View Full Version : An Adventure in Oscillator Theory


VintagePC
05-08-2014, 03:17 PM
Okay folks... here's something I'm hoping some of the seasoned folks here might be able to help with. It's gonna be a wild ride in trying to understand oscillators and some theory, so hold on to your hats! :)

:scratch2::scratch2::scratch2:

Specifically, I'm looking at the implementation of the 6AF4 oscillator on Pg. 4 of this SAMS: http://www.earlytelevision.org/pdf/Blonder-Tongue-99-Sams-259-3.pdf

Here's what I've been able to puzzle out, and I'd like to know if I'm correct in my understanding of the below points (or not). The upper half of the circuit can be ignored since it has no bearing on the LO frequency - if you haven't read the rest of the sams it's just a diode superhet mixer.

1. L9 and L8 - I suspect these function as RF chokes to keep the B+ clean and the RF from back feeding. No impact on frequency.

2. The oscillator design: I suspect it is a colpitts type oscillator based on the similarity to some tube reference designs I have found - but I'm getting confused any time I try to relate components in the SAMS to a reference design (e.g. on Wikipedia, http://en.wikipedia.org/wiki/Colpitts_oscillator) so that I can attempt to calculate frequencies (or component values, based on a known frequency.. which is my ultimate goal, see below.)

3. C7 and C6 are decoupling caps which serve to block the B+ from the chassis, as the other end of the tuning inductor (L3) is attached to the chassis. Since they're in the grid-anode circuit I suspect they still affect frequency by 1/C= 1/C7 + 1/C6 + 1/Ct where Ct is the capacitance of the tuning mechanism, C is the total capacitance of the LC tank circuit.

4. The item labelled as "gimmick" adds some minute capacitance (Ct?) in parallel with the tuning coil to tweak the frequency. If this Ct is several orders of magnitude smaller than C7 or C6, (pF vs F?) it would then dominate those terms in the inverse equation above... and C7/C6 would therefore be negligible in determining frequency?

4a. Similarly, the output "tickler" is off the cathode of the tube.. Colpitts reference designs have the output from anode to cathode though... which is why I'm not entirely sure of (2).


How am I doing so far? I don't have a significant background in EE but it's a hobby and I do understand most of the basics and am trying to build. ;)

Where I'm ultimately interested in going is determining what modifications to the circuit are necessary to lower the frequency range of the LO. As designed it would span UHF 14-83, minus Ch5 (for superheterodyne downmixing) or 394 - 808 MHz.

Target frequency range is cable 14-83 minus ch5, or 44-500MHz.

However... baby steps... I'm hoping to break it down in a way that is logical for my thought process and then work up to the point where I understand it well enough to attempt that. :)


Thanks in advance, and I'm looking forward to the discussion and the fascinating tidbits people usually have to offer around here!

bob91343
05-08-2014, 04:12 PM
I think it's a Hartley. The clue is that the tuned circuit has a grounded tap on the end of the travel of the slider.

VintagePC
05-08-2014, 06:02 PM
I think it's a Hartley. The clue is that the tuned circuit has a grounded tap on the end of the travel of the slider.

That would have been my second guess. I spent a good while looking at tube implementations of both of these and I ended up deciding on Colpitts because it is referenced more often in relation to UHF design.

Is the slice of tuning mechanism that runs from the wiper to the chassis connection the bit that forms the second inductor in a Hartley?

bob91343
05-08-2014, 08:22 PM
The Hartley uses a tapped inductor, and that's what we have here. The tap happens also to be ground, a bit unusual but the concept is similar. Most Hartley oscillators have the tap going to cathode, the resonant circuit between grid and ground.

Colpitts, on the other hand, uses two capacitors for a tap and doesn't tap the coil.

So this circuit doesn't really act like either one. The bottom line is that you need frequency selective feedback and gain.

Electronic M
05-08-2014, 09:12 PM
I think I have one of those Blonder Tongue units stashed away.

Another thing to consider is that at higher frequencies cavities and mechanical structures around that size can have capacitance and inductance at values suitable to contribute/be integral to the operation of the circuit. You may have to look for parts that are not on the list to fully understand the operation.

I recall reading a 50's article on building either an antenna preamp or a UHF converter based on a 'Pencil tube' where the physical enclosure actually comprised a substantial part of the tuned circuit it used.

bob91343
05-09-2014, 12:48 AM
And if you drink more bleach than the next guy, you will have a blonder tongue.

VintagePC
05-09-2014, 07:55 AM
Sounds like it might be easiest to find some way of monitoring what frequency the LO is running at and then tinkering with various aspects to see what effect it has... I don't have a scope but I'm thinking an SDR might work well too.

Valid point on the cavity and structures - in fact quite common as cavity resonators and cavity tuned antennas (quite common in SCADA systems).

jr_tech
05-09-2014, 12:49 PM
Target frequency range is cable 14-83 minus ch5, or 44-500MHz.


But channel 5 is 76-82 mHz, so at some point in tuning, your oscillator will be tuned to the output frequency... there will be a range of channels that will suffer severe interference.

If you add enough "C" across the tuning coils to get the resonant frequency as low a 44 mhz, the tuning range will be greatly reduced.

Are there any cable channels in that range that are NTSC, anyway?

Fun project, but I seriously doubt that it will yield useful results.

jr

VintagePC
05-09-2014, 01:47 PM
But channel 5 is 76-82 mHz, so at some point in tuning, your oscillator will be tuned to the output frequency... there will be a range of channels that will suffer severe interference.

If you add enough "C" across the tuning coils to get the resonant frequency as low a 44 mhz, the tuning range will be greatly reduced.

Are there any cable channels in that range that are NTSC, anyway?

Fun project, but I seriously doubt that it will yield useful results.

jr

Aye, a very valid point. I was doing some experimental calculations to see what would happen when trying things like 2C, 4C, 8C 16C etc... and noticed that at 16C you get a LO of 117 low side, but your high end suffers and is down to 222.

As for interference, that would definitely result in a complex mess to clean up.


Cable and broadcast share frequencies on channel 2-13, then they diverge, except for the high end of cable which overlaps with the low end of broadcast (which is why the converter picks up Ch62-70 cable). http://en.wikipedia.org/wiki/North_American_broadcast_television_frequencies#Ca ble_television has the details.

What's the practicality of inserting a second fixed-frequency LO mixer to simply upshift the incoming cable signal by 350Mhz before mixing with the LO? There'd probably be a detrimental effect on reception but it would boost the channels into rage of the converter... and I do have some more 6AF4s kicking around...

I don't really have any anticipation of this actually working but I figured it would be a neat "let's try and see, just because". :)

old_tv_nut
05-09-2014, 02:46 PM
My guess is that this is actually a transformer-coupled oscillator. The sliding short changes the inductance of the transformer, which is connected to invert the plate signal, thus producing positive feedback when connected to the grid. I think maybe the ground at the end of the transformer is theoretically not required, but serves to equalize stray capacitances on the input and output sides that don't show explicitly in a schematic. How's that for a guess?

http://en.wikipedia.org/wiki/Meissner_oscillator

VintagePC
05-09-2014, 05:21 PM
My guess is that this is actually a transformer-coupled oscillator. The sliding short changes the inductance of the transformer, which is connected to invert the plate signal, thus producing positive feedback when connected to the grid. I think maybe the ground at the end of the transformer is theoretically not required, but serves to equalize stray capacitances on the input and output sides that don't show explicitly in a schematic. How's that for a guess?

http://en.wikipedia.org/wiki/Meissner_oscillator


I think you might be on to something there... :) Definitely explains the use of that tickler gimmick on the grid plate... I can't imagine that being the capacitance part of the tank circuit as it would be incredibly sensitive... one wrong bump and the service guy owes you a new one. :D

That said, the ground in this system isn't actually connected to anything; it's just chassis which bears no connection to either side of the AC line.