joepampel
- Mar 22, 2023
- 15 min read

Load Boxes

Updated: Feb 29

What do we do when our amp is "too loud"?

TL;DR Your amp makes some power, that power may easily be too much to play at home - or even out - the way you would like to. If you live in an apartment, 1W may be too much. You can't just reduce the power with a volume knob, you need to "soak it" - effectively turn it into heat - with a device that ideally lets your tube amp work as it normally would.

I have stayed away from the term "reactive" so far here because it's a funny topic. Amps are designed using load lines. Load lines are really based on ohms law, where volts = current X resistance. A tube's characteristic curves are charts of volts against current - and so you can pick points along a tangent to derive a resistance value for those points (some V at some I equates to some R). So a load line really corresponds to a resistive load for the circuit - it is a straight line. The actual resistance will determine the slope of the line, and then you can graphically solve for things like power out, distortion products and so on.

Now this does not happen in real life; speakers have a load impedance that varies dramatically. A real speaker load is an elipse, and it is generating reverse EMF that is running backwards through our output transformer and into the plate circuit. Oh, and don't forget negative feedback (if we're using it). And there is more, of course.

What does a load have to simulate?

What we tend to focus on is this:

This set of output data is pretty close to the Fender spec. 450V on the plates, driven with 42.5V RMS on the grids (bias is -30V, so the signal is ~60V p-p, and 1.414 x 42.5 is 60.095 as a check there). Fender jacked up the screen volts, but otherwise we're pretty close here. You can see output power and distortion change with effective load resistance - not impedance mind you - but resistance. What they are not showing here is that the mix of harmonics is changing as the load changes. 4K, where we tend to see these amps is on the 2.5% THD line; the lowest THD is at 6K which also has the highest output power (50W). But the mix of distortion products is not going to be the same.

If you had an amp with a 6K output transformer and halved the load, swapped an 8 ohm cabinet for a 4-ohm, it would be reasonable to infer that you would see something like the 3K load spec on this sheet - THD would be close to 3.5% and power would be down at 32W or so. This is not showing the effect of that load line though - would the tubes be overheated? If V is the same and R changes, I will also be changing, and P = I * V.

Here is a more detailed look (from the GE Tube Manual) so you can see the relationship of element current vs dissipation etc. in the tube with a 3.6k load:

The rule of thumb when you design (what I was taught anyhow) is you want 10:1 for maximum voltage transfer (load is 10X the source, or more) and 1:1 for maximum power transfer. The plate resistance of the tubes is not a simple value though, as you can see from the above charts.

And when we're designing, our load line looks something like this. The full load line would continue below, and you would flip a copy of this chart upside down and backwards to show the whole signal swing. Class AB implies that we are conducting less than 360 degrees and more than 180 - so each tube cuts off and the other picks up the signal. 180 degrees would be Class B. Class A would be 360 degrees. We're of course talking about a sine wave having 360 degrees in a full cycle. Any amp where the drive signal can cut off one tube is AB. There are no "special classes". A class AB amp will be operating in class A until one tube is cut off. it's a continuum.

As there is a load line, there is also a maximum dissipation curve (which should be up and to the right), and that is what we do not want to violate when we change load impedances.

Taken from https://www.apexmatching.com/frequently-asked-questions GE AB1 specs.

We'll do a full graphical solution in another post, but the thing to keep in mind is that the lengths of those load lines - from 0 signal to max and from 0 signal to min would be the same length if the system was "perfect". And of course they are not, so the difference in each segments length is our distortion. Even harmonics are asymmetrical (one side of the wave is affected) and odd harmonics affect both sides. The takeaway is that we want to match the expected load, but it not really as an exact a science as we would probably like.

The other thing we focus on is what a speaker "looks like" as a load. Now not every load box does this - many are purely resistive. And there really isn't anything wrong with that. These circuits are designed assuming a resistive load. But if we are replacing a speaker load, it makes sense to try and emulate one as a start.

So here is a typical speaker as a load. You have a peak at the mechanical resonance, and then rising impedance with F. Where is it really the rated impedance? Probably somewhere around 400Hz. Now a lot of focus gets placed on finding the right RC combos to yield these curves and that is all well and good but we're far from done if we really care about emulating a speaker: all we have done is mimic the coil in the circuit on the assumption that it will make the amp circuit - which was designed assuming a resistive load - happy. (confused yet?) That bit looks a lot like this (simplified). Mechanical part is the resonance, electrical part is inductive reactance in the voice coil.

But at this point we have done nothing about the speaker's frequency response, dynamic response, the distortion products and reverse EMF. And there is still more to get to after that.

Recall this is what our Celestion Blue frequency plot looks like: (its on the speaker basics page too)

That response above is just the speaker all by itself; no amp, no cabinet. A perfect speaker would have a mass-less cone that moves like a perfect piston and have a flat response. The reality is vastly different and the cone has all kinds of challenges trying to reproduce what the amp is sending it - and the amp is having challenges controlling the speaker as well. We are trying to wiggle a 12" paper cone thousands of times a second after all with the air pushing back against it the whole time. Where is all this in the speaker "model"? Well it really isn't there at all. Net net, in a traditional load box you are doing some kind of post processing, EQ & compression to get back towards what real speakers in a cabinet would do. We have tried to make the amp happy, but we have not yet addressed our ears.

Fundamental Issues with Loads:

It's not a speaker (duh) and emulating the electrical "look" of a speaker is a fraction of the job (we're missing reverse EMF, and we're missing the frequency response and dynamic response of a speaker pushing against the air, in a cabinet, in a room) We're also missing any distortion products the speaker is adding.
Fletcher-Munson curves * will explain why your amp sounds midrangy and anemic at low volume (how human's hear)
Reduced or no sound field. The sound field our guitar would be in is not there. This impacts sustain, timbre and feedback.
They have a bad rep, I think largely undeserved though. The question always comes up around "they will hurt your amp". Which is a bit misleading, I think. Survival can hinge on having the correct load on the amp (which again, is assumed to be resistive during design) But the largest issue after that (to me) is that since you are not getting pounded with sound, there is a tendency to beat the amps mercilessly like a distortion pedal. That has consequences. You still have 450V on the output tubes, and they are still making 50W or whatever. Like running your car at redline, you are asking it to live on the edge. You should expect there is more maintenance required, more care taken around bias settings, tube aging, fuse values and so on. If you play clean, you may never have an issue. If you are a crazed rocker, well it could get expensive if you do not like regular maintenance. Pay now or pay later, but you're going to pay. I think it was racing star Jackie Ickx who was once quoted as saying. "Run hard, maintain harder". Words to live by.

What is "wrong" with Master Volumes?

I thought you'd never ask. The most basic assumption they are making is that the sounds and responses you really want are all happening in the pre-amp section. There is a fair amount of evidence that this is not the case. For starters, all the amps with tube output sections that are still being made & sold today, over 100 years after the invention of tubes and 76 years after the invention of the transistor. It is not that people have not tried. Most of what makes a tube amp expensive is the high voltage power supply and the big heavy transformers which are required for the power & output sections. People have tried to eliminate these for a very long time, and consistently failed. In a number of cases across the years, amp makers wound up with solid state pre-amps and tube power amps, probably going back to the mid 1960's and the Vox 7xx series. Not sure if anyone else big was doing this so early? (anyone?) And financially it's upside down, although it is a good way to pack more features in.

You are also affected by items 2 & 3 from the list above. The Fletcher-Munson curves describe how you hear more midrange as sounds get quieter. It's why older hifis had a "loudness" switch - to boost lows and highs while listening at low volumes. F&M performed their study was back in the 1930's and it has been superceded by several others since, but the results are along the same lines. The Audio Engineering Society (AES) uses the ISO curves for loudness these days. The tricky thing about them is that it's really not just an EQ thing you can fix. It is dynamic EQ. Your frequency response - literally how you hear - changes with level. Just changing the EQ at all volumes is not an accurate fix.

You also don't have that big sound field to play your guitar in, and it really hurts your sustain. If you go back to your classic rock albums, your Aerosmith, Led Zeppelin, The Who, even AC/DC, there really isn't a ton of gain; the sustain and feedback are from being in the room with a cranked amp, and from the amp vibrating as well.

The net result of losing all the good sustain and gain/distortion from the output stage (tubes, transformer and speakers!) is that we have built more and more gain into the pre-amp to try and make up some of that loss. This has squashed the dynamics and largely erased the sonic differences between instruments. We can chug & djent, but it's kind of generic (I know, my age is showing!) and a bit of a dead end street. How much gain can you add before it's really just a fuzz? How do you sell guitars when they all sound the same? Why are old 4-hole Marshalls and 5F6-A Bassman and amps of their low gain, no reverb, no loops, no master volume ilk still selling for 5 figures in 2023? Why are top players still often touring with 60 & 70 year old amps? Or maybe I am just dating myself. IDK. But something to think about at least.

To be fair, there are modern amps that are designed from the ground up with master volume circuits, and some are quite excellent and subtle, and some terrific players base their sounds around them - which of course is fine. Someone will no doubt flame me about $100K Dumble amps. Hey - whatever floats your goat! But there are still a tragic number of amps that have master volumes which are just "bolted on" and don't perform in an additive way. I am looking at you, 70's Fender combos.

A Q&D Evolution of power soaks

In the beginning, for guitar at least, there was the Altair PW-5; it was purely resistive. It had an in-out switch, a pad and an in and out for the speaker cables with a big attenuator knob on the front. It featured ventilated chassis to help get rid of the heat it build up and it used tungsten resistive elements, possibly to encourage heat dissipation? Not really sure. Bottom line for all of these devices is that power that isn't going to your speaker has to go somewhere. Like brakes on a car, we are turning the energy into heat.

All in all, the Altair was a nice start, it looks a lot like some contemporary units but is over 40 years old.

The Sholz Power Soak, (pictured at top) invented by Boston front-man and MIT engineer gone wild Tom Sholz was also resistive. It was able to switch from 4 to 8 to 16 ohms cleverly and used banks of cement resistors in a ventilated chassis to radiate the heat away. Patent & schematic below. I used this unit to torture test my Marshall (pictured below) while developing the OEI 50W output transformer. I also use it on the test bench a lot to keep from going deaf or crazy. It takes a licking and keeps ticking! Possibly my least favorite sound-wise though.

The next biggie was less popular from a sales volume point of view but it would influence many subsequent devices. Aspen Pittman's Groove Tubes developed and patented their own speaker emulator. This patent was granted in 1990, so based on the 17 year life of these things (afaik) it should have opened up in 2007 to other groups using parts of this concept. There is a link to the patent filing below. Essentially they modeled the impedance curve of a Celestion Blue, and then added 5th order (5 pole in the Patent) Butterworth filters to EQ the resultant signal. Look like a speaker electrically (check), sound like a speaker (check).

The most basic thing to know about filters (high pass and low pass) is that a single RC (resistor + capacitor) combo is called a first order filter and it has an attenuation slope of 6dB per octave after its "knee" (the 3dB down point or half power point. It is also the frequency where we are 45 degrees out of phase). Add another pole and you roll off at 12dB per octave, and another for 18dB per octave (you can see one of these at the output of the Trem/Vib channel on the AC30 - 3 RC poles in a row to keep the LFO artifacts from leaking into the output stage) Butterworth filters are one of the best filter designs, and at 4th order we're looking at 24dB per octave rolloffs. If you look at the frequency response from about 4k to 8k on the Celestion diagram up above, it rolls off very rapidly - and pretty close to 24dB/8av. How about that.

The Trainwreck Airbrake - designed by pioneer/guru/legend Ken Fisher, the Airbrake is a nice simple circuit that has some devoted fans. Dr Z has issued their own version in recent years, and Kendrick had a version out prior to that. It uses a tapped 100W wire-wound resistor as its load, and features a high pass filter on the attenuation control. This is one schematic that has been floating around for many years (>10) I have not built it or checked it but given its age and ease of location I have included it here.

The Marshall Power Brake (pictured at top) licensed parts of the GT Patent - which to be fair I think was pretty broad and based on prior art - the Altair and the Sholz. Additionally, Leslie had already used resistive loads to connect to Hammond console organs such as the M3 for many years (since about 1963). Marshall cleverly used a tapped inductor (an autotransformer) as the attenuator and used the excess power to drive a fan to cool off the soak circuit. #nice It still sounds pretty good today. It works with 8 or 16 ohm loads. I had to re-flow the solder in mine and replace an output jack, but otherwise it is holding up just fine.

The THD Hot Plate (pictured at top) adds a few tweaks, high and low filters and a compressor that uses a bulb. #nice Full disclosure, I have used a THD live and in the studio for about 15 years now. They work really well, and the first few clicks sound and respond fine. You'll never pick it out on a recording. (there is one linked at the bottom of this page) So yes, I am a fanboy. It's also built like a tank. No device is going to work well when it is all soak and no speaker. THD also did a nice job making the chassis the heat sink which helped to keep the size down. Small, bulletproof, works. What else do you want?

The Weber MASS was a slightly different design, which I consulted on. We used a real speaker motor (since we were making speakers anyhow!) rather than just simulate the coil element. Once again, getting rid of heat was a challenge. If you take a peek at the wire, you will see why. They work well but need some post processing, as most do, especially at higher attenuation levels.

There are many more and this is really not intended to be an all-inclusive list (I'm focusing primarily on units I have used), but "old style" tend to be built like the above pioneers, and the "new" style are using digital audio tech to make up for the fundamental shortfalls in the hardware world.

Which brings us to "next generation" devices like the Two Notes Torpedo Captor X and the Universal Audio Ox, (The Ox is pictured at the top, and below) both of which use speaker, microphone & room models to add back much of what we're losing with a soak. The Ox is really more of a recording interface IMO although it can certainly be used live. Some features are even foot switchable. For my own use it's a bit expensive and complex to use at a gig. I do use it a lot though; my home recording setup is UA based so I just take the SPDIF out from the Ox into my Apollo interface. I always have a speaker connected, and use varying amount of it depending on the time of day, what I am going for, etc. If it's late at night, I will fully soak it so I can stay married. If I can make noise, I can nearly always get a better sound by mixing in some real amp + real speaker. When I have a serious project to work on, I will still opt for guitar-->cable-->amp for nearly everything when I have the choice. It's just what I am used to, and works well for my style of playing. But the Ox does a terrific job and is really super handy to have around, and there are lots of sounds it can get that I cannot get any other way. When I recorded my parts for JSR's 2021 release, "Now Is Not A Good Time", I was able to do some of it live and loud, and the rest was at home, in the evening after work so it had to be quiet. It was great to be able to do both without much compromise.

The main thing you are still missing is the sound-field, and I can't help thinking about a sustainiac-like device that would vibrate the guitar using a fed-back signal. As an obvious barrier is the fact it might need a different cable to plug it in and it might need a different amp, or you would need onboard batteries to do something contained. Either way, it's a stretch for standard guitar players to adopt with their existing guitars and amps. Which is a pretty big barrier. Would you attach this to your guitar? https://www.parts-express.com/Dayton-Audio-DAEX19CT-4-Coin-Type-19mm-Exciter-5W-4-Ohm-295-218?

One really cool thing about these next generation units is the ability to simulate all kinds of speakers and cabinets with any amp, from an 8" champ speaker to a 4x12 Marshall cabinet. Running my old 50W Marshall into an 8" champ speaker model recently was a quick trip to the "houses of the holy"as an example. There are also a number of handy effects built in, compression, EQ, echo, chorus & reverb as well as a number of microphone models, panning and so on. To some extent I think it is probably more important for these kinds of devices to have unique things they do well rather than simulations of things that the original devices will do better. Just a general opinion. Devices that contribute something unique, last. Devices that emulate something else do not last, in my experience. Two notes has really run with the speaker emulation piece, and is even able to load IRs (impulse responses) from users in addition to all of their DynIR cabinet simulations. You can even run their cab/room/mic simulations ("Torpedo Wall Of Sound") as a plugin in your DAW so you can run anything into them. Here it is in Logic Pro X: #Kewl

Sound samples:

UA Ox sound samples:

All 3 guitars here are the same guitar, same amp, same settings. Silent mode, late at night. LP & that 50W head (pictured above) using different cabinets to carve out space for things. (1:45 is where the nasally outro solo starts) https://soundcloud.com/dr-banzai-1/nye

And this is an SG-->Marshall->Ox on the lead. (break at 2:20 point you can hear it pretty clearly) From 2021's "Now is not a great time" by John Sally Ride. Also silent mode. https://thejohnsallyride.bandcamp.com/track/my-persistence-vs-your-resistance

THD Hot Plate sound sample

Recorded live, but not loud, in my bedroom closet. I am living the meme life.

Elec Guitar is the LP+AC30 (before I rebuilt the pre-amp) The mic was an SM57 https://thejohnsallyride.bandcamp.com/track/the-nicest-things