Xmas project Tv delay line osc

[dr pepper]

Instead of drinking eggnog I decided to breadboard a project I attempted a while back.
An oscillator based on a Tv delay line, I saw something similar in a 80's electronics mag (now lost).
A continuous osc running at 4.43 mhz colour burst frequency is gated into a delay line, feedback from the end of the delay line is amplified and fed back via the lowest inverter on the schem.
Initially this inverter was a monostable from 2 inverters, but I found it works just as a straight inverter without timing.
I dont quite get how this is working, the burst period appears to be 21.3uS, 1/3 of the delay lines period, why is the third harmonic controlling the feedback, the transistor amp doesnt appear to be clipping, is this just the gain in the feedback nand?
I'm surprised this works with just 3 gates, though it does need something adding to start it, right now touching one of the components starts it up.
I dont have a use its just one circuit I wanted to play with, pics attached of the 'scope trace (top gated drive to delay line, bottom feedback) and rough schem.

 

[betwixt]

Why two oscillation loops? You should be able to do it with the delay line alone. Note that some delay lines are there to hold back the luminance signal so it lines up with the chroma signal. As the chroma passes through a narrower bandwidth filter it tends to lag the luminance so the delay added by the line may not be one cycle at 4.433MHz but some amount considerably longer. Note that 64uS = 1 cycle at line scan rate of 15.625KHz not 4.433MHz. If this delay line is for one line period it would be for delaying chroma information in a PAL color decoder rather than delaying one cycle of chroma clock.

Brian.

 

[dr pepper]

The delay line stores 1 video line at 1/15.625 kc yes.
The lines piezo transducers don't respond to 15.625kc, seems peak response is 4.43mc colourburst freq.
So the xtal oscillator provides the 4.43 tones, and the nand configured as an inverter is needed to give correct polarity pulses, due to the tranny inverting the signal.
We're you thinking of making the thing oscillate at 15.625?
I tried to get the thing to oscillate at 15.625kc but noise caused trouble in the feedback.
It still needs a means to create a startup pulse on power up.

 

[betwixt]

OK, so it is a glass delay line rather than a wire one. They are quite lossy so you will need fairly high gain in the feedback loop and I'm not sure if the phase at the output is exactly 180 degrees or 360 degrees. If it doesn't have a common connection between input and output you could try reversing the transducer wires to invert the polarity. I think, if you are successful, you will achieve is an oscillator with the delay line creating many multiples of phase shift instead of the usual one. In fact there could be more than 280 cycles of delay! If it will oscillate, noise alone should be enough to kick start it.

Brian.

 

[dr pepper]

Yep it's glass, and acoustic.
Both transducers have 2 separate leadouts.
I created a delay memory, it's not oscillating, fiddling and touching the amp section creates a data 'word' and this then cycles through the line over and over.
I wanted one 20us or so burst to recirculate through the line, the feedback inverter needs to be a one shot.
I believe ancient memory's used a shift register for this, I just need 1 bit hence the one shot.

 

[dr pepper]

Yep it's glass, and acoustic.
Both transducers have 2 separate leadouts.
I created a delay memory, it's not oscillating, fiddling and touching the amp section creates a data 'word' and this then cycles through the line over and over.
I wanted one 20us or so burst to recirculate through the line, the feedback inverter needs to be a one shot.
I believe ancient memory's used a shift register for this, I just need 1 bit hence the one shot.

--- Updated Monday at 6:11 PM ---

I threw the following circuit together.
The waveforms this makes are the same as my 'scope pic above.
The circuit is just a bunch of gain and 180 degree phase inversion across the delay line.
I didnt expect the waveform to look like it does, a 21.3uS toneburst of 4.43mc and the start of each tone burst is 64uS between them.
Seems like theres 2 resonances and a harmonic, the 4.43mc is possibly the piezo transducer peak response, the 64uS is the delay time and the 21.3uS tonebusrt length is the 3rd harmonic of the 64uS delay.
All these probably comming about due to the gain and the squared edges from the 'Hc00, fascinating little circuit.
I have a delay line from a tube telly the size of a brick, I might just see if that'll work.

 

[dr pepper]

Not having a good supply of Ic's plus this not lending itself to breadboarding due to high freq's I decided to sim this on Lt spice.
Noise is a problem hence the lowpass C across the amplifier transistor, however the circuit seems to work.
Instead of a 'delay line' I went with a transmission line as its the nearest thing Lt psice has, but provides the same function.
The output is what I expected 64uS of 4.43mhz, followed by 64uS silence.
I did actually did get this from the breadboard setup but it wasnt reliable, probably expecting too much from an Mb4.