Path: ...!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!eternal-september.org!.POSTED!not-for-mail
From: Cursitor Doom <cd@notformail.com>
Newsgroups: sci.electronics.design
Subject: Re: AI
Date: Sat, 11 Jan 2025 00:07:20 +0000
Organization: A noiseless patient Spider
Lines: 127
Message-ID: <ucd3oj5ofvthssar7ki7jtiq9vb4cr2ljc@4ax.com>
References: <vlrssv$5d0i$1@dont-email.me> <vls368$7o9u$2@dont-email.me> <vls7nh$8od8$1@dont-email.me> <vls95v$8sjm$1@dont-email.me>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Injection-Date: Sat, 11 Jan 2025 01:07:25 +0100 (CET)
Injection-Info: dont-email.me; posting-host="5b9487b7abedd9360d40ea2e7caecf24";
	logging-data="317647"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1/+PlQg3yZWL6juwLVEYZMT6G53qS4LrzQ="
User-Agent: ForteAgent/8.00.32.1272
Cancel-Lock: sha1:qVgyqn2q+Gh7vLSaW8anCAH9Pkc=
Bytes: 6635

On Sat, 11 Jan 2025 00:10:55 +0100, Jeroen Belleman
<jeroen@nospam.please> wrote:

>On 1/10/25 23:41, Cursitor Doom wrote:
>> On Fri, 10 Jan 2025 22:28:40 +0100, Jeroen Belleman wrote:
>> 
>>> On 1/10/25 20:37, Cursitor Doom wrote:
>>>> Gentlemen,
>>>>
>>>> I've just been goofing around with X's AI assistant, Grok. After
>>>> feeding in a few straightforward questions and getting surprisingly
>>>> impressive answers, I decided to try something a bit more subjective to
>>>> see how it would cope. Here's what happened...
>>>>
>>>>
>>>> My question:
>>>>
>>>> Which of the common 'classic' oscillator designs is the most impressive
>>>> or ingenious from a technical standpoint?
>>>>
>>>> Grok's Answer:
>>>>
>>>>   From a technical standpoint, several classic oscillator designs stand
>>>>   out
>>>> for their ingenuity, but one that is often highlighted for its elegance
>>>> and versatility is the Colpitts oscillator. Here's why:
>>>>
>>>> Design Ingenuity:
>>>>
>>>>       Simplicity and Efficiency:
>>>>           The Colpitts oscillator uses a simple configuration with a
>>>> transistor (or vacuum tube in older versions) and a combination of
>>>> capacitors and an inductor to create a feedback loop. Despite its
>>>> simplicity, it achieves stable oscillation with just a few components.
>>>>       Feedback Mechanism:
>>>>           It employs a capacitive voltage divider for feedback, which is
>>>> particularly ingenious. The two capacitors in series (C1 and C2) form a
>>>> voltage divider that provides the necessary phase shift for
>>>> oscillation. This method naturally provides the correct phase without
>>>> additional complex circuitry, making it both efficient and reliable.
>>>>       Frequency Stability:
>>>>           The frequency of oscillation in a Colpitts oscillator is
>>>> determined primarily by the LC tank circuit's resonant frequency, which
>>>> can be calculated as f=12?L(C1?C2)/(C1+C2)
>>>>           . This formula highlights how the design inherently stabilizes
>>>> frequency through the interaction of inductance and capacitance. The
>>>> use of capacitors in the feedback loop also helps in reducing the
>>>> impact of transistor parameter variations on frequency stability.
>>>>       Versatility:
>>>>           The Colpitts oscillator can be adapted for use across a wide
>>>>           range
>>>> of frequencies from audio to RF. By changing the values of L, C1, and
>>>> C2, one can easily tune the oscillator to different frequencies, making
>>>> it adaptable for various applications from radio transmitters to clock
>>>> generators.
>>>>       Low Noise and High Purity:
>>>>           When designed correctly, the Colpitts oscillator can offer low
>>>> phase noise, which is critical for applications like communication
>>>> devices where signal purity is paramount. The design inherently
>>>> minimizes noise by reducing the number of components in the feedback
>>>> path.
>>>>       Ease of Adjustment:
>>>>           Adjusting the frequency is relatively straightforward with
>>>>           minor
>>>> changes to the capacitor or inductor values, making it practical for
>>>> both prototyping and production environments.
>>>>
>>>>
>>>> Historical Context:
>>>>
>>>>       Invented by Edwin H. Colpitts in 1918, this oscillator was a
>>>> breakthrough in the era of vacuum tubes, which were later adapted for
>>>> transistor use. Its principles have stood the test of time, still being
>>>> taught and used today in various forms.
>>>>
>>>>
>>>> Comparison:
>>>>
>>>>       While other oscillators like the Hartley oscillator (which uses an
>>>> inductive voltage divider) or the Clapp oscillator (a variation of
>>>> Colpitts with an additional capacitor for better frequency stability)
>>>> are also significant, the Colpitts design is often praised for its
>>>> balance of simplicity, efficiency, and versatility.
>>>>
>>>>
>>>> In conclusion, the Colpitts oscillator stands out for its impressive
>>>> combination of simplicity, effectiveness, and adaptability, making it
>>>> one of the most ingenious classic oscillator designs from a technical
>>>> perspective.
>>>>
>>>>
>> **************************************************************************
>>>>
>>>> Any observations?
>>>
>>> Yeah. The resonant frequency formula is wrong. To first approximation,
>>> it's f=1/(2? sqrt(L C1 C2 / (C1+C2))). (Inline algebraic notation is a
>>> dog...)
>>>
>>> It looks like Grok has trouble reading formulae.
>>>
>>> Of course, it doesn't really know what it's talking about.
>>> It just ladles out some phrases it gathered from various web sites on
>>> the subject.
>>>
>>> Jeroen Belleman
>> 
>> The formula error is not Grok's mistake. The sqrt didn't survive the
>> cut'n'paste I'm afraid. I didn't reckon on someone spotting that so didn't
>> mention it in the OP.
>
>We need a new word for this kind of thing. I propose "plagiamaton",
>meaning something like "agent that steals ideas".
>
>For some of the more awkward phrases, I even found where it got
>them from. For example, the phrase starting with "This formula
>highlights..." seems to come almost verbatim from a particularly
>poorly written web page by Keysight.
>
>Jeroen Belleman

I suppose we're still in the early days of AI in terms of this kind of
usage, so can't expect perfection for anything too demanding (not that
this was, but asking for an opinion from a computer is a bit more
challenging than stuff like 'who holds the record for drinking the
most beer?')