Monday 18 January 2010

More to this than meets the eye...

This weekend I took another look at the IFTs - you may remember I had made one to the new, improved, "taller" design and needed a second. Well, here it is, installed in the Valve Superhet project - now featuring due espressi



Good news - where before I could copy CW on headphones, now I could copy voice in speakers! That's real progress, I guess - but there are limits to how many aviation meteorology reports or Radio Taiwan shows a boy can stomach.

However, I still had a much higher loss to contend with than the writings of other enthusiasts suggest I should expect (I have no "real" IFT to compare against). So, I spent some time chasing down the origins of this disappointing performance. My (interim) conclusion inspired the title of this blog; "there's more to this than meets the eye".

A little reading around the web led me to invest some time in reading more of Max K4ODS's website "Fun with Tubes", amongst which is an interesting idea for IFTs made with axial chokes . The thing that rang the bell with me was the fact that I'd already messed with axial chokes along the way to building the present IFTs, as described here .

I quickly made up an IFT using the "chassis" of one of my original devices, replacing the coils wound on sewing machine bobbins with 1mH axial chokes, as you can see here...



I had already been amazed when playing with the IFTs with the very low levels of coupling between the coils required to get the characteristic "critical" response. The (relatively) small coupling between the two coils in the photo above is about right for critical coupling - amazing!

I built the "axial" IFT more with a mind for experimentation than use and - in playing around with it - I stumbled on another key realisation (for me at least) ... there is CAPACITIVE coupling between the primary and secondary circuits of my IFTs which is almost as significant as the (intended) inductive coupling. This prompted a lot more messing with Spice models but - rather more practically - I decided it was time to re-visit an idea I toyed with on an earlier blog - I would build a Wobbulator.

Don't know what a Wobbulator is? Wikipedia does !

Here's a pic of the quickly knocked together m0xpd Wobbulator



The ugly board (left) includes both the voltage-controlled oscillator and a first try at the sawtooth "timebase" circuit. The circuit on Veroboard is an improved timebase circuit.

Here's the first result from the Wobbulator...



I don't yet know what the peak at the left (low frequency) end of the trace is - I haven't had a chance to get used to it / de-bug. However, you can see the near-critical response of the IFT made of axial chokes (seen earlier in this blog).

Now, armed with a wobbulator, I hope to be able to conduct some more orderly "experiments" with my H/B IFTs and with the Superhet project as a whole. Watch this space!

...-.- de m0xpd

Sunday 10 January 2010

New IFT

Long time no blog...

I've been doing Christmas, the Children's Birthdays (all three of which fall in the first week of January), bad weather, computer problems, work etc.. That's my excuse anyway!

The beloved got me a "Made From Junk" 969 Tuner for Christmas (which shall be the subject of a future blog). It works but - let's just say - it lives up to its name! More of that later. For the moment we return to the Valve Rx project (Blogs passim).

I ordered up an ECH81 over the holiday season to replace the ECH84 I'd been using. The new tube only cost me £4.85 from The Tube Shop so the wallet didn't take too much of a bruising!. For reference, the pin-out of the ECH81 is shown below...



The new mixer valve didn't really change much - the receiver works but only just! As I knew that all the stages worked OK, I turned my attention to the great unknown in the project - the Intermediate Frequency Transformers.

The whole project started with an idea to learn something about valve superhets, within which the IFTs are obviously an important part. I'd made some simple units (described here) , which displayed pretty disappointing Q-factor, so I modified to the ferrite-cored version described here. You can see the last two generations of (attempted) IFTs here...



I decided to look again at the whole theory behind the IFT and soon realized that I'd neglected to control the single most important variable in my prototype devices; the coupling. My IFTs (like most others) have two resonant sections (think of them as the primary and secondary of the transformer). The response of the whole system is critically controlled by the degree of coupling between these two LC resonators.

I was inspired by to make a Spice model of the IFT by Max K4ODS's work, reported here .

Here's my Spice schematic, shown with a high degree of coupling (k=0.1) between the coils of the IFT...



My Spice model showed me that - with high coupling - the "IFT" can give two quite distinct peaks (associated with the two modes of the system)...



Looking again at my second-generation IFT, I realized that this was exactly what I was seeing - I had much too high coupling (I'd managed this as an accidental by-product of the desperate bid to increase the Q of each LC section). So - I needed to build another IFT with variable coupling between the two stages - simply achieved by varying the spacing between the two coils.

I couldn't vary the spacing of the coils in the existing devices (pictured above) as there wasn't too much space available and (more importantly) I'd stuck everything fast with glue!

I quickly made a new IFT, using the same design as before (I had previously etched a batch of the "end plate" PCBs) but with...
1) more "height" (so I've got plenty of space to slide the coils relative to one another - whilst keeping clear of the end "plates")
2) coils on Ferrite formers sliding on a central dowel



Now, I could tune each section to 455 kHz (by driving the other section directly from a 50 Ohm RF generator output) and observing the response on the 'scope, THEN make the crucial adjustment of coupling to the near-critical value. Sure enough, I get the shape predicted by my Spice model (except for losses - which I shall try to understand in more detail later)...



The tuning of the IFT changes as it is mounted in its shield can (made here out of an Illy coffee tin - I've moaned before about the difficulty of getting tins without corrugations, so I've left the brand name clearly visible in thanks to Illy)...



Now we have something that actually behaves the way it should, I couldn't wait to try it in the receiver...



The good news - it really has made a difference - I can clearly copy CW in the contest currently running on 80m (albeit in headphones). I hope that, when the other "IFT" is replaced with something more worthy, the whole receiver will come to life.

That, I'm afraid, will have to wait 'til next weekend.

...-.- de m0xpd