Programmable delay line for analog signals

[jean_jean]

Dear all,

I have a project where I need to delay a signal with about tens of microseconds (max. 50µs) and I can only find digital delay lines or 64µs delay lines for PAL TV systems.

The signal I am working with has a max amplitude of 1Vpp, with a 1.5MHz max frequency.

Can anybody help me?

Best regards,
Jean

 

[BradtheRad]

Color tv's used to send the signal through a certain length of wire, which introduced a delay, so that earlier information in the signal could be combined with later information. (Chroma and luma, I believe.)

The component was literally a coil of wire. I don't know if inductance was a factor.

To produce a delay of 50 uSec, at the speed of light, you would run a signal through 19 feet of wire.

 

[FvM]

PAL delay lines are fixed to one period of the horizontal frequency and have been mostly using ultrasonic delay. They have been superseeded by digital signal processing methods in modern TV sets.

I presume, the easiest way to make a programmable analog delay these days is to go via ADC/digital delay/DAC.

 

[zorro]

You can look for Bucket‑Brigade Devices (BBD). [They use the same concept as Charge-Coupled Devices (CCD).]
Several types were designed for audio processing. Maybe there is some device suitable for your application.
Regards

Z

 

[Warpspeed]

If you only require a fixed delay, the coil of wire method is often used in oscilloscopes to delay the signal so you can see the edge the oscilloscope is triggered to.
Its simple, can have a very wide bandwidth and is pretty low cost.



If you want a programmable delay, that is an entirely different thing.
Bucket brigade devices work at audio frequencies, but I am not sure about 1.5 Mhz.

Your best bet might be very high speed analog to digital, and digital back to analog approach.

 

[chuckey]

Using the 19 foot of coax idea, if you split it in half, then one half split it again, say 4 times you would have sections that put in, 25,12.5,6.25,3.125 micro seconds which you could combine using digital analogue switches. In this example you could get from 0 to 50 microsecs in 3.1 micrsec intervals.
Frank

 

[crutschow]

If you only require a fixed delay, the coil of wire method is often used in oscilloscopes to delay the signal so you can see the edge the oscilloscope is triggered to.
Its simple, can have a very wide bandwidth and is pretty low cost.

View attachment 127745

I believe that coil of wire is a length of coax and the delay is simply due to the propagation delay through it's length.

- - - Updated - - -

...............
To produce a delay of 50 uSec, at the speed of light, you would run a signal through 19 feet of wire.

Your decimal point is a little off. :wink:
The speed of light is about 1nS/ft, so 50us would require about 50,000 feet.

 

[FvM]

The speed of light is about 1nS/ft, so 50us would require about 50,000 feet.

Transmission lines have a speed reduction by factor √Er, about 1.6 for usual coax cables, give 1,5 ns/ft (5 ns/m).
Delay lines of classical analog CROs provided a delay in a 100 ns range, which is manageable with cable delay, 50 µs clearly isn't.

CCD ("bucket brigade devices") have been a popular analog delay solution a few decades ago. I believe that all devices are obsoleted now. There dynamic range was however limited, so professional audio delay units used digital techniques since mid 70th.

 

[zorro]

There are different types of analog delay devices mentioned in previous posts that should not be confused:

a) Simple delay lines made with coaxial cable, like the picture in post #5 used in an oscilloscope for delaying the signal with respect to trigger.

b) Delays made with coils wound over, and/or covered with, coaxial conductive cylinders. This acts like an artificial delay line with high values of distributed inductance and capacitance per unit of length. The propagation velocity 1/sqrt(l*c) (l and c are specific values, in Henry/meter and Farad/meter) is much less than in a "normal" transmission line made with linear conductors, where the wave propagates in TEM mode with speed 1/sqrt(µ*ε) .
They were used in TV sets for delaying a few microseconds; surely BradtheRad was refering to this type in post #2.

c) Delays for PAL line delay, mentioned by FvM in post #3, based in acoustic waves and piezoelectric transducers.

Not mentioned are devices based in long optical fibers for delay and electrooptical transducers.
I agree that a solution based in ADC-digital_delay-DAC is maybe the natural solution nowadays. A solution based in b) can be interesting if you want to experiment.

Z

 

[BradtheRad]

Your decimal point is a little off. :wink:
The speed of light is about 1nS/ft, so 50us would require about 50,000 feet.

You must be right, because 19 feet sounded too low to me as well. Here's the way I calculated it. I still can't see what I did wrong.

186000 x 5280 / 50000000 = 19.6 ft.

- - - Updated - - -

And I seem to recall the tv coils produced 1uS of delay with 100 or 200 feet of wire.

 

[zorro]

In metric system I would say:

length = c*T = 3e8 m/s * 50e-6 s = 15000 m

And I seem to recall the tv coils produced 1uS of delay with 100 or 200 feet of wire.

Maybe, but not this way. Please see point b) in my previous post (#9).

Z

 

[BradtheRad]

You must be right, because 19 feet sounded too low to me as well. Here's the way I calculated it. I still can't see what I did wrong.

186000 x 5280 / 50000000 = 19.6 ft.

Now I see what I did wrong. It should be:

186000 x 5280 x 50 / 1000000 = ... which is 9.3 miles. Now it sounds too much.

- - - Updated - - -

But it's a closer match to your figure.

 

[Warpspeed]

b) Delays made with coils wound over, and/or covered with, coaxial conductive cylinders. This acts like an artificial delay line with high values of distributed inductance and capacitance per unit of length.

There was a brief time, before digital circuitry really took off where lumped LC delay lines were used in equipment that had delays in the order of a few microseconds. This might have been around the mid 1960's.



I can remember seeing one of the very first solid state TV station master sync pulse generators that used tapped delay lines and RTL logic chips.

Various events such as front or back porch widths and sync pulse width were adjusted by soldering wires onto tapped delay lines and so could be moved around in 100nS steps to meet exact specification.

It was a real technical marvel, as "our" TV station SPG with its power supply used dozens of valves, and entirely filled one very tall 19 inch equipment rack.

 

[crutschow]

Now I see what I did wrong. It should be:

186000 x 5280 x 50 / 1000000 = ... which is 9.3 miles. Now it sounds too much.

- - - Updated - - -

But it's a closer match to your figure.

My rough estimate was 50,000 feet, while 9.3 miles is 49,100 feet, about 2% difference -- close enough for government work. :wink:

 

[dick_freebird]

There used to be analog CCD ("bucket brigade") delay line ICs that
hobbyists would use for reverb and so on.

These days I'd go with an ADC, a deep memory and an interleaved
write and read with an offset index, and stuff it to a DAC at the
back end. Your index offset being the delay variable, in conjunction
with clock frequency.

You might even find this could be done with a resource-rich uC
especially if you were forgiving of audio quality at the higher
frequencies.

 

[zorro]

And I seem to recall the tv coils produced 1uS of delay with 100 or 200 feet of wire.

Please see point b) in my previous post (#9).

Please take a look to this DIY delay line for color TV (with distributed L and C):

http://www.hawestv.com/mtv_color%5Cdelayline.htm

For about 1.3 µs, this design has about 2800 turns on RG-59U. That gives about 180 feet of wire. Fits with BradtheRad's remark.

All is very interesting, but... what about the original poster?
Jean, are you there?

Regards

Z