Introduction: Rod Type Tube Guitar Pre-Amplifier: Marshland 800
Hi all! Some time ago, I made an op-amp based guitar preamp loosely following the Marshall JCM900 4100 & 4500 models. Recently followed a "true" JCM800 preamp from reused tubes. Now I finally get to present you another JCM800 style circuit but this time with a crazy peculiarity:
In this build, the circuit is adapted for 1Ж24Б (1J24B) soviet radio tubes that are still abundant and rather inexpensive, have battery-friendly filament voltage and current draw and can be used with low current 100V anode supply that poses less of a safety concern than the >300V supply needed by the original 12AX7 or ECC83 tubes.
You want to try playing this pre-amp but not build it? Guess what - you can do that with the neural amp models I trained on it. Scroll down to Step 15: Neural Amp Models - Try Before You Build to learn how.
Before glorifying soviet products in the face of today's situation, I do want to, but also feel obliged to express my sincere sympathy for people affected by the war in Ukraine. Nobody deserves to die or lose their relatives and friends for reasons of unnecessary aggression for the "gain" of a few maniacs. I am from Austria, the country that brought forth one of the world's worst dictators of all times and I feel a painful shame seeing history repeat itself with such horrible risks being taken and made gruesome realities. I'm further ashamed that my country practices this "we don't care as long as we can look away" politics. I hope this paragraph does not take it too far in terms of political coloring, but I believe in this situation a clear statement of distancing from both, glorification and carelessness of what regards war, violence and oppression is suitable and necessary. Especially as this build directly involves probably war driven technology. Let us hope for those Slavic brothers despite all odds to be soon shaking hands and celebrating both their similarities and diversities.
Next there is a list of credits I want and need to give, as each of those were vital for this project to happen:
- Rob Robinette's guitar amplifier knowledge site, especially How the Marshall JCM800 Works. I will pass on to you a lot of insights and conceptual understanding I gained on his pages, spiced with some flavor of my own experiences, hoping it will add some value. If I may copy some of his statements more or less directly, then because I could not find more suitable words. I do not claim all the findings are mine - no plagiarism intended!
- ThomasH358's Instructable: Battery Powered Tube Amplifier, the first build of such a project I have seen using rod-type tubes, actually this made me aware such tubes even existed. Totally sparked my interest.
- RadioMuseum page giving datasheet-like data on the 1Ж24Б (1J24B) tube measured by devoted individuals. Huge thanks to every one of them! All the characteristics data I used came from graphs posted there.
- Tom Schlangen's notes on triodizing pentodes.
- sovtube.com, a Ukranian store selling tubes and lots of other vintage stuff for good prices. Unfortunately, the 1J24B are no more in stock there. An alternative would be:
- Retrostore.com (Slovenia/EU), or eBay, among many other but more expensive suppliers.
- My Russian colleague at work, Ilya for being kind enough to help me with the Kyrillic data sheet.
!! ATTENTION !! this project involves moderately high voltage (100V) at low amperage (<5mA). However the filter capacitors might supply substantially more current, at least for a short moment. Be advised to work with proper care under suitable safety precautions!
Skill wise this project is well suited for you, if you...
- ...have some experience with higher DC voltages, or there is someone who helps you achieve confidence in handling it safely.
- ...feel confident finding your own layout for the schematic or you can accept some frustrating portion of trial and error while learning. Seriously, I am trying to give you schematics-to-layout tips whenever I can, but most of the time I find myself unable to guide you on those details. If you can't find a spot to start on this, I would recommend watching DIY Pedals' take on beginner breadboarding skills tutorial. There could be better ones, but among what I have seen, I like this best. Its about non-solder breadboards, though.
- ...are able to pack this circuit in your own housing. I used a miniature PC PSU housing as I had it at hand and it was perfectly suiting the space demands of this project. However I have not seen any of those before nor after, so I guess details on working it to fit won't help you much. Sorry to have to say that.
- 1x TL072 (4558 will work as well, even has the same pinout!)
- 4x 1Ж24Б (1J24B) tubes (recommend getting some extra. I feel like they're built like tanks, but still better safe than sorry)
CAPACITORS - better have all of those in the tube-circuit be rated 150V+, also: axial types are preferred in the signal path, especially for cathode bypass caps
- 3x 100 pF
- 2x 220 pF
- 3x 470 pF
- 1x 1 nF
- (1x3.3 nF) V3 cathode bypass cap - removed in final version due to unpleasant over-pronunciation of 2-3 kHz region
- 1x 10 nF
- 2x 15 nF
- 5x 22 nF
- 1x 47 nF
- 1x 220 nF choose higher value if you appreciate bassy tones.
- 1x 330 nF can be left out (shorted), or replaced with even higher values, both might result in more bass
- 2x 470 nF
- 2x 47 uF 400V
- 2x 470 uF
RESISTORS - 0.25W is fine except where stated otherwise
- 2x 10k 1W
- 4x 100k 1W (0.25W should be more than fine even here, still, larger components look more important ;)
- 4x 22R
- 4x 47R
- 1x 300R
- 1x 4k7
- 3x 10k
- 1x 33k-56k (mine as well as the original are 33k, but experts recommend 56k as it increases the amount of controllable treble)
- 1x 68k
- 3x 470k
- 2x 1M
- 3x 5M
TRIMMERS - precise variable resistors, adjusted to read those (values in brackets); optimum settings might vary for your ears, guitar pickups, playing style etc.
- 1x 20k (11.5)
- 1x 100k (62.8)
- 1x 50k (14.4)
VARIABLE RESISTORS - POTENTIOMETERS
- 2x 22k
- 1x 500k
- 2x 1M
- 1x Anything 100k to 2.2M - I tried the extremes and did not notice much audible difference.
- 2x Audio Jack (I like 1/4"/6.3mm guitar jack as input, but more often than not use 1/8"/3.5mm stereo output devices so for me, such they are)
- 2x Switch 4 channel changer
- 2x Switch 2 channel changer
- 4x Battery Holder AA or AAA
- 1x 20-Pin ATX Connector (if using PC-PSU)
- 1x optional capacitor-stick-bar (=2-row header sockets)
- 2-3x Hole Plates
- Some yards/meters of single and twin-wires
- Masking tape for insulation of conductive walls, just in case
- Mounting plate for insulated mounting the PCBs and macrocompoonents
- High voltage supply (adjustable boost converter) able to provide 100V
- 9-24V supply for Semiconductors (! if using also for Boost converter: mind its input specs!); PC-PSUs are near-perfect! Do a little research on whether modding those to fit your needs is for you, though...
- 4x AA or AAA battery cells (highly recommend voltage controlled 1.5V Li-Ion cells, those with USB-C connectors directly on the cell)
Step 1: An Overview of What This Is All About
First of all: don't be intimidated by the fullness of this schematic, we will work the details in small portions so everything will get clearer on the way.
You find the full, non-annotated schematics PDF as vector graphics for download at the end of this step
Please forgive, this text is for understanding the design implications in order for the reader to decide on their own preferences rather than for providing a strict recipe for "mindless following". While ending up less 'instructive' than the site name may suggest, I believe this format serves the reader best.
Specifically, with this Instructable I intend to stir awareness of...
- how vacuum tubes can be used for guitar signal (voltage) amplification
- how pentodes can be dealt with to do the triodes' jobs
- what every part of them do, and how they react to changes
- how far you can get building a "high gain" amp from rod type tubes, leaning on a schematic that was clearly made for different type tubes
- where the limitations of such a design are (what not to expect) and how I would try to expand those limits if I decided to take the design even further.
Name trivia: The Marshlands have been a place that gifted me with musical inspiration while wading them, roaming them, as well as partying and meditating there. It is also a combination of that famous British amp brand and my own name, a shameless combination to form a kind of suitable branding for a build like this.
I tried my best to phonetically type out the "logo" in kyrillic letters, trying to culturally approach its functional core components' origin - just in case anyone asked.
So here we are, building a tube preamp!
It's heart piece - the Marshall JCM800-imitating tube circuit has a number of elements, each of which I insist should be understood separately. Therefore the next steps will emphasize on:
- how a preamp tube stage achieves voltage amplification
- general characteristics of the 1J24B tubes
- filament powering: what, why & how
- high voltage supply
- testing tubes
- the overall routing of the signal path
- biasing: resistors & bypass caps
- pentode specific grids and what to do with them
Tube puritans beware! There is a 2-channel op-amp in the schematic, and it's there for 2 reasons:
- The preamp tubes have rather high-impedance output, even the "buffer" tube V4. Even headphones would load the output down so you don't hear much. Thus I engaged an output op-amp buffer to provide some current amplification/impedance matching without having to mess with power amp tubes and transformer.
- Also, there is an optional (switchable) mid/high-boost stage before the first tube so we can spare the pedals usually used for this. ...or can we?
Now, let's jump into it, shall we?
Step 2: Principles of Tube Amplification Stages
Actually I wanted to omit this to keep theory short, but its importance for the understanding of tube amp schematics made the need for it obvious. There is plenty of information on how vacuum tubes work, with my favorite informational being this ancient motion picture, cut shorter and annotated by Rob Robinette, the tube amp guy.
However, understanding the basics does not automatically make you aware of how an amplification stage actually achieves VOLTAGE AMPLIFICATION. At least to me, this was kind of a missing link towards understanding tube amp schematics. Actually, voltage amplification in general is a topic not so easy to find good explanations for. I'll do my best.
Please take a look at the tube stage schematic (the picture just above).
The primary signal enters from the left. It is pulled to ground by the grid leak resistor, thus has ground (0V) DC potential, which is a little bit more negative than the cathode. But the grid leak resistor is large valued, thus it allows signal fluctuations (AC) to flow without being drained to ground too much.
Imagine the tube's setup as a 3-resistor voltage divider with typical voltages for this tube type:
- ground (0V all values are referenced to ground here)
- bias resistor, cathode (2V)
- tube, anode (65V)
- anode resistor, supply rail (74V)
Seen from the tube's perspective, bias voltage (=grid potential) is mostly referenced to cathode potential. As the grid potential is essentially ground and the cathode is slightly positive, the "bias voltage" is usually written as negative, single-digit Volt values. Don't let this confuse you!
Now, as AC variations of the input signal lead to...
- variations in voltage between grid and cathode, with grid-cathode voltage being the primary leverage factor for allowing electrons to flow through the vacuum tube, this leads to
- variations in the density of the electron beam permitted to flow from cathode to anode, which is equivalent to
- variations in transconductance (cathode-to-anode). Varying transconductance of the tube inversely but still also means
- variations in "transresistance" of the tube (transresistance, as name-giving feat of the transistor is not the physical working principle of the vacuum tube, but can be calculated and treated like if it was). Now, varying resistance of one of the resistors in a voltage divider leads to
- varying partition ratios of the voltage divider, which FINALLY means
- varying output voltage! Wow, quite complicated, isn't it?
And if we are lucky, the variations in output voltage should now be of higher amplitude, than the original one fed to the grid. The ratio of output amplitude to input amplitude is what we cherish so dear as our voltage gain.
Now, as a homework, I would ask you to wrap your head around why the output signal is inverted (vertically mirrored or polarity-flipped) with respect to the input signal - this might matter later on!
Note: I do discourage polarity inversion to be viewed as 180° phase shift. While symmetric signals as sine or square waves do behave identically if 180° phase shifted and polarity inverted, not so for asymmetric ones like saw tooth or played guitar signals, especially the picking transients. Phase shifting those does not really cancel those out with the primary signal, while polarity inversion does!
Step 3: 1J24B - What We Are Dealing With
So this is the amplilfication device we are going to work with, the 1Ж24Б (1J24B) subminiature rod-type pentode vacuum tube.
By size, Tubes can be categorized as
- standard - usual power-amp tubes
- miniature - usual pre-amp tubes
- sub-miniature - anything even smaller
- on the other hand, tubes can also go much much larger for some applications
The rod types seem to have been made for mainly hand held (radio transmitter) devices, taking the working principle of the vacuum tube to the minimum size feasibly mass producable at the time. This also serves us when trying to make an awesome device need little space.
Regarding tube anatomy:
As opposed to standard tube intestines, made of complicated metal parts, each geometrically optimized for its specific job, there is this family of rod-types, where all the jobs are done by smartly positioned metal rods or poles. Here is a depiction of how the electron beam is formed and interacted with in such tubes, the small central thing being both, the
- filament ("Heizfaden"), but simultaneously also the
- "a" means anode, high positive voltage and main attractor of the electrons.
- g1 (a & b) are the control grid, which is slightly more negative than the cathode, thus repelling the likewise negatively charged electrons and hindering them at flying towards the anode. This effect is strongly dependent on the actual potential of g1 relative to the cathode, thus when connected to the guitar, the electron flow is strongly dependent on what comes from your guitar.
- g2 "screen grid" has high (positive) voltage potential, thus is depicted attracting electrons, pulling them towards the anode. Its actual purpose, however, is to shield the control grid from capacitive interactions with the anode.
- g3 "suppressor grid" has cathode potential, repelling electrons again. Actually, high energy "primary" electrons from the cathode itself are not really repelled, but secondary electrons emitted from the anode with way less energy are blocked from travelling to g2 where they would cause a variety of troubles.
You find the...
- anode wire leaving at the top of the tube and the
- grid wire in the center of the bottom.
- cathode/filament (+) & (-) and
- screen, as well as
- G2 and
- G3 are arranged in rows of length 3 and 2 at each side of the grid lead. All the connections have the shape of long leads, so if a tube fails, you will need to solder a lot.
The datasheet advises to not bend the leads closer than 5mm (approx. 1/4 inch) from where they leave the glass.
Sticking them on IC sockets is not advisable - sticking cycle &/ corrosion... Its a constant source of problems such as poor connection, unpredictable faults, shorts when trying to get the bend right... costing lots of mental energy. Just solder them on, please!
Regarding tube modeling (a tangential excursion):
If anybody wants to go deeper into these tubes' characteristics and simulation, I cannot guarantee the quality of the data, but I did some work with the things people have posted on radiomuseum.org:
I extracted coefficients to computationally approximate the tubes' behaviour in order to simulate it. Unluckily, I did not get very far with the simulation thing, but I did at least get a usable characteristics plot, also seen in the pictures to this step, but I made an effort to comment it well.
Full raw data of everything I did is accessible via this link. Going there, tubeSim.r is a script written in open source statistical programming language R. It is not fully functional, but yet able to produce characteristics plots, if executed line by line with care. All of the data is dependent on the filament voltage used when measuring the characteristics. As described later, filament voltage has a clearly noticable, very interesting effect on the tone which leads me to suspect it changes characteristics a lot. With no data ever going to be available... Thus, I went away from carefully designing beforehand --> to doing the interesting things experimentally.
Step 4: Testing Your Tube's Filament
Let's finally get some hands on the hardware!
To disappoint you right up front: the boring thing about the filament of those tubes is: unlike separately heated cathode tubes,
- these don't glow and
- don't get warm either,
so keep them, they're not broken!
That is, as long as filament resistance measures about 30 Ohms cold and draws datasheet-like current of about 10mA at a trans-filament voltage of around 1.2V.
Try resistor values around 20 Ohms in series with the filament, fed by a 1.5V battery, and you should achieve datasheet-approved filament powering.
Polarity is not important as long as you are merely testing the filament for voltage drop/resistance measurements.
--> BUT it gets highly important as soon as the anode voltage is applied to amplify signals, as the filament voltage will interfere with bias voltage!
Thats about all there is to filament measurements.
Step 5: The Amp's Strategy for Filament Power Supply
As indicated before, the cathode itself is part of the filament (or vice versa) and therefore their potential wrt. common ground must equate to the tube's bias voltage. This bias voltage, however, varies (intentionally and greatly) among the individual tubes (V1 through V4). The only way to deal with this are filament power supplies, which are independent from absolute GND potential. While having capacitance-coupled AC supply, or independent transformer coils, rectifiers and ripple filters for each of tube's filament would be theoretically possible, my approach is to use batteries, especially when considering the filament voltage requirements (0.95-1.4V).
The datasheet also provides allowed current flow for the filament. As the heated wire is cold by the time of switching it on and gets warm later on, its resistance is low first and rises with on-time. I found 22 Ohm resistors (incidentally the same as thomas358's) to be the best match for when the wire is warm, but I feel like I saw transient current exceed the maximum value of the datasheet, thus I added another switchable 47 Ohm"warm-up" resistor which you can bypass when the the filament is on temperature, as seen on the annotated detail schematic. Actually, the resulting 69 Ohm resistance lets both the filament voltage as well as its current draw drop below datasheet minimum values. Still, they do work under these conditions.
The scheme of the filament power supply is the same across all tubes, and simultaneous switching is preferable. Thus, 2x 4-channel 2-way switches are needed. I was lucky to find such, but after discovering the interesting effect filament voltage seems to have on the tone, I would probably rather use 100Ohm potentiometers (in series to 22R fixed) for each tube's filament for individual fine-tune-ability.
Of course you can use 4x 2-channel switches to accomplish the switching control with more resolution. Your interface might start to resemble a MiG21 cockpit a lot, however (knobs and switches everywhere!).
In an attempt to find words for the tonal characteristics of each filament supply setting, I called them
- "bubble" (69 Ohm total filament resistor, actually far below datasheet filament current levels) and
- "grzzle" (22 Ohm, about 11mA warm)
All the stated resistor values correspond to 1.5V kind of batteries.
--> What a relief: 1.5V rechargeable voltage controlled Li-Ion Cells are now available in the old AA / AAA format. Testing has proven them to be the best solution for this task.
Testing with 1.2V rechargeables (Ni-Cd, Ni-MH) was not satisfying. You get reasonable tone out of the"grzzle" setting, but "bubble" is hardly present and the whole thing does not sound the way it is meant. Those types of batteries get useless for voltage-drop reasons way too early, making you ask yourself what to do with all those still like 70% charged batteries.
Even 1.5V alkaline batteries need to be quite fresh for strong tone, which is a total waste, unless there are other appliances that make good use of half-drained cells. Still they're only single use and therefore more of an environmental problem.
Regarding battery accomodation, 4x 1-battery holders are probably most convenient. Do align them in parallel, so cable routing becomes ordered and easy.
Otherwise... mine is a single 4-AAA alternating pattern battery holder. I needed to hack it first to get all the leads independently available for my needs. And the orientation inversion pattern doesnt exactly help with tidyness of cable routing. This is one of the many reasons why my build looks so messed-up.
Step 6: The High Voltage Supply
Our tubes won't work without a serious anode supply, hence we will need to gat that right before we can test the tubes' amplification capability.
On the minimum side, there is no chance of achieving any sound with this tube when you're far below 30V of anode voltage. Still, below 40V, you don't get much amplification, which is kind of the point of an amplifier.
On the maximum side, the datasheet states allowed values of 120V for the anode, but 90V for the 2nd grid, which we might want to connect to the anode (more of that later).
Yet, if you follow my schematic, with an initial supply voltage of 100V you get enough voltage drop at the ripple filter (where the 10k 1W R's and 47uF 400V C's meet) to be safe even for the 2nd grid. This way you get the best of both: close to maximum allowed voltage for highest amplification, but still safety to not exceed any rating.
Thus a requirement for your "high voltage" power supply is the ability to achieve 100V at <5 mA of current draw.
Mine is a 12V DC boost converter and it was able to heat a 22kOhm resistor between 0 and 100V with no problem, meaning 0.45W is possible at 100V and more than enough to power the tubes. It did not like 10kOhm, however. While it maintained the voltage with almost no deviation across all resistor values high enough, at too low values, the short-circuit protection kicks in and shuts off totally. Beware of the voltage at all times, even though it looks so innocent being boosted from only 12V.
The larger the filter caps, the more current they can keep on delivering even if the supply itself is shutting down for over-current or any other reasons.
The anode voltage supply is split in 2 groups at the ripple filter: the 1st 2 valves (V1 & V2) group and the 2nd 2 (V3 & V4). This is probably to prevent the signal ripple to pass over from the later, high output-tubes to the earlier ones via the anode supply rail, which could lead to unwanted feedback oscillations.
Also, every tube stage inverts the signal and the tubes V2 and V4 do not produce much voltage amplification.
--> This means that the anode ripple of tubes 1 and 2 as well as 3 and 4 might cancel each other out quite well, so least possible ripple is brought in either direction via the power supply rail! (Side fact: in another tube preamp build I experienced that odd numbers of tubes and/or varying gain distribution per group do not noticably adversely affect feedback stability)
From there, each anode is connected to it's power rail via a 100k resistor - all except for the 4th tube, which is a buffer tube meant for current amplification/impedance matching, not voltage amplification. Its anode is directly connected to the supply rail.
Step 7: Testing a Tube
Now that we know where we want to go with the filament supply, lets make a test circuit to see if the tube can amplify signals. I recommend breadboarding the full V1 schematic as shown. It is way too easy to lose function when going for simpler setups.
If wary of stickboard connectivity, you could solder everything on perfboard, but the tube with all its leads is not so nice to un-solder later on.
Do not point-to-point-twist-leads or crocodile-cable everything. You'd want to check connections all the time - with the risk of unintentional shorts or touching dangerous voltage being disproportionally higher.
Be cautious with the breadboarded circuit as well. Do not follow my bad example in the video, but rather...
- think twice whenever about to switch on high voltage (HV) supply!
- immediately switch off HV as soon as no longer needed
- always check HV supply rail voltage has creeped below measurable before touching any HV or anode related leads!
- Use insulated tools whenever live-checking or connections is inevitable
- Check the potential of any lead you intend to touch, so you know for sure it does not have an ill-connection! Also make sure you do not feed HV to your sound system or guitar
So lets finally get going!
- get your filament supply (as we did in Step 4: Testing Filament Supply - do mind polarity this time!)
- get a DC boost converter for high voltage supply (<90V) connected to the anode via a reasonable anode resistor (100k is a classic)
- A large grid leak resistor (1M is good)
- A reasonable cathode resistor (1-3kOhm)
- A "grid stopper" resistor between the guitar signal and the grid (68k is a classic)
- A capacitor going out from the anode lead (be above 20nF)
- An active speaker system (recommend old, expandable PC-Speakers)
- And don't forget to connect the guitar ground with the tube circuit's ground! (That's why it's often called common ground or COM)
- Better already connect G3 to the cathode and
- G2 to anode, but via a large (few MOhm) resistor
Quite an effort compared to a simple triode, I admit. There's no obligation to test every single tube, if you're confident the amp circuit will work out. But its fun to try and could be a valuable tool for debugging. I think it is good to get to know the tube's implications hands-on a little bit before messing things up in the amp circuit Plus, if you know which tubes might be higher gain than others, you know better, where to choose which for placement in the amp circuit.
Step 8: The Signal Path
The beginning: Omitting the optional boost stage, the guitar signal enters facing a 68k "grid stopper" resistor, after which the signal is loosely pulled to ground by the 1M "grid leak" resistor and enters the 1st tube, V1 at the control grid (G1) pin.
V1 characteristics: the weak primary guitar signal is more or less cleanly/linearly amplified by V1, with the amplified signal leaving at the anode. V1 is paralleled by a 100pF capacitor to cut ultra high frequency transients and oscillations.
We will come to biasing in more detail later but the huge 470nF cathode bypass capacitor boosts all frequencies down to the very bass of the guitar signal and ensures lots of mids and lower mids are getting the extra amplification too.
V1-to-V2 signal processing: After a 22nF coupling cap removes the DC potential and low frequencies, the gain potentiometer is part of a voltage divider, controlling the amplitude of the signal fed to V2's grid between 2/3 of V1's AC output and 0 (ground).
Small value "bright caps" go parallel to the voltage divider, so high frequencies are permitted to pass. You can hear the signal getting brighter when turning down the gain knob.
--> when compensating the lost gain with a boost pedal before the preamp, the gain knob can be used as a kind of pre-distortion tone control!
--> Having lots of high frequencies before entering parts of a circuit that introduce distortion is important for getting this "chiggy-chiggy" sound thing when the plectrum hits the strings, while having less lows prevents unpleasant crackling noises.
V2 characteristics: V2 is next, also called "cold clipper" for its rather "cool" bias (close to cut-off voltage, where the operation characteristics are pretty curvy and non-linear, this seems to play a huge role in achieving what we know as the pleasant tube distortion sound that is loved by many). Cold bias requires a high-value bias resistor, resulting in higher bias voltage and rather little amplification. No surprise, as with very negative bias voltage, the grid blocks too much of the electron beam and thereby hampers the flow of current a lot. But as you might have guessed, the "cold clipper" is made rather for clipping than for amplifying.
V2 bypass cap: these are a very controversial topic (meme by RobRob). I follow his recommendation to have one bypassable there. The tone stays rather dark without one; if you have one with a small value (<10nF), the tone gets very brittle and harsh, while a large value (>20nF) will make the tone sound unpleasantly strained/stressed, like it was just trying too hard. My best choice was 15nF axial type cap (audiophiles will recommend axial types in general, can't say I can hear why, but this particular cap is a very game changing component, I felt like even I could hear the difference).
V2-to-V3 signal processing: the bright cap bypassed voltage divider thing after a 22nF coupling cap again. Those attanuating parts are vital to any nice "high gain" pre-amp sounds, as overstressing valves with too high amplitude signals always make them sound unpleasantly strained and crackling, at times also very fizzy, in any way, very "untubey".
V3 characteristics: we had cold biased V2, now we have hot biased V3 - as we learned in step 2: tube amplification stage principles, the signal gets polarity inverted when passing a tube stage. The cold clipper did some reshaping to the trough of its input signal, which is the peak of its output and equally V3's input. The hot biased V3 with its high gain nature now further does a lot of reshaping to the peak of its input signal and leaves its trough rather linearly clean and 'intact', adding to the asymmetry of the signal distortion. The small 3.3nF cathode bypass cap was removed in the latest design. Its "late distortion" high frequency boost did not exactly add "chiggy-chiggy" and "aggressive top end grit" as intended, but rather over-pronounced awkward 2-3kHz frequencies and added more crackling to the already too clicky transients.
V4 buffer stage: I admit I do not fully understand how this stage works in detail, but as long as you connect it's grid directly to V3's Anode and V4's anode directly to its supply rail, you will get a "buffered" signal from V4's cathode, provided its 100k cathode resistor goes to ground. "Buffering" here refers to a kind of current amplification / un-coupling V4's output impedance from V3's hard overdrive work - so even if V4 gets loaded down hard, V3 will not "feel" it too much, so overdrive/distortion characteristics will not change when dialling in the tone stack. Yes, the tone stack is quite some load for the tubes!
The Treble-Mid-Bass (TMB) Tone Stack: Also here I have my troubles wrapping my dull head around. Still, I decided to stick to it and as of my perception, it magically just works. Might take some time working out the correct pot connections, so they control their thing with correct turn direction. For circuit board real estate reasons (and for easier reference which part does what) I like to solder as many components as possible directly to the pot leads/holes. As all the EQ channels feature unbiased capacitor coupling, a further coupling cap is actually not necessary. I truly wonder why I left mine in...
Master Output buffering: whether you design your own tube power (!) stage or go with semiconductors, you will need a minimum measure of current amplification to be able to power even headphones. Those preamp tubes all have their 100k resistors as an integral part of their operational setup. I guess that's why preamp tubes hardly ever wear out. But effectively, they do not provide any current to back up the voltage for driving even small loads.
Step 9: Bias Resistors & Bypass Caps
We already saw how bias resistors make the cathode potential more positive than the grid potential. The latter is always tied to ground - loosely with high valued resistors to prevent AC signals from being drained to ground too much - but still around 0V DC.
It was also indicated before that the more negative the grid potential gets wrt. cathode (cold bias), the more curved and less steep the lines of the tube's characteristic chart get
Meaning: non-linear transfer characteristics (=signal distortion) and only little amplification gain.
The more positive (closer to 0, hot bias) the grid vs cathode potential gets, the steeper and straighter the lines get, offering more gain, more linearly
--> linear-neutral region.
But in excess, this leads to less pleasant sounding heavy fizzy distortion
--> saturation region
and non-negligable current flow through the grid as soon as it gets even positive (--> low input impedance), which might damage or wear the tube faster than expected ("faster than its spec's said" little tongue twisting giggle intended).
Bypass caps: when you place a capacitor in parallel to the cathode (=bias-) resistor, at first, this does not affect DC bias voltage, but higher frequencies are permitted rather than low ones as they are able to bypass the bias resistor. This loads down the tube anode voltage, and pulls up the cathode voltage, and thereby bias (imagine the "valve-equivalent-resistor" getting smaller in the "3-resistor voltage divider" model of step 2: Tube amplification stages).
To purge away any doubts, I did a verification test run and the results are as expected: even regardless of HOW the signal is boosted, be it by
- setting the filament power from "bubble" to "grzzle" (22 Ohms instead of 69)
- activating a cathode bypass cap
- playing louder with the guitar
- activating a boost pedal,
all those make the "bias voltage" more negative for the time the signal roars. My V2 has a
- silent bias voltage of about -2.1V, which can
- go to -2.6V or even -2.8V when hit hardwithout the bypass cap or boost pedal.
- Now with the bypass cap, and full thrashing on the strings, -3.0V is easily exceeded (--> more negative), while
- with a boost pedal engaged, it even went towards -5V while still delivering substantial output. Strange, huh?
Depending on the bypass cap size,
- with few nF ones, the effect is usually a high frequency boost.
- Getting into the 10-100 nF region, it will boost also mids to lower mids, with the perception of overall loudness drastically increased, but with it also the chances/danger of getting into distortion territory. Unless you know what you are doing, it will most probably not be a pleasant sounding one...
- several 100's of nF will boost all of the signal, again the overall loudness perception will rise. Anywhere but with V1, the high amplitude low frequencies create a crackly, "farty" and "trying too hard" kind of tone. Very disgusting, even for a metal and grind guy like me!
Step 10: Handling of the Additional Grids & Morphing of the Transfer Characteristics
As indicated before, we need to find a useful way to connect the pentode's "surplus grids".
For convenience of easy reference, I repost the link to rod type functional anatomy here. Also, take a look or 2 at the change of transfer characteristics in dependency of grid connection of the 1Ж24Б tube, measured by helpful individuals on radiomuseum.org/1J24B. There, you can see the way of connecting G2 can have quite some effect. But another thing too has quite an effect: filament voltage. Switching the filament resistor does not only change total volume, but also all anode voltages as well as bias voltages and everything in the whole signal path in a way nothing else does.
As batteries drain over time and change their voltage, this effect is hard to control, after all. Not a problem anymore since voltage controlled rechargable 1.5V Li-Ion batteries are emerging. Those do not lose any voltage as they drain, since they are precisely down-controlled all the time. They shut down completely when a certain emptiness condition is reached, thats when you have to charge them using a USB-C phone charger
G3, the "suppressor grid" is the easier one. Just connect it to the cathode, where also the "screen" (don't confuse with "screen grid") goes.
G2, also called "screen grid" (again: do not confuse with "screen"!) is more complicated. It should have rather positive potential to not interfere with the electron beam too much while doing its actual job: shield the control grid from the anode's electric field to prevent capacitances between them. Positive potential is achieved by connecting G2 to the anode. However direct connection I have only found to be useful for V3. All other tubes have large resistors between anode and G2. Need not be exactly 5M, anything close to or better in single digit MOhm region will be fine. I tested all the situations with a 2.2 MOhm potentiometer and went with what sounded best to my ears, which was 0 for V3 and max for all the others.
G2-cathode cap: Again, I just threw around with values until I found sweet spots. V1 did profit from the large 470nF G2-anode cap, all the others are <1nF, with values being important only in terms of orders of magnitude, as far as my ears could tell.
Step 11: The Output Buffer
We learned to love the tubes' transfer curve peculiarities and what they do when genuinely pushed into overdrive, however in this setting, their current amplification capability is very poor. Not even higher impedance headphones can be used to hear much.
Playing through PC speakers or into another active system (like an IR-loadbox) is an option for me, but not if it's the only option. I decided to add a silicon output buffer stage to be able to push some current and play with headphones without external systems.
What concerns the op-amp circuitry, I went with a very simple design, "hard grounding" the inverting input while putting a pot as a voltage divider between signal origin, non-inverting input and output. In such way, we have true master volume control that can even go into op-amp distortion if one feels like blasphemizing all the good tube sound with semiconductor fizz.
By "grounding" I do not mean routing to the ground we had in the high-voltage circuit. Well, with the latest version of the circuit it might coincide with it, but still, they are not DC coupled.
--> Instead, there is a pseudo-ground rail which is established using a 10k/10k voltage divider between the OpAmp's power input pins. Large capacitors (470uF) stabilize the pseudo-ground against hum, buzz and transients.
This way, the op-amp's supply voltage pins can still be chosen freely and independently of the high voltage ground. For instance, op-amp supply can be 0..+12V, but also-12..+12 or a 9V battery, without changing anything else in the circuit.
Do not forget to use a coupling capacitor between the op-amp's output and the load! If you are into bass-rich guitar tones, you might want to choose a higher value output coupling cap, like 470nF or even above 1uF instead of my 220nF one.
Step 12: Input Boost (Optional)
Ok, I have to hide a lot of shame here, as this is the worst designed part of the project.
It is optional - the tube part of the amp works for blues and rock'n'roll tones and gives sweet overdrive spiciness. However, If you're into dirtier, beefier, more juice dripping kind of distortion sounds, it just lacks the gain/input amplitude to get those. Even the original JCM800, although marketed as a "high-gain" amp is known to best be played using a boost pedal before the input when trying to achieve actual high gain tones (by today's standards).
So, using the other channel of the op-amp already needed for the output buffer, I loosely followed the circuit of the 1st amplification stage of the JCM900, but went down with the coupling cap sizes so it would be a highs + higher mids boost as to not trouble the tubes with too much bass.
Initially I had placed cheap chinese TL072 there. Let me tell you: this was the first time I actually HEARD cheap fake components be tonally inferior to original parts. Don't be too cheap on those! However, some kind of squealing oscillation is still there when seriously cranking the boost. Could be caused by the power supply - PC PSU's might not be so well suited for powering audio circuits with lots of amplification after all...
Possibly, the solution still is to use external boost pedals, following well-established circuitry. From what I heard in tests on Youtube, my future go-to might be the ZVex - "Box of Rock" MOSFET circuit, for it seems to have very tube-compatible boost and pre-distortion.
In the Meantime, I had some very good experiences using the clean boost function of a Behringer Super Fuzz SF300. You can hear it featured in the "Brick Make Noise!" video in Step 14.
Boosting too hard even with very clean boosts results in crackling and tube overload. Kind of usable if something is to sound very old / bad vintage, or if youre into power amp distortion, that's quite close!
I did test putting a boss HM2 clone into the signal chain right before the amp, but... it's like a collision of worlds, just too different.
Step 13: Packing It All Together & Debugging
It is good practice to have high voltage circuits enclosed in a grounded housing. Usually, I prefer non-conductive housings for reduced risk of shorting things when they accidentally touch the walls, however, those are not shielded and more vulnerable to noise from external electric fields. Here, I went with a spray painted miniature PC PSU that I had lying around. It already included what I consider a very important part of any metal housing: a well suited plastic piece to insulate anything inside the housing from the conductive walls.
I was lucky to have access to a 3D-printer to design a base plate to mount all the PCBs on. Well, cutting and drilling the same design out of plywood or a sheet of plastic would have worked just the same. Unlike bigger amps' tubes, those sub-miniature rod types don't seem to get hot or even warm, so even wood is probably safe.
Better be sure you provide sufficient space for mounting the high voltage supply and ripple filter well separated from any conductor carrying either the signal or op-amp related voltages.
Regarding the user interface, thus I had my troubles coming up with a functional and any kind of meaningful layout - I'm sure yours will be better!
Some advice from partly my experience, partly common sense, partly established conventions:
- Whenever having a lot of cable-connected potentiometers, it helped me to have the cables be long enough so you can lay the lid aside without cables pulling at them.
- Also, having all the potentiometer bodies adhere to some kind of solid board before mounting them to the actual housing really helps with the structure, mounting and serviceability.
- Before drilling holes, make sure the potentiometer bodies have enough space next to each other, the ultimate objective being actually individually accessible/operable knobs.
- The switches were tricky too. Had to file the slits for the levers to be able to move freely. When tightening the screws too much, the switches stopped working the way they should. Sometimes moderation is key.
- Usually, I like to have my builds running at the very moment of closing the lid - just too often they stopped working just there. With higher voltage circuits as this one, however, I need to strictly discourage this practice! !!DO NOT RISK YOUR HEALTH OR EVEN LIFE to such a stupid thing!! Even if it means everything to you in that little moment. Think twice and leave the power off whenever possible in any way!
- The first and foremost reason for failure were COLD SOLDER SPOTS. There are some component leads that just won't bond with the solder, no matter how diligently you clean them before, how much flux you use or how hot you will let them get - they just refuse. Others will break away from contact even though the spot felt and looked like well done.
- Often enough, watching carefully while mechanically pushing and pulling on components (with voltage off!) gives good indication of where the bad solder spots really are located.
- Minimize buzz/hum noise by providing proper shielding: Connect the metal housing to the ground pin!
Step 14: BRICK MAKE NOISE!!!
Oh, don't miss on having fun with the completed build! This wasn't an all too easy one, or was it?
Hoping for some comments sharing your thoughts!
I made this video to see how well the amp takes over-the-top distortion and boost pedals - evaluating the amp as a pedal platform
The signal chain is like:
- Mediocre & highly improvised playing on a
- Jackson guitar with passive EMG bridge pickup, (you can hear how the action is set waaay too low from all the fret buzz in clean mode - gotta fix that soon!) into a
- Sondery SML-9 "METAL" (a actually very cheap but very capable ProCo RAT ripoff, but unaware which version) this is disabled, true bypass, until obviously activated.
- Behringer SuperFuzz SF300, very cheap very crazy octave fuzz, but it is not used as such - it really is just a clean boost (adding only volume, not any distortion!). It is set to boost mids and high frequencies while cutting lots of bass. You can hear how once activated, this boost suddenly makes the tubes scream. They are pushed very far into a variety of highly non-linear operating regions. What about the amp's input level? "Some like it hot" ;)
- Now THE AMP. The gain knob isn't even set too high - well, with hot pedals in front the "gain" knob really is just a tone knob, actually. But the clean/crunch playing could be way sipicier with gain full up, to be honest.
- old active, "midforward" PC-speakers - as said before, the build is "only" a pre-amp.
- ambient recording using the very same cellphone as for the video.
I am really impressed by how the amp takes all the distortion and boost. It does "chainsaw" a lot, also a lot of "octave-fuzz smell" in it - all in all its way over the top distorted, of course, but still to a great extent by those tubes native to the amp.
Step 15: Neural Amp Models - Try Before You Build
You want to play a software-only version of this amp for free? The good thing about today being 2023 now there is a quite usable - even very good - open source project called Neural Amp Modeler (NAM):
- Download here (make sure you pick the plugin, not the trainer for this purpose) as a plugin or standalone and turn your PC into any amp you can find models for (and the community is already huge).
- Download the Marshland800 models I trained so you can try this very amp build through a few settings by just clicking around a bit! It even contains the "brickMakeNoiseConfig.nam" model, but this one turned out to sound even way creepier than the tone you hear in the vid. Not every model can be trained equally well.
- Feel free to also try all my NAM models or search the wide web for tons of more user trained models.
- Let me even help you with a few guitar cabinet IRs I recorded for yet another Instructable.
- The experience is most authentic only when going into the sound device with low output impedance. Active pickups, DI-Boxes, active bypass of any pedal (Boss, Behringer do a lot of active bypasses) help whenever you don't have a dedicated high-impedance "instrument-in" on your audio interface. Avoid hitting the generic audio interface input directly with the signal from passive guitar pickups, it will sound rather dull.
- Fire up the NAM Plugin/standalone, load a model, pick an IR (or not), make sure audio input/outputs are correct, then adjust input & output levels to your liking and get jamming!
- Leave comments if in doubt!
Step 16: Summing Up the Project
To answer some questions you might have:
(at least those were burning in my head through sleepless nights...)
--> YES it is possible to make a usable guitar pre-amp, and even one with interesting and likeable, mid-forward tone from no-budget rod type tubes.
However, heavy tweaking was necessary to achieve a circuit that yields listenable results. The rod type's characteristics do by far not match the one of the original 12AX7s. Hence the tone is not really a match. I can't rule out having missed out on taking the best steps to make sure to approach the original sound the way a knowledgable engineer would. I'm just a maniac with a soldering iron, trying around randomly, "flying mostly blind" just trusting my subjective ears.
--> NO this is ANYTHING BUT a faithful JCM800 clone.
And, you guessed it,
--> NO it's not even a "high gain" pre-amp by reasonable standards, despite the high-gain circuit it is built upon.
--> YES it does have interesting, unforeseen merits:
- Boost the input substantially to get somewhat of a "farty", "crackling" tone to emulate power amp distortion! This might have to do with the fact that pentode tubes are being used, just like in a usual power amp, but don't pin me down on that! I'm going to use it in music production as a model, whenever I can't push a power amp as far as I feel would suit the tone I have in mind.
- It also behaves a lot like lesser-gain vintage amps (makes me want to play James Bond theme melody or dark themed western movie soundtracks)
- It makes for a really flexible and alive pedal-platform...
- Even with the "cleanest clean" settings you have a lot of bass rolled off and a lot of mids boosted, so it will never sound cold and surgical (you definitely need another kind of circuit if you want a pedal platform to sound more "sparkly fresh").
- Give it any kind of gain related settings to easily trigger tube distortion upon raising your pedal's output level.
- Give it all it has ("grzzle", V2-cap, gain knob full up) and we're at mercilessly raging power amp distortion tone, as mentioned before.
--> YES, with this build I learned a freakin' lot about tubes and
- how they behave under which circumstances.
- How and why people choose which exact bias resistors and
- cathode bypass caps, and
- how the tone changes with varying connection patterns of pentode specific grids
- The effect of tube filament voltage on the tone. I mean come on, show me one single other amp that can change that - and I'm not even yet talking about "at runtime"!
- How hard it is to get such insights captured and to make them understandable for others. I can only hope it will help some of you! Please fill the comments section with anything you want me to test - I can't guarantee anything but I'm going to do my best!
This bit is a true "DIY'ers carreer breakthrough" moment for me.
--> NO, a PC-PSU might not be the best source of power for audio circuits involving a lot of amplification.
--> NO, building boost circuits into an amp might not always be the best option, but
--> YES, op-amp output buffering saves us a lot of trouble and is rather easy to implement and yields clean and precise volume control over a huge range of amplitudes.
I want this build to further demonstrate the following:
--> You cannot just replace a critical component of a circuit with a drastically different one and expect the new thing to work just as the original.
The truth is that (1) you need a very capable amplifying device (e.g. tube) to begin with.
But even then, they only can shine if (2) you find an immaculately great circuit, designed exactly around it.
Only a good combination of (1) purpose-fit components and (2) a circuit tailored exactly for them is able to achieve a specific kind of tone.
So when trying to accurately clone your favorite gear, try to first be accurate with both, the choice of (critical) components and the circuit and only make small changes to see what suffices your needs. Otherwise you might end up like me here, spending months over months of weekly free time hours trying to get something usable out of a messed up circuit. Miraculously it did pay off this time, as something very interesting yet usable was created.
Amplifying and clipping devices in the signal path (transistors, tubes, diodes & op-amps) are critical in this respect, at least to begin with.
Step 17: Improvement Suggestions
...next time on 'what happens if...'
- Rob Robinette cites tube-amp-tech-book author Y. Blais - if he was to design a tube amp, he'd add another switchable / bypassable tube BEFORE V1. I would most probably try that instead of the silicon pre-boost, which currently causes oscillations in a mode that involves at least v1 and v2 when cranked (with cranking being the purpose of the boost, really).
--> In the meantime, I did do that already, with actual 12AX7 tubes this time - see my other tube preamp Instructable.
- If I was into endless laborious testing I'd like to get deeper into the filament voltage thing with individual potentiometers. The effect this has is way too interesting to not be investigated! Still, I do not expect the results to really justify the effort. I like the bubble / grzzle settings quite a lot. I think all tube amps should have variable voltage filament supply options!
- One thing that is known to greatly improve both, signal/noise ratio and visual appeal is to design a proper etched circuit board following good design practices, and ideally SMD components wherever possible. However, it will not change what the tubes can do. Even if that was able to mitigate the boost stage squeal... I'm fine with this crazy amp project looking and sounding as freaky as I still feel about it.
- Commercially available boost pedals work way too well, are way too versatile and way too cheap to not use them. I'd just stop building pre-boost circuits into amps, especially semiconductor boosts into tube amps.
- Once I had fantasies involving 4 separation transformers for the filament supplies because of their reliably constant voltage output. But the now available 1.5V voltage controlled Li-Ion batteries do the same and are way, way simpler.
If there is anything further you want to know or want me to know, please do not hesitate to contact me! I love reading private messages, yet probably more people will benefit if you use the comments section.
I will do my best to help and improve this instructable wherever possible.
Have a nice time building and playing!
This is an entry in the
Make Some Noise Contest