Pretty decent read (thd on why we bias amps too hot)
Results 1 to 1 of 1

Thread: Pretty decent read (thd on why we bias amps too hot)

  1. #1

    Join Date: Sep 2008
    Location: 7th plane of ethereal hell
    ME: Ibanez UV7PWH
    Rig: VHT UL /AxeFX2 /Mesa RK2

    iTrader: 8 (100%)

    Pretty decent read (thd on why we bias amps too hot)

    and for those who shun social media:
    Quote Originally Posted by THD Electronics / Andy Marshall Designs
    A customer, who owns a THD Flexi-50, recently asked me why our recommended bias is so much lower than the "commonly accepted 70% of maximum dissipation."

    The following is my rather verbose response:

    Let’s start out by stating that I am a proper amplifier designer, not a hobbyist who cuts and pastes from other designs, and gleefully repeats whatever he reads on the internet, as if it were fact. I’ve been at this for 50 years and have consulted for everyone under the sun in terms of sound, reliability, manufacturability, and just about every aspect of analog audio circuit design you can think of. I do amp repairs and restorations for dozens of multi-platinum rock bands, and I’ve often gotten that work because they see that the only amps they have that almost never need repair are their THDs. I’ve also helped Philips, Fluke and Hewlett-Packard design and build better test equipment based on my experience. I don’t even take proper textbooks as gospel, preferring to actually measure and analyze real-world situations, often proving PhD engineers and university department heads wrong in their assumptions.

    Whoever started the myth that amplifiers should be biased to 70% of maximum allowable tube dissipation should have sugar put in their gas tank. This is totally wrong in every way. There is no magic percentage of maximum dissipation to which one should bias, but most class AB1 audio amplifiers are happiest in the 35% to 40% range. (For the record, 70% is what is generally recommended for class-B2 radio and television transmitters, and is close to what a class-A1 amp wants.) It all depends upon the impedance of the output transformer primary, the load line of the output transformer, the impedance of the screen supply, and a few other, more esoteric factors that aren’t really worth mentioning.

    People tend to repeat something that the read somewhere as if it were truth, and it gets repeated and repeated until 95% of the hits on the internet are recommending something that was wrong to begin with. At times I find it frustrating.

    That 70% factor also goes out the window when dealing with current production tubes, which are not designed to proper specs, but rather are made by reverse-engineering well-made old tubes and making countless assumptions as to why things were done the way they were. For instance, few of the Russian power tubes in production will work in a class-A amplifier. They tested their “designs” in Marshall 50-watt heads. If they worked for a few hundred hours, they gave the go-ahead to production to make 100,000 of them. Everyone who really understood how to design power tubes has been dead for 40 years, and their expertise died with them. Most people don’t spend their hard-earned money on NOS tubes because they sound better (some do, some don’t), they usually buy NOS tubes because they will sound good for 10 years of daily use as opposed to 1 year of daily use with current production stuff.

    The purpose of biasing a class AB1 push-pull amplifier is to eliminate crossover distortion in the signal, plain and simple. This can be done with a distortion analyzer, but it does not differentiate crossover distortion from other types of distortion, so an oscilloscope is the best way of determining when crossover distortion has gone. When a tube is biased too cold, the signal from one tube (or tubes on one side) stops flowing before the signal from the other tube (or tubes on the other side) starts flowing, and this “gap” is called crossover distortion, as it occurs around the zero-crossing of the signal voltage. You bias a push-pull audio output section by first making it too cold on purpose, applying signal that is well below clipping, usually around 25% to 50% of clipping, and slowly increase the bias current (making the bias voltage less negative) until the gap at zero crossing is gone. When you have reached this point, you have attained the maximum dynamic range possible from the amplifier without the (very ugly-sounding) crossover distortion.

    I got curious about where this point fell, as it could be a little bit ambiguous exactly where the crossover distortion “disappeared”, so I started writing down the no-signal (quiescent) dc current in the tubes where I was sure that it was gone. I documented this in roughly 10,000 amplifiers between 1969 and 1999. What I found was that, in high-powered class-AB1 amplifiers with properly designed output stages that ran their tubes near their voltage limits and had relatively linear load lines in the output transformers, the crossover distortion was sometimes gone by the time the bias was set to 21ma quiescent current per tube, often gone by 23ma, and always gone by 25ma quiescent current per tube.

    One surprising thing is that it didn’t seem to matter at all what the tube type was. This number held true for 6L6, EL34, 6550, KT66, KT77, KT88, KT90, 7027A, 6CA7, 7591, 7868, USA, German, Dutch, British, French, it just didn’t seem to matter what the tube was, or who made it, as long as it was in this family of tubes that could deliver roughly 50 watts from a pair.

    So, in order to make life easier for my techs, and to take some of the “guesswork” out of the biasing process, I determined that with all Marshall and HiWatt 50-watt and 100-watt amps, Fender 40, 50, 80 and 100-watt amps, and similar circuits, that they were to measure cathode current with no signal, and set the bias so that the tubes drew 25ma per tube with no signal. Then they had to do a listening test, to make sure that all was indeed well. On a few occasions the amps sounded bad at that rating, and when that happened, I took over to try to figure out why. Sometimes the two tubes were not matched, sometimes the output transformer was damaged on one side, sometimes the output transformer had been replaced with one that was not correct for the amplifier, sometimes the screen supply had too little or too much impedance. But, when things were as they should be, 25ma always gave a clean signal with no crossover distortion.

    So, if your amplifier has maximum dynamic range at 25ma quiescent current per tube, what happens if you run it hotter, like 30ma or 40ma? Good question. First off, the noise floor comes up and, the tubes run physically hotter, which reduces tube life, the power and output transformers run hotter (due to the extra current running through them, even with no signal), increasing resistance in the windings, reducing overall output levels, and shortening the life of the transformers. As you can likely guess, the filter caps don’t last as long, either. I do know a few people who claim to prefer the sound of an amp biased to 40ma per tube, but these people almost always run full out, maximum distortion, 100% of the time, and never ride that expressive slope where clean gives way to “fat”, “fat” gives way to “crunch”, and “crunch” gives way to overdrive.

    Another side of this is that the power supply and output stage can only deliver so much power before they just can’t deliver any more, and this is the maximum distorted full output. If you bias the output stage hotter than it needs to be, you are reducing the ratio of clean output to maximum overdrive output, and robbing the amplifier of dynamic range. Think of it like the idle in your car engine. If your engine redlines at 8000 RPM and will idle smoothly at 400 RPM, you have a ratio of 20 to 1 in terms of “dynamic range” when it comes to acceleration and general driving characteristics. You always want your idle as low as you can get it and still have it running smoothly. You can set the idle at 2000 RPM, but you will only have a dynamic range of 4 to 1, and will be shifting gears a LOT more often to get to speed. Also, you’ll be burning a lot more fuel and fixing your car a lot more often.

    Biasing your amplifier to 70% of maximum plate dissipation is like setting your car’s idle to 5600 RPM with a redline of 8000 RPM. You’ll be rebuilding your engine (or replacing power tubes, filter caps and transformers) a lot more often.

    I have a lot of customers who have amps worked on (under warranty) by other companies (as I only do warranty work on THD amplifiers), and then bring them to me for biasing, as everyone else seems to bias them too hot, and then their tubes don’t last very long, their amps sound overly compressed and “mushy”, and they hum too much.

    It is perfectly safe for your Flexi-50 if you want to try running your bias a little bit hotter. I would not recommend going above 40ma per tube, though.

  2. Thanked by:

  3. Remove Advertisements

Tags for this Thread

Fractal Audio Systems  Shred Training  Angel Vivaldi
Bowes Guitarworks  NumbSkull Audio Production

VigLink badge