Wow this created a bit of email traffic overnight...
To be honest I must be lucky as mine works, admittedly not first time... I had reversed the windings between earth and antenna initially, and forgotten the 1K resistor between C8 and C10.
However I don't have a copy of EMRFD to have checked circuit theory that I may not have understood anyway (I borrowed a copy once from MI5MTC).
I have never run SPICE, modelled a filter or a tuned circuit either. If I have to do a calculation for a tuned circuit it is usually done long hand on a scrap of paper. I'm not hot on theory as was never taught that much and tend to work things out by doing.
Asked a few things off list so rather than answer individually I hope this answers everyones questions:
The T1 to be honest I just wound all three windings in the same direction, right handed and held toroid in left hand
and passed wire up though the centre from underneath for each turn, no rocket science or theory involved.
The primary wound first then each of the secondaries wound on top one either side of the primary ends. Imagine pairs of wires come out on three sides of a square, primary in the middle (make sense?). I have since heard that the winding of this is the critical bit. See http://www.qrpedia.com/book/200812/flea-catalonias-answer-pixie for a picture, that is what mine looked like, don't think I would have wound it differently though. I'll take a close up image of T1 as asked by a few and it'll be on blog http://m1kta-qrp.blogspot.com/ when I can get back in the shack
The 2N3866 was from a batch from GQRP I cannot remember who donated them but they are not marked as such. I use them a drivers for a PA stage usually.
The trimmer capacitor there was probably only 10% movement between hearing signals, lots and then nothing again so the tuning is sharp.
I might investigate adding a smaller value cap in parallel to see if the tuning can be improved.
C2 and C3 I use a silver mica and a polystyrene, both were salvaged junk box components.
All the 10nF were box ceramics and had them for years.
All the R were 1/4W
The 1N4148 was actually a 1N914 I use whatever I have interchangably.
The LPF caps were either silver mica or disk ceramics (not NPO).
Q1 gets very hot without a heatsink. Ditto if use 2N2222A
SLA battery was providing 12.8V
If I used shack supply that is fine for normal use a little extra audio 'buzz' heard adding a 12uH choke and a 0.1uF to ground into the +ve lead reduced this.
LM386 used was a 386-1 and the headphones were elcheapo 99p ones from Tescos (similar to WallMart for those in US)
10uV signal source was from NorCal by WA6AYQ, W3CD. I had it callibrated for 50uV, 10uV and 1uV (I added an extra output), nothing difficult about building your own, it is just a 2N3904 oscillator, using HC49 crystals, I added an attenuation pad to obtain 10uV.
The manhatten pads were made using a Harbour hand punch http://m1kta-qrp.blogspot.com/2008/09/2n2-40-build.html
I have no idea why these were asked but:
The glue was poundshop regular super glue.
The yellow cable is solid core.
The shack temp (inside) when I built it was well below freezing as the concentrated FeCL etch solution I use for projects was like a slush puppy
and I added a pad or two extra and the super glue I used didn't always bond to the board (I had to heat up the board using a heat gun on the undersde a couple times to get them to 'stick'). Given your's and few others comments I bet if I tried to reproduce now I couldn't, I'll be jinxed now. Anyway boxing it up with a keyer (prob KD1JV simple keyer, no memories etc) and hope to have some trial qso with G3XBM who is sort of local, but 1.3W on 40m should be fine as that's almost qro hihi
>From: Stephen Wandling
>Sent: Mon, 11 January, 2010 20:48:20
>Subject: Re: [emrfd] Re: the flea, a 40m minimalist transceiver
>This is great to hear! As you may recall, I too built the "Pulga" or
>Flea xcvr. My trials and tribulations are to be found here:
>http://www.qrpedia. com/book/ 200812/flea- catalonias- answer-pixie
>I did get it working, "sort of", on both 40M and 80M. Reading through
>my notes there, its obvious that most of the problems were created by me
>and not the circuit. But, since learning was my prime goal, it all
>worked out great in the end.
>I was initially plagued by 'fake' 2N3866 transistors: low Beta. I
>finally got some 2N3553 and the improvement on receive was amazing.
>But, I kept toasting them during the tuning procedure.
>Eduardo had said in an earlier post :
>> Yes, FLEA is not very reproducible, you need all the components
>> and EMRFD book to understand what happens when it does not run.
>> This TRX is like the Arecibo Observatory receivers,
>> they are designed and manually tuned for performance,
>> not for reproducibility.
>So congratulations to you are definitely in order!
>> Built this last night based on information from GQRP SPRAT #137.
>> http://m1kta- qrp.blogspot. com/2010/ 01/flea-40m- cw-minimalist- tcvr.html
>> Nice fun little rig. RX signals from a Vertical 6BTV perfectly readable. The winding direction of T1 is critical though I wired the antenna connecton in reverse initially once corrected adjusting C3 brought in the signals.
>> Managed to use it on 7.025 (FT243), 7.030 (HC49) and 7.040 (HC6) and I have used one half of a turned 8pin DIL socket for the crystal between two manhatten pads so changing frequency is not a problem.
>> Using a small 10uV signal source and step attenuator I could receive signals down to about 0.4uV.
>> Changing C1/C2 and using a different C3 would mean bands 80m-20m should be possible anyone else had success on other bands that they might share with this design?
>> A 2N2222A also works for Q1 but with a much lower output c150mW.
>> Increasing V1 to 50K seemed to shift the RX offset (syntony=tuning? ) a lttle bit more.