Measuring Stuff Digitally Without an A to D Converter

[Richard_S]

This is just a jolly question to celebrate the end of the year.

Many years ago I was fortunate to have been granted a patent for measuring light intensity. That patent raised a question that I was at that time unable to answer. The invention measured light without the use of an analogue to digital conversion process. Ever since then I wondered whether this was unique, or whether there are other analogue quantities that we can measure numerically without analogue to digital conversion.

Then recently it occurred to me that not only is there another such device but I actually own one, a vintage laboratory balance. The balance has an ingenious mechanism where, as you turn a dial, weights of increasing size are placed on one side of the balance arm. These, as you would expect, are used to counter the weight of the subject on the other side until equilibrium is restored. When it all balances, you can read the weight directly from the numeric dial.

So, now that I know of two, I am left wondering whether there are any other cases of turning what we would think of as an analogue quantity directly to digital without using the type of device that we would usually call an analogue to digital converter? If there are, then I bet that the community here would know.

[jbb]

Hmm. I’m sure there are things like that around.

I’m a technicality: how about a good old fashioned moving coil / needle panel meter?

Ooh, a classic: analog ‘spinning disc’ power meter.

[fourfathom]

Then recently it occurred to me that not only is there another such device but I actually own one, a vintage laboratory balance. The balance has an ingenious mechanism where, as you turn a dial, weights of increasing size are placed on one side of the balance arm. These, as you would expect, are used to counter the weight of the subject on the other side until equilibrium is restored. When it all balances, you can read the weight directly from the numeric dial.

How is this not digital?  You are manually adding weight in discrete increments, and determining the measured value by observing when a threshold is passed.  No electronics are used, instead human stimulus and response, but it's not fundamentally different  from many types of A-D converter.

[Richard_S]

How is this not digital?  You are manually adding weight in discrete increments, and determining the measured value by observing when a threshold is passed.  No electronics are used, instead human stimulus and response, but it's not fundamentally different  from many types of A-D converter.

Absolutely, this is an analogue to digital process. Of course it is. I meant "without what we would conventionally call an analogue to digital converter". I did think about making that explicit but decided everyone would prefer brevity. Happy to clear that up though, without the electronic circuit that we would conventionally label an analogue to digital converter.

I hope that helps.

[Richard_S]

Hmm. I’m sure there are things like that around.

I’m a technicality: how about a good old fashioned moving coil / needle panel meter?

Ooh, a classic: analog ‘spinning disc’ power meter.

Yes, of course, a basic moving pointer meter is turning an analogue quantity to a numeric one by literally pointing to the values. Good call. I think the added element of this challenge would be to look for cases where the output was still in electronic numerical form and so unlike the balance which is human-readable but not electronic as such.

[abeyer]

Seems like a geiger counter or anything similar where your measurement is a count of discrete events would qualify.

[ledtester]

A voltage to frequency converter is another kind of A/D converter.

[jbb]

Perhaps stretching the prompt a bit: a Kelvin-Varley divider plus a null meter (eg galvanometer)

[wraper]

Then recently it occurred to me that not only is there another such device but I actually own one, a vintage laboratory balance. The balance has an ingenious mechanism where, as you turn a dial, weights of increasing size are placed on one side of the balance arm. These, as you would expect, are used to counter the weight of the subject on the other side until equilibrium is restored. When it all balances, you can read the weight directly from the numeric dial.

It's still an ADC, it's just that comparator that is part of ADC is largely implemented mechanically, rather than electronically.

[ch_scr]

So the whole premise of this is pretending an SAR-ADC is "not an conventional ADC" somehow, or did I misunderstand?
If an analogue value (a voltage or light intensity or whatever) goes in and an digital one comes out, it's an ADC by definition?
First, the question becomes what is a "conventional ADC", then?
A KVD+null meter is a kind-of SAR-ADC, just the interface is a bit more convoluted (and even though it's comprised of digits, would stretch the conventional meaning of digital)

Edit: Reading the question again, this is about directly converting analog to digits?

[tggzzz]

This is just a jolly question to celebrate the end of the year.

Many years ago I was fortunate to have been granted a patent for measuring light intensity. That patent raised a question that I was at that time unable to answer. The invention measured light without the use of an analogue to digital conversion process. Ever since then I wondered whether this was unique, or whether there are other analogue quantities that we can measure numerically without analogue to digital conversion.

Then recently it occurred to me that not only is there another such device but I actually own one, a vintage laboratory balance. The balance has an ingenious mechanism where, as you turn a dial, weights of increasing size are placed on one side of the balance arm. These, as you would expect, are used to counter the weight of the subject on the other side until equilibrium is restored. When it all balances, you can read the weight directly from the numeric dial.

So, now that I know of two, I am left wondering whether there are any other cases of turning what we would think of as an analogue quantity directly to digital without using the type of device that we would usually call an analogue to digital converter? If there are, then I bet that the community here would know.

All DVMs used to act exactly like that; the basic principle dates from c1840. I own one, a Fluke 893a nominally accurate to 0.005%:



Principle is that the knobs control a Kelvin Varley Divider with a known voltage applied to its "top". The knob settings determine output voltage, and the meter measures the difference between that voltage and the input voltage. When the meter reads 0, the dials indicate the voltage and zero current flows even if the input voltage is 1kV. Modern DMMs have a 10Mohm impedance at that voltage, so 100uA would be flowing.

[guenthert]

I don't know what we would conventionally call an analogue to digital converter.  There are various schemes and they look quite different to me.

One measurement just done digitally (on a high level) is perhaps electron counting, recently made possible: https://www.ptb.de/cms/en/presseaktuelles/journals-magazines/ptb-news/ptb-news-ausgaben/archivederptb-news/ptb-news-2012-2/current-standard-based-on-single-electrons-in-the-offing.html

I believe astronomers counted photons for a while already.

[T3sl4co1l]

You'll have to cite the patent, because it doesn't sound like you're hinting at anything original?

For example, if your light intensity is measured by photodiode, and say you cascode the photocurrent through a BJT just to keep it more stable (wide compliance voltage range), and say you counterbalance it with another current source, this time a binary weighted series: the bit value of that "digital" current source, when bounded on one side with a gain-node voltage that saturates to one side, and the other side on the adjacent code, has performed exactly a digital conversion of the input current, to a digital value between those two bounds.

SAR ADC is the same thing, comparing an incrementally-more-accurate synthetic voltage to the input reference; S-D could be considered the same as well but time-averaged through an integrator; Flash ADC is doing the same thing but the bounds are explicitly spans of myriad window comparators; etc.

Or if we put a capacitor on the gain node, the current difference gives a dV/dt, which we can measure with a clock and counter, and, same thing: the instant the voltage crosses the reference threshold, occurs at some instant between two edges of the clock; the analog time is bounded.

Tim

[tggzzz]

You'll have to cite the patent, because it doesn't sound like you're hinting at anything original?

Patents are like Schreodinger's Cat: you don't know whether they are valid until they are challenged in court.

A patent that applies a known technique in a new context can be valid, provided it isn't "obvious to one skilled in the art". Obviously "obvious" is the key word there Recursion: see recursion.

[Richard_S]

All DVMs used to act exactly like that; the basic principle dates from c1840. I own one, a Fluke 893a nominally accurate to 0.005%:



Principle is that the knobs control a Kelvin Varley Divider with a known voltage applied to its "top". The knob settings determine output voltage, and the meter measures the difference between that voltage and the input voltage. When the meter reads 0, the dials indicate the voltage and zero current flows even if the input voltage is 1kV. Modern DMMs have a 10Mohm impedance at that voltage, so 100uA would be flowing.

Wonderful!

[Richard_S]

You'll have to cite the patent, because it doesn't sound like you're hinting at anything original?

For example, if your light intensity is measured by photodiode, and say you cascode the photocurrent through a BJT just to keep it more stable (wide compliance voltage range), and say you counterbalance it with another current source, this time a binary weighted series: the bit value of that "digital" current source, when bounded on one side with a gain-node voltage that saturates to one side, and the other side on the adjacent code, has performed exactly a digital conversion of the input current, to a digital value between those two bounds.

SAR ADC is the same thing, comparing an incrementally-more-accurate synthetic voltage to the input reference; S-D could be considered the same as well but time-averaged through an integrator; Flash ADC is doing the same thing but the bounds are explicitly spans of myriad window comparators; etc.

Or if we put a capacitor on the gain node, the current difference gives a dV/dt, which we can measure with a clock and counter, and, same thing: the instant the voltage crosses the reference threshold, occurs at some instant between two edges of the clock; the analog time is bounded.

Tim

Wow, amazing.

My patent was a lot simpler and less clever. It was just a photon counter. Well, to be honest it was a way of using up CCD sensors that had failed inspection. A CCD from a (then) high-resolution camera would be put beneath a diffuser. The diffuser, and so the CCD, would be exposed to the light source. The contents of the device would then be clocked out into a capacitor. The voltage on the capacitor would be read by a comparator. The process would continue until the threshold was passed at which point the count would stop and provide the digital value. A value that was essentially a count of the photons. Not individual photons granted, given the noise performance of the CCD sensors, but still that is the essence. Dodgy cells could be skipped which was why reject CCDs could be used. The patent has long since expired. No idea if the company ever made one. It is irrelevant now, there aren't surplus reject high-quality CCDs anymore I would assume.

[Richard_S]

I don't know what we would conventionally call an analogue to digital converter.  There are various schemes and they look quite different to me.

One measurement just done digitally (on a high level) is perhaps electron counting, recently made possible: https://www.ptb.de/cms/en/presseaktuelles/journals-magazines/ptb-news/ptb-news-ausgaben/archivederptb-news/ptb-news-2012-2/current-standard-based-on-single-electrons-in-the-offing.html

I believe astronomers counted photons for a while already.

Goodness! That is incredible technology and spot on for the challenge. A very elegant solution to the issues they have in losing electrons too. Thanks for sharing that.

[Richard_S]

Seems like a geiger counter or anything similar where your measurement is a count of discrete events would qualify.

A geiger counter is exactly the type of thing yes. Great suggestion. Thank you.

[coppice]

A voltage to frequency converter is another kind of A/D converter.

Most voltage to frequency converters are purely analogue, converting one analogue measure to another. If you then quantise the frequency you have an overall analogue to digital converter.

[wraper]

Then recently it occurred to me that not only is there another such device but I actually own one, a vintage laboratory balance. The balance has an ingenious mechanism where, as you turn a dial, weights of increasing size are placed on one side of the balance arm. These, as you would expect, are used to counter the weight of the subject on the other side until equilibrium is restored. When it all balances, you can read the weight directly from the numeric dial.

It's still an ADC, it's just that comparator that is part of ADC is largely implemented mechanically, rather than electronically.

BTW most of modern high precision balances are made in similar way, the difference is them using electromagnet instead of weights and optical sensor to detect equilibrium. Load cells are used in less precise stuff. Also there is tuning-fork type.

[T3sl4co1l]

Wow, amazing.

My patent was a lot simpler and less clever. It was just a photon counter. Well, to be honest it was a way of using up CCD sensors that had failed inspection. A CCD from a (then) high-resolution camera would be put beneath a diffuser. The diffuser, and so the CCD, would be exposed to the light source. The contents of the device would then be clocked out into a capacitor. The voltage on the capacitor would be read by a comparator. The process would continue until the threshold was passed at which point the count would stop and provide the digital value. A value that was essentially a count of the photons. Not individual photons granted, given the noise performance of the CCD sensors, but still that is the essence. Dodgy cells could be skipped which was why reject CCDs could be used. The patent has long since expired. No idea if the company ever made one. It is irrelevant now, there aren't surplus reject high-quality CCDs anymore I would assume.

Ahh, I see.  So, there's a way in which it could be said to be -- skipping an A-to-D by always being D in the first place -- but, since it's still going through a capacitor, it's not, really, but "wouldn't it be cool if it were?", I guess is more where you were going with this thread?  Yeah, I can see that.

Such systems would be better called "quantum", since QM isn't really digital itself, but is stochastic in practice, so, it's not sharply timed digital, it's noisy, it's uncertain, and really these are all just alternative ways to say "analog noise" in the limiting (smallest signal) case, give or take exact physics of the system in question of course (not all systems exhibit full shot noise, for example).

Which, put another way: digital is a strict subset of analog where we define thresholds for '1' and '0' (or any other values), but if we dial those thresholds down into the noise floor, it doesn't really matter, does it, we're counting statistics of bits equally as well as analog spectra down there.  Quantum is a superset of analog, I suppose is the point then.

There are indeed methods nowadays, where for example, an avalanche photodiode array can be paired with individual (per-pixel) counters, thus resolving photon counts (assuming illumination low enough that chance of overlapping events is arbitrarily small) directly.  This gives an image nearly identical to a regular one, in the long-exposure limit, but the interesting part is what's enabled by more involved statistical methods, like L1-norm filtering, which, I still have to read up on, but loosely speaking and as I understand it, it's not something that's been done much before because it's computationally expensive, but in the mean time we've trivialized that, so, we can now get shockingly good image formation from rather modest photon counts.

Tim

[TUMEMBER]

You'll have to cite the patent, because it doesn't sound like you're hinting at anything original?

Patents are like Schreodinger's Cat: you don't know whether they are valid until they are challenged in court.

A patent that applies a known technique in a new context can be valid, provided it isn't "obvious to one skilled in the art". Obviously "obvious" is the key word there Recursion: see recursion.

The perfect pick for every possibility.


This is an interesting example of a natural A/D converter. It reacts with a change of state upon stimulation and secondary emission. Until it is stimulated by UV-A, it has a "weak green" color, but a small amount of 365 nm radiation causes an intense "YAG glow". Synthetic rubies (corundum) also have this (red).

[coppice]

Patents are like Schreodinger's Cat: you don't know whether they are valid until they are challenged in court.

Its more accurate to say its not valid until it is no longer being challenged in court. These days pretty much every court outcome is challenged until someone's spirit or cash runs out.

[MK]

Well photomultiplier tubes can be used both ways, count the pulses of electrons, or turn the pulsed current into an average voltage across a burden resistor at the bottom of the chain, or with a current to voltage converter at the output of the PM tube. see AofE for both modes of use.

[joeqsmith]

I have a ruler.