Re: How to explain how negative feedback lowers noise?
Gentlemen,
I have been following this splendid discussion with great interest. There are so many ways you can discuss amplification and noise, negative feedback or no feedback, hopefully not positive feedback. However, in all statements and comments made so far one important term is never mentioned and that is Transfer Function, the relationship between input and output.
When you have an input signal plus noise you wish to predict what you would have at the amplifier output. And the answer should be to use the amplifier transfer function. I believe most amplifiers would have a transfer function describing a low-pass filter.
If you have an amplifier using negative feedback and the gain of is set to +1, then the input signal will appear at the output unaffected. Thus no noise attenuation. There is no cancellation but instead a continuous balance occurs keeping the output the same as the input. Low frequency components track better than high frequency components do because the open loop gain drops at higher frequencies.
If you wish to design a multistage amplifier with low phase noise factor you should maximize the first amplifier gain and the noise factor contribution from the following stages will be small.
Does anyone want to comment on this?
G?ran
|
Re: How to explain how negative feedback lowers noise?
Barry,
Yes! In fact AGC is negative feedback.
GOD Bless and Thanks,
rich!
toggle quoted message
Show quoted text
On 3/25/2021 4:16 PM, n4buq wrote: So is negative feedback more analogous to AGC?
Thanks,
Barry - N4BUQ
----- Original Message -----
From: "Jeff Dutky" <jeff.dutky@...>
To: [email protected]
Sent: Thursday, March 25, 2021 4:04:37 PM
Subject: Re: [TekScopes] How to explain how negative feedback lowers noise?
Barry,
The sighs were because the idea that negative feedback in an amplifier
reduces noise in the output signal, whether that noise is present in the
input, or is introduced elsewhere in the amplifier, is erroneous. In the
example of noise cancelling headphones the noise is a separate source from
the main signal being amplified. In that case converting the noise signal to
an inverted signal and adding it to the signal of interest does, indeed,
reduce the noise in the summed output, but that bears little relation to the
erroneous idea that negative feedback just reduces noise.
The sighs were specifically because there had been several posts debunking
and bemoaning the error, followed by a post, presumably from someone who had
not read the rest of the thread, modeling exactly the erroneous belief that
had just been debunked and providing a purported explanation that only
served to promote the error.
-- Jeff Dutky
|
Re: How to explain how negative feedback lowers noise?
So is negative feedback more analogous to AGC?
Thanks,
Barry - N4BUQ
toggle quoted message
Show quoted text
----- Original Message ----- From: "Jeff Dutky" <jeff.dutky@...>
To: [email protected]
Sent: Thursday, March 25, 2021 4:04:37 PM
Subject: Re: [TekScopes] How to explain how negative feedback lowers noise?
Barry,
The sighs were because the idea that negative feedback in an amplifier
reduces noise in the output signal, whether that noise is present in the
input, or is introduced elsewhere in the amplifier, is erroneous. In the
example of noise cancelling headphones the noise is a separate source from
the main signal being amplified. In that case converting the noise signal to
an inverted signal and adding it to the signal of interest does, indeed,
reduce the noise in the summed output, but that bears little relation to the
erroneous idea that negative feedback just reduces noise.
The sighs were specifically because there had been several posts debunking
and bemoaning the error, followed by a post, presumably from someone who had
not read the rest of the thread, modeling exactly the erroneous belief that
had just been debunked and providing a purported explanation that only
served to promote the error.
-- Jeff Dutky
|
Re: How to explain how negative feedback lowers noise?
Barry,
The sighs were because the idea that negative feedback in an amplifier reduces noise in the output signal, whether that noise is present in the input, or is introduced elsewhere in the amplifier, is erroneous. In the example of noise cancelling headphones the noise is a separate source from the main signal being amplified. In that case converting the noise signal to an inverted signal and adding it to the signal of interest does, indeed, reduce the noise in the summed output, but that bears little relation to the erroneous idea that negative feedback just reduces noise.
The sighs were specifically because there had been several posts debunking and bemoaning the error, followed by a post, presumably from someone who had not read the rest of the thread, modeling exactly the erroneous belief that had just been debunked and providing a purported explanation that only served to promote the error.
-- Jeff Dutky
|
Re: 485 super weak brightness control
Just wanted to say thank you for continuous support. I haven't abandoned
the ship just yet but I realised I spent way too much time on a scope
project and haven't done enough paid repairs to bring the money back home
to the family. I need to put it aside for few days. Fresh start with bit of
a distance between the attempts won't hurt.
toggle quoted message
Show quoted text
On Thu, 25 Mar 2021, 13:00 Raymond Domp Frank, <hewpatek@...> wrote: On Thu, Mar 25, 2021 at 06:50 AM, Ozan wrote:
In this video I think I know what is happening. You have variable
horizontal
knob out (time 0:13 in the video). Your 5ns B gate is 300ns wide instead of ~
130ns Raymond measured so this is supporting the theory of variable time
scale. When you switch to 10ns it is the same RC time constant as 5ns (2x is
in horizontal amp) but because variable time is disabled for B other than
1/2/5n, B-gate suddenly looks like it became shorter. If you push in the
variable time scale knob my guess is you will see 5ns width same as 10ns
width, not jumping to a smaller time. I tried it and can replicate what you
are seeing with variable time scale set.
I noticed this a few days ago, when I observed the unlocked variable
timebase knob and had to fine-tune my settings in order to get exactly the
screens that Ondrej showed in his video, so I wrote and suggested:
*** QUOTE ***
Gentlemen,
First of all, in order to make life as easy as possible, it would help
me/us if we would "standardise" our settings:
- A-ramp at top
- A-gate below that
- B-gate below that again
- Variable timebase knob in the "calibrated" position (pushed in)
- Delay Time Position to minimum (0)
- Observation 'scope (2465B?) in non-delayed mode
- Trigger on A- gate trace
My observations match Ozan's latest ones.
** UNQUOTE **
I thought that things were made unnecessarily complex by using the delayed
display on the observation 'scope and variable time base in the video.
Raymond
|
Re: How to explain how negative feedback lowers noise?
Yes, I understand how these work. My question was more about why the explanation where noise from the first stage of an amplifier is fed back (negated, inverted, or, ?) to the input which seems much like a noise-canceling headset to me. I inferred from the sighs that the previous explanation was incorrect and (sorry) but I don't understand why that's the case.
Were the "sighs" meant to indicate that the example given was incorrect or, perhaps, just more of the same already-covered explanations? Sorry - sighs can a bit more meaningful when combined with an eyeroll.
Thanks,
Barry - N4BUQ
toggle quoted message
Show quoted text
----- Original Message ----- From: "G?ran Krusell" <mc1648pp@...>
To: [email protected]
Sent: Thursday, March 25, 2021 3:18:58 PM
Subject: Re: [TekScopes] How to explain how negative feedback lowers noise?
Hi Barry,
No sighs from my desk, your question is a good one. I think these modern
headphones work in the following manner, a small amount of noise comes
through the leather in your headphones. At the same time the external noise
is picked up by a small microphone in the headset, amplified by an inverting
amplifier and added to the direct noise to you ear in equal amounts. The
noise is thus attenuated. And it does work, I have one such headphone from a
well-known company.
G?ran
|
Re: How to explain how negative feedback lowers noise?
Hi Barry,
No sighs from my desk, your question is a good one. I think these modern headphones work in the following manner, a small amount of noise comes through the leather in your headphones. At the same time the external noise is picked up by a small microphone in the headset, amplified by an inverting amplifier and added to the direct noise to you ear in equal amounts. The noise is thus attenuated. And it does work, I have one such headphone from a well-known company.
G?ran
|
Re: How to explain how negative feedback lowers noise?
On Thu, Mar 25, 2021 at 08:29 PM, n4buq wrote:
Is there an analogy between this subject and the way noise-canceling
headphones work? I'm sorry if that's a silly thing to introduce into this
conversation but it might help me understand this a bit better (or not).
There certainly is, in a way, between the title of this thread and noise-canceling headphones: In noise-canceling headphones, a fraction of the outside environmental noise is in anti-phase added to (effectively subtracted from) the intentional sound (music, speech) and the result is produced by the speakers in the headphone; a nice example of negative feedback. The system works within a limited frequency-dependent amplitude band. The electronics' automatic adjustment usually is audible for a short period after putting on the headphones and when changing its position on the head, causing some sort of audible "sinking in" impression. The noise-canceling headphones' circuits can not remove oreven reduce noise present in the signal as fed from the source. OTOH, many noise-canceling headphones have a limited frequency response, effectively forming a low-pass filter for all audio, giving a noise-suppressing effect but that's got nothing to do with the noise-canceling "mechanism" itself. Raymond
|
Re: Persuading a 7S12 to play nice with a 7934.
The solution with reduced value of R693 seems to have disadvantages also, otherwise Tek would have done so I think.
That might be the reason that P600 has been introduced to enable/disable interdot blanking. In the open position the PNP 2N2907A is shut off completely. In the closed position the situation is as before. (I avoid Q690/Q694 numbering.)
The next version of P600 is wired differently. When closed it it ties the base of the 2N2907A to GND. The connection of the base to R593 is deleted. In the closed position interdot blanking works, and works always, i.e. independent of what happens at A16 and B7. When P600 is open the situation is the same as with the previous P600 version. This version explains why I see the dot blanking in other traces so clearly. About 50% blank of between samples period is completely blanked.
Albert
|
Re: How to explain how negative feedback lowers noise?
Is there an analogy between this subject and the way noise-canceling headphones work? I'm sorry if that's a silly thing to introduce into this conversation but it might help me understand this a bit better (or not).
(I think I hear the sighs already but I'm still going to send this...)
Thanks,
Barry - N4BUQ
toggle quoted message
Show quoted text
----- Original Message ----- From: "Raymond Domp Frank" <hewpatek@...>
To: [email protected]
Sent: Thursday, March 25, 2021 2:12:48 PM
Subject: Re: [TekScopes] How to explain how negative feedback lowers noise?
On Thu, Mar 25, 2021 at 07:14 PM, Tom Lee wrote:
Sigh.
I never knew sighing in chorus existed, defies relativity's simultaneity
concept.
Raymond
|
Re: How to explain how negative feedback lowers noise?
On Thu, Mar 25, 2021 at 07:14 PM, Tom Lee wrote:
Sigh.
I never knew sighing in chorus existed, defies relativity's simultaneity concept. Raymond
|
Re: TEK465 Horizontal trace
On Thu, Mar 25, 2021 at 06:45 PM, Dave Peterson wrote:
If it's just the "A-fire" diode dead, then I have 7 to pick 4 from. I'm hoping
I can use our hacked curve tracer to match two pairs. It certainly allowed me
to see the variation in the 151-0212-00 vert amp xtors.
Those TD's should work. Just go through the applicable adjustment procedure in the SM, no special equipment needed. Results will be better than by matching. Raymond
|
Re: TEK465 Horizontal trace
On Thu, Mar 25, 2021 at 05:25 PM, Dave Peterson wrote:
I found it likely _not_ merely coincidence that this transistor has failed
_and_ has some form of manufacturing defect. An engineering axiom brought to
my attention some time ago is that engineers/designers/manufacturers rarely do
or add anything without good reason. I doubt the "goo" is there only to secure
the leads. Mind too that these particular components were manufactured by
Tektronix in the late 60's or early 70's. I don't think the industry had quite
settled down at that time. Wink. My use of passivation seemed an apt term.
Perhaps I should have stuck with goo. Wink.
I really don't see a manufacturing defect. Your image shows the underside (base) of the package, with its glass(-like) lead feed-throughs. Packages are purchased by the transistor manufacturer complete with leads and feed-throughs. There is a plateau on top, where the semiconductor chip (die) is affixed by the transistor manufacturer. The chip itself is passivated/glassivated during chip production. A cap is placed on top of the base during manufacturing, completely enclosing the die hermetically. TO-5 cases' bottoms often look similar. With TO-39's, the glass is usually just visible as a thin "ring" around each lead (usually base and emitter), flush with the package bottom. Raymond
|
Re: TEK465 Horizontal trace
On Thu, Mar 25, 2021 at 05:32 PM, Leon Robinson wrote:
Could he mean the glass lead seal where the leads exit?
He does, according to the legend with the image showing the glass seal with the leads protruding. Of course, that's got nothing to do with passivation of the chip inside, apart from providing a hermetic seal to the case. Raymond
|
Re: How to explain how negative feedback lowers noise?
Sigh.
Tom
Sent from an iThing; please excuse the terseness and typos
toggle quoted message
Show quoted text
On Mar 25, 2021, at 6:56, "Torch" <tekscopes@...> wrote: 1 + -1 = 0
Unwanted noise is a side effect of amplifying a desired signal. The ratio of internal noise to desired signal is greatest at the first stage of amplification and magnified in further stages. If a portion of the signal coming from the first amplification stage is inverted and fed back to the input, it will tend to cancel the internally generated noise more so than the desired signal, improving the ratio seen by the input of the next stage.
|
Re: 0.25R resistor in calibrator ground
My guess would be to save the calibrator from a backed signal (feeding an external signal into the calibrator, AC or DC).
Don Bitters
|
Re: 0.25R resistor in calibrator ground
?Exactly. High-gain plugins (Type B, D, etc) were seeing hum on the calibrator. The resistor made the bottom ranges usable. In the 535 Mod Summary, it's Mod #1412, effective S/N 5001. Dave Wise My "S/N 75" 535-535A prototype needs this mod. ________________________________ From: [email protected] < [email protected]> on behalf of keantoken via groups.io <keantoken@...> Sent: Thursday, March 25, 2021 9:28 AM To: [email protected]; [email protected]Subject: Re: [TekScopes] 0.25R resistor in calibrator ground I noticed the same resistor, also open, on my 561B! On Thursday, March 25, 2021, 11:18:00 AM CDT, cmjones01 <chris@...> wrote: I've just noticed something odd about the calibrator outputs on my 535A scope and 127 plugin power supply. They both have a 0.25R resistor in series with the calibrator socket's ground connection. The reason I noticed is that the 127's resistor is open-circuit, so connecting the calibrator to another piece of equipment left the ground disconnected, with the expected symptoms. I couldn't find any information in the manuals as to why this resistor is there. Why is it? To avoid ground loops? Chris
|
Re: TEK465 Horizontal trace
I found a set of 4 on eBay that are supposed to be NOS and tested good:
These match what's called out in the SN<250K service manual: 152-0125-00, 4.7mA, 15pF.
Not that I have much of a clue what I'm talking about. However, per something I found, that I'm going to have to find again, probably the horizontal debug manual I've referred to before, you can test for basic function in-situ. And the A-sweep fire diode tested open - high voltage. So I suspect that one is dead. The other three seemed to be ok.
If it's just the "A-fire" diode dead, then I have 7 to pick 4 from. I'm hoping I can use our hacked curve tracer to match two pairs. It certainly allowed me to see the variation in the 151-0212-00 vert amp xtors.
I also have sets of plug-in sockets coming in this Mouser order. I'm going to re-build the tunnel diodes on this scope to be plug-in instead of soldered in. That'll give me more opportunity to fix and tune this aspect. I.e. I may well be buying more diodes soon.
A little knowledge, a little guesswork, a little luck. In other words, typical repair procedures. Thanks for the info. I'll let you know what comes of it.
Dave
toggle quoted message
Show quoted text
On Thursday, March 25, 2021, 10:31:28 AM PDT, Jeff Dutky <jeff.dutky@...> wrote:
Dave,
I've had a search through the CDPC (Common Design Parts Catalog) for the tunnel diodes (Tek PN 151-0386-00) in the 465. The service manual says that they are selected from 1N3718, but the Tek part number is not listed as a selected part (as with transistors, whose part number appears to change from 151- to 156- when selected by Tek internal processes, it looks like diodes change part number from 152- to 153- after selection, but I'm just inferring this from what's listed in the Matched or Selected diodes table on page 12-6 of the CDPC). The CDPC also indicates a different vendor part number, STD927 rather than 1N3718, so I don't know what's up with that.
The specs given for STD927 are a 10 mA part with 25 pF capacitance, which appear (based on the Russian Tunnel Diodes list on TekWiki, ) to be matches to 3I306M, 3I306N, 3I306S, 1I305A, 1I305B, 1I308V, 1I308G, 1I308J, 1I308K, 1I308I, 1I308E, or 1I308D (where the rated current meets or exceeds 10 mA and the capacitance is around or less than 25 pF). Most of those are available on eBay.
I also looked for the Tek PN on eBay and Amazon with no success.
-- Jeff Dutky
|
Re: TEK465 Horizontal trace
?Dave P, that bottom view of a TO-72 is only epoxy sealing the annular joint between header and can. If that's a Silicon transistor, most likely the can is full of inert gas and the die is passivated with a grown layer of SiO2. Dave Wise ________________________________ From: [email protected] < [email protected]> on behalf of Michael W. Lynch via groups.io <mlynch003@...> Sent: Thursday, March 25, 2021 9:49 AM To: [email protected]Subject: Re: [TekScopes] TEK465 Horizontal trace On Thu, Mar 25, 2021 at 11:25 AM, Dave Peterson wrote:
Pardon if I was speaking too off the cuff and using language or wording that
was inaccurate, misleading, or wrong. I'm a newb at this TekScope thing.
By passivation I meant the material used to fill a TO-72 can. See picture:
/g/TekScopes/photo/262237/0?p=Name</g/TekScopes/photo/262237/0?p=Name> ,,,20,0,0,0
Dave, A picture is worth a thousand words. Thanks for clearing up the questions. -- Michael Lynch Dardanelle, AR
|
Re: Recapping the 535 HV (was Re: 535A HV issue)
ADDENDUM
Grid/Feedback/Cathode caps that are short enough to lie on their
backs can be installed that way. The baseplate is 70mm wide.
T1 is 26mm + 3mm for wire bends. The feedback resistor
string is 5mm wide. That leaves 12mm per cap, which opens
up more part numbers, especially if you include 2kV.
We're still single-sourced on the tripler.
Grid/Feedback/Cathode:
3kV Wima MKP1W021005B00JSSD $.89 6x15x27mm
Panasonic ECW-HC3F822
ECW-HC3F103 $1 8x16x27mm
2.5kV Kemet F464BP103J2K5A $1 9x15x18mm
2kV Wima MKP1U021004D00KSSD $.70 7x14x18mm
MKS4U021004D00KSSD $.68 7x14x18mm
FKP1U021006B00MSC9 $1.45 11x21x32mm
Epcos/TDK B32672L8682J $.67 6x11x18mm
B32672L8822K000 $.70 6x12x18mm
B32613A2103J008 $.84 11x17x27mm
B32652A2682K000 $.94 11x19x18mm
B32653A2682K000 $1.30 9x17x27mm
Kemet R76UI210050H7J $.76 12x13x18mm
R76UI182050H4J $.67 6x12x18mm
R76UN2100SE30J $1 6x15x27
PHE450SB5100JR06 $1 7x13x18mm
F463DB103K2K0Z $1 6x15x26mm
Nichicon PPB2201680KGL $1.30 6x15x27mm
On the left, place Grid and Feedback on their backs.
Place the feedback resistor string to the right of Feedback.
To the right of that, place Cathode on its back.
Near the upper right, place T1 flat with leads facing left.
Above the upper right strips, hang T3 vertically.
Below the lower right strips, hang T2 at an angle, head down/leads up,
avoiding the ground lug and straddling the filament wires.
|
G?ran Krusell
Raymond Domp Frank
Dave Peterson