I have found old radios that lived in Florida for 50+ years (before we had air conditioning) with resistors that measured 100+ percent off. The really old stuff (1920's and 30's) almost always went up in value, often 2X or more. Electronics often used grid leak biasing back then with resistor values in the 1 to 5 meg range. Find me one of these that's even close to its original value.
Yes, a handful of nH is a non issue at audio. However a 5842 or D3A will gladly oscillate at 200+ MHz if you give it a chance.
That is the property that will have the greatest effect on sound / distortion, especially as a plate load in a guitar amp. I plan to explore this some day. First some static measurements, then a typical 12AX7 gain stage, mosfet buffered for near zero load. Swap in some resistors and measure distortion VS drive voltage under several different operating conditions.
Not necessarily. You need to avoid having a parasitic resonance in the grid circuit, or reduce the "Q" of that circuit such that conditions for oscillation do not exist. There are no hard and fast rules that can be applied when all of the variables are not known, particularly circuit layout. The typical case is a series resonant circuit being formed by parasitic inductance in the grid circuit and the grid capacitance of the tube. This creates a high grid impedance at the resonant frequency. Any stray capacitance coupling with the plate (including interelectrode capacitance in the tube) and a similar high impedance at the plate near the same frequency can create an oscillator. Add enough series resistance in the grid circuit and you can lower the "Q" of the resonance. The few extra nH of inductance from a film resistor is usually not an issue, and it's what I often use, but in the wrong place it can be an issue.
I have that book somewhere. It's still in a box from moving twice in two years. I'll dig it out as my basement lab gets built out. First priority is insulation.....it got cold in here last winter.
Remember you need a big voltage swing to see the effect.
There's two easier ways to do it since you have an amplifier or two or ten lying around.
1) Take any amplifier and add a rotary switch for the plate resistor. Do an A/B on carbon composite and metal film. If you really want to be thorough, add a C for carbon film. Cycle through the values and take measurements.
2) Switch the plates on one channel to magickal carbon composite and the other to metal film.
Feed mono to both channels driving a purely resistive load. Overlap output signals.
Last edited:
The "mojo" I am referring to is distortion added by CCR. I don't think Houdini is in a warehouse putting a spell on resistors.
I will repeat myself; Believe it or not in guitar amplification you don't want a squeaky clean amp, they are designed to add distortion. I know I know some engineers won't be able to sleep tonight after reading this. Relax, take a step back, take a deep breath and come to the realization that maybe if we can build an amp with components that add distortion it can help us to our end goal.
So its not just the circuit its also the components that act as colors we can use on our canvas.
I will repeat myself; Believe it or not in guitar amplification you don't want a squeaky clean amp, they are designed to add distortion. I know I know some engineers won't be able to sleep tonight after reading this. Relax, take a step back, take a deep breath and come to the realization that maybe if we can build an amp with components that add distortion it can help us to our end goal.
So its not just the circuit its also the components that act as colors we can use on our canvas.
Ummm, nobody would argue with any of that. Except for the fact that the mojo only shows up at the plate.
You are absolutely correct: electric guitars sound horrible without distortion. When I was 18 and bought my Stratocaster (first electric guitar, sunburst, of course, paid for by my job building gizmos) it sounded bad: flat, lifeless, and with no warmth. Nothing like what I heard on the albums. (I'd played acoustic before that.) What was missing? Distortion! Oh, and some talent. Right. Can't forget the lack of talent.
So, of course, guitar amplifiers are different and no engineer would argue that point. They would, in fact, tell you every single place distortion can arise in an amplifier and help you to add it. Engineers are helpful people because engineering is all about problem solving. So if the problem is adding distortion, they'll help. Analog engineers are amazing at this because working hard to remove non-linearities they can accurately discuss them in ways nobody else can.
I noted that as, " Deliberately putting distortion into an amplifier is fine (else we'd have no electric guitar sound) but call it what it is: distortion."
My comments were about golden-ears HiFi people who rip out tone controls wailing, "my ears, my ears", who use speaker cable that costs more per foot than an NYC apartment's rent claiming "my ears hear the difference", or pontificate about how their tube substitution uses lower-noise and greater-linearity devices because the black-plates made in the early part of March of '57 offered the most accurate reproduction, and then go and willy-nilly toss in components with high noise, thermal instability, voltage variation, harmonic distortion, and weird tolerances that move the voltage levels and load lines all over the place, and then brag about how they "upgraded" the amplifier and "improved" the sound. Yeah, sure, whatever.
But, again, the only location a CCR could have any effect (guitar or HiFi) is at the plate, because of the requisite voltage swing. I have seen mojo advertised for transistor amplifiers with a 15 V swing, and that's just impossible and downright fraudulent. So even for guitar amplifiers throwing CCRs everywhere is not a good idea because it just makes things out of tolerance, without adding beneficial distortion. If you want it out of tolerance, make it exactly how you want it to be out of tolerance, not some mystical value that varies component by component.
I love distortion in a guitar or bass amplifier. But in a HiFi amplifer? Ummmm, not so much. Well, sometimes the second order harmonics for cathode bias. (Because who doesn't love a nice, warm second-order harmonic that kisses you like it hasn't seen you in six months and missed you ever single day.) But that's about it.
Last edited:
You almost certainly already have a resonance, as a UHF resonator is a piece of wire, so the main need is to add some resistive loss to push the required loop gain for oscillation above the available loop gain. Equivalently, cancel out the negative resistance presented by the valve at certain frequencies.Tubelab_com said:
I'm not saying that a CC stopper might never be needed. What I am saying is that in most cases a CC stopper is not needed, and in some cases a CC may be worse than a film resistor. Loss is the main mechanism; a little bit of reactance may shift the frequency, but who cares if the oscillation has been stopped anyway? I suspect that if CC were now cheap and plentiful and not preferred for guitar amps due to their distortion, everyone would be saying 'use film resistors for stoppers as their inductance helps lower the frequency and make it easier to damp away'.
Yes, agreed, and in many cases minimizing the inductance will push the resonant frequency above the range where the tube has enough gain to support oscillation. Many tubes will gladly oscillate well into the UHF region. The venerable 12AT7 has no problem oscillating at 400 MHz. It was designed for use as the front end tube in an FM radio.
Adding resistance, lowering the "Q" increases the loss through said resonant circuit, thus lowering the loop gain.
I'm saying the same thing. The results of my testing in one of my amps tend to support that.
I am an engineer, actually an analog RF engineer that spent 41 years working ay Motorola designing cell phones and two way radios. In it's prime Motorola had 5 facilities in our area with over 5000 people. About 1000 of them were engineers in some capacity (electrical, mechanical, chemical, software, industrial design, manufacturing, process and quality control, etc). Out of that pool there were 6 or 8 people who played with tubes, and 4 of us were guitar players......So the engineers among us here are already "different."
I already have a unique guitar amp breadboard in the works. Damn near every component in the amp can be swapped via reed relays. It is however a one channel design.
Excellent idea......This got me thinking about a valid experiment, the kind of stuff we had to take classes for at work. Every engineer had to take 40 hours of in house coursework each year, minimum. This was to learn the stuff NOT taught in engineering school. It ranged from basic soldering and SMD rework, to advanced DSP, software, process control, and "design of experiments" (remove ALL other variable except the one you want to observe).
Some outside vendors gave classes, and gave out free stuff. Microchip gave a week long seminar once or twice a year.....lots of free stuff from them. Linear Tech came in and taught LT spice, I asked for help with one of my designs.....yes, tubes, and yes, they put the design on the screen for about 50 people to see......lots of blank stares from the audience, but the Linear Tech guy understood it.
I think we need two identical single 12AX7 gain stages, built next to each other using the two halves of the same tube, driven from the same source. Start with a hand full of resistors of different types (CC, MF, CF)selected for identical resistance readings. Put two identical MF resistors in the plate circuit, connect the scope to each plate, drive the stage to maximum undistorted output voltage, set the scope to subtract one channel from each other (I have an old TEK that will do this), and roll through a box full of tubes to find a near zero output on the scope (both outputs identical). Then test different resistors in one plate circuit while observing the scope. I will try to set it up after the resistors arrive.....UPS will take at least a week to get here.
But this, again, is the core of the problem. Because a tube has behavior in one set of circumstances (i.e. high frequency RF) it is widely conflated into having this behavior in all circumstances (i.e. at audio) and this imagined behavior is then endlessly repeated as if it were fact.
Back in high school some friends and I were building fuzz boxes and compressors using optoisolators. The analog engineers we talked to at work were always happy (uniformly so) to answer our questions tell us how to get the whackiest non-linear behavior out of normally linear devices and share stories about unstable circuits and how awful germanium transistors were for real things, but how great they'd be for our purposes. The really old guys would say, oh, what you actually want is some tubes.
Any dual device would do, because it corrects for variations between tubes in the DUT. I'd suggest a dual triode or a tetrode/pentode strapped as a triode only because it is an easier experiment.
Yes, exactly so, that was what I was suggesting.
One could also run the output of both devices (with and without resistor distortion) into a tube configured to subtract one signal from the other (one of those hexode mixers comes to mind) and then run that signal (difference) through an audio transformer (and then an amplifier) so the distortion, and only the distortion, would be audible. That will show how audible it will be in practice.
I would test with combinations of voltage, temperature, and frequency. The CCR is non-linear with frequency, and, as I noted, the grains have peculiar diode effects albeit all jumbled together so it conducts in both directions. This accounts for the second and higher-order harmonic distortion it introduces. Sort of like a miniature crystal radio using a crude diode (rusty high-carbon razor blade vs. carbon grains and clay) for rectification.
Correct. But wouldn't that larger signal at the plate be present across the resistors in the tone circuit after it?
What if it that large signal at the plate fed another common cathode gain stage, the input of that stage is between grid and cathode and the output is between plate and cathode. The cathode is AC referenced to ground via a cap and the input grid is referenced to ground via a grid leak resistor, wouldn't the larger signal from the preceding stage be present across that input grid leak?
What if the next stage was a follower stage with the load on the cathode, wouldn't the distortion characteristic of that stage change? I mean if it didn't why would people be using a CCS for follower loads?
I agree those people are the worst. I hope I don't appear as such a person by saying that I think I can hear/feel a difference when I built a guitar amp with all CCR and I felt things were different given that I had built the same circuit with the same transformers, chassis, cabinet, caps, wire, solder, tubes, and speakers etc..... but the only difference was using CC resistors.
The equation given for the gain of a stage has R load in it which is usually the next stages grid leak resistor right? So if that grid leak was a CC wouldn't it too have an effect?
Regardless of resistor composition I always measure it's value before installing, I don't know why anyone would skip this step.
Amen brother. I love my single ended triode amp, it has a nice touch of second harmonic that I love.
Last NOS CC's I bought were nearly 100% off value so they not only need a sort they need to be measured.
You're forgetting that the coupling capacitor (or whatever coupling technique is used) is removing the DC component. We only want the AC component. Otherwise there is blocking distortion aka bias shift. The resulting signal going to the grid is low voltage. Must be.
A few words about coupling.
In a resistive-coupled amplifier, to contrast with the DC removal done by a capacitor in capacitive coupling, multiple power-supply voltages are used to remove the plate voltage and thereby shift the output into the proper range for the next stage's grid.
A transformer-coupled amplifier passes AC through the transformer but not DC.
So in each case (capacitor, resistor, transformer) the DC component is removed. There are exceptions, of course, when the signal is not feeding the grid but is undergoing current gain, such as in a phase splitter. That's because we want the voltage to remain. So that is clearly not what we're talking about here. We're talking about driving a grid from a plate.
This removal is why a negative clamp (aka DC restoration) is useful: it moves the entirety of the signal into the desired range by shifting the DC baseline.
Again, the plate voltage must be removed from the next stage or it would overdrive the grid, and even drive it positive.
I'm not saying you can't hear a difference, I'm saying the only place that difference creeps in is at the plate and it depends upon how large your voltage swing is.
Otherwise you just have an out-of-spec resistor!
For example. If you have a voltage divider it always outputs the same voltage. (More or less, given that sag in the B+ occurs at high load.)
So if the ratio is off, the voltage is off. But consistently off. And the fluctuations in B+ are not major enough to generate the wide voltage swings needed to make that voltage divider behave in magickal ways. It is just a bad divider because the ratio is wrong.
Because the tinkerers completely lack even the most superficial understanding of the issues involved, or any grasp of how a tube amplifier actually works. All these tinkerers "know" is the internet lore—i.e. half-truths, exaggerations, fabrications, and outright nonsense—about carbon-composite resistor mojo, and if someone's handing out free mojo, well, they're going to reach for it with both hands. It's the same reason they spout nonsense about different tubes and swapping them in and out, not realizing the Miller Effect is shifting (sometimes subtly, sometimes not) their RC filters all over the place and that the differences in mu/Gm are causing sonic shifts, instead the fact that the getter is purportedly the right diamond shape instead of the oval or, horror of horrors, rectangular.
Not all distortion is created equal: some distortion is more equal than others. All hail second-order harmonic, the one true harmonic.
The tube doesn't know what circuit it's in. Even in an audio amp, if the conditions are present for oscillation at 50 MHz. it can and will oscillate at 50 MHz (or whatever), while still passing audio. It is possible for this oscillation to go undetected if it is not strong enough such that audio + RF causes saturation.
Some of us are old enough to remember the Japanese two (germanium) transistor radios from the late 50's that used the "reflex" circuit. It ran the RF through transistor #1 for RF gain, then through a diode detector, resulting in audio, which ran back through transistor #1 again to be amplified again. The audio then went to the volume pot, and to transistor which drove the speaker.
Yes, but the average tone stage has 10 to 20 db of loss.
If a hypothetical component causes a "bending" of the transfer function in a given stage, and the next stage causes a similar "bending" some of the effect will be cancelled because the two stages are out of phase with each other. Of course there are higher amplitude signals in the second stage, so for "optimum cancellation" some gain reduction needs to be applied......This is the reason for the "distortion cancellation" amp designs. In practice this cancellation is never complete even if all the magic mojo was properly aligned. I have seen near perfect cancellation over a small frequency and amplitude range.
Guys, I really don't take this stuff that seriously. I look at the $20 box and see some more parts to torture, blow up or whatever.....it's just like a big box full of tubes I drag home from a hamfest......hook them up and see what you can squeeze out of them. Maybe something good, maybe nothing at all.....but you never know until you try.
I even got some VFD displays from the same company about a year ago when Korg announced the NuTube......They probably work about as good as the Korg tubes, Maybe a good thing for a fuzz box or a drive pedal, but I could barely squeeze any gain out of it and I blew the filament out of one in an attempt to get 1 mA of plate current.
Attachments
I thought we'd all been there. Back when forums were all stories and no proof everyone was saying it. People (myself included) would change all resistors in an amp for CC and notice a difference, then through experimentation discover that it really comes down to just one or two of them. Then came the proper explaination, then the discussion of whether the 2HD level can or should be tuned to craft an acceptable overall distortion profile, and whether it can be maintained at different levels.. Then back to simple amp archtectures in the hope of removing the whole issue as much as possible 😉
http://www.geofex.com/article_folders/carbon_comp/carboncomp.htm
"Grid resistors in all but output stages also do no good, because the signal level is typically too low."
That's what I would think, you got a DHT amp and there is 70-100Vrms present at the triodes input. Maybe we can set up an experiment where the CCR loaded 12AX7 drives a film resistor as a load and compare results when driving a CCR as a load to see if the distortion increases or maybe decreases as George mentioned distortion cancellation between cascading stages.
"Guidelines 1 and 2 are simply the recognition that the voltage coefficient of resistance is not very big. In fact, although the coefficient is small, it was specified to be small by the makers and controlled tightly, indicating that it was a recognized problem. In the Radiotron Designer's Handbook ( 4th edition, pg. 1345) they list the JAN-R-11 specification for CC resistors as less than 0.035% per volt for 1/4 and 1/2W resistors, and 0.02% per volt for higher power ratings. Given that the max voltages for these parts was 1/4W- 200V; 1/2W - 350V; 1W and 2W - 500V, that works out to a 7% change in resistance for a 1/4 W part used at its max voltage, a 12.3 % change for a 1/2W, and a 10% change for bigger resistors. That's one of the thrusts of guideline 4 - pick the smallest dissipation resistor you can, to maximize the coefficient.
Of course, that's as big as the effect can get, and you would have to carefully set up the situation to get that much resistor distortion. In an amp, you probably won't be able to get that close to max voltages or signal levels. Realistic levels might be 200V across a 1/2W resistor, and a 75V signal swing. That would give you a 2.6% distortion - enough to be audible as sweetening. That's the point of guideline 3 - you have to have a big enough signal swing across the resistor to have the signal distorted significantly by the voltage coefficient.
But with a 10V signal, you only get 0.35% distortion, and it starts down the slippery slope to inaudibility. More importantly, these percentages represent the maximum beyond which a resistor would have been rejected in the 1950's. Today's CC resistors are much lower distortion. From IRC's web site, we find some numbers. A typical resistor voltage coefficient can be seen at [url]http://www.irctt.com/pdf_files/IBT.pdf[/URL] - which shows carbon comp at 0.005%/volt for that company's products. Another was 0.008%/V. These are smaller than the max allowed under the JAN military spec."
"Grid resistors in all but output stages also do no good, because the signal level is typically too low."
That's what I would think, you got a DHT amp and there is 70-100Vrms present at the triodes input. Maybe we can set up an experiment where the CCR loaded 12AX7 drives a film resistor as a load and compare results when driving a CCR as a load to see if the distortion increases or maybe decreases as George mentioned distortion cancellation between cascading stages.
"Guidelines 1 and 2 are simply the recognition that the voltage coefficient of resistance is not very big. In fact, although the coefficient is small, it was specified to be small by the makers and controlled tightly, indicating that it was a recognized problem. In the Radiotron Designer's Handbook ( 4th edition, pg. 1345) they list the JAN-R-11 specification for CC resistors as less than 0.035% per volt for 1/4 and 1/2W resistors, and 0.02% per volt for higher power ratings. Given that the max voltages for these parts was 1/4W- 200V; 1/2W - 350V; 1W and 2W - 500V, that works out to a 7% change in resistance for a 1/4 W part used at its max voltage, a 12.3 % change for a 1/2W, and a 10% change for bigger resistors. That's one of the thrusts of guideline 4 - pick the smallest dissipation resistor you can, to maximize the coefficient.
Of course, that's as big as the effect can get, and you would have to carefully set up the situation to get that much resistor distortion. In an amp, you probably won't be able to get that close to max voltages or signal levels. Realistic levels might be 200V across a 1/2W resistor, and a 75V signal swing. That would give you a 2.6% distortion - enough to be audible as sweetening. That's the point of guideline 3 - you have to have a big enough signal swing across the resistor to have the signal distorted significantly by the voltage coefficient.
But with a 10V signal, you only get 0.35% distortion, and it starts down the slippery slope to inaudibility. More importantly, these percentages represent the maximum beyond which a resistor would have been rejected in the 1950's. Today's CC resistors are much lower distortion. From IRC's web site, we find some numbers. A typical resistor voltage coefficient can be seen at [url]http://www.irctt.com/pdf_files/IBT.pdf[/URL] - which shows carbon comp at 0.005%/volt for that company's products. Another was 0.008%/V. These are smaller than the max allowed under the JAN military spec."
History repeats so they say🙂
I am young and wish I was around the forum when all this was going on. But still it's fun to do some research personally as a learning experience, it sinks in more than just reading a bunch of threads online.
I am still happy to buy the box of resistors and I know I can use them for my guitar amp projects and tinkering. If they break or drift in a few years all the better, it gives me something to do😛
Nothing beats blowing up a few parts for a learning experience.
When I was starting out at playing with tubes there was no forum because the internet didn't exist....because home computers didn't exist....because the microprocessor didn't exist......but tubes, transformers and all those carbon comp resistors were essentially free. All it took was a trip to the trash dump with a pair of wire cutters and a screwdriver. Yes, it was frustrating when something blew up and you didn't know why, and there was nobody to ask, but it was cool when you made a guitar amp out of junk without spending a cent, and your friends wanted it. The only problem was that I had no idea why some of my amps rocked, and some just smoked and fried. That would come a few years later......
The whole "box full of resistors" thing won't really matter if they aren't repeatable.....but we will know soon. I got a notification that my box should be delivered on my 64th birthday......UPS is not that accurate, and it usually takes longer. The USPS is usually right on, or it comes one day early. The box is coming UPS.
Last edited:
I will say we do have it easier with the interwebs around and all the information a few mouse clicks away. Nothing beats melting sh1t in the name of science though 🙂
I checked my shipment and USPS will have it here Wednesday. I'll report in when I get the goods.
I checked my shipment and USPS will have it here Wednesday. I'll report in when I get the goods.
That is true, but it is a different issue.
The claim is that the inductance of a metal film resistor affects the performance in audio. This is false and provably so. The fact that a poorly designed circuit will oscillate in the RF range, and thus the inductance of the metal film will alter that oscillation in some way, isn't relevant and is a distraction. That entire behavior should be suppressed with better design and some Zobels.
People think that because tubes (or transistors) intended to operate in RF circuits at RF frequencies can be affected by the inductance of a resistor, even lead length matters in those circuits, that all tube circuits, including those not intended to operate at RF, are affected. This is pernicious nonsense.
In short, it doesn't matter what that metal film does to the RF response, it doesn't matter because that RF response itself is undesired. What matters is what happens at audio frequency signals, and they are totally unaffected.
RG is a well-respected authority on this subject, as well as many others. Heed his words. His citations are authoritative, as well.
Again, the variations with voltage only matter with large swings, and that means the plate.
The one thing we learn from history is that you can have people's carbon-composite resistors when you pry them from their cold, dead amps.
It's ok to like distortion, but one must call it what it is: alteration of the relationship between input and output signal. (This is not what the posting I am quoting is saying, this is just a general observation about being honest about one's HiFi or guitar preferences.)
Yes, I completely agree that amplifiers should minimize unintentional distortion. When distortion is the intended goal, which is particularly true in a guitar or bass amplifier, one should engineer exactly the type of distortion one wants, instead of having it be a random, and misunderstood process that is difficult to reproduce.
- Status
- Not open for further replies.