Do resistors really have voltage ratings?

[Boogie]

Hi,
I've read a few times online where people said to watch the voltage ratings of resistors. Is this necessary? I thought that since the resistance of a resistor is fixed (since that's the whole point of a resistor) then doesn't that mean by P = V^2 over R that assuming the power rating is not exceeded, the voltage must be okay? And too high a voltage across a resistor would draw too much current (and therefore power), so for e.g. say a 100 ohm 5 watt resistor , at maxium power would have sqrt(100x5) = about 22 volts across it? And any more voltage than that would be too high?

So (summary of above paragraph) the voltage rating is built into the wattage rating? And other than that there's no need to consider voltage ratings of resistors?

Or am I missing something?

That was my thoughts, then looking online again I found conflicting answers: eg "A resistor can be used at any combination of voltage (within reason) and current so long as its “Dissipating Power Rating” is not exceeded with the resistor power rating indicating how much power the resistor can convert into heat or absorb without any damage to itself.",

which is what i thought, and then this, which is the opposite

"The voltage rating is for the resistor series typically and specifies the maximum peak voltage you can apply without danger of damaging the resistor due to corona, breakdown, arcing, etc.The power rating is completely independent of the voltage rating. It specifies the maximum steady state power the package is able to dissipate under given conditions.
You have to conform to both specs. If placing the maximum voltage across the resistor results in more power than the spec allows you have to reduce the voltage until you meet the spec. Likewise you can't increase the voltage above the rating just because you're not hitting the maximum power limit.
"

So which is true?

The second answer seems counter intuitive to me, since, if you assume there is a resistor with a voltage rating that can be exceeded without exceeding the power rating, then doesnt that also mean it would be impossible to run that resistor at its actual power rating, since that would imply that at the power rating there would be a higher voltage across it than its voltage rating would allow. Does that make sense or am i talking cr#p?

[Chuck H]

Looking forward to the higher minds answering this one. And thank you for posting this question. As far as I can tell, what you propose about voltage across the resistor being the relevant figure makes perfect sense to me. So maybe we're both full of crap.

[eschertron]

The voltage rating is for the resistor series typically and specifies the maximum peak voltage you can apply without danger of damaging the resistor

I've heard that too. Now finding this voltage rating spelled out in black and white has eluded me so far. I have stumbled across generic rules of thumb that indicate the lower the power rating of the resistor (1/8W, 1/4W, and so on) then then lower the voltage rating - and I'm presuming that it has something to do with the robustness of resistor construction for the typical film resistors. When in doubt, I use a resistor with a wattage rating comparable to the ones I see in mass-produced amps.

Resistor noise as it relates to wattage ratings is another can of worms. As is the trend to receive resistors rated at 1W from Mouser that are physically smaller than the 1/2W resistors in my bins. So are they good for a higher voltage, or not?

I'm thinking that I'll make 1/2W resistors fly for anything that doesn't require a higher rating for dissipation purposes. Other, more skilled members, may offer correction on that. Gladly accepted.

edit: Snap! I thought Chuck was one of the higher minds!

[Chuck H]

edit: Snap! I thought Chuck was one of the higher minds!

Sorry to disappoint

I'm just an amp nerd. Not even an amp "tech" and certainly not an electronics engineer as some here are. I do like to think a little out of the box, which is why I liked this question. And I strive for the simplest and most eloquent solutions. That may make me seem bigger than I am sometimes. But I struggle, devote a lot of time and rely quite a lot on online services like SPICE and SPICE based calculators. If you look into it you can find posts where member more humble in nature than I am have corrected me on some pretty basic stuff!!! I'm all smoke and mirrors. Pay no attention to the man behind the curtain

[Boogie]

Thanks guys. I was also thinking about reducing resistor noise, specifically whether its a good idea to swap wire wound for carbon 100k plate resistors. I've got some 5W ones which is more than the original carbon ones (1 or 2 W I think). I read that inductance is not an issue in this application and in general for high resistance values (usually or always?)

The only other idea I had after posting that was perhaps there's a voltage limit associated with transient spikes, where there may be a really high voltage (higher than that given from using ohms law on the power rating) for a very short time, not long enough to heat anything up enough to reach the power limit. However in that case, it seems to me that the "voltage rating" would exceed the usual power rating and therefore using the power rating as if it was the voltage rating would be fine since it would still be within the limits.

[Tom Phillips]

...and then this...
"The voltage rating is for the resistor series typically and specifies the maximum peak voltage you can apply without danger of damaging the resistor due to corona, breakdown, arcing, etc. The power rating is completely independent of the voltage rating. It specifies the maximum steady state power the package is able to dissipate under given conditions.
You have to conform to both specs. If placing the maximum voltage across the resistor results in more power than the spec allows you have to reduce the voltage until you meet the spec. Likewise you can't increase the voltage above the rating just because you're not hitting the maximum power limit."

That one is true.

Take the case of a ˝ Watt 10 MegΩ resistor. You don’t reach ˝ W dissipation until there is 2,236 V across the resistor. By then something is going to arc over inside the resistor.

The voltage specs are readily available. You just need to look at the manufacturer’s full data sheet. Most of the time, the voltage spec isn’t listed in short form catalog. In addition the capacitance and the inductance of the resistor is often not listed but those specs do exist and people who design high frequency equipment (radio & microwave frequencies that is) are interested in those specs.

[Boogie]

Take the case of a ˝ Watt 10 MegΩ resistor. You don’t reach ˝ W dissipation until there is 2,236 V across the resistor. By then something is going to arc over inside the resistor.

Thanks Tom, that does sound logical about 2,236 V arcing, but then, if you can't ever reach that much voltage across it, then how can the resistor ever dissipate 1/2 a watt? Meaning, in my thinking, that its not really a 1/2 watt resistor?

[Chuck H]

The only other idea I had after posting that was perhaps there's a voltage limit associated with transient spikes, where there may be a really high voltage (higher than that given from using ohms law on the power rating) for a very short time, not long enough to heat anything up enough to reach the power limit. However in that case, it seems to me that the "voltage rating" would exceed the usual power rating and therefore using the power rating as if it was the voltage rating would be fine since it would still be within the limits.

Interesting that you would observe this BEFORE having a problem... Because "I" didn't! I managed to cook a couple of resistors in customer amps because they were in service as OT primary shunts in series with a capacitor (sometimes called a "conjuctive filter". whatever that's supposed to mean). Using a 1500pf 600V cap and a 10W/10k resistor on a circuit with 355Vp. The cap, which was self healing, eventually healed itself into a short and the resistor turned quite brown in color before it was replaced, with no gross amp failure other than performance I replaced the circuit with a 2kV cap and a "high energy" resistor. Sorry I can't remember the specifics. Right out of the Mouser paper catalog. No future failures. Having said that...

The only critical plate resistors for thermal (Johnson) noise would be the first and maybe second preamp stage (in an uber gain type preamp). Carbon comp resistors introduce another type of noise called "shot" noise. Which, as I understand it is actually tiny arcs of electricity across pockets and inclusions inherent in the this type of resistor. Interestingly, any advantage to the carbon comp mojo, which has to do with some level of variable resistance with power resulting in a sort of compression, only occurs in higher voltage type circuit where they cause the most noise!?! At least that's my understanding. I use to build amps with metal film plate resistors and carbon comp for the rest of the signal path. I have since abandoned this practice and now just use metal film because I found myself subbing metal film for all high voltage circuits. No high voltage, no advantage. So why use a resistor of inferior construction? I don't know how carbon film stacks up here, but I've wondered because all the pots we use are carbon film. Anyhoo...

For those first two preamp stages there is A LOT of spike suppression by the time any voltage gets to them. Several resistors and filters in series in the HV rail. And those stages are amplifying lower gain signals in class A operation. There's really not much opportunity for spikes that need to be considered. Don't worry about spikes here.

I've read that wire wound resistors will be the quietest WRT thermal noise. And that it IS actually helpful to over rate, to a point. 1W wire wound is supposed to be about as good as you can do. Don't worry about inductance. It just doesn't apply at the frequencies in question.

Metal oxide resistors generally have the highest voltage rating for a given power spec. and have been recommended here as HV rail resistors. They are also said to be a good choice for plate resistors on the early preamp stages. But wire wounds are said to be best, will almost surely be up to the voltage on those stages, spikes don't really need to be considered and neither does inductance. So... 1W wire wound is my suggestion.

I use the standard 1/2 rating in metal film and don't have a problem. But I haven't tried 1W wire wounds to see if there's an improvement either.

NOTE* See, eschertron? Most of what I just wrote is a regurgitation of knowledge I've gleaned here over the years. I have a decent memory and after almost 10k posts even "I" can remember some of this stuff I'm no hero, but I like to help.

[Boogie]

The other answer I thought of to resolve the voltage limit issue is that perhaps, for high ohm resistors, the power rating isn't really the power rating, in that 1/2 watt doesn't mean "this resistor is capable of handling 1/2 a watt through it", but actually means "a completely different valued resistor of similar construction is capable of handling 1/2 a watt through it" ???

In which case for high value resistors, the power rating would seem to be completely meaningless and the voltage rating would be the only rating that meant anything?

Thanks Chuck. So 5W 100K wirewounds should be fine then, assuming they fit physically?

I originally bought them for a red knob Fender twin, which had bad reverb noise that I eventually traced to some smaller passive components which I replaced and now its fine. But while I was reading about noise I read enough to order the wirewounds. And then decided the amp isn't really that noisy enough to bother swapping the plate resistors.

The amp I'm thinking of now is a Mesa Boogie Studio .22 which is really noisy. I tried using coiled heater type tubes (JJ's) which I read can improve noise in that amp, but made no difference. Since I've got the 100k resistors it seems logical to try some in the Boogie in the early gain stages and see if its much quieter. Its got a heap of gain (3x 12AX7s just for gain plus one more for the reverb and one more for the phase inverter). The first gain stage has both halves of the 12ax7 parallelled (I'm not sure why), and there are another 4 more gain stages cascaded (two more 12ax7's) which is a heap of stages . The first (parallel stage) tube has a 68K plate resistor so i have to buy one for that. Though, interestingly , the clean channel is more noisy than the lead channel - suggesting its one of the later resistors thats causing the noise (if its even a resistor at all).

I've also recently got a Mesa Boogie SOB which is pretty quiet really. I'm about to put new PSU caps in it. It doesn't really need to me much more quiet, though it also has 100k plate resistors so i could experiment, perhaps.

[Chuck H]

Nothing suffers with over rated resistors unless the resistor rating is intended to partly act as a fuse. Which actually happens in some designs! But that's not the case here, so freak out man.

Regarding the Mesa clean channel noise... Some Mesa designs don't adequately silence gain channel noise when switched to the clean channel. It's been a common flaw in their designs for decades. They will simply ground a grid at the beginning of a cascade circuit and leave the output coupled to a mixer stage. I had a Studio Preamp that did this. In this case it's a design flaw and not a circuit flaw. Any small measure of improvement to a circuit like that won't make a noteworthy difference in the hiss of a high gain circuit bleeding onto the clean channel. I don't have the Studio 22 schematic in front of me for evaluation though, so this may, or may not apply.

[Enzo]

You are confusing two specs. the power handling rating of the resistor is an average. If 1ma if flowing through a 100k resistor, 100v drops across it, and you can calculate the dissipation. If all the tubes did was run steady current through the resistors, then the assumptions would be valid. But we have large signal swings, even hundreds of volts. The average current stays the same, at least as long as the waveform is remotely symmetric, but the voltage across the resistor can swing large. And if it exceeds its voltage range it will break down in some fashion. Maybe arc between turns on a film resistor. The dissipation won't change though.

[Boogie]

In my first ever full time job at age 18 I discovered that capacitors can make good fuses. I was a PC (as in computer) tech assistant/trainee and I was testing power supplies by plugging them into a motherboard to see if they worked. One of them went BANG with sparks and smoke and I was like "I don't think this one works!". But - oopsy - it was my fault as the edge connector (or whatever its called) was capable of being plugged in with one pin offset from the proper way, so all the pins were going to the wrong connections on the board. The board appeared to be fried, then we changed all the tiny capacitors on it (there were about 10, maybe, it was a long time ago), and then it worked again

Thanks v much about the preamp noise, I didn't know that - I think the studio preamp and studio 22 are v similiar (at least if the same vintage). Though I'm looking at the schematic with my limited knowledge and I can't see anything that looks to me like a grid being switched to ground. The channel switch seems to just add in the first gain stage (which is the parallel both halves of the tube), and add some tone shaping in the tone control section, and a level/volume change at the end of the gain stages before the master volume.

Even the gain channel is still too noisy for my liking, also, though i was surprised the clean channel could be worse. There is actually more noise in the clean channel, so if its bleeding from the gain channel, more than 100% of the noise is bleeding (like its being amplified more when on clean)? Or perhaps its just adding to the clean channels own noise and giving twice the noise? or something like that...

Assuming that's the issue is there any way of modding it to make it quieter? Like maybe adding another LDR to cut the signal at the other end of the unused gain stages also (so its cut at both ends)? It looks like that's what is in mine already (there's two LDRs one at each end of the first gain stage)

http://www.tremolo.pl/Firmowe/MESABO...20Studio22.gif

[Boogie]

Thanks Enzo! That means what I kind of thought about transients is basically the answer then. (?)

and what Chuck said about not worrying about spikes ("For those first two preamp stages there is A LOT of spike suppression by the time any voltage gets to them. Several resistors and filters in series in the HV rail. And those stages are amplifying lower gain signals in class A operation. There's really not much opportunity for spikes that need to be considered. Don't worry about spikes here.") means its all good for the plate resistors presumably

[Boogie]

But what I said a few posts above about the transient thing still meaning that staying within the power rating meant you couldn't exceed the voltage rating was wrong. I think I actually get it now

[Tom Phillips]

Thanks Tom, that does sound logical about 2,236 V arcing, but then, if you can't ever reach that much voltage across it, then how can the resistor ever dissipate 1/2 a watt? Meaning, in my thinking, that its not really a 1/2 watt resistor?

That line of reasoning is not bad but it is still a 1/2 Watt resistor. It comes about because there have been standards chosen so that the manufacturer doesn't need to make an unusually large selection of sizes. For a line of commercial off the shelf parts manufacturers would not vary the physical size of the resistor so that no extra material was ever wasted. An exception to that would be when the absolute best performance to weight ratio is required for a specific application such as for a satellite. In that case you could order a custom part if you really need it AND you have the budget.

There are more things at play too. For example, if the circuit was required to operate in a high temperature environment then you would need to apply a derating factor. I'm looking a spec sheet right now that shows 70˚C as the maximum allowable ambient temperature at which a ˝ Watt resistor can safely operate at ˝ Watt. The derating curve shows that at 135˚C the “˝ Watt” resistor can only be used as a 0.1 Watt resistor. I don’t see any derating of the voltage spec with elevated temperature operation. My point is that there are lots of design tradeoffs and reasons to have extra margin for certain specifications.

When someone is optimizing a design for reliability, all the parameters must be considered and when you hit the limits for any one parameter then you must change your design or use a better part. When you hit the voltage limit the unused wattage rating doesn’t really help you. The same goes for vacuum tubes. Let’s say you have set a bias current at 22mA and the plate voltage is 425V. That’s 9.4 Watts dissipation and a 6V6GTA would be comfortable at that setting. What would happen if you set the bias current to ~7mA used a plate voltage of 1,275V? That’s still ~9.4W plate dissipation but the 6V6 would not last long. You would probably just get a short fireworks show. The resistor, just like the tube, has multiple ratings and whichever parameter reaches its limit first determines when you must stop.

HTH,
Tom