggestion that the sections on modeling Crystals and Oscillators might be useful.  And in my morning reading I read that when a signal passes through a crystal there is a 180 degree phase shift.  Voila!  In the Flea, the input signal is coupled by the antenna winding to the primary tank circuit, where it then comes off capacitive voltage divider and passes through the crystal before it reaches the base of the 2N3866 transistor.  This was the missing link, or at least one of them to explaining the circuit's performance.

No contacts yet with the current version.  I still need to fine tune the secondary windings, where it's a tradeoff between output and local oscillator level for receive.  Then I can experiment with the capacitive voltage divider values.  One question that is hounding me is why there is such a difference between the output voltages in the Spice model and those obtained from the rig on the bench.

And in the back of my mind, I wonder if figuring out what's going wrong with my use of the Capacitive Transformer Calculator will open up new levels of performance.