Mixer

The worst excesses of Christmas being over, I returned today to the Valve Receiver Project (Blogs passim)...

the last stage to be constructed is the Mixer - a modified version of the schematic on Harry sm0vpo's site is shown below...



I had a slight panic on looking at the schematic - for I realized that I had mis-read Harry's original instructions and not noticed that the valve is a HEPTODE (not a Pentode).

I had a rummage through the junk boxes and came up with only two heptodes - one an ECH84 which looked as if it had been to hell and back and the other an EH90, which looked as if it never had been used. Fate made the choice for me, as I had already drilled the chassis for a B9A valve base (as used by the ECH84) whilst the little EH90 has a Miniature 7-pin base. For the record, here's the pin-out for the ECH84...



This valve is similar to the ECH81 which Harry used - another Heptode/Triode, the Triode section of which I'm ignoring.

The mixer stage is simple to build - here's a shot of the ECH84 doing its stuff...



Readers may recall that I needed to find a twin-gang capacitor to use as local oscillator tuning AND first stage resonant grid circuit - the best I could do was the ugly triple-gang unit seen below. Actually I like it - it looks appropriately "old-fashioned" and clunky.



Whilst I was in metal-working mode I fitted the remaining controls to the "front" - an RF gain potentiometer (extreme left) and what will eventually be the band-change switch...



The receiver to date is seen in all its (limited) glory in the next photo...



The good new is - it works! It doesn't work very well (yet) but I can hear CW on the 80m band - which was my target all along.

Next steps will be adding trimmer caps to align the two tunable circuits and investigating the low sensitivity. For now, I'm just going to bask in the warm glow of thermionic radio reception!

...-.- de m0xpd

Local Oscillator

What better way to hide from preparations for Christmas than to build the local oscillator for the valve receiver project...

The schematic (modified from the original on Harry, sm0vpo's site is shown below...


Like Harry, I used Zener Diodes to derive the stabilized voltage for the Hartley oscillator (although I did find a couple of suitable voltage stabiliser valves (CV287) whilst sorting through a box of valves this morning).

The underside of the local oscillator circuit is seen below...


It worked first time - but it now works a whole lot better after I learned something really significant...

Whilst sniffing around in that box of valves mentioned above, I was adding to my catalogue of valves "in stock", in which I like to include links to a data sheet. Whilst doing that, I noticed something I had never realized before.

The ECC81 used in the local oscillator is also widely know by the equivalent name 12AT7 (as mentioned in describing the BFO). What I never twigged before is the obvious fact that the valve should use a 12V heater supply (obvious because I knew all this from the 6SK7s etc used in my Paraset - I was considering replacement with 12SK7s to make battery operation easier). Accordingly, I had wired up the heaters for the two 12AT7s in this project for only 1/2 the rated heater voltage - amazing that they still worked at all (and another possible explanation for the fun and games I had with the BFO, as described here.)

The clever thing is this - the 12AT7 heater is "centre-tapped" so it can be powered from 6.3 V, if wired appropriately.

There really is no excuse for my ignorance - apart from the clue being very much in the name "12AT7", it is there as bold as brass in the data sheet...



Note to self (and any others who happen to be reading): this is also true for the double triodes ECC83 / 12AX7 and ECC82 / 12AU7, widely used in audio frequency gear.

Now that the heaters are running at full tilt (and glowing in a wholly more satisfactory manner) the oscillator runs fine...



I'm hoping to give this receiver 3 bands - but mostly I'm interested in the 80m amateur band, so here it is running at an appropriate frequency...



Readers won't be surprised to hear that my coils were wound on the sewing machine bobbins I've used previously...



... you can see the coil at bottom left. You can also see that I only have a single gang capacitor - which is going to be a problem (I need a twin-gang device; one for tuning the LO and one for tuning the resonant circuit at the input). That will have to be dealt with in the future - for now the carol service from from King's has just kicked off...

..../.-/.--./.--./-.-- -.-./..../.-./../.../-/--/.-/...

...-.- de m0xpd

Toroid Library for EAGLE

*** Stop Press ***

My Toroid Library has been published at CadSoft's Eagle Website.

It seems to me that one of the components we radio homebrewers and QRP enthusiasts use most often is an inductor(/tapped inductor/transformer) wound on a Toroidal core. I was surprised to see that these components are poorly catered for in the Eagle PCB software suite (Eagle is of special interest to misers like me as it is free). So - I knocked up a library...

The Library includes several devices (inductors, tapped inductors (/autotransformers), transformers, trifilar coils, etc)...



and the packages are dimensioned so as to be useful with 0.37 and 0.5 inch diameter rings...



You are welcome to try the Library although there is no guarantee that it is worth anything or suitable for any purpose. All I can say is "it works well for me".

If you do use it, please drop me a line and tell me how you get on.

...-.- de m0xpd

Funster PLUS

I'm taking a break from the valve receiver project (Blogs passim), to concentrate on a little 40m QRP rig I've been playing with this year - the rig was working but needed some kind of enclosure. Just the sort of job to take my mind off valves & the cold snap which is gripping Manchester at the moment. I need the enclosure to get ready for the rig's debut operation in this year's Winter Sports.

Here's what I produced yesterday and today...



The rig is a development on the excellent "Funster 40", which was my first homebrewed rig. This simple, rock-bound transceiver is featured on Todd Gale’s excellent website.

I built Todd's little rig when I was a foundation licensee and benefitted greatly from the experience. My version of the Funster 40, seen in the picture below, followed Todd’s “words and music” quite closely, but I added a discrete alternative to the LM386 AF amplifier (which wasn’t in my junk box) using the 2n3904 device Todd views as “the ultimate popcorn part”.



I wasn’t (as I understand clause 7(2) of the amateur license) legally entitled to operate the rig on air until gaining my intermediate license, when I found it to be fun yet frustrating to use.

After a first qso with Billy, m0jha, whose qth is only a couple of miles from the two 2n3904s sweating in the Funster’s 1.5 Watt final stage, I managed qsos with PA and DL stations. After initial elation, the frustration soon set in and the rig was destined for “improvements”. These improvements in no way impute shortcomings to Todd’s original concept – indeed, the transmitter section remains exactly as Todd (and Wes before him) prescribed and the 2n3904 “popcorn” device is used extensively in the rig I now call – in honour of Todd – the “Funster PLUS”.

I found the frustrations of the original Funster to focus upon two aspects; the rock-bound local oscillator and the receiver.

Being limited to operation on one frequency has its limitations, as I know from operating my Paraset. However, I had been frightened off liberating the rock-bound Funster by talk of the problems encountered in making stable VFOs. This rumoured difficulty was briefly supported by some of my early experiments with Colpitts oscillators but immediately dispelled when I built a copy of the VFO designed by Roy Lewallen, w7el, for his “Optimized Transceiver”. This circuit, which uses the Hartley oscillator architecture, immediately gave me a very stable VFO and opened up the Funster to operation across the entire CW section of the 40m band. The schematic of my VFO is shown below...



and here's a picture of the VFO (with its lid off, so we can see "the works")...



The lid (indeed the whole mechanical assembly, which is very sturdy) is important to achieving the low drift I wanted. The measured turn-on drift of my version of w7el’s VFO is shown below, after which the oscillator is effectively as steady as the original crystal LO – until I turn the tuning control!



The eagle-eyed reader will notice that the frequency data in the graph above is quantized to a resolution of 10 Hz - yet I can measure frequency more accurately than that. There is an explanation - read on...

In order to enjoy the stable local oscillator to the full, I needed a frequency meter to proudly display the constant, unerring output. I had already bread-boarded the excellent, simple solution offered by Francesco Morgantini, ik3oil, so I produced a PCB and made a dedicated version for inclusion in the Funster PLUS project. Here's what it looks like at the moment ...



With the local oscillator now free of its rock-bound shackles, the restrictions of the receiver section of the original Funster became harder to ignore. Todd makes no excuses, acknowledging that the crowded 40m band is a “worst-case scenario” for the Funster 40’s single ended BJT product detector.

I found the “open door” lack of selectivity distracting and wanted a receiver with slightly more sophistication, whilst preserving something of the “Funster” spirit. I needed to look no further than w7el’s Optimized Transceiver, already mentioned.

This uses a double-balanced diode ring mixer in a DC receiver, with some “selectivity” won by active filtering in the AF path.



Performance of a first ugly prototype and a later PCB-mounted version was surprisingly good, and the Funster PLUS was now a genuinely usable rig, on which it was suddenly easy to make contacts all over Europe.

My Funster PLUS includes two more features which make it easier to use, which have been described in an earlier blog . The first is a simple PIC-based keyer, designed by Eamon “Ed” Skelton, ei9gq, and published in Ed’s homebrew series in RadCom (also available at Ed's website). I modified the schematic and the PIC program, to add an interface for a straight key and a sidetone oscillator. Unfortunately (as an amateur musician and professional acoustician) I am rather fussy when it comes to CW sidetone generators – the crude buzzing of an astable irritates me, as does any gross de-tuning between the sidetone frequency and the nominal transmit frequency offset. Accordingly, I “borrowed” another sub-system for the Funster PLUS, which I had used previously in my SDR rig...

We all pay attention to the spectral consequences of keying our RF emissions, but few seem to pay similar attention to the keying envelope applied to the sidetone. Without such attention, even the best sidetone oscillator is compromised. G Forrest Cook, wb0rio, has published a nice circuit in which an oscillator signal is switched by a voltage controlled amplifier, capable of applying an exponential rise and fall rather than the crude switching directly from the two-state keyline. Here's my ugly prototype version...



This gives a clean sidetone envelope, shown in the LTSpice simulation of my version of wb0rio’s idea below, which sounds great – even to my fussy ears!



The ugly development version of the Funster PLUS is shown below...



I made PCBs for all the main sub-systems in the rig and had my first QSO on the Funster PLUS with Stefan, DL4ST - he gave me a 579 report - not bad for a tangle of wires on the bench...



Finally - yesterday and today - I squeezed it all in some kind of enclosure (or, at least, half an enclosure - I still don't have a lid).

The "box" is made from single-sided copper clad F/G board, soldered together. I have two AF gain controls, main tuning, RIT and the keyer speed control on the front...



It doesn't matter what rig I try to plug into what antenna - I always have the wrong connector. So - this time - I put two antenna sockets on the back - a BNC and an SO239...



There are (right to left) 4mm power sockets, a single power connector (not yet fitted), a 3.5mm jack socket for a paddle and one for a straight key (not yet fitted), a 3.5mm socket for the output of the AF amplifier (for speakers or phones) and a phono socket for line-level AF out. Then the extravagant pair of antenna connections!

Looking from the side, you can see the stack of three PCBs (Tx at the bottom, then keyer/sidetone generator, then Rx on top) - at least you can see the top two! The Frequency Counter PCB lives atop the VFO box. The whole thing is almost impossible to put together - let's hope nothing breaks as it isn't designed for service!



Now we're all set for winter sports - and, as I look out of my shack window, I am pleased to say we certainly have enough snow HI HI.

...-.- de m0xpd

IF Amplifier

Now we have some IF Transformers (OK - they're not as good as commercial items but the price was right and I learned again some tiny fraction of the wisdom of the ancients) we can go ahead and make the IF Amplifier stage.

Here's the schematic (modified from the original on Harry sm0vpo's website)...

It went together very easily - and I could confirm operation by applying a signal to the primary of the first IFT (via a 470 kOhm resistor) and looking at the response at the secondary of the second IFT. All very interesting, but it would be much more fun to connect up the detector and LISTEN to the response...

I wired it all up, only to discover that the BFO was not working. It used to work - you know it did because I showed you pictures. What had gone wrong?

After some messing about I discovered that the problem was quite a subtle one...

Although the BFO was working fine before I added the IF Amplifier stage (a single EF80 pentode), the additional valve pulled down the heater voltage slightly. This drop in heater voltage stopped the BFO - I could get it to run again by dropping the HT supply to around 180V (where it would start and run) but if I pulled the HT back to the usual 230V, it stopped again. Similarly, if I lifted the heater volts for a few seconds, it would start, but stop again when I returned it to nominal value - interesting!

Swapping the 12AT7 with another from the "junk" box produced an oscillator that would run robustly.

Vincent 2e0zvl came up with what sounds to me a credible explanation; cathode - heater leakage. The (Colpitts) oscillator circuit runs with quite a high impedance cathode network, which could be swamped with such leakage between cathode and the heater (effectively at ground).

OK - now it's running I ought to be able to hear a 456kHz input to the IF amp modulated down to AF. Yes - sure enough - it is there, a nice tone which moves with the BFO control as expected and which allows me to hear the "selectivity" of the system by changing the RF sig gen frequency around 455 kHz. Fine - until I take the 'scope probe off the BFO oscillator output, when it goes dead again...

Time to make a fool of myself - it took me ages to realize that the 'scope probe was changing the tuning of the BFO to such extent that removing it took the demodulated AF tone outside the passband of the IF (and/or my ears and/or the AF amp/speaker combination). It was still working fine - it just needed tuning without a probe hanging on it. FOOL!

It is a sort of laziness which I guess many of us are guilty of - I am so used to throwing a buffer stage on an oscillator that I wouldn't dream a 'scope probe would disrupt things - FOOL! A buffer in a modern rig might be just another transistor, soldered in place in a matter of seconds. But a buffer in this rig would mean another valve, more heater current, metalwork, space, time - forget it!

The greater part of the work - and the fun - in this project has been in learning exactly this kind of thing. Stuff that was common, everyday practice to the old masters, now waiting to be re-discovered by fools like me!

Here then, finally, is the project to date with two homebrew "IFTs" proudly sitting on the side of the chassis...


The screening cans are made from baked bean cans - it is difficult to find cans without "corrugated" walls nowadays as I discovered walking around Tesco's at the weekend - why is that (rhetorical question - please don't write).

Here's another view (just can't help myself)...



The IFT toward the front of the chassis is the IF Amplifier's load, whilst that to the rear loads the mixer - I wanted to leave the area in the "middle" of the chassis clear for the VFO's tuning arrangements.

OK - next it is the VFO - but I don't know if that will happen immediately. The festive season and "Winter Sports" might push this project onto the back-burner for a while.

...-.- de m0xpd

The "best laid plans"....

OK - the trimmers arrived promptly, courtesy of Peter G7JAB's excellent service (visit his website here - usual disclaimer). I quickly got them soldered atop the waiting IFTs - only to be disappointed ...

The measured Q was pathetic - I was going to be making a receiver with no more (lack of) selectivity than the "wide-open door" of the Paraset or one of the direct conversion Rx's on my QRP rigs at this rate!

I decided to investigate the problem one coil at a time, rather than try to play with the complete transformer (that way I could eliminate the central glass-fibre rod or the PCB end plates as the origin of the pathetic Q).

I took a single coil, wound on one of the sewing machine bobbins, and resonated it with a 100pF capacitor...


For a benchmark, I also took the nearest axial choke I had in stock (a 470uH device) and paralleled it with another 100pF capacitor...


I drove both networks from a signal generator, through a 470 kOhm resistor and monitored the response on a 'scope (via a conventional X10 probe). The resulting (normalized) responses are shown below...

(sorry about the frequency axis - Excel is very bad at handling logarithmic graphs)

This - it seems - is why my IFTs were as selective as thirsty man at Oktoberfest. But why is the Q so poor - what's wrong with my coils?

The first straw I clutched at was the idea that the plastic of the bobbin was causing the loss (hardly likely at such reasonably low freqs but I was clutching at straws), so I tore off one side of the "bobbin" and re-measured...


Result: no observable difference (yes - I realize that more loss may be created by the "middle" of the bobbin - but this wasn't nearly so easy to tear off).

There then followed a series of "I wonder if..." type conjectures, (during another trip to HB-land, when all I could do was speculate). I wondered if it was the coil geometry (but it looks close to the ideal "Brooks" form), if it was the wire (but the d.c. resistance was less than 2.5 Ohms and I'm too clumsy to go much finer), if it was the...

I turned in frustration to those who know more than me - in this case that's a large group! I had some interesting and inspirational discussions with Richard G8UNO and Albert G3ZHE, both of whom pointed to capacitance (self- or stray-) and so, on my return to the UK I made some similar coils on the same bobbins but with fewer turns. The results (see graph below) are an improvement - but nothing to write home about. It wasn't until I tried some ferrite cores that I started to get improved Qs, approaching (but not equaling) that I'd seen with the axial choke.

You can see my original bobbin coil (with 300 turns) and a 100 turn air-core on the same type of bobbin. Also shown are the results for the same coils with a ferrite core (a "tube" from the junk box, with O/D conveniently close to 1/4 inch and length about 3/4 inch).

I decided to run with 100 turn coils on a ferrite core for a new batch of IFTs. The graph below compares the response of my original IFT to one of the newer ferrite cored models...

This isn't "the final solution" - it is just the last one I recorded!

Now for a little more tin-bashing to make some covers and we can move on to the IF Amplifier

...-.- de m0xpd

IF Transformers

OK - change of plan (this is a hobby, after all). I was going to build a "Wobbulator" - but it occurred to me that I wouldn't have anything to wobbulate. So - I decided to go right ahead with building the IF Transformers.

I intended to use the sewing machine bobbins sourced by my wife in a layout following that proposed by Harry sm0vpo and others (on Harrys website, look under "Data Technique" for the page "DIY IF Cans" - I won't give a direct link 'cos Harry's going to be moving stuff around, judging by his current problems with Telia).

The first task was knocking up some end pieces - I decided to etch some PCBs as it would be easier in the long run - here's the board layout...

and here is the first batch of PCBs (made using my standard "toner transfer" process)...

The coils, wound on the sewing machine bobbins, were going to sit on 1/4 inch fibre-glass rod - here's a dry assembly...

Glass-fibre rod (and PCB material) is a pig to "machine" - I've got itchy hands from the little splinters!

The coils were wound onto the bobbins - Harry's words and music suggest that you should "wind 335 turns of 32 SWG wire for both the primary and secondary" - I knew (from yesterday's fun and games) that the bobbins hold about 300 turns of #30 wire - so that will have to do.

The resulting assembly starts to take shape...

I fitted a 100pF capacitor to each coil and measured the response with an RF generator and 'scope - the devices peak at about 550 kHz so, once I add the trimmers, everything should be fine. Here are the finished IFTs (sans trimmers - which I'll have to mail-order)...


I got an old tin out of the dustbin - no, credit where credit's due, my WIFE got a tin for me - and I roughed out a screening can - easier than I expected (although it is still a work-in-progress)...


Now we're all set to build the IF stage and the rest of the receiver... guess what I'll be doing next weekend!

...-.- de m0xpd

Beat Frequency Oscillator

Having completed the audio stages, it is now time to look at the first high-frequency stage - the BFO.

Harry's circuit (edited from the full version on his website) is shown below...

The new challenges in this circuit are the coil and the variable capacitor ...

I found a 75pF variable in the junk box and removed some vanes to make it closer to the specified 10pF. However, the coil was going to be harder...

Harry specifies "205 turns on a 1/4" (6mm) former fitted with a ferrite slug". I thought about how I was going to make the IF transformers - they would need some formers too. Inspiration struck in the form of plastic sewing machine bobbins - my wife has loads of them pre-loaded with different cottons, so I asked her for some. Instead, she stopped by the sewing shop and got 8 for me - at 10 pence a throw they seem a bargain...

I lashed up an impromptu "coil winder" on my lathe and found I could squeeze about 300 turns of #30 enamelled Cu wire on the "former". Rather than take off a hundred turns to get into the right ballpark for the BFO, I went right ahead with the fully-loaded coil (of course I had to change the tuning caps in the (Colpitts) oscillator to get to the required 455 kHz).

You can see the system being tested below...

(couldn't resist turning the counter/timer on and showing 455kHz!)

The output from my BFO is about 1.4V - see the 'scope trace below...

Harry warns that it may be neccessary to reduce the 22k cathode resistor in his schematic to get the oscillator to run - I reduced mine to 6k8 (it probably could have been higher - but I found the 6k8 resistor on the bench and fate intervened).

Now I have the BFO, it is time to turn my attention to turning those sewing bits into IF transformers. But, before I do that, I'm going to make a "Wobbulator" - watch this space!

...-.- de m0xpd

AF Pre-Amp

Back to some more work on the radio receiver - Harry specifies an additional gain stage before the final EL84 output. The stage uses half of a double triode - in this case an ECC81 (/12AT7).

Before doing anything more, the pin-out of the tube is as shown in the figure below...


Harry's schematic (seen complete on his excellent website - be sure to use his updated address) is reproduced in "edited" form below, to show just the single AF Pre-Amplifier...

This stage was simple to build - I described the enclosure last week and I did all the hardcore metalwork - so this stage involved no significant mechanical engineering !

I took the opportunity to put the inverter into a box - for convenience and safety. I used another of the "chassis" enclosures from Maplin (order code XB56). The entire system to date is shown below (couldn't resist stuffing the empty valve bases with their respective valves - although only the two on the right are "doing anything" in this picture).

Now the audio section of the radio is completed - time to get on with some higher-frequency antics...

...-.- de m0xpd