Sunday, 30 June 2013

Tranfixed by Transverter

Last night, after one of my quite regular bouts of total inability to sleep, hatred of the covers, and before the Nytol had chance to kick in, I decided to think back over the receive sensitivity problem. So, I sat and thunked. Suddenly, the folly of my argument hit me. If I have to have the capacitor fully meshed, then my resonant frequency is too high, and no amount of widening the inductor turns spacing will make it lower! What I need is more turns, or more capacitance.

Tonight then, once the boys were in bed, I slunked my way back out to the workbench. Having already put the new coil on, and having during the day found a few moments to build the LO Buffer and put the SBL-1 DBM in place, I decided to properly connect the IF and Antenna connectors, and to run the tests again with a view to seeing what needed to be done.

Well, tonight I had to put a stonking -67dBm signal on in order to even hear the received signal, let alone measure the SINAD. Very odd. Both AM and FM were the same, and couldnt hear a thing in SSB. So, in frustration i disconnected the audio out from the scanner - and bang, stonking loud received audio tone!. It seems the funny issue I had with the scanner must be some sort of ground problem due to it being on battery and all the rest of the kit being off the mains. Test set up works fine until I connect the audio lead from the scanner to the 2955, which kills the received audio.

So, doing an FM SINAD test by ear, I found a perfectly acceptable -117dBm for 12dB SINAD. With the test set in AM mode and the modulation off, the slight frequency reading error of the scanner allowed me to hear the high pitched difference tone down to -127dBm just discernable, thats a pretty decent 0.091uV!. The audio tone was perfectly obvious at -120dBm/0.27uV.

But, heres the real surprise! I adjusted the test set to zero beat the scanner, and then backed it off a little to give a strong to my ears 800Hz tone. This was at 70.401MHz on the test set, 28.400MHz on the scanner. I then peaked the input tuned circuit by squeezing the coil turns together a bit, and adjusting the trimmer. It peaked best signal with the cap just short of fully meshed. So, I started turning down the signal, and down, and down, until the test set wouldnt go any lower! At -135dBm the tone was still very readable, had it been CW it would have been reasonable if not easy copy. Thats 0.039uV! I even swapped to the BNC output on the test set, but all that happened is it got better! (which shows that my N to BNC adaptor needs cleaning). It was even obvious back in the house listening from the open kitchen door!

video

I have to say I am amazed. Such a simple circuit, just a tuned circuit and an NE602 IC, and its sensitivity is awesome, even when used with lets face it a pretty poor IF receiver. The video above shows the set-up and the circuit under test. The warble is the signal from the Marconi 2955 test set.

It does of course mean that I owe Roger an apology, as I suggested that perhaps there was a typo in his schematic, as the diagram says a 35pF trimmer in this circuit, but his photo shows that his transverter has a 65pF one. Mine has the 35pF. I will stand by however my ascertion that 35pF is too small. Mine is very close to fully meshed, I think with a 65pF it would be more like half meshed, which I think is preferable in this case, I dont like things to be at the limit. I probably wont change the trimmer though, but instead add a 27pF or so fixed cap in parallel. So Roger, if you happen to read this, my apology on record, but also my gratitude and admiration for an amazing circuit!

Strangely, when I tried to check the LO Buffer, I couldnt get a frequency reading on my counter, neither could I from the oscillator (which was quite clearly running properly!). This is probably an issue with the counter rather than anything else, something to look at another day now.

Saturday, 29 June 2013

CaPs LoCk

After the 12 or so hours of preparation yesterday at home and today at the village hall, the couple of hours of Toms birthday party were brilliant. On arriving home afterwards, there was a small packet awaiting my return, marked Saffron Electronics, residing within which were 10 65pF trimmers.

Eventually, I managed to slip out to the workbench for a short while. Much of my time was spent putting a small cable entry chassis mount BNC socket on the end of a bit of RG-174, and sticking F-connectors onto a length of RG-59. Those two jobs done, I added one of the new trimmers to the crystal already mounted on the 10m WSPR transceiver board. I can at least now start building the LO proper, and hopefully have it running on the 14MHz fundamental soon.



Being rather knackered, I didnt wish to work on the LO buffer amp on the 4m transverter tonight. I have a particular aversion to dual gate MOSFETs as they are rather expensive four legged fuses, and with my antistatic wristband being 'somewhere' in a box under the bench, decided best leave that task for when im a little more awake. Instead, I decided to rebuild the input coil on the 70MHz receive side. This is a 7t 4mm dia air wound coil, tapped 1t from ground. I had built the original from 18swg wire, very sturdy, but found its associated trimmer capacitor had to be fully meshed for best results. Hmmm, I much prefer it when I can peak a tuned circuit and tune out either side. Being so sturdy, I couldnt open the turns up to reduce the inductance a fraction, and hence increase the resonant frequency. So, a new coil was would with 22swg wire, still sturdy enough, but now more manipulable. I suspect this inability to adjust both coil and cap, and hence not be able to tune the pair up to a high enough frequency, was responsible for the -87dBm receive measurement. I also this evening, took the liberty of running a 12dB SINAD test on the MVT-7100 at 28.4MHz, and it came in at better than -122dBm, so it does look like the loss is in the converter stage.





The photo above shows the transverter so far. The new coil can be seen on the right above the antenna changeover relay. On the left, roughly in place but not yet soldered on, is the SBL-1 double balanced mixer. Being a hot carrier diode ring, this can work at a higher level than the NE602 Gilbert Cell mixer used in the receive chain, and will form the heart of the transmit up-converter. However, before it can be fitted, the MOSFET buffer amp needs to go just above it.

Before going any further on the transmit side, I think I will connect the nice new coax connectors and retest the receive.The bulkhead BNC sockets mean I wont have the mismatch of a pair of crock clips during the test as was the case last time. But for now, im turning in for a good kip.

Thursday, 27 June 2013

If its not one thing, Its yer mother

The Transverter is coming along, added the antenna changeover relay, ex- of an RC-690, and have started the 10m WSPR transceiver. When I say started, I mean ive soldered the crystal on, then come to a shuddering halt at the realisation ive no damn trimmer capacitors left in the right values. Not even anything I can bodge into the circuit.

So, ive had to order some. Bloody hell! Might as well be ordering a kilo of Columbian Gold! I really must find some junk with lots of trimmers in to strip, because the price of these things is getting extortionate.

So theres nothing more I can do on the WSPR side, until these arrive. I can do a bit more on the transverter though, although not much. Looks like i'll have to instead finish of some long overdue old projects.

4m Transverter - Receiver success

After a bit of lack-of-sleep brain addled cocking up, I have managed to get the coils for the receive side of the transverter wound properly. A quick, but not exactly accurate, test was set up, using the Marconi 2955 test set as signal generator on 70.4MHz, and the MVT-7100 scanner as the receiver, tuned 28.4MHz. With a source signal with 1kHz 45% AM, this little circuit worked nicely, giving a 12dB SINAD (roughly, by ear) of about -87dBm. With the modulation off, and the scanner in USB mode, a tone could be detected by ear with a signal of about -102dB. The set up was far from ideal, the input from the signal generator was connected with a croc-clip lead, and the scanner connected with an array of various adapters to connect to the TNC connector used to wire up the converter output. The scanner didnt even need tuning a couple of kHz away, as being a scanner its frequency resolution isnt great to start with. It was also a bit intermittent, did some odd changes when I touched the case. I suspect a proper test set up will be much better. I did notice that the trimmer capacitor in the receive circuit was best set at max, I may try a bigger value in here to see what effect it has, or maybe widen out the turns of the coil.



Onto the antenna and power switching arrangements now, which I want to have done before starting on the transmit side.

Before however, im going to pinch yet another of G3XBMs designs, and make a start on the 10m WSPR transceiver. I have the necessary toroids and crystals now, so first job is to make a rough draft layout, and then build and test the oscillator and doubler.

Wednesday, 26 June 2013

Old and New

Spent an almost pleasant couple of hours salvaging components from junk equipment this morning. Mostly the usual suspects - electrolytics, small transistors etc, but since the big beast im dismantling is an old analogue TV transmitter, theres some real nice RF components as well coming out, one board has a lot of small ferrite binocular cores, and several nice Toko coils. Sadly, most of the really nice parts are very well soldered to the groundplane, through hole plated, and surrounded by masses of SMT parts. I really think I need a bigger rated soldering iron for taking these off. I did try with the heat gun, but its not precise enough and just torches the boards.

Whilst out doing this, I missed the post lady, and one coming back into the house found that a much sought after parcel from Graham G3MFJ had arrived - G-QRP club parts! A sexy collection of toroids and Spectrum coils (modern Toko 10K), some needed semiconductors, and of course - the crystals! I can now not only carry on with the 4m transverter, but can start the WSPR 10m transceiver as well!

The only parts im desperate for, sadly Graham doesnt stock - trimmer capacitors!

Oh well, still a few in the junk boxes, and theres a rally in the not too distant future...

Off now to cook up some Frankfurters, shrooms, onions and cheese. Mmmmmm

Friday, 21 June 2013

Tub Thumping

Butter tubs, that is. Dozens of them, mostly 500g, the occasional 1kg, and a couple of ice cream tubs.

Why? I'm adding to my component storage. I use butter tubs, in particular Sainsburys or Flora, because they stack well and are quite strong.  It makes separating different component types easier, and keeps them tidy. Which is something I really need to do better! Containers added to my shelves now for toroid cores and ferrite beads, trimmer capacitors, coil formers (ive included Toko/Spectrum 10Z and S18s in there), battery holders, relays, plus some others.

Its quite amazing just what a huge variety of different components there are. Its easy to lump, say, small signal transistors into one box, but thats still awkward - what about bipolar/FET? general purpose/RF? All I can say is I need a lot of butter tubs!

Anyway, enough about my storage difficulties. Some RF parts on order from G-QRP club sales today, mostly for the 4m transverter, but some to make a start on a 10m WSPR beacon transceiver. Just realised though, that if I decide to follow Roger G3XBMs design, im going to need half a dozen T37-6 toroids. I have T50-6 and T50-2, and could have ordered the 37's today, if i'd bothered to check the schematic! At least I have the crystals ordered!


Ive also had an idea regarding the secondary antenna for the show. I had planned some form of vertical array, but im wondering now if a cobweb might be the answer? It will go on the 5.4m mast, so not sure if the height is good enough.

Wednesday, 19 June 2013

Clock On!

After repairing the broken former, regluing pads that came loose during experimenting, and discovering the Spec An is 4MHz off, the local Oscillator for the transverter is now running where it should be - 42Mhz.

Despite being fairly certain the series inductor is the same type as specified, and after removing several turns to get it to 10t, it still wouldn't oscillate correctly. I then took it down to 5t and was wayyy over! So, a deft rewind, followed again by stripping off one turn at a time and testing, plus the expedient of changing over to the Frequency Counter from the Spec An, led to the discovery that free running around 42MHz needs 7t on this former. Adjusted to within a few hundred Hz, and took the shorting link out -Bang on!

Next step, then, is to build the Receive Down Converter. This is a simple NE602 circuit. The only down side here, is that other than feeding the Signal generator into it, I have no horizontal antenna for 4m with which to try and receive an off-air signal.  I feel a halo or dipole coming on...

I still need to verify the instrument readings against listening for the oscillator on a GCR.

4m Transverter

The Larkspur mast is nearly refurbished, just one section left to repaint, so ive turned my attention to another project - a transverter for 4m. This is the G3XBM QRP design.

Its taken me quite some time to even get started with this, as the 42MHz crystals are not exactly cheap nor easy to come by. And once I had the crystal, I found I didnt have the series inductor. However, a kind chap at the vintage forum has supplied said part, and so on we go.

Working with 'Manhattan' construction, I have the Local Oscillator constructed. As is often the case, first time out it failed, and in my tired state trying to fault find last night at gone 23:00, I confused myself with the display on the spectrum analyser, and so the one time I did get it to oscillate, I thought it was just a false reading.

Having correctly started alignment this morning, I find it free runs at 27MHz. A quick check of the inductor shows it has far too many turns on than specified, so next task is to take a few off and get it free running around 42MHz, then the short can be taken off of the crystal.

Just started removing the excess turns, and have cracked the damn former off the base! Out with the cyanoacrylates...