Re: A collection of monographs on high accuracy electronics

On 7/5/24 01:41, JM wrote:

On Mon, 17 Jun 2024 12:40:01 +0200, Jeroen Belleman <jeroen@nospam.please> wrote:
 

On 6/16/24 23:20, JM wrote:

On Tue, 11 Jun 2024 19:11:51 -0700, john larkin <jl@650pot.com> wrote:
>

On Wed, 12 Jun 2024 02:50:19 +0100, JM <sunaecoNoSpam@gmail.com>
wrote:
>

On Mon, 10 Jun 2024 15:14:40 -0400, Phil Hobbs <pcdhSpamMeSenseless@electrooptical.net> wrote:
>

On 2024-06-09 21:43, Phil Hobbs wrote:

On 2024-06-09 20:55, JM wrote:

On Mon, 10 Jun 2024 00:29:17 -0000 (UTC), Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:
>

JM <sunaecoNoSpam@gmail.com> wrote:

On Sun, 9 Jun 2024 18:09:24 -0000 (UTC), Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:
>

Jeroen Belleman <jeroen@nospam.please> wrote:

On 6/9/24 19:02, ehsjr wrote:

On 6/7/2024 9:14 PM, JM wrote:

A collection of monographs on high accuracy electronics written
by Mr.
Chris Daykin, following his career predominantly in metrology.
>
Unfortunately Chris will be unable to complete the unfinished
monographs (having started end of life care) but there is plenty of
interest to any analogue engineer.
>
https://1drv.ms/b/c/1af24d72a509cd48/EZhO_rP5-glDmxtc4ZHycvYBhrsqmyC5tuZjt2NFFsS0gQ?e=Wq2Yj0
>

>
>
Thanks!
Ed

>
I have an issue with his definition of resistor noise power
as the product of open-circuit noise voltage and short-circuit
current. That makes no sense.
>
There's more than that, probably, but that just jumped out at
me.
>
Jeroen Belleman
>

>
It?s four times too high, for a start.
>
Cheers
>
Phil Hobbs

>
"It is shown elsewhere [1] that the noise power is four times the heat
energy which would flow down the conductors
from a warm source resistor to a matching cold resistor."
>

>
Which, if true, would solve all our energy problems, except that
thermodynamic systems would all be unstable.
>
The thermal noise power produced by a resistor into a matched load is kT
per hertz.

>

>
Sure, which is what he states.  By mentioning a hot and cold resistor
he makes it clear that net energy flow is from hot to cold, and that
the T refers to the hot source.
>

But apparently he says that it's four times larger than that.
>
I'm not making a microsoft account just to download the PDF, so if you
want to discuss it further, you could email it to me.
>
Cheers
>
Phil Hobbs
>
>
>

Bill was kind enough to send me a copy (thanks again, Bill), and right
there on P. 374, the author says,
>
Pn = 4kTB
>
which is a factor of four too high.
>

>
No it isn't. He is calculating the thermal noise power dissipated in an unloaded resistor - something (or at least the related noise voltage) which is actually required in the design process of a transducer/amplifier low S/N system.

>
What does that mean? Do unconnected resistors get hot?
>
A box of resistors could start a fire!

>
And why would that occur.  In thermal equilibrium there is no net transfer of energy either from or to the resistor (when averaged over any time interval of interest appropriate to the bandwidth of current electronic circuits).
>
The so called resistor thermal "available noise power" KTB implies there is a net power delivery from a source to a load.  In the case of maximum transfer the source must dissipate within itself exactly the same as it delivers to the load (due to having the same resistance).  However if the so called load is at the same temperature as the source it also delivers KTB to the source and dissipates KTB within its own resistance.  Thus there is no net transfer of energy between the two resistors in thermal equilibrium.  If one is at a lower temperature than the other there will be a net transfer of energy, but this will be completely dwarfed in any practical system by the energy transferred due to thermal conductivity between the two resistors.
>
So the power dissipated in a system of two equal value resistors is 4KTB.  But this also holds if the two resistors have different values, including the situation where one of the resistors is a short or open circuit (i.e. leaving a single open or short circuit resistor).  So it is entirely reasonable to state (as many engineers do) that the thermal noise power of a resistor is 4kTB.
>

>
I've never seen it stated like that. It doesn't strike me as very
useful to state it like that.
>
It's very kind of you and of Chris too have made this text available
for all, thank you. However, there is some work to be done to finish
it. For example, in 3-5 5 "Noise matching transformers", it is stated

 You're lucky it's not one of my drafts, in which case there would be typos on every line!
 

that the noise resistance goes down with the root of N, the number of
parallel transistors. That is not correct: It goes down with N.
>
It is true that the noise voltage drops with sqrt(N), but the noise
_current_ rises with sqrt(N). So Rn = Vn*sqrt(1/N)/(In*sqrt(N)) =
Vn/(N*In).
>
I also have an issue with the use of passive matching transformers
for platinum resistance thermometers. This will obviously not work
near DC.

 You're not seeing the wood for the trees.
 Daykin's industrial experience was gained at ASL whose primary products were thermometry bridges.  They
had a resolution of about a mK.  To achieve this the resistance thermometers are part of an ac bridge (the other
leg being a ratio transformer) followed by a synchronous detector.
 These monographs all basically discuss the electronics used in these bridges.
 

>
Jeroen Belleman

An AC bridge. That wasn't mentioned in that section, but yes, that
would work. That's another suggestion for improving the text then.
Jeroen Belleman