Path: news.eternal-september.org!eternal-september.org!.POSTED!not-for-mail
From: Don Y <blockedofcourse@foo.invalid>
Newsgroups: sci.electronics.design
Subject: Re: "Colorimeter"
Date: Wed, 21 May 2025 05:42:46 -0700
Organization: A noiseless patient Spider
Lines: 93
Message-ID: <100khoc$2r1fh$1@dont-email.me>
References: <100ao53$hkhu$1@dont-email.me>
 <1rchzv3.ijlk1h1prnmzmN%liz@poppyrecords.invalid.invalid>
 <100bsmh$rkc8$2@dont-email.me>
 <1rcjue0.nnrdfm7rw09cN%liz@poppyrecords.invalid.invalid>
 <100ievd$2bk2r$1@dont-email.me>
 <1rcoljh.ueditn2zlduuN%liz@poppyrecords.invalid.invalid>
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Injection-Date: Wed, 21 May 2025 14:42:55 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="602d21e6d16a06c08f09f8ac8f619b45";
	logging-data="2983409"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+W3au4eO52230D/dsHSrxh"
User-Agent: Mozilla Thunderbird
Cancel-Lock: sha1:e3Do5nXO5vhULm83h1DTedyghZc=
Content-Language: en-US
In-Reply-To: <1rcoljh.ueditn2zlduuN%liz@poppyrecords.invalid.invalid>

>>>> That might be better than a varied filter.  But, probably require finer
>>>> control (or sensing) of its current orientation.
>>>
>>> If it is spinning steadily, all you need is a synchronising pulse at
>>> some point once per revolution and a wide spectrum photocell with an
>>> optical slit and a lens.  Software can work out the wavelength from the
>>> rotational speed and the known characteristics of the prism.  The
>>
>> Of course.  But, if spinning faster than your integration interval,
>> I suspect any jitter in your angular resolution might be difficult
>> to factor out of the mix.
> 
> Mount the prism on a a flywheel and spin it rapidly.  The only jitter
> might come from errors in the timing pulse (or knackered bearings!).

But each wavelength (or, wavelength window) strikes the detector
for a certain portion of time -- before the NEXT "wavelength window"
comes around.

You have to sample the detector's output within that window and,
ideally, at the same time in that window as the presented wavelength
is continuously varying *in* that window.  Then, move on to the
next "wavelength window" immediately thereafter.

Sampling jitter within a window corresponds to spectral resolution;
the more jitter, the wider the range of wavelengths potentially
involved in the sample (over time).  As sampling the detector
is a discrete time event (the interval between samples being the
width of the window), how frequently you do this further defines
the spectral resolution.

Without a real design, I'd be leary of making any commitments as
to what could be achieved, there.

By contrast, a slow spinning prism has the sampling interval determine
the window width with each wavelength within that window seeing the
same "exposure time".

The drawback would be the *entire* spectrum would be sampled at a lower
rate, corresponding to the slower rotation of the prism.  You don't see
any wavelength again until a complete "cycle" has occurred.

> One way of obtaining a jitter-free timing pulse would be to reflect a
> known pattern of light off the faces of the prism into the photocell;
> use the software to recognise it and make corrections for any long-term
> speed drift.

If the motion is smooth and stable over a revolution (cycle) or two,
you could likely servo the drive to maintain a frequency lock.
Trying to adjust the sampling strobe might leave you with more
*apparent* jitter (because you don't have infinitely fine resolution
in time; the sampling event's motion in time looks like jitter)

>> This would, instead, suggest a slower rotation so the prism feeds
>> the detector a single wavelength for a longer (continuous) period.
>>
>> That means the time to get a sampling of the spectrum is multiplied
>> by the integration interval.  If, instead, you could get "quick peeks"
>> at each wavelength "quickly", and the more precise integration "later",
>> you have more data to work with, sooner.
> 
> If it spins faster you can simply integrate multiple 'passes' for as
> long as you want until the noise is negligible.  The frequency response
> of the photocell and head amplifier is likely to be far wider than any
> mechanical system needs, so the physical narrowness of the slit and the
> distance from the prism will set the resolution limit.  .A narrow and
> distant slit will give higher resolution at the expense of a worse S/N
> ratio, which can be overcome with a longer integration time.

I'll have to do some back-of-napkin figuring before I present
the idea.  And, as always, see what other pertinent details they've
not provided.  I'm not keen on being dragged into a design effort...

>> [This is the approach I have historically taken with data acquisition
>> as it lets me trade response time for resolution, dynamically]
> 
> Yes, it has many advantages.

Particularly when you *do* want data of differing qualities (and
can leverage that to your advantage).

>>> The same hardware could be used for an expensive high-resolution device
>>> or a cheap and cheerful version - the software and the time to reach a
>>
>> "cheerful"?
> 
> "Cheap and cheerful" is a slang [UK English] expression meaning  a quick
> rough estimate or goods that aren't intended for serious long-term use.

"Quick and dirty", "spit and baling wire", "bubble gum and shoe strings",
"good enough for government work", "mickey-mouse", "jury-rigged",
"jerry-built", etc.