This is just me thinking out loud

With recent discussions here about test
equipment and which scope to buy, I was
thinking how much fun it was building our
own stuff back in the before times.
(This should probably be under Nostalgia )

You can do some cool (and fun!) stuff with
a 555 and hand full of other components.
(eg I built a capacitance to voltage converter
which gave every appearance of being accurate
to fractions of a picoFarad)

A sampling scope used to be an analog critter
that grabbed a piece of signal and recirculated
it through an analog delay line then took succesive
samples every time it came around.

So you'ld grab, say, 100ns of signal then sample
1ns every 101ns and display that sample for 101ns.

There have been similar schemes for eg Apple, C64
to digitize pictures that used a VCR in place of
the delay line.

You could, in principle, convert your signal to,
say, black and white NTSC (or something close)
record it on your VCR find the right frame,
digitize it and have your C64 (or what ever)
display it as if it were a scope trace
(probably only be good for a couple MHz bandwidth).

I recall seeing plans for a DIY sampling scope
that used a door spring as an analog delay line

Now adays I'm sure you could do better, 20 ft
of fiber optic cable or something, maybe.

So any way I was thinking wouldn't a relatively
simple DIY sampling scope module be a cool project.

Something you plugged into your C64 joy port
or a NES cartridge or something

C64 has it's own DA but it takes so long I don't
know if it would be practical (have to maintain
you recirculating signal sample integrity long
enough to digitize it)

What could you use for a cheap delay line?

I opened up a video2000 VCR once and found several solid-state 64-microsecond delay lines inside:

http://en.wikipedia.org/wiki/File:Ultra ... ayline.jpg

I have done sampling on my 6502 workbench computer many times, like to do spectrum analysis with an FFT, but it's really only good for audio. The computer has fast A/D and D/A 8-bit converters.

If you only needed digital and not analog, it wouldn't be too tough to have counters feed a memory IC's address pins and make 8 channels of input go to the memory's data lines. Without having a processor directly involved, you'd have a much higher speed limit.

That would be the point of sampling in that way you'ld (in effect) be
slowing your multi mega Hertz sample down to audio speed, the
problem is you have to maintain it long enough.




That'd be cool too. I think it would be more complicated in some ways
though (basically you'ld have a bunch of stuff that all had to run at high
speed).

The "long enough" part is limited by memory size, as with any other way of doing it.


The speed bottleneck might still be the memory speed, since 74xx counter ICs, tristate buffers, etc. would be faster. Getting to a few dozen MSPS should not be hard. As for complication, my hesitation in spending time doing it would be that in spite of the complexity, you still get a single-purpose piece of equipment, unlike the home-made workbench computer that can be tons of different things.

I think you're mis apprehending what I'm suggesting.

Originally this was all analog. The memory is an analog delay line*.
You run your signal through the delay line. At some point you
start feeding what's coming out of the delay line back in so you've
got a bit of your signal "trapped", recirculating through the delay line.
Every time (or every other time or every tenth time or what ever)
it comes around you take a sample and digitize it.
So you've got the time between samples to digitize.
Of course you can't recirculate it for long, errors creep in and it degrades.
So the question is: Can you keep the signal integrity in the delay line
long enough to digitize all the samples you want?

(How long can we keep Scotty trapped in the pattern buffer? )

*Of course, what I'm suggesting is to digitize it and stuff it in RAM

Back "in the day", the fastest 8 bit ADC's were half successive approximation, half track and hold, and I think maybe reached a frequency of a couple hundred KHz. 16bit ADC's existed, but maybe had a Fmax less than a couple hundred Hz.

Now look what we've got! A dream come true if you've checked ADC's recently. Cheap, fast, and hi resolution. Some up to 24bits with sample times up to 2.5MHz @~$50US. The low res 8bit 100+MSPS ADC's are less $10 each.

Here's something to keep you dreaming. I saw it the other day, which is why I decided to add my input...

_________________
65Org16:https://github.com/ElEctric-EyE/verilog-6502

OK, this is a slightly frivolous topic, but I'm sufficiently frivolous that I'll take the bait! Re: delay lines, I'm reminded of several things, such as...

...the old, spring type reverbs, as used in guitar amplifiers, for instance. There's a photo, "Spring reverberators," 3/4 of the way down this Wikipedia page. Unfortunately these aren't suitable for us because they are built to produce a complex jumble of multiple reflections, not a single, clean, delayed output.

...the mercury delay lines that were used in early electronic computers. This is more to the point!

...Also (quoting Bogax): "A sampling scope used to be an analog critter that grabbed a piece of signal and recirculated it through an analog delay line [...] Now adays I'm sure you could do better, 20 ft of fiber optic cable or something, maybe."

Fiber optic, yes, but coaxial cable (or other transmission line) will work, too. I bet it was coax that formed the delay line in the sampling scopes you mention. I say so because there's a coax transmission line in my trusty Tektronix 465, even though it's used for a different goal and the signal isn't recirculated. It's a coil of 50 feet, maybe, tucked in the bottom of the chassis, and its job is to delay the vertical deflection to the CRT. The delay is useful when you're using triggered sweeps -- because of the coax "memory," you end up being able to view what the signal was doing an instant before the trigger was recognized and the sweep commenced.

But I like the door spring idea better!

-- Jeff

I think a reverb spring (adapted so you don't get the reflections that cause reverb, just the delay) would probably mess it up too much in phase response; so even if the frequency response was reasonable, the waveform might be pretty messed up. I fooled with with reverb springs decades ago but phase was not an issue in audio reverb. How about a bucket brigade? Something like a long piece of coax will gradually lose the high frequencies, and putting a high-frequency boost on it for each round to compensate would again change the phase relationships. The analog-storage spectrum analyzers I've used had some kind of mesh behind the screen that you could alter the charge on, to, in effect, make the phosphor's persistence super long. I don't really know how it worked.

Similar to what you're saying, if aliasing (ie, images) isn't a problem because you're sampling a relatively narrow band, you can, in some situations, undersample and get away with a very slow sample rate as long as the jitter spec is not loosened up.

I seem to recall that bucket brigades built from discrete components
were used. I don't know about the integrated ones, but I don't think
they were fast enough at the time. I'm pretty sure coax was used.
Basically any thing that could serve as a delay was tried.



Yeah, I'm thinking digitize and put in RAM, but the old sampling scopes
wrote to a storage screen.

Wikipedia mentions using sampling scopes on repetitive signals,
but it doesn't mention using a delay line to get the repetition.

http://en.wikipedia.org/wiki/Oscilloscope_types#Analog_Sampling_Oscilloscope

Oh, for sure! -- yes, you're right. I wasn't seriously proposing to use these. Spring reverbs would be absurd for this.

Bogax, are we certain that some early 'scopes did in fact recirculate an analog signal? (I thought the signal had to be repetitive to begin with.) I understand about sampling a repetitive waveform; I'm just scratching my head about creating that repetition using analog technology. As you point out, after successive recirculations "errors creep in and it degrades." So I'm not being a smart-alec but I wonder if there's a misunderstanding or mis-recollection of what was actually achieved. (This stuff is before my time.)

Mercury delay tanks had various applications, and I gather in some cases were used to recirculate signals, but I think it was only digital signals that could be preserved for long. In that case error didn't accumulate because you could resolve the received signal back to 1s and 0s, and that let you retransmit a signal that was always clean.

LOL! Okay, regarding this point my credulity is seriously stretched. Were these "plans" purported to have led to a working device? Hats off to anyone who succeeds with this approach! Meanwhile I had to check the date of your original post to make sure it wasn't April 1st.

-- Jeff

I don't know about "we" but I am. (unfortunately that doesn't necessarily mean it's so )
My recollection is that they were getting something less than a hundred
samples and this was the only way feasible for signals up around 1 GHz.
The limiting factor was how long they could keep a signal recirculating
with out too much degredation.
I don't recall ever seeing mercury delay lines mentioned but that doesn't
mean they weren't
I seem to recall an article about how SAW devices might revolutionize
things if the problems could be over come.

This was late 60's, early 70's If I can get a few hours in the university
library I'll see if I can find anything. I really don't even know where I was
reading about them (I doubt if I ever actually saw one but I might not
have recognized it if I had, I was just becoming aware of such things as
O-scopes) Most likely it was Electronics Design. It may be that they were
not so prevalent as I thought (kinda like 100Ghz DSOs these days)



I have no idea where that was. That was sort of my reaction except it
was some source that had enough credence with me that i thought
maybe it could be true. Like maybe something I read in QST.