Now you are talking my language . Bearings indeed !
The VLF noise I was thinking of was small generators at a different frequency albeit a small difference .
As for noise distribution in resistors is concerned , is it different to two transistors ? Is it not the fact they are not synchronized that makes it work ?
No, all generators on an AC grid are either phase locked (=identical frequency) or severely broken!
Two transistors can be added in parallel to reduce signal-to-noise. They do this by increasing noise by 3dB (exactly the same as a pair of resistors) but signal by 6dB. This works best when most of the incoming signal power can be dumped in a resistor rather than the active devices - commonly the case in audio.
Two transistors can be added in parallel to reduce signal-to-noise. They do this by increasing noise by 3dB (exactly the same as a pair of resistors) but signal by 6dB. This works best when most of the incoming signal power can be dumped in a resistor rather than the active devices - commonly the case in audio.
With generators that was true in the past . Self excited or externally excited . Most are externally excited ( synchronizing to the grid ) . The generators I am thinking of are using a semiconductor oscillator that is tied into the mains . People who do this can sell electricity to the grid . I doubt it will be anything like as good at a few hundred tons of power station rotor . I only say this because someone wrote to me from Scandinavia saying it is a known problem there . Doubtless a phase lock loop can be used and the problem is imagined ?
The grid, even at local level, has a very low impedance. You can't connect to it without damage unless you are already locked, or very close. For a generator, connection produces locking. For a power oscillator failure to lock means fairly instant destruction? Harmonics or sub-harmonics might occur, but beats are most unlikely.
In fact the solution is easy to this problem . The inverter master oscillator can be a Schmidt trigger from the mains itself ( filtered to give a sine wave or trigger the inverter ) . I notice modified sine wave inverters are allowed which should by the impedance of the mains be swapped by the largest generator .
If the error is small little current will flow . Thinking about it, it is easiest in engineering terms not to have an error . Just like a externally excited alternator it is easy to engineer and doesn't require even a phase lock loop .
I seem to remember when there was a major grid failure in the UK a few years ago it took some hours to get the grid back . I was told panic set in at the GEGB ( National Grid these days I think ? ) . When a generator falls out often it is reluctant to feed back in , worse when the grid is lost completely . It is said few engineers even understood how the grid worked when this happened , de-nationalization means less engineers trained in the subtleties . I remember something about the USA having problems in the early days , a power station damaged . If the grid becomes long enough there is potential for a 1/4 wave resonance ( multiples of just over 780miles ? if 60 Hz and just under 1000 at 50 Hz ) . I suspect reconnecting a grid must be done so as to avoid unfortunate cross country cable lengths ?
If the error is small little current will flow . Thinking about it, it is easiest in engineering terms not to have an error . Just like a externally excited alternator it is easy to engineer and doesn't require even a phase lock loop .
I seem to remember when there was a major grid failure in the UK a few years ago it took some hours to get the grid back . I was told panic set in at the GEGB ( National Grid these days I think ? ) . When a generator falls out often it is reluctant to feed back in , worse when the grid is lost completely . It is said few engineers even understood how the grid worked when this happened , de-nationalization means less engineers trained in the subtleties . I remember something about the USA having problems in the early days , a power station damaged . If the grid becomes long enough there is potential for a 1/4 wave resonance ( multiples of just over 780miles ? if 60 Hz and just under 1000 at 50 Hz ) . I suspect reconnecting a grid must be done so as to avoid unfortunate cross country cable lengths ?
Last edited:
A Black Start takes time - you have to balance generation and load to keep sufficiently close to 50Hz/60Hz otherwise automatic trips will make the whole thing collapse again.
It was the CEGB - my former employer. Nowadays National Grid would supervise; various private generator and distribution companies would hopefully do what they are told. I assume they keep their plans up to date and run exercises from time to time.
It was the CEGB - my former employer. Nowadays National Grid would supervise; various private generator and distribution companies would hopefully do what they are told. I assume they keep their plans up to date and run exercises from time to time.
+1
long leads will likely add more inductance
If resistor inductance is issue, why not to use smallest smt size instead
dvv, been there, done that ...Also, Scott, if you haven't seen it, do take a look at his somewhat lengthy test of op amps. I did and agree on some points, but disagree on others; a good example of the measurements saying one thing and practical experience and human ear disagreeing.
Frank, I also expect you to study that same text, I'll drop around and question you on this.
I thought two resistors in parallel have 3dB less thermal noise because the resistance is halved. Same mechanism behind parallelling active devices.
+1
long leads will likely add more inductance
If resistor inductance is issue, why not to use smallest smt size instead
What has been useful here is for me and others to discount inductance as a sonic trait even up to 50 kHz in typical circuits . The PU I use doubtless has high inductance and must be the dominant effect ? Low inductance resistors by coincidence work well sonically . This is because if someone is going to charge between $8 and 20 for a resistor is has to be low inductance . Reading many text books etc wire wound has a good reputation . Inductance will be high . A friend said don't think of hiss as a bad thing , sometimes the right type of hiss is good . For example the hiss from small power transistors like BD 135/6 looks very nice on the spectrum analyzer . Almost like snow and about 4 db above the resistor noise ( or whatever it is ) . John is right . The resistor noise looks jagged by comparison . For a small degradation it might be a good trade off ? How many times have I used a transistor so as to be technical when a resistor should be OK and thought the transistor version sounded better . Perhaps it was added noise and nothing else ?
I will try to do resistors in paralell again . It is possible the " snow " effect is better ? I must have graphs of this somewhere . I looked the other day and didn't find them . They clearly show the snow .
Maybe even analog circuits need dither ? Transistors being what they are may like a bit of truly white noise ?
One thing I suspect is true . Two op amps ( transistors ) in parallel not only have lower noise they have nicer noise ? To my ears it is a gentle noise which is not exactly like the single version . If I raise the volume so as to mimic the noise level it doesn't sound the same . There are many factors involved so hard to say other things are not happening . The noise of resistors to the op amp will be doing something which might produce even more of the snow effect .
My friend Peter Danial was a high voltage man for the National Grid . When privatized the new bosses were getting rid of items in storage . When challenged about it the new bosses said the French have promised to supply anything NG needed . This boss had not understood the unique nature of some equipment . This grew into a conflict . Secretly the supplies were hidden for emergency use ( still hidden I beleive ) , the technicians feeling they needed to take responsibility for this . The stupidity of this was the boss had decided storage was expensive or similar as his only reason . The fact the guys found somewhere to hide it means it was the caprice of a control freak and nothing to do with common sense or economics , the hiding place another NG building in a remote location said to be used for something else . The madhouse no less . The London Underground is different . Many fixtures and fitting date to circa 1910 . These are made as required . As they work well enough they continue . Fukushima proves this is not a uniquely British disease .
My friend Peter Danial was a high voltage man for the National Grid . When privatized the new bosses were getting rid of items in storage . When challenged about it the new bosses said the French have promised to supply anything NG needed . This boss had not understood the unique nature of some equipment . This grew into a conflict . Secretly the supplies were hidden for emergency use ( still hidden I beleive ) , the technicians feeling they needed to take responsibility for this . The stupidity of this was the boss had decided storage was expensive or similar as his only reason . The fact the guys found somewhere to hide it means it was the caprice of a control freak and nothing to do with common sense or economics , the hiding place another NG building in a remote location said to be used for something else . The madhouse no less . The London Underground is different . Many fixtures and fitting date to circa 1910 . These are made as required . As they work well enough they continue . Fukushima proves this is not a uniquely British disease .
Two resistors in parallel have 3dB less thermal noise voltage but 3dB more thermal noise current - so exactly the same thermal noise power. This is because their noise is uncorrelated.vacuphile said:
Paralleling active devices is different to the extent that the wanted signal (often a current) is correlated so is boosted by 6dB while the noise current is only boosted by 3dB.
Only in Essex.nigel pearson said:
In large organisations providing essential public services it is sometimes necessary for the staff to creatively ignore their bosses. Bosses tend to believe what salesmen/politicians/consultants tell them to believe, or what they were told in business school. The working level staff often know better. Their commitment was/is(?) to serve the community (e.g. keep the lights on); the boss just wants to ensure he doesn't get the blame when the lights go out.
Last edited:
I have and sent some of the samples.
...
Good man!
Hope you'll agree it was worth the time and trouble, Groner did some nice work.
Since I'm well stocked with AD op amps, you don't have to send me any.
There seems to be some value in close coupling of resistors to reduce inductance . The same trick in series will raise inductance if that is desired . I had to use inductors rather than resistors to get meaningful results . The exact alinement seemed to make little difference . To get a 25 % boost if wanting it seems excellent . The reduction seems more or less identical saying to me the measurement is reliable . As a resistor is on a ceramic former I doubt the inductance change will be high ( < 5% ? ) . I return to my original conjecture that the noise is nicer if from two identical noise sources .
You may be confusing 'ability' to reduce inductance with 'value' in doing so.nigel pearson said:
You have shown that two inductors in parallel can couple to give slightly less than the uncoupled inductance (212uH vs 250uH). Exactly what theory predicts.
You have also shown that the alignment details makes a significant difference (1061uH vs 1323uH for series) - contrary to what you say. Exactly what theory predicts.
I can see no connection whatsoever between this set of measurements and your entirely false conjecture that combining two identical white noise sources gives something 'nicer' than white noise.
That's the point . The alignment can be important especially if real inductors . There was one other version I didn't include .
There is a small possibility that lets say 2 x 100 K might be better than 1 x 51 K . The reason can not be inductance . The long shot is that noise from two resistors is not the same as one of the same product range . I gave the examples of inductance as the changes were larger than I expected . Although reasonably obvious keep them apart . With many PCB's smaller than a packet of cigarettes these days it might not be easy .
I did not confuse the reduction in the bigger picture . It is mostly a lost cause and I know have doubts it mattered . What may be important is the carbon composition and foil resistors whilst not measuring very differently even in an MC pick up circuit do have very different noise . The standard metal film resistor seems less good except when in pairs . The latter I have no easy explanation for . I do accept imagination can be part of it . For CC it is not imagination , they have that 1950's sound .
There is a small possibility that lets say 2 x 100 K might be better than 1 x 51 K . The reason can not be inductance . The long shot is that noise from two resistors is not the same as one of the same product range . I gave the examples of inductance as the changes were larger than I expected . Although reasonably obvious keep them apart . With many PCB's smaller than a packet of cigarettes these days it might not be easy .
I did not confuse the reduction in the bigger picture . It is mostly a lost cause and I know have doubts it mattered . What may be important is the carbon composition and foil resistors whilst not measuring very differently even in an MC pick up circuit do have very different noise . The standard metal film resistor seems less good except when in pairs . The latter I have no easy explanation for . I do accept imagination can be part of it . For CC it is not imagination , they have that 1950's sound .
Well we knew that alignment matters for mutual inductance. One point at issue was that swapping ends makes no difference because it leaves the helix winding sense unchanged. The other point is the resistor inductance is of no consequence for audio frequencies.
Why would 100K||100K be better than 51K? I could postulate that green is a better colour than red, and sincerely believe it, but that does not make it so. Two different resistors may have different noise spectra, but two identical resistors are likely to have similar noise spectra unless one is faulty. CC are known to be noisy and nonlinear (measurably so!) - which may be why some people prefer them. You need measurements of your alleged 'parallel resistor sound improvement' - anecdotes don't carry much weight when balanced against known physics.
Why would 100K||100K be better than 51K? I could postulate that green is a better colour than red, and sincerely believe it, but that does not make it so. Two different resistors may have different noise spectra, but two identical resistors are likely to have similar noise spectra unless one is faulty. CC are known to be noisy and nonlinear (measurably so!) - which may be why some people prefer them. You need measurements of your alleged 'parallel resistor sound improvement' - anecdotes don't carry much weight when balanced against known physics.
- Status
- Not open for further replies.