Carbon Comp ?

[Steve Conner]

Oh stop it JM. It is a scientific fact that carbon comp resistors generate second harmonic distortion. There is a plot of it in The Art Of Electronics.

At the signal level seen at the plate of a tube amplifying stage, the effect should amount to a few percent distortion, which could well be audible. I'm not sure why Loudthud's experiment didn't show it up. Here are some possibilities:

The distortion from the tube itself swamped it.

The noise and distortion from the scope front end swamped it. (Scopes are designed for high bandwidth, not low distortion.)

The signal wasn't big enough.

The resistors were too good. Maybe some resistor maker figured out a way to reduce the voltage coefficient. What was the tolerance band? If there was even such a thing as a 1% carbon comp, then it would need special treatment to stay within its tolerance band at the voltage corresponding to rated dissipation.

The "carbon comp" resistor wasn't really a carbon comp.

I personally prefer metal film and metal oxide resistors for their lower noise and higher reliability. I think my circuits sound fine with them.

I once joked that metal film resistors would make the amp sound "metallic" but it turned out that a few respected boutique amp builders actually thought I was serious and agreed. I think there's a synaesthetic thing going on, where people imagine what the material would sound like when tapped, and then assume that the flow of electrons through it somehow brings out that same sound. This is not at all true in general.

[J M Fahey]

Sorry, MEF crashed in mid post and apparently I double posted.

[J M Fahey]

Dear Steve: you are completely missing the point .

I couldn't care less about carbon composition resistors.

Preaching *against* them is just as stupid as preaching *for* them .

What pisses me off is the incredibly poor way the "experiment" was conducted, the clear bias shown by the "experimenter" and his despise of the Scientific method
Even though it might prove his point .... simply he doesn't accept it.

You should re-read my posts where I show no prejudice and claim I will accept the results of the experiment .... if properly made.
Won't boringly self-quote, just reread my post #5

Back to your reference , which you had already talked about in other posts:

1) please don't misquote Horowitz; and if you do , do it in full and not just a section out of context

2) instead of just mentioning a graph, *read* it and use it with the values we are talking about

otherwise: "to look is not the same as to see"

Ok, let's look and see.

I'll save us all some time and do it for you:

a) Horowitz talks about parts used GROSSLY (and that's an understatment) beyond their rated specs.
Can you complain about them not behaving as expected?
I wouldn't. At least not in good faith.

In this case, he subjects 250V rated resistors to 1500V . GROSS .

But that same graph you quote show deviation is NIL when used under guitar amp conditions.

I know you design Power Supplies for a living and work at a Very High Voltage Lab
I respect you for that, in fact Horowitz is talking about Power Supply design in your example.

Simply don't apply those very different design constraints here, where they don't belong.
I mean, unless we feed 1500 or 3000V to our 12AX7

Am I making this out?
Let's check:





Nope, I guess I read Horowitz right.

[bob p]

Here are some possibilities:

The distortion from the tube itself swamped it.

Well, if the tube swamps it, then it's contribution is insignificant and the argument for CC resistors becomes moot.

The noise and distortion from the scope front end swamped it. (Scopes are designed for high bandwidth, not low distortion.)

Now you're opening Pandora's Box. The Golden Ears people like it when engineers give them reasons to refute the objective measurements of test equipment.

The signal wasn't big enough.

In the real world of guitar amplifiers, the signal voltages just aren't big enough. Sure, the theory predicts effects at ridiculously high voltage swings, and people who aren't good at math insist that their Golden Ears hear these theoretical effects, even though the reality of the situation is that the operating points in use aren't in the range that's required to enable these theoretical effects to occur. Yet some people insist on believing in the mojo.

Looking at this objectively, let's think about the votlage swings. I don't see any possible circumstances where the signal swings are going to be large enough on a preamp tube for the voltage coefficient deviations to be meaningful. The math says that there is no mojo to be had from using CC resistors in the low voltage sections of a circuit. Period. OTOH, it is possible that the voltage swings could be high enough to be meaningful on circuits that use select output tubes that have extreme voltage swings, but these types of tubes aren't used in guitar amps. In a guitar amp the voltage swings would have to be HUGE, and by the time these voltage swings occur, the tube itself is already producing enough distortion to render the contribution of a resistor voltage coefficient as insignificant.

The resistors were too good. Maybe some resistor maker figured out a way to reduce the voltage coefficient.

If the resistors were too good, then the logical conclusion is that some people should want to use lower quality resistors in their builds. That just doesn't make sense to me.

What was the tolerance band? If there was even such a thing as a 1% carbon comp, then it would need special treatment to stay within its tolerance band at the voltage corresponding to rated dissipation.

I have thousands of 1% CC resistors. They're all labelled as a much wider tolerance limit, but I've hand selected them to bin those that have 1% tolerance. The problem is that even though these resistors might start at 1% tolerance, they won't stay there. They drift. While some people may like that, drift limits their utility for me.

I personally prefer metal film and metal oxide resistors for their lower noise and higher reliability. I think my circuits sound fine with them.

I agree with you completely. To me it makes more sense to design a circuit around stable components and parameters that you can control, rather than relying on mysticism, unstable components and parameter drift that you cannot control.

I once joked that metal film resistors would make the amp sound "metallic" but it turned out that a few respected boutique amp builders actually thought I was serious and agreed.

Now that is funny. Unfortunately, it goes hand in hand with the mojo mindset.

[J M Fahey]

Sorry, Forum crashed while posting and apparently I doubleposted.

[nashvillebill]

Thanks for "keeping it real" as some folks say. I'm in the middle of a bare-chassis rebuild of an old Epiphone (even made a new eyelet board); I was tempted to be different and stick a CC or two in a couple of spots. But, considering every CC in there that I've measured has drifted upwards (some of them considerably), I'll stick to metal film/oxides.

Besides, I figure there's no mojo in crackling plate resistors.

[Mick Bailey]

Measurement has limits; CCs have psychoacoustic properties that cannot be measured. Like mustard caps, or solder touched by the hand of Leo.

Next you'll all be telling me that my concrete Elvis statue doesn't cry tears of milk. But I won't have any of this.

"A man convinced against his will,
Bears the same opinion still"

It does cry tears of milk.