Tube Matching, Bias, and Breakup

Tube Matching, Bias and Breakup – How They Work Together

If you have an old-school tube amp without a master volume, then you know the only way to get breakup is by turning up the volume and moving some air. These old amp designs stretch far into the past, to the early days of amplified instruments. And you know what? They didn’t want distortion. They just wanted to be heard by the audience. Rock and Roll wasn’t a thing and distortion wasn’t cool yet.

As musical styles evolved from the 1950’s onward, the sound of overdriven electric guitar became a big part of the musical landscape. High quality P.A. systems weren’t available so amplifiers were pushed to the breaking point. When treated this way, tubes start to work outside their designed parameters, compression occurs, harmonic frequencies are generated and the tone gets fat and aggressive. It just so happens that a lot of people like this sound. Rock music was at one time considered nasty and aggressive, so the sound of overdriven amps was the perfect marriage.

Most of this glorious tone was delivered by pushing those amps hard and causing the output tubes to distort, or breakup. In the 1970’s, amplifier design started to change. People wanted their amps to breakup at lower volumes, and more severely than they had in the past. Engineers developed effect pedals that would generate thick distortion and amp designs shifted to having a master volume and multiple, cascading gain sections. The main idea was to create the breakup before the output section and use the output to amplify the already distorted signal. Mesa Boogie was the true champion of this cause and in the 1970’s and 1980’s many people took their Fender amps to the shop for modifications to try and get the Boogie sound. That sound has continued to evolve. Metal took off and heavy distortion was here to stay.

I recently had a discussion with my friends at thetubestore.com. It seems that there are people looking to get early breakup in their guitar amps and they want to buy tubes that will give it to them. At least 1 company advertises tubes as being matched based on breakup characteristics. An interesting idea, and perhaps valid, but not something that I had thought about when reviewing tubes or building amplifiers. Are tubes designed to distort? Can they be designed to distort?

Vacuum tubes are mass produced devices that have electrical and mechanical design properties. The physical structure of a tube and the materials used to make it, have a big impact on the success or failure of meeting those designed properties.

Let’s look at a typical specification sheet for an output tube.7189A Vacuum Tube Data Sheet

There are a lot of numbers that describe the tube’s properties in terms of dimension and electrical ratings. Things like the height, width and type of base are quite common. The electrical properties are more diverse. They specify maximum levels for voltage, current, capacitance and output power. They also specify typical values for these criteria when the tube is operated in different modes. As single ended or push-pull pairs, the tubes will react differently, and the manufacturers provide guidance on how to use their devices. Some tubes are created to intentionally have different gain characteristics, but this is typically found in small tubes like dual triodes, such as the 12AY7 and 12AX7. They are essentially interchangeable but provide different amplification factors. You will never find a tube manufactured as clean or dirty sounding. All of that good stuff happens after production. You will never find mention of breakup or onset of distortion. All you will find are values for Total Harmonic Distortion (THD) at different operating voltages and current levels.

Tubes are produced in batches, fairly large batches at that. The only goal is to produce them so that they meet the requirements of the tube data sheet, which is a type of engineering specification. This is accomplished by careful control of the production process and the individual components used in the tube. Any mistake in the components or assembly process will create an unusable tube. Any defective tubes should be detected and removed from production at the factory, but this isn’t always the case. Because of the number of tubes produced, 100% inspection becomes too expensive. Instead, statistical sampling plans are used. I won’t get into a lot of detail on this subject but a bit of background will help you to understand how tubes are sorted and matched.

In a manufacturing process there is always variation in the parts being made. There are two types of variation. I’ll refer to them as common cause variation and assignable cause variation. Assignable cause is something that you can identify as being abnormal. For example, a failure caused by using the wrong assembly method, bad components or improper test equipment settings. Any of these issues can produce tubes that don’t meet the design requirements, or subject to abnormally high failure rates. They can all be identified and corrected.

That leaves us with common cause variation. This is variation in the product inherent in the manufacturing process. Remember the tube data sheet with all the ratings and specifications? Those are targets, and nothing is perfect. Tubes are produced that meet these specs, more or less. Because tubes are assembled in stages by a combination of people and machines, each one will be slightly different. When assignable causes of variation have been eliminated then you often find that the remaining variation follows something called a normal distribution. A normal distribution has properties that allow you to make statistical predictions about the variation in a production characteristic. Here’s a diagram.

Nominal Distribution Diagram
Nominal Distribution Diagram

This shape is often referred to as a normal, or bell curve. It shows the pattern of process variation in whatever key characteristic you are studying. The centre of the curve, point 0 is the average value of the characteristic you are measuring. The average variation from this point is known as the Standard Deviation. 68% of the values will be +/- 1 standard deviation. 95% will be +/- 2 standard deviations and 99.97% will be within +/- 3 standard deviations. This is all part of basic statistics and process control. It has nothing to do with design specifications. Hopefully, this normal distribution of values falls completely within the design specifications. If it doesn’t, then you are actually producing something that should be scrapped.

Tubes are measured in terms of current, resistance, capacitance and voltage. The physical assembly of all the individual components has an effect on the electrical characteristics as well. This is one reason that tubes are purchased in bulk and then screened by resellers into groups with similar characteristics. Just like the diagram above, there is variation in the manufacturing of tubes. You could look at any important, or key, characteristic you wanted. A stable production process, in a state of statistical control, should usually show variation that follows the normal curve. There are exceptions, but not in this scenario.
Because of the nature of tubes, the testing of many characteristics is accomplished just by turning them on and seeing if they pass current. In most cases, this go/no-go stuff is sorted out at the factory. A reseller knows that barring a hidden defect, everything that comes in should power up and work. So how do they “Match” the tubes and why?

From this point on, I’d like to confine this discussion to common power output tubes. These are the ones that people ask about frequently. I have kept the math simple, and used more complicated formulas in the background, to generate the numbers used in the example below.

When testing or setting up output tubes, the most often tested and used characteristic is plate current at idle, with no input signal. It’s a good baseline. That’s why vendors pay so much attention to it.  In order to consistently measure this when testing, each tube is placed in a test circuit with a set plate voltage and a fixed bias signal on the input grid. The idle current of each tube is measured under the same conditions. If you changed these settings during testing, the results would be invalid.

If you were to take a large number of one tube type, measure and record the idle current and then chart the results, a familiar picture should appear. Here’s a simulation of test results from a batch of 500 tubes.

Idle Current
Idle Current.

 

Knowing this natural idle current is going to give you the best information when trying to match tubes. I say this for several reasons. In amplifiers, a tube can be operated in different modes, with different plate voltages and different output loads. Refer once again to the tube data sheet shown previously.  All of these factors also have an effect on signals passing through the tube, in terms of gain and distortion. Every amplifier has a circuit designed to regulate, or bias the output tube plate current in operation. The tubes natural plate current at idle, is the starting point. The plate voltage and biasing system then adjust this natural current level up or down as required, for the operating level you want. The rated power dissipation of the tube is the limiting factor. For a 6L6GC the maximum plate dissipation is 30 watts. If you can measure the plate voltage and plate current, then you can calculate the plate dissipation. The power P, is equal to the voltage E multiplied by the current I. { P = E x I } From this batch, tested at 400 volts, a tube with 18 ma idle current would be dissipating 7.2 watts {400 x .018 = 7.2}. A tube at the other end of the current range, say 30 ma, would be dissipating 12 watts.

So the natural idle current really isn’t an issue when using a single tube but it becomes very important if you are trying to get multiple tubes to operate together. If the difference in idle current becomes too big, then you will start to generate noise and it becomes impossible to get both the output tubes biased to optimum levels.

In this test batch we can also make determinations based on the normal distribution discussed earlier. Applying some statistical calculations to the test results tells us that the average idle current is about 24 ma. 68% of the tubes produced will range from 21.4 ma to 26.9 ma. About 96% of the tubes will range from 18.6 ma to 29.7 ma, and 99.97% of the tubes produced will have an idle current between 15.8 ma and 32.5 ma. Only 3 out of 10,000 tubes produced will have an idle current outside of this range.

You can also make other determinations. For example, in this batch you can confidently say that 16% of the tubes will have an idle current less than 21.4 ma and 16% will have an idle current higher than 26.9%. Only 2% of the tubes will have an idle current above 29.7 ma.

You may be thinking “why does this matter?” For the person assembling matched pairs or quads it matters quite a lot. You don’t want high current and low current tubes being combined into matched sets. Ideally, matched tubes used in a push-pull configuration should be within 1 or 2 ma of each other and not above 5 ma apart. The reason, is that the tubes run cooperatively in push-pull sets. As one tube turns up, the other turns down. This action allows the signal to be continuously amplified while cutting down the current requirement, generating less heat and reducing the unwanted electrical noise generated in the tube. The hum from one tube will be equal and opposite of the noise from the other tube and effectively cancelled. For an advanced explanation you can search the term common mode rejection.

Another reason why we want to know this information is how the tube will behave when used in an amplifier. As stated earlier, every amplifier is designed to run the output tubes within their operating limits. There is a biasing system in place to ensure that happens. You can use this bias system to set an operating point for the output tubes… to a point. In a cathode bias system you have a fixed, cathode resistor that establishes the operating point of the output tubes. The resistor value is determined based on design averages for plate voltage and current. Amps with an adjustable (referred to as fixed) bias system, input a bias voltage on the input grid of the tube to regulate the current. This voltage is raised or lowered to establish the operating point of the output tubes. In some amps components must be changed to alter the bias voltage, but most today have a simple circuit that provides a range of bias voltage so you can dial in an exact value.

But what happens if you pick a tube from the small percentage that run hotter, or colder, than average?

In a cathode biased amp, the natural idle current of the tube directly affects how much plate current will flow, and by extension, how much power the tube is dissipating. If you use tubes that naturally run hotter, you will be driving the tube harder and it will reach its operating limits sooner. Signal compression and resulting tube distortion will kick in earlier. The increased plate current will actually pull down the plate voltage and help you get that “brown sound” earlier than if the tube was running colder. This is one reason why cathode biased amps like a Fender Tweed Deluxe or a VOX AC-30 have that raw, aggressive sound. They really push the tubes hard, with lower voltage and high current. Distortion, compression and breakup come early and easily. The thing to watch out for is exceeding the rated plate dissipation of the tube(s). If you do exceed this value your only option is to select tubes that naturally run colder, or change the bias resistor to a better value.

In fixed bias amps you have more options. Typically you set the bias voltage to the maximum and check the current. It will likely be fairly low. You then reduce this bias voltage allowing the tubes to draw more current while raising the plate dissipation. The bias circuit is again designed around average values of plate voltage and idle current. If you select tubes that have very low natural idle current they may run too cold for the bias circuit to compensate for and you will get a nasty sound produced by crossover distortion. If the natural tube current is very high, the bias system may not be able to lower it enough to prevent exceeding the rated plate dissipation. If you find that your bias adjustment does not run hot or cold enough bias circuit of the amp must be modified to change the range of bias voltage it can produce. Or, you could ask the vendor for something with a higher or lower current rating. Distortion has yet to enter the discussion.

Fortunately, most tubes can be biased in most amps without any issues. Between the bias system in your amp and the natural idle current of the tubes you use, you have the ability to tailor your sound. If you have a cathode biased amp and find that the sound is never clean enough, or too mushy, you can try using tubes that have lower idle current to roll back the onset of distortion and cool things down. If you have an adjustable bias circuit you can dial in the output tubes to match your style. Dial it down for sparkling cleans at high volume or push the tubes harder to get more breakup at lower volume levels. As long as the tubes run hot enough to prevent crossover distortion and cold enough to prevent exceeding the rated plate dissipation you are good to go.

The concept of matching tubes by distortion is an odd one for me. I always ask myself “distortion under what conditions?” You’ve seen in the tube data sheet that tube manufacturers only address distortion as an absolute value under pre-set conditions of voltage, current and output load. Changing any of these values will change the distortion rating, and after all, they are making the tubes so that they all have the same distortion rating. If we use this measurement as an absolute guide to matching, problems occur. What if 2 tubes were tested and found to distort a signal equally, would they be a matched set? They might be, but what if one of them was drawing 18 ma of plate current and the other was drawing 32 ma? All the issues previously described would come into play.

I think the better way is to match based on idle current and bias the tubes accordingly to get the breakup characteristics you want based on their plate dissipation. This is why most vendors have a matching number on their tubes. Once you have the amp operating the way you like it, replacement is as simple as ordering tubes with the same natural idle current or matching number.

Other tricks you can use to effectively change the breakup characteristics of you amp would include substitution of preamp tubes as described in this article, or by changing the impedance or type of speakers you use. You can buy tubes that are matched by whatever system you believe in but be sure they’re matched for idle current as your primary consideration.

 

I hope you enjoyed this look into the inner workings of vacuum tubes. If you want to learn more, I encourage you to spend some time looking at tube data sheets and the other articles available online through thetubestore.com. If you don’t find this type of information interesting and just want your gear to run well, find yourself a good technician who can worry about it for you. With modern amplifier technology you’re never stuck with one sound.

Thanks for tuning in.

2 Responses

  1. Brian wheeler
    Brian wheeler at |

    So I have a 1month old Marshall Jvm205H with one of the two power tubes red plating and getting way hot. I get that it needs to be replaced, but does the amp need some work as well?

    Thanks,brian

    Reply
    1. Sandra @ thetubestore
      Sandra @ thetubestore at |

      It sounds like you need to do a bias adjustment ASAP. Both tubes should not be red-plating after this short of time and I suspect the bias is out and possibly the cause. Find a local tech f you do not know how to check the bias yourself. The tubes may require replacement due to running so hot. The amp itself should be fine but better safe than sorry.

      Reply

Leave a Reply