Design of a Korg Nutube Amplifier Part 1: Tube Basics
This is the first in a multi-part series exploring the design of a “guitar pedal” for amplification and distortion, going all the way from project start to a finished working design and making use of the Korg Nutube. We will be using only free design resources such as DesignSpark PCB so that it is easy to follow along and add your own touches to the design.
Why a guitar pedal?
A guitar pedal is the simplest and least restrictive amp we could pick. The term guitar covers so many devices that whatever effect is added, there would probably be a pedal to cover it.
Also, not all guitar pedals have pedals — some simply amplify, some distort and some have a pedal to control the effect, whereas others a potentiometer. This leaves some leeway, allowing us to design around the device. While this may seem counter-intuitive, it will become useful later when we discuss some of the limitations of the amplification device.
Audiophiles love tubes
Sound and music are so subjective and something that would sound perfect to one person may sound awful to another. So it’s quite unlikely that we’ll come to a consensus regarding tubes vs. transistors. Tubes do change the sound as they amplify and it’s often said the sound is “warmer”. Without getting too deep in the rabbit hole this comes from many factors; some tube amplifiers add noise, while others have a low-frequency response and amplify the bass more than treble. Maybe it’s simply the electrons are more angry in a tube than in a transistor, who knows!
Tubes are valves
Almost everyone has heard of a tube but may not know how they work. Just like semiconductors there are/were many types of valve, however, the staple was the triode and below is its symbol.
The triode is a simple device and likely much easier to understand than its modern replacements, transistors. It consists of three elements and you may recognise two of the names, anode and cathode:
The cathode (k) also known as the filament, simply put is an electron source. In a hot cathode tube the electrons “boil off” — technically “thermionic emissions”, but we like the image of very angry boiling electrons. This effect is created by heating the filament to very high temperatures (thousands of degrees), as it gives the electrons the energy to escape the work function of the metal in the cathode, or in other words “boil off”. Coincidentally this is one of the reasons why old tubes get very warm and use lots of power.
The anode (a), also referred to as the plate, is where those boiled off electrons are accelerated to. This acceleration of electrons necessitates the plate is at a high potential of up to several hundred volts (or more in some cases). This anode/cathode arrangement was the design and is inherited from the first valves, the diode (yes it works the same as the silicon version).
The grid (g) is where things get interesting: by adjusting the voltage by a tiny amount on the grid with respect to the cathode, the electrons can be attracted or intercepted at the grid. This causes a much bigger change at the anode/plate, creating voltage amplification.
Voltage amplification vs. current amplification
Having got basic valve theory covered it’s time to look at everyone’s favourite amplification device of choice, the NPN transistor.
Note that while the end effect of these devices is similar — and in some ways their construction — they do not operate in the same way. The tube amplifier is a voltage amplification device and a MOSFET may be closer to a tube in actual function. A bipolar transistor is a current amplification device, which is subtly different.
In the case of the tube, if we apply a voltage at the gate the voltage at the plate will change by the gain multiplied by the input. I.e. Vp=Gain * Vg.
In the case of the NPN above, if we provide a voltage to the base it will act like a forward biased diode, rise to 0.6-0.7V, remain there and the base current will increase accordingly. If the source is not current limited the transistor will “pop” quite literally. This is why we have a base resistor which will limit this current. When we use a bipolar transistor amplifier we have to control the current into the base and vary this, not the voltage.
The gain again is quite simple and Ic= gain *Ib. Given a fixed impedance this can become a voltage (V=IR). This difference is an important one, not to be forgotten and means these circuits are not directly comparable. This is also one of the more tangible audiophile differences, driving your speakers via voltage will produce different characteristics than current.
So why tubes?
For the first time in a while (a long while) there is a new tube on the market, the Nutube (144-8943) and (144-9016). They aim to behave just like the triode, but without a few of the more problematic aspects of a standard triode. While the Nutube behaves like a pair of triodes, the construction is different. and more reminiscent of a Vacuum Fluorescent Display (VFD). There is a good reason for this and they are based on VFD technology.
A VFD is not a lot different to a tube; traditionally they would have a hot cathode, an anode/plate and a grid to control electron flow. The difference being there is a phosphor on the anode which glows when electrons bombard it. Noritake Itron Corp has found a way to create some gain during this process and hence the creation of the KORG Nutube.
In modern electronics, power has become a major concern and unlike the VFDs of old, the modern equivalents have adapted with the times and are very power efficient. The working voltages have dropped meaning the Nutube can work on as little as 5VDC, right up to 80VDC — although we wouldn’t recommend going so high as creating that kind of voltage is tricky, dangerous and comes with a risk of shock. 12V would be a much more reasonable voltage to work with.
At this point, we should have a quick look at the design considerations/pros and cons of the Nutube.
The Nutube has the following benefits:
- Low power
- Safe operating voltages
- High reliability
It also has the following problems that must be considered: remember it’s a VFD, not a power amp!
- Low gain
- High source impedance
Comparing the KORG Nutube to a traditional triode like the 12AX7 its gain is pretty small; the 12AX7 typically has a gain of 100, whereas the Nutube has only around 5. This will limit the gain we have in a single Nutube to around 25 maximum (assuming no other factors and using both stages). This is likely enough for most purposes and will be fine for our pedal application.
High source impedance
In addition to the low gain, the plate (anode) source impedance is around 300kOhms. This means that there is very little current available from the amplifier (e.g. 30-40uA level for 12V plate).
In comparison, the 12AX7 would be 62.5kOhms but with 250V plate voltage, leaving the drive around the mA level (at least 100 times more than the Nutube). Also since power = volts * current there power increase is a lot more than the Nutube can manage.
This makes the device unsuitable for most output drive stages without some form of buffer.
Now we have the fundamentals of tubes under our belts we need a basic amplifier circuit. Below is a typical triode circuit for a device such as the 12AX7, which would be a good starting point for our Nutube amplifier.
In the case of the Nutube, we have both ends of the cathode to connect. The cathode voltage should be 0.7V. Luckily for us in the Nutube datasheet, there is a reference schematic and it shows the cathode voltage being generated by connecting it to 3.3V via 150R. This means that the cathode is taking 17.3mA, which via Ohms law gives us a cathode resistance of around 40Ohms.
With this information, we can either leave the circuit as per the reference or create our own cathode driver. The default 300k anode resistor is probably a good start, but we’ll look into further this as we progress with the design.
Finally, unlike the circuit above, the grid needs a bias and so placing a pot to 3v3 will allow us to bias the tube accordingly. Consider this pot as a variable bias voltage. From this, it’s a simple task of breadboarding up the design for testing. Note that the bias current is quite high so this circuit must be a low(ish) impedance.
Having had a good look at both traditional tubes and the Nutube, we are in a good position to start looking at the amplifier design and other considerations — even building up a test circuit. If you are following along now would be a good time to get the scope out and look at the tube characteristics for a given input. At this point, it’s a good time to say the dreaded words: to be continued…. Next time we’ll look at this basic circuit and what we need to do to turn this into a practical amplifier.
See you next time.
CommentsAdd a comment
November 20, 2017 15:22
I cannot help suspecting there is a lot of voodoo nonsense around valve amplifiers and vinyl recording ("angry electrons" indeed!). Is it not likely that the characteristic imperfections of these technologies simply trigger nostalgia? (Although personally, I prefer my music without hisses and pops.) I wonder has anyone tried measuring the transfer functions and noise characteristics and implementing them in a DSP? I bet that if you did that the audio snobs would be unable to tell the difference in a double-blind test! The feature could then simply be added at no cost as an optional software filter to any digital audio player.
I get the "steampunk" appeal of a traditional radio.
November 14, 2017 08:32
Hi Karl, your basic guide to the transistor is stretching the truth somewhat. Transistors are ALSO voltage operated devices, BUT base current flows as a result of that voltage, beta of hFE vary wildly with many operating parameters but the voltage operation is much more predictable over a large range of parameters and will allow you to realise much higher gain if you design for this. Adding a base resistor will help stop parasitic oscillations, but for an amplifier circuit will almost always add distortion. (So for a guitar amp that might be a plus...) As far as I know circuit simulators based on the SPICE softare use the voltage operation to perform simulations and use the Ebbers Moll equation and perhaps more modern derivatives rather than just the beta value. Yes, I accept that there are those circuits whose design and operation are purely based on beta, but that is a design choice rather than a fundamental facet of transistor operation. Ok, that said, these new valves look really cool :) Thanks for the article!
November 6, 2017 07:46
I'm very excited about this! I've experimented with other low-voltage tubes (Raytheon JAN-6418 and JAN-5636), with less than satisfactory results.