NuTube Overdrive - Spice simulations

Disclaimer

This is just an amateur pedal with no intention for profit or to infringe any trademark.

Tube Screamer is a trademark of Hoshino Gakki Co.

Nutube is a trademark of KORG INC.

This pedal is not for sale. If you want to buy a similar pedal I recommend you Ibanez NTS Screamer.

NuTube Overdrive Schematics

NuTube Overdrive is basically a Tube Screamer TS808 classic circuit (with a couple of mods I will explain later) where two NuTube triodes amplifier stages have been added between the preamp and the tone circuit, the original Tube Screamer output of the preamp is mixed with the output of the two triode amp sections by means of a 100K linear potentiometer, the buffered output of the mix is connected to the original tone circuit.

The following figure shows the LTSpice schematics of the NuTube Screamer pedal:

NuTube Screamer LTSpice schematics

NuTube Screamer Mods

The following mods have been made to the original schematics:

1. Bass boost mod: 220nF capacitor added in parallel to 47nF on the high pass filter connected to the preamp section inverting input. This mod can be remove by just removing a series 0R resistor
2. Tone mod: 20K tone potentiometer has been replaced by 5K linear potentiometer providing a more gradual response of the tone potentiometer.
3. Asymmetric clipping: The original pedal had symmetric clipping, here an additional diode has been added in series to the positive output cycle.
4. Output buffer: Resistor biasing has been slightly changed using 1000K resistor bridge from +9VDC instead of a 510K single resistor to +4.5VDC. It is basically the same.

Some components have been changed due to availability, performance or space gain:

1. Original diodes were silicon MA150, they have been changed by dual BAV199 silicon diodes for space gain, and easier asymmetrical circuit construction.
2. Original opamp JRC4558 has been replaced by LME49723 which has better performance in terms of harmonic distortion, noise, bandwidth, offset voltage and slew rate. See comparison here
3. Original transistors 2SC1815 used for buffers  have been replaced by MMBT5089 in a small SOT23 package

Frequency response

The following figure shows the gain response from 0 to 10 with tone at 5 and mix at 10, maximum gain is 56 dB and minimum gain is 36 dB:

NuTube Screamer Gain = 0 to 10, Tone = 5, Mix = 10

The following figure shows tone response from 0 to 10 with gain at 10 and mix at 10:

NuTube Screamer Tone = 0 to 10, Gain = 10, Mix = 10

 The following figure shows Mix response from 0 to 10 with gain at 10 and tone at 5, maximum gain is 56 dB and minimum gain is 38 dB:

NuTube Screamer Mix = 0 to 10, Gain 10, Tone = 5

The total gain of NuTube triode amps section is 17.7dB, 8.2 dB for the first section, almost flat from 10 Hz to 20KHz (82mdB loss at 20 kHz) and 9.5dB for the second section (with 141dB loss at 20kHz)

Time response

Time response shows the pedal output signal with different settings with a 60 mVp-p 30s-1 decaying 440Hz sinewave input signal. Clipping is always on.

The following figure shows Gain response from 0 to 10 with Tone at 5 and Mix at 5:

NuTube Screamer time response with Gain = 0 to 10, Mix = 10, Tone = 5

The following figure shows Tone response from 0 to 10 with Gain at 10 and Mix at 5:

NuTube Screamer time response with Tone = 0 to 10, Mix = 5, Gain = 10

The following figure shows Mix response from 0 to 10 with Gain at 10 and Tone at 5:

NuTube Screamer time response with Mix = 0 to 10, Tone = 5, Gain = 10

Frequency response at +9V and +18V

The same circuit could be powered at +18V, to increase around 10dB the gain, here are the simulations at +9V and +18V

NuTube Screamer Freq response at +9V and +18V with Tone = 0 to 10, Gain = 10, Mix = 10

Dirty Little Secret MOSFET 3.3V version: LTSpice analysis

Why JFET and not MOSFET?

JFET is a rare bird these days, you just have to go to a distributor and check the number of references of each of them, for example at Digikey there are less than 900 references of JFET devices and more than 40K of MOSFET devices. Transistor curves are quite similar, but it's much easier to find a proper MOSFET device that works well at 3.3V, N-JFET devices are biased with Vgs<0, while MOSFET devices are biased with Vgs>0. There are lots of MOSFET devices called "logic level gate" with low Vgs biasing values that are easy to bias from 3.3V power, while the offer of JFET devices that can work at 3.3V is quite limited, since as we saw in a previous post, JFET devices with low pinch-off voltages, and close to 0 Vgs have to be chosen.

These are the typical curves Id vs Vds for different Vgs values of Fairchild  FDV301VN:

The basic MOSFET amplifier only has one transistor compared to the 2 used on JFET circuit, because it is easier to bias the device:

The small signal frequency response shows a maximum gain of 37.5dB at 1kHz and 37.4dB at 10Hz for bias potentiometer set at 0.3

Dirty Little Secret 3.3V MOSFET version

The figure below shows schematics of the DLS MOS 3.3V circuit using Fairchild FDV301N:

The figure below shows the time response for a decaying 600mVpp 440kHz sinewave and gain from 0 to 1 in 0.1 steps:

The figure below shows the frequency response with gain from 0 to 1 in 0.1 steps and tone controls set with bass=0.3, mid=0.6 and treble=0.6:

The figure below shows the frequency response with gain=0.5, bass=0 to 1 (0.1 steps), mid=0.6 and treble=0.6:

The figure below shows the frequency response with gain=0.5, bass=0.3, mid=0 to 1 (0.1 steps),  and treble=0.6:

The figure below shows the frequency response with gain=0.5, bass=0.3, mid=0.6 and treble=0 to 1 (0.1 steps):

Audio test samples

Listen to the audio generated by the LTSpice simulations:

Dirty Little Secret 3v3 clone: LTSpice analysis

The Catalinbread Dirty Little Secret is an overdrive/distortion pedal that tries to emulate super-lead o super-bass Marshall amplifiers using JFET transistors instead of tube valves.
The tone filter section is identical to the one used on Marshall amplifiers.

Catalinbread Dirty Little Secret overdrive/distortion pedal

Starting with the Klon Centaur boost/overdrive pedal I am trying to show that guitar pedals circuitry can be powered at lower voltages than the standard 9V, using the appropriate components and circuits.
When I started Spice simulations the task seemed quite challenging, it was quite difficult to make the JFETs work at 3.3V and even when they worked at 3.3V it was quite difficult to properly bias the JFET to avoid distortion.

The secret to make a JFET work at low voltages is to find JFETs with low pinch-off voltages. The pinch-off voltage is defined by the dotted line on the left of the figure above, this line represents the boundary between the ohmic region and the active (or saturation) region. A JFET transistor should work in the active region. The problem is that the selection tools that offer distributors like Digikey or Mouser to choose a transistor don't provide the pinch-off voltages, they usually provide the Vgs cut-off (Vgs at which the JFET start conducting) and the drain current at Vgs=0. I finally find a JFET manufacturer specialist (InterFET) that provides these pinch-off voltages.

The JFET amplifier selection guide offers audio and general purpose tables. In these tables we can find that the tow N-channel JFETs models that have lower pinch-off voltages (Vp) are the J201 and 2N4117 (or MMBFJ201 and MMBF4117 in their SMD version). 2N4117 has Vp min = -0.6V and Vp max = -1.8V . The J201 has Vp min =  -0.30V and Vp max = -1.5V.

Apparently the MMBFJ201 should be more suitable for its lower pinch-off voltage, but as we will see later, the MMBF4117 has lower drain currents which are more suited for working at lower voltages and higher resistor values.

JFET characteristic curves

The first thing is to find Spice models for those JFET transistors, actually I found different models that showed different behaviors during simulations.

Then I traced the JFET characteristics, Id vs Vds (from 0 to 3.3V) at different Vgs values (-1.65 to 0V at steps of 0.1v) using the following circuit:

The following figure shows the Id vs Vds curve for mmbf4117:

Maximum pinch-off voltage is 0.97V at Vgs=0V and Id= 70uA

The following figure shows the Id vs Vds curve for mmbfj201:

The pinch-off voltage is already as high as -2.3V at Vgs=-1.65V, For Vgs=-075V the pinch-off voltage is 3.3V. As we can see, this device is not suitable to be used at 3.3V, the drain currents are quite high for very negative Vgs voltages (9mA at Vgs=-1.65V) so it cannot be properly biased, and the pinch-off voltages are very high making impossible to work in the active region with them.
These curves from a Fairchild Semiconductor J201 Spice model do not correspond with the values announced in the InterFET page but they actually correspond with the curves found on the datasheet, so I think that the InterFET J201 model it's quite different form the Fairchild Semiconductor J201 device that it's currently found through distributors.

I actually found a J201 Spice model that had very good characteristics corresponding to those announced by InterFET:

with a maximum pinch-off voltage of -0.54V at Vgs=0, and Id=600uA.

But as the device that is readily available is the Fairchild one, MMBFJ201 device had to be discarded.

JFET biasing

In order to better understand why low pinch-off voltage is important at low power voltages, let's analyse a simple circuit that shows JFET biasing and voltage margin.

In order to have maximum voltage margin, the JFET has to be biased in a way that Vds is set at half the power supply voltage, so if power supply voltage is 3.3V, Vds has to be set at 1,65V. In the following circuit and for a given drain resistance Rd, we must find the biasing value of Vgs at which Vds = 1.65V. At Id=0, Vds=3.3V for any value of Rd, but we must find an Rd value that at low values of Vds still keeps in the active region to avoid saturation, hence the lower the pinch-off voltage, the easier to find a low enough Rd that cuts the JFET curves at low voltages in the active region. The slope of the straight line is inversely proportional to Rd, we could always find a value of Rd that is high enough to keep in the active region, but if Rd is too high, currents are too low, our circuit becomes too sensitive to noise and it cannot accept any load.

The red curve below shows the drain current Id for different values of Vgs from 0V to -1V in steps of 100mV. The green lines represent the current through Rd vs Vds, the higher Rd value the lower the slope. To find the proper biasing point, we find a point where a green line crosses a red line at approximately 1.65V, but where the green line crosses a red line in the ohmic region at quite low Vds voltage. For Rd=250k we find a biasing point at Vds=1.65 and Id=6.6uA, for Vgs= -700mV

Let's do a time response with a 300mV sine 1kHz input and biasing DC at -0.7V, Rd=250K and output resistance at 1Mohm.

The output voltage has been amplified to 2.78Vpp

Another important characteristic is that the biasing Vgs voltage is as close as 0V as possible, N-channel JFETs are biased at negative voltages, if Vgs is very negative it will be difficult or even impossible to bias it. The way to bias Vgs to a negative voltage is adding a source to ground resistor (Rs) that will keep Vs higher than Vg. But this resistance Rs will add to Vds affecting the output voltage, unless that it is coupled with a capacitor in parallel so that Rs is only seen at DC biasing but small signal sees a short to GND through the capacitor.

A Rs=120k will bias Vgs to 700mV when Id=6uA.

We obtain a similar time response with a 300mV sine 1kHz input but no input bias, with Rd=250K, output resistance of 1Mohm and Rs=120k in parallel with 1uF.

Two JFET amplifier biasing

Let's analyse the biasing of the two JFET amplifier circuit that will be the basis of DLS pedal.
The same biasing current will circulate through both JFETs, which means that Vgs bias will have to be identical for both, then the voltage on the gate of the top JFET will have to be Vin+Vout, where Vin is the biasing voltage of the bottom JFET.

The obtained curve  is shown below. The biasing point will be selected from a point where blue and cyan curves cross and the current of both JFETs is identical as close as possible to Vds=1.65V (power supply mid-point) The selected blue curve will give us the Vgs biasing point (160mV)
Vds=1.65V, Id = 94uA and Vgs=158mV.

The small signal frequency response shows a maximum gain of 34.5dB at 1kHz and 32.7dB at 10Hz for bias potentiometer set at 0.4.

The time response of a decaying 1kHz sinewave 600mVpp shows a  maximum output voltage of 2.26Vpp (gain of 3.8 = 11.5dB)

Dirty Little Secret 3V3

The figure below shows schematics of the DLS 3.3V circuit using MMBF4117 JFET:

The figure below shows the time response for a decaying 600mVpp 440kHz sinewave and gain from 0 to 1 in 0.1 steps:

The figure below shows the frequency response with gain from 0 to 1 in 0.1 steps and tone controls set with bass=0.3, mid=0.6 and treble=0.6: 

The figure below shows the frequency response with gain=0.5, bass=0 to 1 (0.1 steps), mid=0.6 and treble=0.6: 

The figure below shows the frequency response with gain=0.5, bass=0.3, mid=0 to 1 (0.1 steps),  and treble=0.6:

The figure below shows the frequency response with gain=0.5, bass=0.3, mid=0.6 and treble=0 to 1 (0.1 steps):

In order to be able to compare below are the frequency response with gain from 0 to 1 in 0.1 steps and tone controls set with bass=0.3, mid=0.6 and treble=0.6 for the original DLS circuit powered at 9V using MPF4393 JFET transitor. Almost no difference using a 2N5457 model:

And the time response for a decaying 600mVpp 440kHz sinewave and gain from 0 to 1 in 0.1 steps:

Klon Centaur 3v3 Clone: LTspice analysis

Klon Centaur is one of the most famous, controversial, loved and talk-about guitar pedals and also one of the most cloned and copied ones, probably because it's quite expensive. Basically it's a Boost / Overdrive pedal with a gain, a treble and a volume control. It's subtly mixes the original signal with an amplified version of the signal that also goes through a germanium clipping section, before everything gets again amplified and filtered by the tone section that can considerably reinforce the mids.

Klon Centaur guitar boost/overdrive pedal

The first dual opamp section is powered between GND and 9V, but the second dual opamp section is powered between -9V and 18V to have more headroom before the opamps saturate. The circuit has a lot of gain and the signal that can get into the guitar power amp is huge, to breakout the guitar amp preamp section.

The main reason for making this project is to verify this idea I will expose here:

The output of most guitar pick-ups is usually several hundred millivolts peak-to-peak, 1Vpp is already a loud signal for a guitar amp. At the same time, guitar amps and specially valve amps are biased in a way that 2Vpp already saturates the input.Why then a guitar pedal should need more than 2Vpp or 3Vpp of headroom since those levels would be already enough to saturate the input and "breaking-up" the amp?

On the other hand, "breaking-up" or saturating the signal inside the pedal gives more control than "breaking-up" the guitar amp, because each model of guitar amp may have different input headroom and hence saturate earlier or later than others.

9V battery voltage is the "standard power" on guitar pedals, but that does not mean that the electronic circuits inside the pedal must be powered at 9V, they could be powered at a lower voltage and still have enough headroom for feeding the guitar amp.

Nowadays there exist very good audio opamps with very low harmonic distortion that can be powered at lower voltages and have rail-to-rail inputs and outputs.

If an efficient step-down converter is used, battery charge could be maximized and guitar pedal could have the same performance even at the end of the life of the battery. At a lower power we also could expect less power consumption and longer battery life. There exist cheap, efficient and stable step-down converters with a small footprint, they just need a small inductor, input and output capacitor and a couple of resistors to set the output voltage. Switching frequencies can be very high, so that any ripple in the power supply will not be audible, and additional LC filtering can be added to minimize ripple and noise.

With all that in mind I decided to do a Klon Centaur "3.3V clone" pedal.

Klon Centaur LTSpice analysis

I will start by doing a LTSpice simulation of the original circuit, and then I will try to implement a 3.3V circuit that tries to match the response of the original circuit.

Klon Centaur LTspice schematics

Klon Centaur Frequency response. Gain =0-10. Treble = 10

First we do a frequency response analysis between 10 Hz and 20kHz with treble set to 10 and varying gain from 0 to 10. Klon Centaur produces a high gain to small signal from 47dB @ 810Hz up to 75dB @ 674Hz reinforcing mids with peaks around 670 - 810Hz. Gains are much smaller at low and high frequencies.

Klon Centaur Frequency response. Gain =10. Treble = 0 - 10

Then we do a frequency response analysis between 10 Hz and 20kHz with gain set at 10 and treble varying from 0 to 10. Klon Centaur amplifies the signal at mids with a peak around 650Hz from 52 dB to 75 dB.

Klon Centaur Time response. Gain = 0-10. Treble = 10

Now we do a time response analysis during 25ms (only 6ms are shown) with gain between 0 to 10 and treble set to 10 using four 300mVpp sinewaves at different frequencies: 82Hz, 440Hz, 2kHz and 4kHz. The output obtained is a large amplitude signal more than 24Vpp that saturates the op-amps between -12 and 12V.

Klon Centaur Time response. Gain = 10. Treble = 0-10

Finally we do a time response analysis during 25ms (only 6ms are shown) with gain set to 10 and treble between 0 and 10 using the same four 300mVpp sinewaves. The output obtained only saturates the op-amps for high treble values.

Klone Centaur 3V3 clone LTSpice analysis

The Klone Centaur circuit is adapted to 3.3V using 3.3V rail-to-rail op-amps and passive values have been changed to provide a similar level of saturation, which means lower level of gain.

Klon Centaur 3V3 clone LTspice schematics

Klon 3V3 clone Frequency response. Gain =0-10. Treble = 10

Again we do a frequency response analysis between 10 Hz and 20kHz with treble set to 10 and varying gain from 0 to 10. Klon 3V3 clone produces a high gain to small signal from 20dB @ 1kHz up to 39dB @ 970Hz reinforcing mids with peaks around 970 - 1kHz. Gains are much smaller at low and high frequencies.

Klon 3V3 clone Frequency response. Gain =10. Treble = 0-10

Then we do a frequency response analysis between 10 Hz and 20kHz with gain set at 10 and treble varying from 0 to 10. Klon 3V3 clone amplifies the signal at mids with a peak around between 735 Hz to 970Hz from 21 dB to 39 dB.

Klon 3V3 clone time response. Gain = 0-10. Treble = 10

Now we do a time response analysis during 25ms (only 6ms are shown) with gain between 0 to 10 and treble set to 10 using four 300mVpp sinewaves at different frequencies: 82Hz, 440Hz, 2kHz and 4kHz. The output obtained is a large amplitude signal, but in this case op-amps saturate between -1.65 and 1.65V for a maximum signal of 3.3Vpp.

Klon 3V3 clone time response. Gain = 10. Treble = 0-10

Finally we do a time response analysis during 25ms (only 6ms are shown) with gain set to 10 and treble between 0 and 10 using the same four 300mVpp sinewaves. The output obtained only saturates the op-amps for high treble values but again maximum signal level is 3.3Vpp.

As we can see, frequency response gain has been reduced from a maximum of 75dB to 39dB, 36dB less in order to obtain a similar level of signal saturation in the time domain. The signal that will get to the guitar amp will be different in both pedals, specially at high gains, Klone Centaur pedal provides large signals of 24Vpp, while the 3v3 clone will provide 3.3Vpp signals at most. Will it sound very different? The only way to really know it is doing a real implementation of the effect.
Stay tuned for next post on effect practical implementation.