Tube Amp FAQ

A comprehensive FAQ on tube & solid-state amplifier design & modding.

Q: Some people suggest that in a cathode biased EL-84 amp, it is better to increase the bias resistor than to change the screen resistors if you want to extend tube life?

A: Proper screen-stop values guarantee life extension of the tube, where simply reducing idle heat does not. if you are going to change only one part of the circuit, the screen-stops are the first area to amend.

Taking care of the screen of the tube is especially important in cathode-biased amplifiers, as TUT2 details.

Q: Does the "70%" bias rule apply to cathode biased amps?

A: No, that is a guideline used for biasing class-AB tube amps.

Cathode-biased amps are usually designed for maximum power in that mode, so the tubes idle at their full dissipation rating. This does not mean that idling them cooler is "against the rules", in fact it is a good thing but will reduce the amount of class-A power the circuit delivers.

Q: Are there any new tube amp circuits? One amp builder claims they do not copy or clone anything. Seems hard to imagine.

A: The basic circuits for pure tube preamps and power amps have not changed in many decades. If you look at modern guitar amp schematics, you will see odd values sometimes, but the circuits will resemble those of older amps. It is only in the area of hybrid technology tube circuits where you find new ideas, such as with Z-B-X and ZBX-2, or with David Berning's and Lars Lundahl's resonant transformer output circuits (from the 2000s and the 1960s respectively).

Everyone building amps - or any other product - thinks they are doing something new and incredible. For the individual, the mere fact they could assemble it and have if work IS new and incredible! In the bigger picture, we see most people re-inventing the wheel. But ... in an even bigger picture we see that we each individually create whole universes, including our own interpretations of other people in our universe, who have created their own universe with an interpretation of us in theirs. So, there are lots of things that are newly created, but the patterns for these items were set a long time ago by others.

Q: I read about the Power Scale-TT technology. Isn't this what some other amp builders do?

A: Power Scale-TT is essentially Two-Thirds Power Scaling, which we call PS-TT. PS-TT is the type of circuit used by designers who do not want to tackle managing the heat from the regulator in a fully Power Scaled amp. Part of their decision is economic and they almost universally make related choices that alter the amp tone even at full output compared to a fixed-power tube amp.

True or "full" Power Scaling, we term "Full-PS". As our tech articles state, Full-PS provides maximum tube life extension with the most accurate tone from full power down to a whisper. The heat pouring off the Power Scale regulator is heat that would otherwise be handled by the tube. A hot regulator means the tube is colder and will last much longer. PS-TT extends tube life also, but not as much. The amount of heat handled by the regulator is lower, so finding a place for it on the chassis is easier.

London Power set the standard for Power Scaling methods and results, and those who copy it usually do it poorly. They change aspects of the circuits thinking the changes do not matter and hope it otherwise makes their "version" look more original. In doing so, the dynamic performance of the amplifier is compromised. Using genuine London Power circuits, kits and methods as outlined in The Ultimate Tone series provides full amplifier performance for both Full-PS and PS-TT approaches.

Q: The notes with the new Power Scale kits show two ways to wire it. How do I choose which to follow?

A: Sonically both wirings will achieve the same result of truthful cranked tone at low-SPLs. The main difference will be in the amount of heat handled by the Power Scale circuit. The Full-PS wiring runs hotter but extends tube life by decades. If there is no room for a fan or heatsink, and the amp is cathode-biased, or the amp is very high power, the PS-TT wiring may be better to try first.

Visit Our Catalogue

Q: Is it true that the SV-TT Power Scale kit can be used in every amp made?

A: Yes. SV-TT can be used in nearly 100% of production tube guitar and bass amps. It would be a clean sweep of 100% were it not for the 650W bass amp built a few years ago using 812A output tubes. These operate at 1500V, nearly twice the 800V rating for the SV-TT.

The SV-TT Super Versatilte Two-Thirds Power Scale kit can be used with fixed-biased amps, cathode-biased amps, amps with up to 800V plate and/or screen voltage, and/or amps up to 700W using no more than the chassis as a heatsink.

Note that SV-TT is optimised for high voltages. Some tweaking may be required for supplies under 500V, where in reality the SV1 or SV84 + TBS would work better and be smaller fitments.

Q: An amp builder is claiming that operating tubes at low voltage and high current will make the tubes last longer. Is this true? And if it is true why doesn't everyone else do this?

A: No, that is not true.

Tubes can withstand voltage stress for their entire life and do not lose this ability unless the glass seal is compromised. At that point, the tube won't function anyway. Tube manufacturers consider a tube to have reached the end of its useful life when the cathode current emission drops to 50% of the rating. This is a true measure of the tube's capability.

The tube manufacturer's own recommendation will always be to allow high voltage within the ratings, and to keep currents low to extend cathode life. This is exactly what we see in old amps and modern amps alike, except in cathode-biased amps where current and heat are maximised - both of which erode tube life.

The only way to extend tube life by lowering voltage is to also reduce current at the same time. This reduces heat from the tube. Of course, this is exactly what happens in a Power Scaled amp and tone is retained.

Q: Badcat amps has just released their "K-master" master volume. Is this actually new technology?

A: Not at all. The description oif the circuit suggests that there is gain after the splitter, which has been done since the beginning of tube electronics. Think of this as a standard preamp followed by a standard MV, then the splitter for the push-pull power amp followed by more gain, then the power tubes and output transformer. The extra gain in the PA is controlled by the "k-MV". You can run the preamp clean or distorted independent of running the power amp clean or distorted.

There are quite a few misleading and outright incorrect statements in John Gilmour's (owner of Badcat) post on TheGearPage. They do not reflect the mind of someone who knows how to design tube amps. The post is reprinted here in italics, with our comments in between:

"The K Master IS confusing. It is so new and different. It confused ME when I first saw the circuit. George Klimek (The K in K master) came up with the idea. What it does is, separate the pre amp from the power amp. There is a circuit between the phase inverter and the power output. So think of the preamp as nothing more than how much grit or distortion you want."

This just describes as above, gain between the splitter and output stage.

"Ordinarily the output from the preamp dictates how much voltage goes to the power tubes after the phase inverter. A regular master is a pot that sits after the phase inverter ..."

He is describing a post-phase-inverter MV (PPI-MV), not a standard MV that precedes the splitter.

"...and takes your signal to ground. It should actually be called a master attenuator. We have placed a controllable voltage output to the power tubes ( The K master) that can be pushed to full output potential even with very minimal output from the preamp. With even a small (very clean) preamp output the power tubes can be pushed into clipping."

Again, just describing the fact that their is gain after the splitter.

"Let me illustrate,
On a normal master volume when you want to see how much clean headroom you can get you put the master wide open and use the preamp to push up the amp volume. That is the full output of clean headroom you have available."

So far correct

"In the same scenario with the K master you set the preamp how you want it to sound and drive up the power tubes to their full potential , which is well beyond how loud the amp would be with what signal is coming from the preamp. Think of it as signal booster after the phase inverter prior to the power tubes."

This is inaccurate. Since the k-MV is the last control in the signal chain that determines overall loudness, it is simply a MV - it makes no difference if it is passive or active. To experience the clean output of the amp, you can set k-MV to max then slowly increase the preamp volume until clipping occurs. Clipping will occur first at the output stage, just as with a Fender amp, a plexi, a Hiwatt, or most other amps. There is also an assumption of a low-gain preamp, such as is found in the Route-66 (Dr.Z) , or as found in many of the limited tube count 18W amps. Those amps, as designed originally by the big OEMs, were intended as "tuning" or "practice" amps - not for "real" preformance or stage amps. Real amps designed for real players have more gain in the preamp.

"It also eliminates phase cancellation that is experienced in a normal master volume. Which is why you lose bottom end and gain."

Phase cancellation only applies to cross-line MVs, but is not the reason for tone chage over control sweep with this MV wiring. As with a standard MV, the reduced resistance at low MV settings changes the RC time constant as the control works against the  circuit capacitance. TUT showed the fix for this, as it applies to standard MVs. For a cross-line MV, see the Soma-84 project in TUT5.

The reference to a regular MV being after the splitter assumes that there is no feedback loop around the power amp. In the low tube count 15-18W amps, the gain that must be sacrificed to have a feedback loop around the power amp would make the amp much less useful. Look at the Fender Champ as an example of too-little gain because feedback was incorporated. The "loopless" copies such as the ValveJr still have too little gain, which modern amps like the Marshall C5 address by adding an extra preamp stage. Similar mods should be made to the 15-18W amps and the k-MV is an attempt to do this. The extra gain is added after the EQ though and whther it is placed before or after the splitter is irrelevant to the result.

"You will hear your exact full tone all the way down to whisper quiet."

Any properly designed MV will do this over the clean range of the amp.

In another post on the same thread, John states:

"most 15 watt amps will not give you continuos 15 watts of clean headroom"

That would have been a good place to differentiate his products - they must not meet their power rating either. In conventional circles, power output is clean sinewave power. Peavey rates their amps at 10% THD, which shows visual flattening of the peaks but nothing extreme.

Further along, John says:

"As for whether this is unique or not, we will let the patent office decide that."

John, the patent office does not care about "originality". They will take your money regrdless of whether the idea is public domain or not - which it is. Many companies play the patent-pouch game to have bragging rights that they have this many patents, or more patents than so-an-so. Better to stroke your ego with huge sales rather than makng your attorney and the USPTO rich.

Q: A friend of mine builds hi-fi amps and suggested using an 845 for guitar. Is that crazy?

A: The 845 power triode, and similar tubes such as the 211, 811A, 812A, 572B, 300B and 2A3 are touted by some audiophiles as having the best tone for audio. They are quite linear tubes often used in single-ended circuits at high-voltage creating just a few or tens of watts. The 211 used in push-pull easily produces 200W+ as a pair. Similarly, the 811A can produce up to 340W per pair at 1500Va. These tubes do sound quite warm and transparent.

Any tube that can amplify can be used in a guitar amp circuit. For guitar, "best linearity" is not a prerequisite. So, this opens the door to all kinds of experimenting and opportunity to create something especially pleasing.

Q: A while back I Power Scaled my Fender Twin Reverb amp using the SB-1 kit. It works really great! Recently I tried adding one of the Sustain kits, the SUS-3, but can't get any compression. What should I look at to get it working?

A: Note that this answer also applies for modern installations using SV1. SB-1 was discontinued in 2012.

Within the SUS kit itself, there are a few resistors that can be tweaked to change the attack, release and depth of effect. Keep in mind that the Sustain effect only changes the attack of notes, rounding them to be less abrupt. R21 changes the basic depth of the effect, but R17 and R18 also contribute to raw depth.

However, the first thing to check is the raw bias voltage being fed to the SB-1. If you are just using the bias tap on the power transformer, odds are that the raw bias voltage is not high enough to provide the best performance from either the SB-1 or the SUS-3. Measure the raw bias voltage at W3-5 on the SB-1, then the voltage fed to the bias pot from W3-4. If these are very similar, the bias regulator operation is marginal at best and the SUS will provide no effect at all near full power.

To fix the bias voltage issue, an alternative raw bias supply should be installed. The RBX Raw Bias Auxiliary supply kit is designed for situations just like this one.

Q: I added the GMX kit to my amp and it really did make it sound bigger. Cool! I'm trying to add Power Scaling now but there seems to be some interaction with the GMX circuit. The GMX makes the sound bigger while Power Scaling makes the sound smaller, so are they sort of battling each other? The amp makes funny noises when I dial it down.

A: The problem here is that you are Scaling the supply to the GMX control circuit. That circuitry needs a fixed supply. Move the "B+" connection for the GMX card to the Va or Vs filter cap directly. Assuming you are using a DC Power Scaling kit such as the SB or SF series, the centertap of the output transformer will be Scaled but not the feed to the GMX board. The other GMX circuit connections remain as for a fixed-power amplifier, and the GMX will track the changes to the Power Scaled output stage.

Q: I just got my new London Power guitar amp and it... well... sounds crappy. What should I check? Or is it supposed to sound like this?

A: Prior to plugging in and playing, you have to bias the tubes. We set the tubes back to minimum idle current before shipping as this helps avoid certain problems during the first power-up after transit. The output tubes will be idling at a very low current if you turn the amp on without setting the bias the first time. The output will be low and dry sounding - not very musical at all.

The bias procedure is in the manual and is very simple to do. Once the tubes are biased up to a normal current, the amp's tone opens up and will be very dynamic. You do not have to reset the bias until you change a tube, or if you merely wish to explore how the bias effects tone.

Follow-on: Wow! That's all it took. The amp sounds as amazing as I expected it to when I ordered it. Thanks!

Q: I'm trying to figure out how to Power Scale my Blackheart amp using the SB-2. The screen voltage is about 50V less than the plate voltage, so which one should I use for the Power Scale pot?

A: Most EL-84 output stages, including the Mesa F-30, the Valve Juniors, and all of the copies like the Blackheart, have significantly lower voltages on the screen than on the plate. The standard wiring for the SB-2 will cause a reduced maximum output power but will go down to unusably quiet levels.

The alternative wiring for the PS pot will introduce a dead-spot in the control sweep, where the top end portion does not vary output, although this allows full output to be achieved and unusably quiet output when dialed down.

The fix requires three steps: First, power the PS pot from the Va node. Second, control the Vs regulator using a voltage divider tied to the Va regulator output. Use 1k/V for the divider - often resulting in a 330k-1W to ground and 33k-68k between regulators. Third, add a 10k series resistor between the PS pot 'X' and the Va cap positive. Then add a 22uF-450V cap to ground from the PS pot 'X'. This will provide a cleaner voltage to the Power scale pot, which in turn feeds a cleaner voltage to the regulator gates.

The divider fix maintains the stock proportion of screen and plate voltages over the full range of full-power down to unusably quiet.

In these lower-voltage amps, one could also use a 24mm ganged pot for the regulators. The pot should be rated for 500Vdc. This application of direct control is not universally applicable, as TUT4 discusses. The Drive Compensation function should be left on a separate control and not ganged with the Power Scale function.

Note that SB-2 is a discontinued kit replaced by SV2. SV2 allows for differences between the screen and plate voltages and there are no "dead spot" issues.

Q: I was biasing some tubes in my new Aurora amp - great amp by the way! - and set the meter to milliamps by mistake. Will that damage anything? I noticed the manual said not to use a milliamp meter, just to use a volt meter.

A: No harm will come to the amp or your meter. The bias test points on all of London Power's current amplifier models are designed for use with a voltmeter of any type. Current-limiting resistances for the jacks themselves protect old-style galvanic meter movement volt-ohm-milliamp meters from potential damage if their range is not set correctly. This disallows the use of direct current measurement.

As the manuals for the Aurora, Studio-10, Sustainor, Studio Power Amplifier and Raven state: "The jacks are specifically designed for use with a volt meter and will give erroneous readings with an ammeter".

Q: I assembled one of your kits but noticed later that I was supposed to leave the 1W resistors up off the board. Do I need to get new resistors and mount them like the kit note says? Will the kit still work if I don't?

A: Power resistors are used in circuit locations where the product of the voltage across the part multiplied by the current through the part will cause some heat to be generated. A resistor rated for 1W is physically larger than one rated for 500mW, so the 1W part can radiate more heat if required, or radiate the same amount of heat with a lower temperature rise.

The best way to get this heat away from the part is to allow for airflow around the part. This is easily achieved by not pushing the body of the resistor all the way down to the board. Instead, the leads are left a bit longer so there is 6mm or so (about 1/4") between the board and the resistor body. Apart from providing better cooling for the resistor, elevating the part will keep it from burning the printed circuit board if it gets very hot.

Ohm's Law will tell you how much current will pass through the resistor for a given voltage across the resistor. However, in most cases, you can calculate a "worst-case" heat level by making simple approximation. Most power resistors used in our kits are across the supply rail - that is, one end ties to the supply and the other is at ground, or very nearly so. In such a case, you can simply use one of the power equations (See our book Ready Set Go) and take the square of the supply voltage divided by the resistance.
(V x V) / R

If there are resistors in series, they share the voltage in proportion to their resistance values, and thus the power is shared as well. Notice where this occurs in the given circuit to save yourself from having a "too worst-case" approximation that is exaggerated.

So, you have to figure out how much heat the parts will dissipate to determine how "risky" it might be to leave them flat on the board. The circuit will no doubt function well for years without any changes to how these parts are mounted.

Q: The LP-MV (London Power Master Volume) kit seems like a very simple solution to controlling volume without tone change. Will I still be able to use my 'presence' and 'resonance' controls?

A: Yes. Their function as power-amplifier feedback frequency shapers is unchanged. However, the depth of effect must be kept proportional to the signal size, and the LP-MV takes care of this automatically.

There is a trend away from power-amp 'presence' controls to preamp 'presence' controls. This simplifies wiring and board layouts within the amp. The presence effect is achieved in a different way, but it is difficult - if at all possible - to hear a difference.

Q: Can I use the GMX kit as a power amp? If I drive it from a 12AX7 will it sound like a real tube amp?

A: Yes, you can use the GMX (Transconductance Multiplier) kit as the complete output stage, or merely to augment the tone and/or power of a conventional output stage. There are a few things to consider.

Assuming you wire the 12AX7 as if it were two triodes driving the output transformer (OT), the sound will be that of a triode amplifier: clean, transparent, and very linear. The GMX circuit will just make this sound even larger and more transparent. In this arrangement, the plate supply must be kept within what a 12AX7 can handle, so something less than 270V - more like 200V to 250V, in turn requiring the use of a custom OT. This will accommodate the OT flyback voltage. Note that the current sample and sense resistors would have to be significantly altered in proportion to keep the current through the tube within its limits.

If you were looking at a more "modern" application, the 12AX7 becomes a gain stage and a concertina splitter, or is wired as a Schmitt splitter, with the split outputs driving the GMX inputs. In this case, the current sample resistor is increased further in value even from the above approach.

In any application, the GMX circuitry requires appropriate heat sinking.

Q: Can I get the bias voltage to remain on the tubes when my Power Scaled amp is on standby?

A: Standby (SB) switches are not really needed on guitar amps, but your question brings up two points. One, is that in the Power Scaled amp the bias voltage tracks actual screen voltage. If Vs disappears, then -Vb disappears, too, unless the standby switch is wired according to the PSK notes. These notes suggest using the SB to interrupt the voltage feed to the tube screen element - and preferably to ground the screen during standby - while the voltage feeding this switch is tracked by the bias regulator. The second point is that the SB can be used to create a "bias standby". In this circuit, the raw bias supply must be very high - at least -80V in most amps. Most amps do not have this much bias voltage available, but an easy mod will fix that: either adjust the stock supply or add an auxiliary bias transformer as shown in The Ultimate Tone Vol. 5 (TUT5). The SB is simply wired across the pass element of the bias regulator. The 'operate' position has the switch open and the bias tracks as it should. In standby, the regulator is bypassed and the full -80V is applied to the tubes, turning them off.

Q: Your new amp line looks really cool. What happened to the Zen and the Mini-Marshmallow?

A: The Zen and Mini-Marshmallow circuit essentials have become available as preamp kits. The sounds of the Zen model are available from Aurora, so Zen itself became redundant. TUT6 outlines some Mini-Marshmallow options in its Dumble chapter, while the High Gain Preamps chapter details circuits that will provide the Zen experience.

Q: I installed one of the old Power Scale Kits and have another amp I wish to Power Scale. Should I use the old PSK circuit or one the new DC Power Scale Kits? Is there an advantage to the old ones?

A: Since you installed the old style circuit, you already have some experience with it and would have an easier time installing it a second time than a first-time tech would. On the other hand, the new SV-series Power Scaling kits are a lot simpler to install.

The Power Scale control itself can be located anywhere on the chassis as it carries filtered DC and therefore cannot interfere with any audio signals. DC-Power Scaling is better suited to multi-channel amps than the old-style PSKs are. This is especially true if multiple PS controls are desired, or when Power Scaling is to be used only with one preamp channel. Power transitions are instantaneous with the SVs, whereas the older design required a fast-transition circuit to get a "reasonable" but not-as-quick transition.

The old-style PSK-1 for fixed-bias amps and PSK-2 for cathode-biased amps operate with pulsating DC, which is very noisy. Wiring layout and component placement is trickier, and can highlight faults already existing in the amp's wiring. This takes a very talented tech to install, and our list of recommended PS-installers have a lot of experience and creativity. The old-style PSKs work well for single-channel amps. The new style SV1 is for fixed bias amps, and the SV2 is for cathode bias. Small amps can use the SV84 on its own if the amp is cathode-biased, or add TBS if it is fixed-biased.

Note that most amps have barely adequate bias supplies even for their stock function, and are enitrely inadequate to support the Power Scale kits. Add RBX in most fixed-bias installations. See the guide "Selecting a Power Scale Kit" to determine if you need RBX or not.

Q: I see your new amps use electrolytic capacitors. In your books you say those are not as good as plastic filter caps, so are your amps as good as they used to be?

A: Careful reading will show that there are performance differences between electrolytic and plastic caps, but there are very good examples of both. The longevity and near perfect performance of plastic caps is often outweighed by their bulk and cost. Modern electrolytics are very good quality, and surprisingly, it is not the most expensive samples that perform best.

We wanted to make our new amp line affordable to more players, and electrolytics were a step in that direction. The 4U amps use a mix of brute-force electrolytic filtering followed by a sophisticated active hum filtering circuit using polypropylene caps. The overall "sound" of the supply is that of the poly caps. High-quality components and proper circuit layout and design give the new amp line a level of performance that was not possible in the past.

The 2U Studio-10 model has an all-plastic supply and extensive choke filtering (three chokes). For this low-wattage amp, an all-plastic supply is still affordable and performs very well.

Q: Are any of the new amps in your line hand wired?

A: Yes, but not in the coloquial sense. They are all built on high-quality printed circuit boards, which are hand-stuffed and hand-soldered, then hand-wired to each other. Our PCBs have heavier copper traces than the industry standard, which makes them more reliable. Serviceability is greatly improved over hard-wired models from the past - and particularly compared to mass-produced PCB amps - because of thoughtful physical layout. PCB construction allows circuit refinements that would be a nightmare to wire by hand even for your personal amp, so overall performance is silky-smooth.

Hand-wired amps can take days to wire given a free schedule, but that often expands out to weeks or months when trying to handle day-to-day business activities. Printed circuit boards change all of that. PCB assembly allows amps to be final-assembled in a short lead time from pre-assembled modules, improving product throughput.

Q: Are the amps on the market that have "Variable Power" or "Power Drive" really Power Scaled amps?

A: No. Most are simply master-volume (MV) amps; some with a post-phase-inverter MV, and some with a conventional MV. The out-of-production Viking amp, for example, is a master-volume amplifier despite their claim. Like any MV amp, power output varies with the MV setting BUT waste heat in the output stage is the same as any conventional fixed-power amplifier. Tube life is NOT extended and the tone MAY vary with the MV setting.

Other types of splitter-embedded "power" controls are also just MVs, regardless of being called "power", "variac", "limit", "power dampening", "wattage", etc.

A true Power Scaled amp extends tube life when power is dialed down. There may be waste heat in the Power Scale regulator, but that is easily managed. Besides, if the regulator is running hot it means the tubes are running cool and will last longer. Properly implemented Power Scaling does not change the tone as you dial down loudness.

A MV can be set up for minimal tone change versus setting, too, as we saw in TUT3 and TUT4. But... we still cannot extend tube life with just a master volume.

Q: I have a VHT stereo power amp. One channel was making a crackling sound, so I looked inside and noticed that one power tube was glowing orange. I replaced it and the new tube glowed orange, too, but it was the non-glowing tube that was causing the crackling. When I replaced that tube, the tube that was glowing orange wasn't glowing orange anymore. How can the one tube affect the other? I had the amp set to 'class-A'. Everything worked and sounded okay once I replaced the crackly tube.

A: The key to the answer is the fact you were in the 'class-A' setting for the amp.

Like most amps, any reference to "class-A" is truly saying "cathode bias". Like most amps, the cathode-bias resistor is shared by both tubes in the output stage and is sized accordingly. If one tube is missing or not conducting its share of current, the other tube will bias to a point that is above its plate dissipation rating, and the plate will glow orange.

The crackling tube in your amp must have had an intermittent internal connection to either the cathode, screen or plate. Interrupting the primary current path from the cathode to plate takes that tube out of the circuit. It does not conduct any current and is basically 'off'.

An intermittent screen connection causes the tube's internal resistance to rise significantly when the screen voltage disappears. Again, that turns the tube 'off'. The intermittency of this connection will be heard as crackling through the speaker.

In the 'class-AB' mode of this amp, the tubes are operating with a fixed bias voltage and therefore idle independently of each other. If the crackling tube opens and conducts zero current, it has no effect on the other tube. You might hear crackling and possibly a small increase in hum until the tube reconnects itself.

Q: What is "fixed" bias? If you make it adjustable, is it no longer "fixed", or is it "broken"?

A: "Fixed bias" refers to the bias condition with respect to the signal cycle and means that the bias condition of the tube does not change over the signal cycle. To understand this, we must first look at cathode bias.

Cathode bias is also called "self" bias. A resistor is placed between the tube cathode and ground. The grid of the tube is also tied to ground through a resistor, which for practical purposes looks like a direct connection to the bottom of the cathode bias resistor.

As we said in the TUT-series of books, the tube is "cathode-centric". This means that the center of the tube's universe is its cathode. It measures every influence with respect to the cathode. If the grid is negative compared to the cathode, the tube will conduct less current from its cathode to its plate. If the grid is at the same voltage as the cathode - or more positive than the cathode - then the tube will conduct as much current as possible.

With the resistor between the cathode and the grid, and the grid effectively tied to ground, the tube starts off with similar voltages on both elements when power is first applied. As the tube heats up, the cathode conducts current which is pulled through the cathode resistor. This creates a voltage difference between the grid and the cathode, where the grid looks negative with respect to the cathode.

As the tube continues to heat, and as current through the tube rises, there is eventually a point where the negative grid restricts the rise of current. The tube finds a "balance" or "quiescent" point. You could say that the tube has found its "happy place" in this specific voltage environment with this specific cathode resistor value.

Now we apply a signal to the grid. Over the signal cycle, we can monitor the voltage between the cathode and ground. Instantaneously, the voltage might rise and fall with the signal. In a push-pull amp, this resistor might be shared by two tubes that alternately draw more and then draw less of the idle current. Over all, the cathode voltage averages to a value different than the idle value with no signal.

To counteract this effect, we add a capacitor across the bias resistor. This allows the signal-dependent current changes to be pulled around the resistor through the cap. We still find that the voltage across the resistor changes over the signal cycle, even with a very high value cap.

In cathode bias, then, we see that the bias condition of the tube changes with the signal. The idle condition is not constant.

The idle condition can only change if the voltage between the grid and the cathode changes. The intuitive "fix" is then to remove the bias resistor and force the tube to conduct a specific current by applying the appropriate amount of negative voltage to the grid. We can still idle the tube at high currents if we want, and that is an issue discussed in the "class-A" Q/A elsewhere in this FAQ, and in the TUT-series.

With our negative control voltage applied to the grids, we find that the idle condition is more constant over the signal cycle. We then describe the idle condition as being "fixed" with respect to the signal. Colloquially, most techs and designers refer to the presence of negative grid control voltage as "fixed bias" or the related power supply as the "fixed bias supply".

We can achieve the fixed bias condition in other ways that do not require a negative voltage in the circuit, just as we can achieve class-A conditions without using cathode biasing. The use of the negative supply to control the tube is merely a bias method and should not be confused with the desired bias condition.

Q: How can I tell if an amp is really class-A?

A: There is a lot of hype about "class-A", a lot of misunderstanding about what it means and a lot of misrepresentation of products claiming to be "class-A".

First, we should understand what "class-A" means, and there are several definitions as we saw in our books TUT, TUT2, TUT3, TUT4 and POP:

1) All output devices remain conducting during the audio cycle
2) Power consumption is constant from idle to full output
3) All output devices contribute to the audio signal at all times
4) Output distortion is predominantly even-order

The most important point to remember, is that class-A is a bias condition. This means that the tubes are running hot - often at their maximum plate dissipation - and it does not matter how we achieve this condition. "Cathode bias", "self bias" (same as cathode bias) and "fixed bias" are all bias methods or specific circuit approaches used to get the tubes to idle at the desired current. These have been explained in detail in our books and many other texts.

It is obvious that a single-ended (SE) amp must operate class-A. It has only one output device, which must be conducting through the entire audio cycle if the entire signal is to be reproduced. This fits definitions 1 and 3. If we place an ammeter in series with the power supply, we find that power consumption rises at full output, so definition 2 is not met. The output distortion is mostly even-order, so definition 4 is met. In general, everyone agrees that an SE voltage amplifier and/or power amplifier operates class-A.

Push-pull class-A amps are most often cathode-biased, but many hi-fi amps are fixed-biased The usual design approach is to idle the output stage at the limits of the tubes, and hope that this is slightly higher than maximum output required. Traditionally, this is achieved by idling the tubes at half the peak output current required for the desired output power. We should look at a popular example.

Amps with quads of EL-84s are popular, with the iconic AC-30 as their progenitor. The EL-84 is rated at 12W plate dissipation; four of them idle together at 48W. If we have the recommended (and typical) load of 4k-aa, then each half of the circuit "sees" a 1k-ohm load. 30Wrms is 60W peak, requiring about 245V peak at 245mA peak swing per circuit half.

According to the traditional approach, we should idle the output stage just above half the peak current, or about 130mA. Dividing this evenly over the four tubes means that each tube conducts 32.5mA, which at 12W means B+ could be 369V. Most AC-30s and copies operate at 320-345Vdc, and the idle current is a little higher. However, at the point where one side of the circuit conducts the whole 130mA or so, the other side is conducting zero current. This is a specific point known as "limiting class-A". If we drive the tubes no harder, then we can still call this a class-A amplifier. But, we do drive the tubes harder to get to the 245mA peak we need for full output, and so we have gone out of class-A.

The power draw in our example rises at full output, and this has been measured by many techs over many decades. When running such an amp into a bench load with a clean sine wave adjusted to just below clipping, it is obvious that the output stage runs cooler at full output than at idle. This is because the voltage across the tube is swinging between a low of about 55V to a high of about 565V. At the minimum voltage point, the tube is conducting the maximum current (245mA). At the maximum voltage point the tube is 'off', conducting zero current. If we take the average of the signal current (173mA) and multiply by the supply voltage, we have a maximum power consumption of about 60W, where idle was 48W.

Overall, the AC30 fails to meet the first three class-A definitions. Its distortion output is quite high and predominantly even-order, so on that basis only might we call it a class-A amplifier. Even here, there are two mechanisms that contribute to the THD that are not related to class-A-ness. The first is that the circuit is not dynamically balanced, partially due to the use of a cathode-bypass cap for the common bias resistor for the output stage. Dynamic balance is better without this cap, reducing both output power and distortion but then skewing the THD spectrum to a balance of odd and even harmonics.

Second, the high idle current causes a lot of thermal agitation within the tube, and thus generates a lot of thermal noise, which modulates the signal, creating intermodulation distortion (IM) that is far worse sonically than any THD.

To make a true, pure class-A amplifier, we must idle the output stage slightly above the peak current required. If we want the 245mA swing into the same 4k-aa load as above, then our idle current should be at least 250mA. With a 320V supply, idle dissipation is 80W, or 40W for each half of the circuit. This would require eight EL-84s in total, or four of EL-34s, 6L6GB/Cs, or just two KT-88s. Since the peak audio current is less than the idle current, current pulled from the supply will be constant, and thus power consumption will be constant. The traditionally quoted efficiency for a push-pull class-A amp is achieved, at slightly less than 50%. Distortion will be low if signal balance is maintained, and the amp will be class-A by all definitions.

So, now that we've seen so many reasons for amps not living up to what they are said to be, how do we tell which are which?

The short answer is to look at the number of output tubes in the amp, and their types. Add up all of the plate power dissipations, and then divide this by two. The result is the theoretical maximum class-A audio power output, although the true value is slightly less due to the imperfection of tubes.

For example, looking at the AC30 again, we have four EL-84s rated at 12W each, thus 48W total. We could theoretically get 24W of class-A output from this tube set, which is about what was predicted in TUT3.

Example 2: Suppose we have a pair of EL-34s, rated at 25W each thus 50W total. We would hope to get 25W class-A, but likely a bit less.

Example 3: Suppose we have four 6L6GCs (real ones), rated at 30W each. Therefore, we have 120W total and would hope for 60W class-A.

Example 4: Many newer boutique amps use a mix of tubes. Suppose we have a 6V6 working against an EL84 in push-pull. Both are rated for 12W, so the total is 24W, and the maximum class-A output would be 12W, or less.

Example 5: Further to example 4, suppose we have an EL-34 working in push-pull against two EL-34s. All the tubes are the same, but we have a dissimilar number for each side. Therefore, we use the lower of the power ratings for the two halves of the circuit, and divide that by two for maximum class-A. The circuit will produce more power than this (likely) but it will be asymmetrical and distorted, but will not be class-A.

It is an easy matter to achieve the sound of a class-A amplifier. It is much harder to find truth in advertising about just how class-A an amp really is.

Q: You used to offer a switching kit using PVAs. Why did you discontinue it?

A: Photovoltaic relays, or PVA-series devices from International Rectifier, and built by other manufacturers, are very neat devices. They conduct current bilaterally, so they can handle AC signals. They offer very high signal voltage handling possibilities. The control side is basically an LED, so it is quite easy to drive. With optical coupling between the LED and the bilateral switch, the voltage difference between the control circuit and signal path can be hundreds of volts.

It seems like an ideal alternative to relays. But... it has one big flaw: variable switch capacitance that depends on the voltage across the switch.

When the PVA is 'off', there is capacitance between the two ends of the switch. This capacitance rises as the voltage across the switch is reduced. It is good practice in audio circuits to provide a DC path to ground on either side of an audio switch - particularly series switches - to reduce DC shifts in the circuit when the switch opens and closes, which comes through as nasty pops. Unfortunately with the PVA, this is the worst voltage condition for it to be in, as its capacitance is the highest - 100s of pFs - so there will be high-frequency feedthrough. This will at best be merely annoying, but at worst can cause circuit oscillations.

In low-impedance circuits, this capacitance might not be a problem. In the high-impedance tube circuits of guitar and hi-fi amps, the PVA cannot be used in the usual way. It can be used where there is a standing DC voltage in the off-condition for the switch AND where its switching function will not cause an audible glitch at opening and closing.

In guitar amps, the places you might think to use a PVA are better controlled using miniature relays. Relay quality has improved in recent times.

Q: In a magazine Q-A, a player wanted to pull tubes to reduce power, but the "expert" said this would cause a meltdown of the remaining tubes. Of course, it was suggested that the expert's attenuator product was the preferred way to go. Is any of this true?

A: This is a person who should know better!

Removing tubes from a multi-tube fixed-bias output stage is never a problem. You can remove any number of tubes, and yes, that means you can take one tube out of a two-tube amp; one, two, or three out of a four tube stage, et cetera. This sounds heretical to techs stuck in the mire of convention, but it is something that has been known since tubes were invented.

The even-number tube extractions reduce power symmetrically. Neither the tubes nor the transformer will be damaged. Power will be reduced and so will frequency bandwidth - you will lose some bass and some treble. This is the point that switching the impedance selector to a less-than-load setting is supposed to correct, but it is completely subjective whether you should. The only 'should' of the matter, is do I like it this way, or do I like it that way?

In the uneven tube extractions, asymmetric power reduction occurs. Conventional thought says "the one tube on one side of the circuit will be trying to match the output of the two tubes on the other circuit half". This is wrong. The single tube can only produce so much power, and that's all it does. It doesn't melt down. The transformer does not blow up.

So, what's missing from conventional thought? The realization that tubes are "self-limiting power governors", which was stated in The Ultimate Tone (TUT), and explored in more detail in TUT2 and TUT3. TUT4 explores all of this in great detail. Our "expert" should get a copy.

In the end, you can pull tubes to reduce power, unless the amp is cathode biased - then you have to split the bias resistor. In any case, you do not have to worry about the impedance selector either.

Q: I thought impedance matching was critical. Some designers say the output transformer must be changed if you want to use different output tubes. That seems awfully expensive.

A: It is awfully expensive, and awful that such things would be suggested. There are two issues here, though; one is the notion of "impedance matching", and the other is simple design preference.

As stated throughout the TUT-series, speaker load impedances and reflected loads to the output tubes are all "nominal". An 8-ohm speaker may actually look like anything from 6-ohms to 100-ohms, depending on the frequency, since the reactive impedance changes with frequency. This means that the reflected load to the tubes is varying widely over the frequency range.

A nominal 8-ohm load may reflect 4k to the plates of the output tubes with a given transformer. The amp might be designed to produce its maximum power into this load, with a designed frequency response. This is the "power bandwidth". If we change the load to 16-ohms, the reflected load doubles and the frequency response shifts upward. We lose bass but have a brighter sound, and also lose power. If we change to a 4-ohm load, the reflected impedance drops to 2k, into which the tubes produce less power, and the bandwidth is again narrowed.

The reason for the confusion, I believe, is that people think tubes will try to behave the same way transistors do. Into half the load impedance, a transistor will try to deliver twice as much current. The device may overheat and destroy itself in the process. Tubes, however, simply don't behave like transistors.

The design issue for impedance matching comes into play when a designer takes the approach that "everything is critical". In some circuits, this may be the case. Tubes don't really care. There is no optimum load for a tube unless you are going for minimum THD, and this then depends upon the other operating conditions. For guitar, criticality is purely aesthetic. The designer says "this is good", "this is bad" and in that decree believes it to be so. He is correct in his subjective impression, but should not confuse the subjective and objective.

Design approaches are dealt with in TUT4.

Q: An "expert" suggested that I change my speakers to ones that match the highest impedance tap on my amp. How do I do this and still have the option of using a second cabinet when I play out? I think I would need three cabinets to achieve this.

A: Yes, and what a waste of your money.

Not too surprisingly, this is the same expert as in the tube-pulling/power reduction question. H

About London Power

London Power was founded in 1990 by Kevin O'Connor, after two decades of research into innovative audio amplification techniques. Kevin is an audio designer, author and speaker, known for his proprietary methodology called Power Scaling, and for his eleven books on audio subjects. Whether you seek amps, kits or information, you'll find useful answers here on not just the "how" but the "why" of audio amplifier design. Please enjoy the information on this site, and don't hesitate to contact us with questions.

Contact Us

(807) 473-0952
11 am - 6 pm, M-F
PO Box 10012, Thunder Bay, ON P7B 6T6 CANADA