We Haven't Had A Radio Thread Here In A While, Yup, about that time again

I haz an almost-radio!
http://webpages.charter.net/dawill/Images/Radio1.jpg
http://webpages.charter.net/dawill/Images/Radio2.jpg

Layout:
Antenna Input
2MHz HPF (see pic 2)
Wideband amp (a few dB gain, mostly buffering; probably good up to 50MHz, don't have any way to sweep up there yet)
DBM (1N4148s, or did I use BAT64s, I forget!) with BNC connector for external L.O. (local oscillator) input

IF strip: 4 x tuned collector, grounded base amplifiers, center around 500kHz. Chose this for stage-to-stage isolation and handy input impedance; common emitter may work as well but I haven't tried yet.
Detector: half wave doubler, BAT54
AGC: controls bias on the first IF amp. Unusually long time constant (10k ohms into 10uF), which keeps it from squegging (AGC is a control loop like any other; I opted for slow, dumb old, dominant pole compensation).

Overall IF strip: approximately 100dB gain -- with this much attenuation between function generator and antenna input, I can just barely hear the gen's AM outout over system noise (generic QRZ). With AGC, the maximum input before compression is quite high (around 0dB). I'm pretty sure the IF stages could be tuned for more gain (there's four stages!), and more shielding and isolation between stages will allow the increased gain without oscillation. Something to play with later.

AF amp: emitter follower drives the 10kHz LPF, which goes through a gain stage (common emitter) and follower to the headphone jack.

Overall system response: without a 500kHz filtered input bandpass, sweeping the L.O. finds many images. A couple AM SW stations show up particularly strong in the 2-5.5MHz range. Relative to earlier experiments (before I built the IF strip), it's like hearing for the first time -- earlier I had only picked up a little noise and maybe local AM BCB stations; it's now like watching the night sky with dark-adapted eyes for the first time!

Future: unless I want to rip up the existing IF strip (ha), a dual stage superhet is likely. I've got ideas for VHF work, involving wide range oscillators (try this: an RC oscillator that works from 10MHz to 300MHz in a single range), crystal-locked PLL with bands, high frequency IF (i.e., something to filter my images apart!) and more.

Tim

Gotta schematic, Sch3mat1c?

(pssst.... QRN, not QRZ)

huh?

at least it isn't QRM .
(i get enough of that around home)


yep ... trying to get the images and spurs to fall outside of where you want to be is tough .
do you mean basically a "dual conversion" IF ?

if you could find some of those collins mechanical filters they would be great for changing the selectivity/BW .
there are some other filter manufacturers but not near as many available as there was 40 years ago .

I thought that was unofficially, "QRW"

Quite surprised you're not taking the SDR route for radio design Tim

My guess is when he has the theory down, he'll have an advanced computer controlled DDS and PLL driving a pentode RF/IF chain

Nahhhh... too cheap to get an ADC + FPGA dev kit Analog does as well, and besides, you still need the front end.

SDR is mainly just running a digital sampling scope (which I'd like to make some day anyway, for S&G), except at phase lock rather than triggered time sync.

But speaking of sampling scopes, I wouldn't mind building one of those either. Could make a transistor avalanche pulser for probably around 500-1000MHz bandwidth; getting jitter low, and generating trigger, will be more difficult.

I digress No schematics yet, pieced together as I go. I may draw some up.

QR-whatever.. forget exactly what designator. Specifically, the generic buzzing radio sounds you pick up in the low MHz, which I suppose are from myriad switching supplies and whatnot.

Tim

Cool, I just heard NHK Radio Japan

Now, according to the table,
http://www3.nhk.or.jp/nhkworld/english/rad.../all_201203.pdf
(map http://www3.nhk.or.jp/nhkworld/english/rad...wave/index.html )
that should be 6110 kHz (Canada relay). Thing is, my LO is 2.5-5MHz, and with a ~500kHz IF strip, the highest image I should get is 5.5MHz. Seems like I found two images of this station towards the low end of the dial, which would be 3055 +/- 500kHz (or whatever the IF is actually tuned for), which would actually be the LO harmonic doing the work!

What surprises me about that is, the LO to begin with is kind of weedy (~10dBm) to drive a diode balanced mixer -- I get much better conversion gain with my function generator (which is able to saturate the diodes easily with a few volts drive, maybe 20dBm). The LO's harmonics make it behave like a square wave with duty cycle other than 50%, so the conversion efficiency at harmonics should be even lower still (maybe 10dB down, maybe worse?).

Is this impressive? I don't know... how strong is the transmitter? If it's coming from CBC/RCI Sackville, it might be 300kW, which should still be pretty strong down here. Anyone else have similar experience?


I've also noticed that signals come and go, and it's really annoying. I can't tell if this is unstable AGC, or frequency stability, or atmospheric disturbance or what, but it feels like I can track it better by playing the frequency back and forth. Maybe I'm fooling myself, like sitting at a bonfire and waving the wind away when it blows smoke at you: you know it doesn't do any good, yet somehow you do it anyway.

It shouldn't be AGC -- I improved the step response by testing with an ASK waveform, adjusting it until it behaves over the full ~100dB range of sensitivity. The time it takes to adjust to the new power level varies widely (because 1st IF gain is very nonlinear vs. AGC voltage), which means the detected audio bandwidth also changes (slower = more bass, more sensitivity to disturbance), but it's not breaking into oscillation* and it looks reasonably well damped. Doesn't mean it's correct on real world signals, of course...

*Squegging, sort of. Actually, this is a funny situation: it's squegging in the sense that, bias is varying due to the HF envelope, and excess gain and phase shift cause the envelope to oscillate. But it's not the conventional circuit you'd see squegging in: it's not self-excited, it depends on an external signal to generate nonlinear signal levels. So it's not squegging, in the sense of a self-excited oscillator, but it is in another sense. Cool!

I'm also confident it's not actually a frequency thing, because when I built the LO, I checked it out a lot against the function generator, and established that the function generator, which, being an RC oscillator at heart, has more phase noise and drift.

That's not the whole story though: the IF strip may have nonlinear frequency response, which I would attribute to the ferrite core inductors. In testing, I've seen this manifest as a variation in IF bandwidth and peaking versus signal level. This should only affect the 1st IF, if at all, though, with AGC present. And under those conditions, I didn't notice a significant change in center frequency, and certainly not a time-varying change.

As for atmospheric instability, is it just a bad week? Any radio should have the same problem though, but I've seen much cheaper radios without this behavior so I don't know what exactly is going on.

Tim

Enjoy the broadcast! The Sackville transmitting site will soon be shut and sold for scrap

Addition for the radio stuffs:

http://webpages.charter.net/dawill/Images/TCXO.png

Made a temp controlled crystal oscillator. Didn't take any pictures; wouldn't do much good now that it's all sealed up. Layout isn't anything special, at least, just dead bug.

Construction:
- Cut a strip of 0.02" thick (~16 oz) copper, long enough to wrap around the circuit
- Cut a piece of copper clad PCB for layout
- Solder the PCB onto the copper strip, using the kitchen stove
(hey, I cleaned up after myself)

Purpose of the PCB is because I'll never be able to solder to copper that heavy with my iron. The PCB still sinks heat quite nicely, but is solderable.

The transistor and regulator are soldered to the copper (also during the kitchen step), and the thermistor is also attached as near as possible.

The temperature seems to be reasonably stable, over conditions of insulation (open air vs. sitting on the table vs. wrapped in tissue paper) and supply voltage. Don't have any instrumentation to determine the spectral output, of course. The waveform is noticably distorted, probably because the LC tank isn't quite tuned, but harmonics are okay.

Tim

12Meg, eh?

Duo and I were talking about an ultra stable clock for a DAC project... this might work

That's an OCXO, not TCXO right?


I was thinking of doing something similar with the bigass 100kHz xtal I got from pat... for the all toob clock, if I ever get to it.

Seems most the tube era high test stuff, like HP, was crystal in container, then wrapped in nichrome or.. whatever, then insulated, and another can. I can't remember now, but they must have used bimetal switches for quasi-regulation? Maybe the ~1960 HP stuff wasn't temp regulated at all, and just being considerably above ambient made it more resistant to changes in ambient temp.. Perhaps a more stable part of the temp curve..? I'm not really sure..

What, is TCXO variable..?

I wanted 10MHz like the classic timing references, but don't have one on hand

I specifically compensated the feedback, so the power supply isn't varying all over while it runs. Should keep stability a notch better I suppose.

Tim

Seem to think that TCXO is temperature compensated, so circuit monitors device temp, and pulls the crystal up/down to compensate. Whereas OCXO controls the temperature, instead of just compensating for it... more accurate.

WOW!

I haven't even seen this thread! OMG! Sorry Tim, I love this stuff.

And you actually went the road I'm infamous for going...the "just build as we go", forget schematics...man...you...mr, document everything, mr. formula himself, now H*** really has frozen over.

Looks like a winter project though, radios are cozy to build inside a warm shack while it's storming and snowing outside, best enjoyed with a warm cup of chocolate milk though and abundances of components.

I can probably mail you a 10mhz TXCO, I grabbed a few when the electronics shop in Copenhagen went out of business. Want one?
Oh..and remember your address.

Oh, thanks But that's okay, I'll manage. Maybe later on I'll develop enough tools to measure things to that precision, and then we can see if my OCXO stacks up to it.

Tim

Alright, just say the word then.

Cheers.

http://t3sl4.dnsdynamic.net/Images/Radio4.jpg
Antenna bandpass, 6-7.5MHz, 3rd order

http://t3sl4.dnsdynamic.net/Images/Radio5.jpg
First IF bandpass, 4.5-6MHz

http://t3sl4.dnsdynamic.net/Images/Radio6.png
IF strip, three amps, first one has AGC

http://t3sl4.dnsdynamic.net/Images/Radio7.png
2nd IF mixer (1st is similar, on another board).

I also set up a dipole for ~40m, could use some trimming as it's a little low, but good enough.

Since these pictures, I added a 1.5MHz injection locked oscillator (locked to the 12MHz OCXO) and BFO (mixer tapped off the raw IF), and soldered together the bits so it makes an ugly stack instead of random board strewn about (though the cables between sections are just as messy).

Tim