Re: Current Noise


Patrick Turner wrote:
Andre Jute wrote:

west wrote:
There are several different types of noise to avoid and reduce, if we can,
in designing tube circuits, from what I been studying. We can discuss these
different types of noises in general but for now I would like to concentrate
on 1 type of noise, current noise, the kind that is generated in a plate
load resistor. I believe that this is especially important in preamp
circuits.
Some tubies advocate that boutique components are all fluff and a waste of
money and I think they have a point, to a degree. However, will a low noise
resistor make a difference in reducing plate current noise? If yes, then
they can make a difference.
The question is ...can we quantify this difference and how? I think it would
be cool to measure this noise and simultaneously hear what difference a low
noise boutique resistor can make. I may be off base with this idea, if so
chalk it up to an inexperienced but eager to learn Rodent. Thoughts? Thanks.

Cordially,
west

West:

This thread has already spiralled off into contending experts ego-
tripping. The two items of useful information in it is where Patrick
states the practical problem and where I tell you how to solve it.

Here Patrick states the problem: "the number of ohms was what
determined noise, and it mattered not one bit what kind of resistor it
was as long as the contacts to each end were mechanically tight"

Allow me to add to this.

Say you have a 60 kohm R plate load R feeding a 1/2 6SN7, Ia 5A,
bypassed
Rk, then noise in the 60k is shunted by the 10k Ra of the triode, so the
resistor noise
is very much reduced,

Yes, that is what I said, probably several days ago.

but then the triode noise will be greater.
So the 60k load does little to increase noise.
If the very well chosen 6SN7 has 3uV of noise at its input even with
grid grounded,
and gain = 15, then there is 45uV of noise at the anode, and I leave you
to work
or how much noise would be in 60k at say 30C, and how much would add to
the 45uV.

If such a gain stage has a following coupling to a gain pot, or bias R,
the noise output
needs more calculation.

But ppl say a choke is the quietest element to use.

Expensive. One of the most appealingly silent amps I ever built had a
choke as grid leak. Unfortunately I didn't save one of those vintage
chokes for reverse engineering. But an SS constant current load is
also has a pretty good noise performance.

But I find an actice device such a san MJE50 as a CCS is NOT noisier
than the R,
and probably because it is shunted the triode noise which has Rout = Ra
= 10k.

Rueful grin. I should read the whole post before I waste my breath
replying!

And here I show how to solve the problem: "The solution is not fancy
resistors (you usually can't get them in the right ratings) but to use
standard oversized metal films in smaller values in series, say 6x10K
rather than one 62K." Thus you get both the silence of the smaller
ohmic value of each resistor (6x10k makes less noise than one by 62K)

Considering how much consternation this statement caused among the
unwashed and the illiterate, I should perhaps have put the next phrase
in capitals to shout it from the rooftops:

and the silence of the oversized resistors running cooler.

Yes, that's right, punters, this works by NOT reducing the rating of
the resistors when you reduce the values and series them to make up
the total.

*That* is
why on all my amps the power resistors are, in strictly cost-
accounting engineering terms, overspecified; ditto for all the best
audiophile-designed tube amps. (The idiot Worthless Wieckless made an
offensive song and dance about overspecced resistors in my T39 Ultra-
Fi once because he is bog-ignorant -- I very carefully didn't explain
the real reason to him; he's such a wretched little man that nobody
else helped him out either. But those big ballast resistors account
for a good part of the silence of the amp.)

Perhaps you may have to think this through again.

Nah, I got it right first time. And I had Bill measure and calculate
the total noise when I first did this, a decade or more ago. It works.

OK, say you have 6 x 10k in series.

Each one makes a noise = noise of 60k x sq.rt of ( 60/10 ).
10k obviuously has less noise than 60k.

In fact the noise with 6 resistors in series = sq.rt ( sum of the
squares of each amount of noise )

And you should find that 0ne 60k resistor of 6 watts rating has exactly
the same noise at 6 x 10k x 1 watt in series.

If the 10k were all paralleled, noise would indeed be lower, but the R
wouldn't be right
for the anode load.
There are no free lunches on resistor noise to be had by series
connection
of well rated quality resistors, unless the connection leads to the R
operating
without any T rise. T rise makes noise worse; hot is noisier than
cooler.

Tres exactement. That is why I treated the attempt by the usual
hairsplitters to separate excess noise from thermal noise in a
practical application with derision. I understand why West wants to
separate them: the texts he's been reading do. But experienced DIYers
know that in practice they are interrelated for the reason Iain
Churches pointed out so bluntly: that we work with 2W resistors by
choice even when the cost-engineering answer would be an 0.6W
resistor.

A 60k anode load for a 6SN7 which comprises of say 10 x 600k x 1 watt
all paralleled to get
60k won't have any noise difference to ONE 6 watt 60k, or 10 x 6k x 1
watt in series,
providing the temp is the same, which is should be if the total 60k
wattage is the same,
or whatever is used to makes sure T remains the same.

But in practice bigger resistors are used when you split the value and
build it up again by series addition of resistance.

All the rest is, as I say, audiophoolery or "experts" tripping over
each other in their eagerness to show off.

Nobody stops you experimenting for yourself though; discovering
whether you like one sort of component better than another is half the
fun of building your own tube amps. There was a time when I swore by
Dale "non-inductive" cathode resistors on 300B and 845; today I just
use the 50W ali-cased jobs out of the RS catalogue. And I still like
Solen polyprops better than any other kind; you can get Solens in the
widest possible range of values in high voltage ratings and they sound
right in my amps and don't take up as much space as the (more
expensive) motor run polys I can buy locally.

I like whatever nice metal film 1% R I can find; I use Welwyn from RS
Components mostly
ever since the Reistor Mafia, Vishay, took over Beyschlag, and the local
distributor was bought
by a darn american company.
The result of these changes meant Beyshlag price of 10c per R for a
genuine 1 watt went to 45c,
and you were compelled to buy 1,000 as the minimum number.

When you want to buy say 50 pcs of all standard values between 1ohm and
4.7M, its very expensive.

Just watch out with voltage ratings on resistors; even high value like
470k and 1M don't
like more than 250V sometimes across them. Welwyn sure don't, and I had
3 x 470k in parallel for a
tail between commoned cathodes in an LTP to -400V, and in both monoblocs
( Quad-II, very revised )
one of the 3 went to a short!

I've gone right off kilovolt amps, among other reasons because it is
so difficult to get components of the right ratings. (It is years
since I published a kilovolt amp; I stopped doing that when I
discovered people were rushing out to buy their first soldering iron
and their first DMM -- with the intention of building one of my
kilovolt amps!)

I like Wima polyprops; I cannot find anyone who can tell me better than
chance will predict,
ie, 50 - 50, which channel has all Wimas and which has all polesters, or
other types/brands of polyprops.

I don't think it matters which polyprops you use, as long as you use
polys (and other films where suitable). I like the Solens because I
have a big stock and because they are small. I have custom-made 50uF
1200V which are (without finding them and measuring one) about an inch
diameter by about four inches high, very modest when you consider that
a standard good quality Italian Comal 47uF 450V motor run electro that
I can buy locally in bulk (for about the same price as a rare and
wonderful Solen!) is two inches diameter by about eight inches high,
and requires many caps and much real estate to make the value/rating
required. Even the standard motor run polys out of RS would require at
least four caps, much more real estate, and many more spondulicks than
the Solen. That's why Solen rules.

Sometimes i find the size of tube-capable components so overbearing
that I consider going over permanently to the Dark Side (where there
is only SS).

Andre Jute
Visit Jute on Amps at http://members.lycos.co.uk/fiultra/
"wonderfully well written and reasoned information
for the tube audio constructor"
John Broskie TubeCAD & GlassWare
"an unbelievably comprehensive web site
containing vital gems of wisdom"
Stuart Perry Hi-Fi News & Record Review

.

Relevant Pages

  • Re: High-ohm grid resistor in tube (pre)amp design
    ... 100K) in series with one of the first 12AX7 preamp tube grids. ... OK, I fully understand that this resistor, together with the Miller ... introduces noise -- a LOT of noise. ... OK, but in this case, the resistor is found in series with the grid of the ...
    (sci.electronics.design)
  • Re: 300B versus other trioded octals.
    ... I try and aim for 0.5mv hum and noise at the terminals as well. ... Microphony is a form of noise, ... Usually the input tube is the source for microphony, ... but also in line and power amps. ...
    (rec.audio.tubes)
  • Re: Current Noise
    ... on 1 type of noise, current noise, the kind that is generated in a plate ... resistor make a difference in reducing plate current noise? ... design of an OTL to drive Stax electrostatic headphone. ... This is a tube audio conference. ...
    (rec.audio.tubes)
  • Re: Amp Noise
    ... more than 50% of the time this noise is from a bad plate load resistor. ... Good advice to check the tube/s first, isolate the circuit where the noise is coming from, then test all components in that circuit. ... Find the gain tube for that channel... ...
    (alt.guitar.amps)
  • Re: Obama likes science ...
    ... Don't ignore the Johnson noise of the resistor. ... voltage sensing resistor off the positive rail, ... Single supply op amps tend not to work when the inputs are biased ...
    (sci.electronics.design)
Loading