Tube Amps and Resistor Composition

[Sadface]

Hi guys,

I have been told in the past that tube amps much prefer carbon comp resistors to metal film types.

From what I can see of other diy and commercial offerings; it seems like it is normal to use carbon comp resistors for the plate load resistor and the cathode resistor. However I often see grid resistors using metal film types.

Would it be fair to say that best practice is to use carbon comp resistors for tube biasing (Rp and Rk) and metal film for resistors in or near the signal path?

 

[MarcelvdG]

Except for resistors that need to survive temporary overload conditions and resistors in antique equipment that need to look original, I can't think of any logical reason to use carbon composition resistors anywhere. Then again, maybe there are illogical reasons to use them.

Their main disadvantage is their much larger 1/f noise (that is, slow random variations in resistance) compared to metal film or (even better) wirewound resistors. That's mainly an issue for resistors used for biasing.

 

[analog_sa]

There is a very logical reason to use them as grid stoppers: the lack of inductance.

 

[MarcelvdG]

Good point, even though I never had problems with carbon film grid stoppers. For grid stoppers, the 1/f noise is not an issue, as there is usually practically no current flowing through them.

 

[Eli Duttman]

Carbon comp. parts contain resistance material that's both non-inductive and non-metallic. Superior performance in the grid stopper role comes from those properties.

 

[GoatGuy]

One thing for sure… the 'classic look' of Allen-Bradley carbon composition resistors, in their remarkably invariant mocha brown shell, shiny finish, precisely laid down color rings and uniformly tinned leads … were sure easy to work with.

And man could they take a wallop of over-current, when needed. I've had carbon-film and metal-film resistors spark-fail repeatedly when CC wouldn't have even started to smoke. Excellent for bleeders for the big reservoir capacitors in a power supply.

Just Sayin'™,
GoatGuy ✓

 

[kodabmx]

Ohmite CC resistors look the part. That said, I use MF and MO types almost exclusively. MF 1% for 1/2W and MO for power resistors 3W or 5W usually). MF in the screen grid and cathode circuits can act like fuses to protect the tubes if biasing fails or something.

 

[DF96]

Use CC resistors if you need good performance under severe overload or low inductance, and you can tolerate excess noise, distortion and poor long-term stability. Note that grid stopper duty does not require low inductance, although there is a persistent meme to the contrary.

 

[disco]

I have been told in the past that tube amps much prefer carbon comp resistors to metal film types.

Don't go there mate as no passive part compensates for sub-optimal circuit design. Only in situations of very high amplification (like a microphone or pickup pre amplifier,, where noise is much more of an issue) there might be the slightest of differences to be noted. What is often the case, is that the part used is much larger and suffers from increased coupling to ground or to other parts of the circuit.

Use CC resistors if you need good performance under severe overload or low inductance, and you can tolerate excess noise, distortion and poor long-term stability. Note that grid stopper duty does not require low inductance, although there is a persistent meme to the contrary.

There's that and allowed working voltage of course.

 

[GoatGuy]

Use CC resistors if you need good performance under severe overload or low inductance, and you can tolerate excess noise, distortion and poor long-term stability. Note that grid stopper duty does not require low inductance, although there is a persistent meme to the contrary.

Just as a note and perhaps nucleus of further discussion, your 3 points enumerated:

№ 1 — excess flicker noise
№ 2 — distortion
№ 3 — poor long-term stability

I imagine № 1 is, like all thermal-resistance related noise, proportional to the value of the resistance and the current being passed thru.

In what few iterations I've done with making ribbon-microphone preamplifiers (which are tasked with amplifying amongst the weakest full-range signals in all audio source conditions), I can attest to metal-film resistors being superior, when needed, in minimizing noise.

But … one day out of curiosity, having a pair of 'preamps in progress' for a friend, I started out with all-film on both. Tested them carefully. Determined they were essentially indistinguishable. Then, I replaced all the critical MF resistors with CC, of the same values. Cherry picked the values to match from my large collection of NOS Allen-Bradley ¼W. Did the A:B test again, and sure enough … a tiny (but clearly audible) increase in the no-signal noise floor. With even modestly quiet source (I chose a night-time outdoor crickets-in-summer venue), the difference in noise-floor was almost inaudible.

Still … this is a good point, and worth the money for high-amplification demand situations.
________________________________________

Pardon my likely ignorance, but № 2 (increased distortion) seems mathematically to have a single attribution: non-linear resistance as a function of current. If nominally,

V(t) = R(I(t))I(t)​

And the 'R(I) = k' is linearly invariant ('k'), well … there's no distortion, regardless of its composition. If however R(I) is a function of I

R(i) = k + ai + bi² + ci³…​

Then it is obvious that V(t) will not be related to I(t) by Ohm's usual resistance constant.

So, the question is DF96, what are the conditions under which one might actually be able to measure or 'experience' the nonlinear resistance-as-a-function-of-current? In your experience … of course.
________________________________________

Lastly, perhaps again out of uncommon good luck, or nearly total exclusion of all CC brands except Allen Bradley from my Big Box of Resistors, I have not experienced long-term value stability issues with carbon composition, so long as the things (in circuit, or out!) aren't stored where water vapor condensation or immersion is likely, especially in proximity to somewhat water soluble solder rosins or other ionic salt bearing coatings, varnishes, paints, dope, the like. Kept dry, and kept well within their nominal watt allowance, I have not found substantial resistance drift — be it for low, low values found in emitter-balance resitors in solid state amplifiers, or cathode-bias series resistors, or even for high-tension anode load resistors.

But again, it could be uncommon good luck, or simply the unusually robust manufacturing standards of Allen-Bradley.

Your thoughts?
________________________________________

Respectfully,
GoatGuy ✓

 

[Eli Duttman]

AB did shine, but Stackpole made some darned good parts too.

Baking followed by checking allows at least some OS CC parts that have picked up atmospheric moisture to become usable.

FWIW, I've seen/heard remarks to the effect that noise and drift are more prevalent among higher wattage rated parts.

 

[Hearinspace]

I have been told in the past that tube amps much prefer carbon comp resistors to metal film types.

Opinion: I'm saying all this with the thought of promoting exploration , if you just want to go with a turnkey design then don't bother reading any more of this.

Tube amps don't prefer carbon comps, the fellow who told you that does.

As long as you meet the technical requirements of the circuit; resistor value, power rating (2 to 4 times the calculated dissipation is often recommended), voltage withstand rating, etc you can pick what you like. As long as the manufacture is of good quality , price then becomes the deciding factor.

There are many other possible considerations , non-magnetic, non-inductive, low noise, etc., but for a beginner who is putting together their first circuit I would recommend you don't worry about them.

Also, if you're putting together your first experiments or breadboards it's not a bad idea to remember that you can create values by putting others in series or in parallel. Doing that , you can sometimes take advantage of surplus pricing , getting a Qty 100 or even 1000 lot for the same price as 10 in the regular price aisle.

My take on carbon comps in the audio realm is that they're often sold to people who are trying to get a smoother sound, but in those cases it makes more sense to learn about the sources of the harshness they hear and take care of them.

Lastly , I wouldn't use boutique parts until you have your basic circuit up and running, then maybe try a fancy part or two in a key place and hear if you can tell a difference. If there's no difference , then no problem, if there is, well more questions and exploration .

 

[MarcelvdG]

Note that grid stopper duty does not require low inductance, although there is a persistent meme to the contrary.

Why not?

As a grid stopper is essentially meant to spoil the Q of whatever parasitic inductors are connected to the grid, I would expect a resistor that hardly adds any inductance of its own to be preferable to some extent, even though the inductance of carbon or metal film resistors is usually small enough not to be an issue.

 

[DF96]

Yes, the important feature of a grid stopper is its resistance - which adds loss to the resonator we have accidentally attached to a valve electrode. A little inductance does no harm. In fact, when making RF PAs it is quite common to suppress parasitic oscillation by using stoppers which consist of lossy inductors (e.g. some thin wire scramble-wound over a CC resistor - CC is used because it offers bulk resistance and so adds more loss to the coil). Hence it is known that inductance does not normally stop stoppers from working. Of course, there will always be the rare example where inductance makes things worse but this is likely to be less common than the examples where inductance helps.

 

[GaryM]

Some reading material on the subject - Using the Carbon Comp Resistor for Magic Mojo

Back when I was an aerospace Metrologist, I did an experiment in our calibration lab with carbon comp resistors on a calibrated Fluke impedance bridge. I tested a variety of brands and wattages of CC resistors and film types. Test results were most significant with high value resistors so I chose 270K resistors, as I had that value in all the types.

Measuring the Q factor aith a 1KHz signal, CC resistors were superior to all others, with Allen-Bradley resistors being the top performer. Some other test types were Ohmite, Dale and ITT resistors. But I've since lost my test results document.

 

[schiirrn]

There is a very logical reason to use them as grid stoppers: the lack of inductance.

That's an internet myth that gets repeated over and over and even if "lack of inductance", which is not possible anyways so let's rephrase that to "minimum inductance" calms the nerves than there are alternatives to carbon composition resistors that share none of the shortcomings.

 

[kstagger]

Back when I first got into tube amps, the word was that if you replaced all the resistors with metal films you would get a more "metallic" sound. I never noticed the difference myself having used all sorts for rebuilds.

I still use carbon comps as grid stoppers - they're inexpensive - but it's usually KOA carbon films everywhere else, or wirewounds if I need the wattage.

 

[DF96]

Measuring the Q factor aith a 1KHz signal, CC resistors were superior to all others, with Allen-Bradley resistors being the top performer.

Difficult to understand what this means. Q normally means either ratio of reactance to resistance (in which case high Q means high inductance) or ratio of stored energy to dissipated energy (which means much the same thing).

 

[Duke58]

Resistors are relatively cheap. Buy some metallic and carbon comp resistors and experiment. Let everyone on the forum know what your results are.

 

[triode_al]

as to the resistors having no magnetic material in their casing nor in the legs is very important.
most metal film reistors I know have both.
The iron clad wires will suffocate any signal.
the wire rings holding the resistor element against the connection wire is also a choke.

Better look for a few high grade ones, compared with expensive tubes there is little extra cost .
I now have a magnet laying around.

by the way I have Corning English glass resistors with a tin oxide, and they are notorious unstable and unusable in a feedback loop. I tested not in a line stage but in a shunt supply: verything shows up clearly. (bought several boxes . . and threw most away last year.)