Q1 is the common current source - so I do not understand your question about its output.
The story behind this circuit is quite old.
It started in the sixties with building myriads of fuzz-boxes using transistors and diodes. The discussion about a smooth transition from linear to distorted was already present, and I recognized by learning semiconductor basics that diode fuzz transition is inherently abrupt.
About 1972 I put my hands on a RCA databook of integrated circuits. During the dark ages of Pregooglium information was sparse so I read this databook from cover to cover.
Among others I discovered the new CA-3080, an "operational transconductance amplifier" (OTA) and somewhere its transfer curve - a smooth, perfect S-curve! In other words the the long sought choice to build a distortion unit.
From that moment on CA3080 was the way to go. Besides being a bit noisy this chip never disappointed me.
In the eighties and nineties I lost interest in guitar amps. The amp worked satisfactorily so there was no need to change anything.
About 2010 I re-started guitar amp engineering aiming at small battery powered solutions based on class-d-amps.
But, alas - CA3080 was obsolete meanwhile.
So I had to simulate CA3080 with parts available now -
and the simplest approach was the long tail differential amp.
End of the story
The story behind this circuit is quite old.
It started in the sixties with building myriads of fuzz-boxes using transistors and diodes. The discussion about a smooth transition from linear to distorted was already present, and I recognized by learning semiconductor basics that diode fuzz transition is inherently abrupt.
About 1972 I put my hands on a RCA databook of integrated circuits. During the dark ages of Pregooglium information was sparse so I read this databook from cover to cover.
Among others I discovered the new CA-3080, an "operational transconductance amplifier" (OTA) and somewhere its transfer curve - a smooth, perfect S-curve! In other words the the long sought choice to build a distortion unit.
From that moment on CA3080 was the way to go. Besides being a bit noisy this chip never disappointed me.
In the eighties and nineties I lost interest in guitar amps. The amp worked satisfactorily so there was no need to change anything.
About 2010 I re-started guitar amp engineering aiming at small battery powered solutions based on class-d-amps.
But, alas - CA3080 was obsolete meanwhile.
So I had to simulate CA3080 with parts available now -
and the simplest approach was the long tail differential amp.
End of the story
More or less the same here. Keep them out of clipping, and solid-state Hi-Fi amps reached audible perfection decades ago (the remaining flaws are too small to hear). Now you can buy audibly perfect stereo receivers at the local thrift store for under $50. Who cares if they are a couple of decades old and don't have HDMI inputs?
So, for me, all the interest went out of Hi-Fi audio electronics long ago. It's a done deal. Been there, done that, nothing new to see here, time to move along.
Guitar amps, on the other hand, are still very much an open challenge, unless we are satisfied with 50 year old designs, with all their pluses and minuses.
-Gnobuddy
We share a lot of similar goals. The immediate purpose of my "de-nastifying filter" was to get one specific pair of boombox speakers to sound acceptable for guitar, but the bigger purpose was to have an off-the-shelf solution that would let me plug a guitar signal into any nominally flat-response, wide-range loudspeaker: boom box speaker, P.A. system, home Hi-Fi speaker.
In 2014 I found out by accident that 10 microwatts of 1 kHz sine wave audio driving an old Jensen 12" speaker was clearly audible a few feet away.
I thought I'd made a mistake calculating power, until I realized that 10 uW is (-50 dBw), and if you drive a speaker with (+95 dBw) sensitivity with (-50 dBw) of power, you get an SPL of +45 dB@1 metre.
Forty-five decibels is somewhere between "babbling brook" and "light traffic" ( Noise Level Chart: dB Levels of Common Sounds ). Quiet, yes, but most certainly clearly audible.
Finding out that microwatts of power was enough to make audible sounds in a guitar speaker triggered me to do a few more similar calculations. I picked 70 dB SPL (typical vacuum cleaner) as a benchmark for the loudest guitar playing likely to be acceptable in an apartment. Even with extremely inefficient bookshelf speakers (82 dB/W @1m), it would only take 63 milliwatts (!) of audio power to get there. With a more typical guitar speaker (92dB @1W @1m) guitar, it would take ten times less power, only 6.3 milliwatts, or 0.0063 watts!
A typical 6-watt Champ, then, has a hundred times (20 dB) too much power for overdriven apartment use - if you're driving an insensitive bookshelf speaker. If you're driving a guitar speaker which is already 10 dB louder (92 dBw sensitivity), then your Champ is now approximately one thousand times, or 30 dB, too loud for typical apartment use!
Some preliminary experiments since then have supported the idea that, in a typical apartment, using a guitar speaker, well under 100 mW is enough to get unpleasantly loud for heavily overdriven guitar waveforms.
Similar experiments showed that clean tones are subjectively quieter, and have more dynamic range, so somewhere between half a watt and one watt of power is nice to have.
After some tinkering, I found I got better sound quality if I designed all the distortion (both clean and overdrive) into the preamp, and then fed that to a more or less clean power amp (in my case, a 2-watt push-pull 6AK6 design.) That seemed to work better for me than trying to make a good-sounding 100 mW output stage.
Of course, your ears and brain may not agree with mine. There is no doubt that the emotional impact of 70 dB SPL is very different than, say, the emotional impact of a live Kiss concert at 136 dB ( Loudest band - Wikipedia ). There is also no doubt that we hear differently at lower SPL (i.e. Fletcher-Munson equal loudness contours.)
It certainly seems that way - there isn't a whole lot of room in between "too clean" and "clipping". I can't fully bypass the source of the input JFET, or the guitar signal will drive the input stage all the way into clipping. If I don't bypass the input JFET source, then the amount of second-harmonic distortion becomes less than I'd like to have.
Same here, at least these days. When I was young and energetic I designed my own PCBs with the most primitive of tools (graph paper, pencil, eraser, a sharp nail, an old ball-point pen with the ball knocked out of the tip and filled with oil-based household paint for resist).I'm a "through hole perfboard" kind of guy.
Now it takes me a month to find the time and energy to build a pine guitar cab. I would never get any solid-state project completed if I was also having to design and fab my own PCBs. So perfboard and protoboard it is.
SMD? I need close-up spectacles from the dollar store to work with 0.1" pitch through-hole components now. It's hard to imagine wanting to take on the additional frustration of working with much tinier SMD stuff!
It's entirely up to you, but you are most welcome to post your experiments here, if you wish. It would be nice if this thread ended up being a repository of good ideas about how to make "SS guitar amps that don't sound nasty"!
If the negative supply rail voltage is large enough, it swamps out manufacturing spreads in FET Vgs, and one no longer has to tweak resistor values for individual JFETs.
As an example, a hypothetical JFET might have a Vgs that varies from 1 volt to 3 volts to achieve the same Id. This is a 300% variation between samples, so if you use a simple source resistor, it has to be tweaked for each individual JFET, or the circuit won't be anywhere near the designed operating point.
But if you use, say, a -20V supply rail, that same Vgs variation from 1V to 3V translates to a voltage across the source resistor that varies from 21V to 23V. The biggest value is only 9.5% bigger than the smallest, so Id is now guaranteed to stay within 10% of the design value. No adjustment needed.
I'm not yet sure I understand KMG's circuit fully, in particular, I don't know exactly what his criteria was when tweaking the operating point for every stage. But if I/we can figure that out, we can make life a lot simpler by using a higher voltage negative supply rail.
Keep in mind: the Super Champ XD has very nice valvey cleans, but the clean channel has no overdrive at all. (There is even a signal attenuator at the input of the valve power amp section, which seems to ensure the opamps will clip before the valves go into overdrive.)
There is a second "drive" channel, but all overdrive there is simulated with digital signal processing in the preamp. I found it less nasty than clipping diodes, etc, but not very good sounding. I did find it was usable buried in a mix with slatherings of reverb and delay on it to hide the worst warts.
So I think the SCXD shows that you don't need to emulate preamp grid current flow to get good clean tones. But I don't know if the same is true for good overdriven tones. (I suspect it isn't.)
There is now a well-established and honourable tradition of using 70-volt audio line transformers as DIY valve amp output transformers. It seems to have started in Australia:
1) http://home.alphalink.com.au/~cambie/6AN8amp/M1115.htm
2) Cheap Output Transformers
Australia is lucky enough to have 100-volt audio line transformers, which have deeper step-down ratios for the same power tap. (And unlucky in that "real" valve output transformers are prohibitively expensive there.)
However 70V audio line matching transformers available in the USA are also quite usable (depending on the primary impedance you need), and very inexpensive for low-power amps. For example, here is a ten-watt OPT for $5 (!): 70V 10W Speaker Line Matching Transformer
The big issue with these cheap transformers is primary inductance, which may not be enough to support full (bass) guitar bandwidth at high impedance settings. Solution, design your amp to need a lower primary impedance!
Personally, I have my doubts about whether a transformer in the signal path is actually needed to get the sound we want, however, it is always nice to have independent data from someone else to either support or refute my opinion.
Some of the reasons why I doubt we need a transformer:
(a) There are a few examples of good-sounding output transformerless (!) valve guitar amplifiers.
(b) There is some evidence that any distortion generated by the OPT is buried deep under the distortion generated by the valves, i.e., is probably too small to make much of an audible contribution.
(c) I have a home-brew preamp with a small signal beam tetrode as the output stage, running single-ended into a resistive anode load as usual. Feeding the signal from this into a clean solid-state power amp sounds pretty "valvey".
While I'm skeptical that we need an actual transformer, I do think we may need to include the transformers frequency response into the speaker emulation filter - particularly if the OPT is limiting bass bandwidth.
It's less likely that the OPT is limiting treble bandwidth on a guitar amp (since 5 kHz seems ample!), but if it is, we need to emulate that too.
I also went looking for through-hole Schottky diodes, and ended up buying some of these:
https://www.digikey.ca/products/en?keywords=SD103BGICT-ND
I have the same intention as you, i.e. to start out copying KMG's work. However my current project is taking far too long. It is very likely you will start your KMG-related experiments before I start mine, and once again, you are most welcome to post your findings and observations here, if you like.
-Gnobuddy
We all know that BJTs have an almost perfectly accurate exponential characteristic. When wired up into a long-tailed-pair, the exponential characteristics of the two devices combine to create a complex tangent (tanh) function. Which is, indeed, a smooth, S-shaped curve: ( Hyperbolic Tangent -- from Wolfram MathWorld )
This characteristic of the long-tailed pair has been used for decades in audio function generators, to round off the tips of triangle waves, turning them into a crude approximation of a sine wave.
IIRC JFET differential stages (also a smooth S-curve, but not a tanh() function) are used in Boss's BD2 and OD3, maybe in others too.
I'm not sure why the BJT long-tailed-pair hasn't been used in more guitar distortion pedals. Possibly just ignorance on the part of most FX pedal designers. Also possibly because the setting of the input signal level is fairly critical, and smaller than the output signal of a typical instrument-level device.
-Gnobuddy
Q1 is the common current source - so I do not understand your question about its output.
The story behind this circuit is quite old.
It started in the sixties with building myriads of fuzz-boxes using transistors and diodes. The discussion about a smooth transition from linear to distorted was already present, and I recognized by learning semiconductor basics that diode fuzz transition is inherently abrupt.
About 1972 I put my hands on a RCA databook of integrated circuits. During the dark ages of Pregooglium information was sparse so I read this databook from cover to cover.
Among others I discovered the new CA-3080, an "operational transconductance amplifier" (OTA) and somewhere its transfer curve - a smooth, perfect S-curve! In other words the the long sought choice to build a distortion unit.
From that moment on CA3080 was the way to go. Besides being a bit noisy this chip never disappointed me.
In the eighties and nineties I lost interest in guitar amps. The amp worked satisfactorily so there was no need to change anything.
About 2010 I re-started guitar amp engineering aiming at small battery powered solutions based on class-d-amps.
But, alas - CA3080 was obsolete meanwhile.
So I had to simulate CA3080 with parts available now -
and the simplest approach was the long tail differential amp.
End of the story
It is just the 'coil' point that I was wondering if it went somewhere or was just equivalent to a test point, I barely scratched the surface of spice simulations way back when. I may have to jump in the virtual world on this one.
The explanation makes sense on reproducing the section of the IC circuit. You must admit it is not the first thing that comes to mind when you (well others) get to designing a limiter or clipping unit. I could not see any lineage from pedals people come up with. I am guilty of poring over data handbooks when I was younger. It beat going to classes in school.
Agreed, I always considered this attempt a trivia and wondered why it was widely ignored.
ah, now I see, the "coil" point is in reality "col1", i.e. the collector of Q1 - just a bias measuring point.
Last edited:
Certainly driving the differential input is a bit more complex than driving a diode-limiter.
But it is no rocket science.
A couple of years ago, I came to the same conclusion. I found Simon Bramble's well thought out LTSpice tutorials extremely helpful: LTspice Tutorial | The Complete Course
For most basic circuits, I got pretty good a long time ago at estimating values in my head or with a calculator. But for more complicated things (such as the "de-nastifying filter" described earlier in this thread, which needed to incorporate bass boost, an 800 Hz notch, a treble peak, and a steep high treble roll-off), I've found LTSpice invaluable. It's allowed me to design things that I could not have designed without it.
-Gnobuddy
Certainly driving the differential input is a bit more complex than driving a diode-limiter.
But it is no rocket science.
I did do some rocket science for half a year. I was suppose to come up with a filter to protect a sqib (ignitor) from accidentally igniting the rocket. The biggest problem was on shipboard use as you have meda-kilowatt radar transmitters blasting the deck. The people working on the deck wear mesh metal clothing to keep the rf from cooking them. At times when they load the rockets on the aircraft a huge arc is generated between the rocket and aircraft, difference in potential when everything is acting as an antenna. Also suppose to protect the squib from EMP, ESD, and lightning strikes. Not a direct strike but when lightning hits the ground or aircraft the current and its magnetic field induces voltage potential that can fry a little 1/4 W squib.
I ended up specifying a feed through pi filter with ferrites, nothing too exotic just the pi filter and only for military applications. The big problem was the arcing between the person loading the rocket in the rf environment. I theoretically solved that problem using a 90V spark arrestor. Normally it acts as an open until the voltage across it ionizes the atmosphere in the device. But looking at the curves they are not an on-off device they have a semi-conduction area where they have a intermediate resistance. And the 90V one had just the right characteristics to eat up the rf without clamping right down.
Anyway long story short, it was pretty cool sitting in on engineering meetings with chemist and mechanical types discussing the strength of the rocket body, the exponent of the propellant being developed. I was their sparky as they said. The exponent is important, it controls the burn rate. The higher the pressure the propellant is under the faster it burns. You want it to burn just below the strength of your pressure vessel, the rocket tube. If the exponent is too high the pressure builds up too fast and... ...BOOM!
I remember being at my desk and two of the mechanical people came in all excited, 'we blew up a rocket, we blew up a rocket'. Yeah, all the production and testing buildings are separate by a good distance and have berms surrounding them. If one blows up is best that the others don't go boom either.
We did come up with a nifty idea to test the propellant, not sure if they went on with it after I left. The idea was to use a diving cylinder and place four strips of the propellant in it. There would be an igniter wire on one side of the strip and a break wire at the other end which you would test for continuity. Then it was just measure the difference between the igniter time and the break time. The good thing about the tank was you could place the goodies inside, pressurize the tank to 2000lbs, fire off the one, bleed off the tank till it is at 1500lbs, do the next test, then again at 1000 then 500. Once I found a feedthrough with the right number of wires that could take the pressure and fit in the hole all that was left was getting a frequency generator/counter.
It was fun working there, had my NATO Secret clearance, they asked if I wanted the certificate, I didn't take it. Would be neet to have it now though.
My rocket science experience.
You need to get your hands and ears on a Marshall 18W, or a reputable clone of one. Sparkling, chimey clean tones and creamy, harmonically rich overdrive. One of those most gorgeous sounding and satisfying amps I've ever played. And with a change of tail resistor value (as well as the socket change), you can run 6V6s in an 18W as well.
The spec sheets actually show 14W for 6V6s. 12W is for 6BW6s (the 9-pin version of a 6V6) and EL84s. So you're safer than you thought.
So far, the smarter, lightweight solutions I've tried (both analog and digital modelling) haven't seemed very impressive. Even when they provide a reasonable facsimile of real tube amp tones, they aren't very touch sensitive, and don't respond to how the guitarist plays in the same way. A real tube amp feels like a living thing, and so far that's completely missing to me in SS.
Last edited:
Chicken cacciatore, beef lasagna, or pork vindaloo? Quite likely your favourite of those three won't be the same as mine.
It's the same thing with guitar amps. There's no particular reason to expect your favourite amp sounds will be the same as mine.
My experience has been that what most people call EL84 "chime" usually sounds like "grit" to me, harsh rather than pleasant. Hearing that sound produces an unpleasant visceral feeling in my mouth, very much like the feel of sand between my teeth ( I wonder if I have a touch of synesthesia. )
mjd_tech mentioned that EL84 amps can sound mellow. I agree; I once tested out a Laney Lionheart ( Amazon.com: Laney L5T-112 Lionheart 5 Watt Class A Guitar Tube 1x12 Combo: Musical Instruments) in a friendly music store after reading many rave reviews.
The Lionheart is a pretty expensive (north of $1000 USD) 5-watt SE class A amp with various "boutique" touches (gold contacts, etc.) It was mellow, all right; it sounded like a guitar amp with several pillows piled on top of the speaker, muffled and unattractive. It didn't do anything for me, musically speaking.
So far I've never enjoyed an EL84 amp. Perhaps, as you suggest, that's because I've never heard the right one. Or perhaps it's just the way my particular ear/brain perceives sound, and I really don't like the harmonic spectrum of these particular pentodes.
To be clear, I don't take my personal perception of EL84 guitar amp sounds as evidence that everyone else is wrong, any more than I would think someone else was "wrong" for preferring, say, the chicken cacciatore.
But for me, a 6V6 in the simplest of single-ended circuits immediately produces sounds that I prefer to even boutique EL84 designs.
I had the same experience with a 6AG5 in a preamp. I immediately loved its sound. Guess what, it is also a beam power tube, albeit a low-power one. The datasheet says "pentode", but inspection with a magnifying glass shows only enough support rods for two grids.
Why would my ears prefer beam tetrodes to true pentodes? I don't know for sure, but I have a clue. Somewhere in the 45 pages of O. H. Schade's white paper on the design of the first practical beam power tetrode (6L6, http://www.clarisonus.com/Archives/TubeTheory/Schade 1938 Beam Power Tubes.pdf ), there is some discussion about beam power tubes typically generating considerably more second-harmonic distortion, but less third harmonic distortion, than true (3 grid) pentodes.
In Schade's world (lower distortion is always better to an audio engineer), that is a bad thing, so he points out that even harmonics cancel in a push-pull output stage, so that a pair of 6L6 in push-pull can end up producing lower distortion than true pentodes of the same era.
For us in electric-guitar-amp world, things are different. Large amounts of distortion, enough to be distinctly audible, are good. To my ears (and oscilloscope), up to 35% of mostly 2nd harmonic distortion produces singing, shimmery cleans. Turn down the 2nd harmonic, turn up the 3rd harmonic, and (again, for me) the sound changes to a more aggressive growling tone.
And it turns out that true pentodes (like the EL84) are naturally inclined to produce that growl. Beam power tubes are naturally inclined to produce those shimmering cleans. I happen to prefer the latter. You? Cacciatore, or lasagna? Or maybe vindaloo?
In the mental stack of projects I hope to build some day, something along these lines has been waiting patiently. Most likely it will look a little different, with a pair of either 6LY8 or 6HZ8 in the power amp.
The 6LY8 contains half a 12AX7, as well as a little beam tetrode. The 6HZ8 contains a somewhat lower ra triode, and a bigger beam tetrode. Either type should make for a fun little 2-bottle "18 watt" amp, with a lower and therefore more usable amount of output power. (Quite literally, I have no use for the usual ear-bleedingly loud amps.)
I agree, and have had much the same experience.
And yet...there is real butter in the supermarket, as well as many different kinds of fake butter. Real butter tastes better to me, no contest.
But I buy fake butter, for two very good reasons: my wife has a dairy allergy, and my cholesterol levels are higher than they should be. After trying a lot of quite disgusting fake butter products, we found one or two that taste reasonably good. (Though not as good as real butter.)
It's the same thing with guitar amps. For example, the amp project that started this thread had specific requirements: it had to be small and light, because it's going to be gifted to a senior citizen with a physical disability. It had to be loud enough to use at our weekly all-acoustic jams (no drummer). It also had to be cheap, because that's all I can afford.
I cannot build an all-valve amp within these constraints. I thought about it, but found I also couldn't build a hybrid valve/SS amp within the time and price constraints. It had to be all solid-state, or nothing.
So it's time for fake butter. It may not taste as good as real butter, but it's still better than having no butter at all.
KMG's fake butter seems to taste pretty good. I'll have to get some and try it out in person.
Will there ever be fake butter that is 100% as good as real butter? Who knows. Maybe, maybe not. "Ever" is a long time!
-Gnobuddy
The spec sheet in my computer says 12 watts. It is however from 1955. It is for the 6V6GT, not the 6V6GTA and uses the "design center" rating system. I just downloaded the 6V6GTA sheet, which does say 14 watts, but uses the "design maximum" rating system. There is 10 to 20% difference between the ratings depending on the tube's intended application.
I tinkered with both of those back in the days of the HBAC. Those little pentodes have a lot of Gm and like to make good transmitters.
The little $50 amp I designed for the HBAC makes 3 clean watts (3% distortion) and 4 watts at the edge of clipping. I can only dime all 3 knobs when no one else is home (91 dB non guitar speakers).
Counting the volume pot in the guitar, there are 3 gain / volume controls and plenty of available gain, it is possible to get all the distortion from the preamp, power amp, or both. If the knobs are set so the power amp just starts to round the tops first, then the preamp starts to distort, it is possible to go from pretty clean to quite fuzzy just by how hard you hit the strings.
The topic is about tube emulation and EQ, but with instrument amplifiers those things don't exist in isolation. Speakers and cabinets can affect sound sometimes in ways that seem to surpass small differences in amplifier distortion. In particular, many open back combo amps sound very boxy, and not in a good way.
So, my question would be what speakers and enclosures do you guys use to evaluate and compare amplifiers?
So, my question would be what speakers and enclosures do you guys use to evaluate and compare amplifiers?
You made a point, this certainly is an important question.
In my sound-snippet above I used the 12" Celestion Century Neo in a half open baltic birchtree combo cabinet, all in all a standard build.
For my 10l practice amp I prefer a 6.5" Jensen.
The thread was split off and named by diyAudio moderators. I would say the real topic is closer to "How can we build a SS guitar amp that doesn't sound nasty?"
I just found out from one of the moderators that I can actually edit and rename the thread. But now that the thread is doing well, and we are getting lots of interesting contributions from lots of people, a name change might be a bad move. I think it's best to leave it alone.
I agree with you. In fact, this project has been delayed by over a month because I could not make the original intended cabinet (an old Sears speaker enclosure from a thrift store) sound good enough, and had to start over and build a cab from scratch.
The original cab had the same depth and width (top and bottom were nearly square at about 8"x10"). The length was close to twice the width (19"). Simple whole ratios (1:1 and 2:1) are very bad choices for a speaker cab. No wonder it sounded unpleasant.
Internally, the new pine cab I built comes close to the Golden Ratio (~ 1.6:1) for both length:width and width:depth. It sounds quite a bit better than the old 1:1:2 Sears cab to me, and this is with exactly the same two 6.5" drivers mounted in it.
The boxiest commercial guitar amp I've ever played through was a little AC4-TV. No surprise - the sealed cab is nearly a cube. I found the boxiness quite unpleasant.
The second-boxiest guitar amp I've ever heard was a Fender Blues Junior. The front panel is nearly square, and the cab is deep enough to be approaching a cubical shape. The boxy sound is one of the reasons why I never really liked this amp.
I don't know of an electronic way to fix a truly bad cab. I think the only solution is to start with a better-designed cab.
If speaker emulation is built into the amp, as in my current project, it should sound reasonably good into any more or less flat-response speaker.
I tried a couple of different thrift-store boombox speakers, the stock Eminence and cab of my Princeton Reverb reissue, a 2x8 cab I built a while ago, and most recently, the new cab I'm building for this project. While there were audible differences (the PRRI cab had the most bass, for example), they could be largely dialed out with normal tone controls, and I didn't find one vastly superior to the other.
I think this just shows that guitar is an instrument very tolerant of speaker imperfections. Any speaker good enough for even mid-fi music listening (i.e. boombox, etc) is likely more than good enough for guitar.
Most of my projects have involved inexpensive speakers that were not sold explicitly for guitar.
-Gnobuddy
That sounds like a healthy, OSHA-approved workplace.
Cool story!
-Gnobuddy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.