Friday, 9 June 2017

Tuesday, 6 June 2017

Korg Nutube 6P1 vs 12AU7 tube: Hybrid Amplifier

My next project is an Hybrid guitar amplifier: Triode preamp + Class D 25W amplifier.
The preamp section can be populated with a Korg Nutube 6P1 triode
 or alternatively a 12AU7 tube
The amp is powered by an external AC/DC 24VDC out wall adapter.
It includes headphones, line out and 8 Ω speaker with 25W output.
Everyting is packed on a compact 1590J enclosure:


Friday, 26 May 2017

Korg NuTube 6P1 vs 12AX7 starved tube / valve: Gain and frequency response

The following figure shows the Schematics used to compare the frequency response of NuTube 6P1 vs 12AX7 in starved mode:
12AX7 has a 1M pulldown on grid input while NuTube 6P1 has a bias circuit to adjust the bias voltage on the grid between 0V and 3.3V, connected to the grid via 33K series resistor, the typical grid current of 6uA has been added, since the model does not include this bias current.

Anode/plate output resistor load is 500K in both circuits.

Both circuits have anode/plate connected to +24V via a series resistor and a potentiometer to adjust load.
NuTube 6P1 has resistor values multiplied by 10 to be able to sweep around the maximum gain point.

The following figure shows the starved 12AX7 frequency response between 10 Hz and 20 kHz with anode load potentiometer varying from 0 to 1. A maximum gain of 24.6 dB is obtained with a load resistance of 50K

The following figure shows the NuTube 6P1 frequency response between 10 Hz and 20 kHz with anode load potentiometer varying from 0 to 1. A maximum gain of 15.9 dB is obtained with a load resistance of 400K.
NuTube 6P1 application note shows a gain of 14dB with anode powered at +12V and 17 dB with anode powered at +30V, which actually corresponds to simulations.


Bypass capacitors are also compared between 10nF and 10uF. NuTube 6P1 bypass capacitor value must be multiplied by 15 in order to have a similar response at lower frequencies:

Thursday, 25 May 2017

Korg NuTube 6P1 vs 12AX7 starved tube / valve: SPICE models

I purchased some KORG NuTube 6P1 triodes samples and I wanted to build a guitar preamp circuit to get the characteristic tube sound and distortion.

I could not find any guitar preamp amplifier schematics that would suit me so I decided to make my own circuit based on the reference circuit and add some gain, tone stack and volume controls

I am more of a "SPICE simulation" kind of guy than a "bread-boarding" guy, also because I like to use SMD components and try to be closer to the final guitar pedal than just a prototype. I have quite some confidence in SPICE results provided that models are accurate.

I wanted to simulate my NuTube circuit before building a guitar pedal or amp. The problem is that there were no NuTube SPICE models available.

I found that Koren had obtained a method for obtaining tube SPICE parameters from datasheet curves plate current (Ip) vs plate-cathode voltage (Vpk) for variable Vgk (grid-cathode voltage). Actually the models I have from traditional triodes, all come from using this method.

I downloaded Koren's MATLAB program and followed his instructions for Finding SPICE tube model parameters.

The task showed to be more difficult than I thought, convergence of the method did not work very well, and I had a lot of tuning to make.

These are the current-voltage curves from the datasheet:

The following figure shows the curves I obtained in MATLAB using Koren's approximation method with the points taken from NuTube datasheet:

There is some dispersion in the curves from the datasheet points but that was the best I could get.

This is the NuTube 6P1 SPICE model I got:

.SUBCKT NU6P1_l  1 2 3 ; P G C (Triode)
X1 1 2 3 TRIODE MU= 18.10  EX= 4.080  KG1=4270851.9  KP=451.94 KVB=   4.2  VCT=  0.00  RGI=330k CCG=9.1P CGP=2.5P CCP=4.3P
* http://www.nutube.us/downloads/Nutube_Datasheet_31.pdf  13-May-2017
.ENDS

And this is Koren's 12AX7 model:

.SUBCKT 12AX7 1 2 3 ; A G C (Triode)  OLD MODEL AKA ECC83
* Original Koren Model
X1 1 2 3 TRIODE MU=100 EX=1.4 KG1=1060 KP=600 KVB=300 VCT=0.00 RGI=2000 CCG=2.3P CGP=2.4P CCP=.9P ; ADD .7PF TO ADJACENT PINS; .5 TO OTHERS.
.ENDS 12AX7

Mu is 18, much lower than 12AX7 with a value around 100.
The exponent EX is quite high, around 4, compared to 1.4 on the 12AX7
A high KG1 value corresponds to a low plate current.
KP which is used in the high plate voltage region is 452 compared to 600.
KVB knee voltage is 4.2, much lower than 12AX7 value of 300.
RGI, CCG, CGP and CCP are obtained from the Datasheet.

Anyway, I had my NuTube 6P1 SPICE model and I could start designing and simulating my guitar preamp.

Then the user Teemuk from DIYstompboxes forum posted an interesting comment on this thread, where he believed that KORG NuTube 6P1 response was not much different from a traditional tube (like 12AX7 or 12AZ7) in "starved" mode, that is, with a low plate/anode voltage.

There are several advantages praised by KORG about this NuTube: smaller size, higher reliability and low power voltage.

But if a similar response is obtained from a traditional "starved" tube powered at the same low voltages, one of the most important advantages does not exist anymore.

So I decided to try to make a comparison between them using LTSpice.

Unfortunately, according to some reports, it appears that existing SPICE models do not work well in the starved region, these are approximated models, and the starved region is just a tiny area in the curves that go up to 400V.

But I knew how to get SPICE parameters models from curves, so if I could find accurate current/voltage curves in the starved region I could make a model suited for those low voltages.

I then found this study on the net:
Improved vacuum tube models for SPICE simulations
that showed these curves on the starved region obtained experimentally:

Again, using the Koren's method I got the following current-voltage approximated curves for 12AX7 triode:

And again, there is some dispersion in the curves from the experimental points. The worst dispersion happens at the knee at Va=1V for Vgk=0V, where plate current is around 60uA instead of 100uA.

This is the obtained 12AX7 starved model:
.SUBCKT 12AX7_l  1 2 3 ; P G C (Triode)
X1 1 2 3 TRIODE MU= 81.48  EX=0.626  KG1=1865.5  KP=248.15 KVB=300.0  VCT=0.00 RGI=2000 CCG=2.3p CGP=2.4p CCP=0.9p ;
* http://www.valvewizard.co.uk/Triodes_at_low_voltages_Blencowe.pdf  24-May-2017
.ENDS

Mu is lower, Exponent EX is lower, KG1 is higher (lower plate current), KP is lower and KVB is the same.

These same curves can now be simulated using LTSpice. These are the schematics:

Starved 12AX7 plate current is then drawn with plate-cathode Vpk voltage (V2) varying from 0V to 24V in 0.1V steps, using grid-cathode voltage Vgk (V1) as a parameter varying from -0.5V to 0V in 0.1V steps:
The following figures show the current-voltage curves obtained from LTSpice compared to scaled values from datasheet:






Friday, 19 May 2017

Korg Nutube Triodes for Guitar Preamp

Just received in the post two samples of Korg NuTube 6P1 triodes based on VFD technology, with a traditional valve response but much lower voltages.
I am planning to make a guitar preamp.
More info here:
http://nutube.us/  (US site)


Monday, 5 December 2016

Dirty Little Secret: Assembly 2

DLS Assembly

Figure below shows a top view of the effect baseboard with 5 potentiometers: Treble, Mid, Bass, Gain, Volume, LED and toggle switch; 3PDT footswitch connected to baseboard with a 6-wire flat cable and 9V battery. 
The 3PDT toggle switch is oversized and it's much taller than potentiometers, in the following version I will probably try to replace it with a lower profile 3PDT slide switch
The enclosure is a 1590B, it is too narrow to properly fit audio jacks and DC jack on the top side, I discovered that a 1550B enclosure is probably more suited to fit the baseboard. I had to file the corners in order to fit the baseboard closer to the top side but still there is a gap.
DLS top side view with enclosure. Baseboard, 3PDT footswitch, 9V battery and 1590B enclosure
 The figure below shows the bottom side with the JFET mezzanine board mounted on baseboard. I had to make some slots in order to plug the mezzanine because of the audio jacks were too deep.
DLS bottom side view with JFET mezzanine mounted 
 The figure below shows the baseboard bottom side with audio jacks (stereo for input and mono for output). FB1 and FB2 are input and output ferrite filters, CF1 is an EMI 9V power supply input filter, D3 is a reverse 9V voltage input protection, U3 is the 9V to 3.3V DC-DC regulator, L1 is switching inductor and L2 is an additional inductor of an LC filter. U4 is a mid-point 1.65V reference voltage. Some resistors and capacitors are part of a Marshall type equalizer filter with treble, mid and bass potentiometers soldered on the opposite side. JP2 is the 2-wire 9V battery connection and JP1 is the 6-wire connection to the true bypass 3PDT footswitch using a 0.1'' pitch flat cable. The two board-to-board connector are used to plug in the effect mezzanine board.
DLS base board bottom side view
The figure below shows the DLS JFET mezzanine board with 3 one turn potentiometers for top JFETs biasing adjust. The DLS JFET effect contains 3 amplifier stages with 2 JFETs each (MMBF4117)
DLS MOSFET mezzanine board
The figure below shows the DLS MOSFET mezzanine board with 3 one turn potentiometers for MOSFET biasing adjust. The DLS MOSFET effect contains 2 amplifier stages and one output buffer stage (Fairchild FDV301VN or Diodes DMG301NU-13)
DLS JFET mezzanine board
The figure below shows a view of the two DLS effect mezzanine boards: JFET on top and MOSFET on bottom.
DLS: JFET mezzanine on top, MOSFET mezzanine on bottom


Sunday, 20 November 2016

Dirty Little Secret: Schematics, PCB layout Assembly and Test

Schematics

In order to test both versions of the pedal, JFET and MOSFET, threee different PCBs are built:
  • Base board including power supply, equalizer, conectors, potentiometers, switches and LED
  • Mezzanine board with JFET DLS effect
  • Mezzanine board with MOSFET DLS effect
Schematics first page is the base board that includes input stereo jack with ferrite filter, gain potentiometer, equalizer including capacitors, resistors and potentiometers, volumen potentiometer, output mono jack with ferrite filter, 3PDT switch for super lead / super bass option, +9 VDC input to +3.3 VDC regulator, effect on LED, connector to external 3PDT footswitch, 2 connectors to mezzanine board and the layout for an optional +1.65V mid-point voltage reference that is not used on this particular effect.

This baseboard can be used as a generic base for other effects.
Baseboard Schematics
Second page schematics contains MOSFET DLS effect
Three MOSFET amplifier stages are used. MOSFET has a biasing resistor network that includes a potentiometer to adjust biasing voltage on the MOSFET gate.
Gain is placed between the first and second stage.
Equalizer is placed between the second and third stage.
Third stage is just a buffer with no gain.
MOSFET DLS effect schematics
Third page schematics contains JFET DLS effect.
Three dual JFET amplifier stages are used. Top JFET has a biasing resistor network that includes a potentiometer to adjust biasing voltage on the top JFET gate.
Gain is placed between the first and second stage.
Equalizer is placed between the second and third stage.
JFET DLS effect schematics
The figure below shows top layer PCB layout . A panel has been made containing the three PCBs: baseboard, MOSFET mezzanine and JFET mezzanine. V-cut separation method is used.

Top layer PCB layout
The figure below shows bottom layer PCB layout. A panel has been made containing the three PCBs: baseboard, MOSFET mezzanine and JFET mezzanine. V-cut separation method is used.
Bottom layer PCB layout

Test

Both effects (JFET and MOSFET) have been tested using a 300mVpp 440 Hz sinewave signal on input.

JFET effect

The figure below shows the signal observed at the output of the first amplifier stage (Q1D1). A 1090mVpp sinewave is obtained which corresponds to a 11dB gain.
440Hw sinewave 1090mVpp signal at the first amplifier stage output (gain=3.6, 11dB) (q1d1)
The figure below shows the signal observed at the output of the second amplifier stage (TONEIN2). A 2110mVpp distorted sinewave is obtained which corresponds to a 17dB gain.
2110mVpp signal at the 2nd amplifier stage output (gain=7, 17dB) (tonein2)
The figure below shows the signal observed at the output of the equalizer and input to the 3rd amplifier stage (Q6G). A 850mVpp distorted sinewave is obtained which corresponds to a 9dB gain.
850mVpp signal at the equalizer output (gain=2.8, 9dB) (q6g)
The figure below shows the signal observed at the output of the 3rd amplifier stage output (FXOUT2). A 1730mVpp distorted sinewave is obtained which corresponds to a 15.2dB gain.
1730mVpp signal at the third stage amplifier output (gain=5.8, 15.2dB) (fxout2)

MOSFET effect

Three different biasing values have been used: low, mid, high.
The figure below shows the signal observed at the output of the first amplifier stage (Q1D) with low bias. A 1730mVpp distorted sinewave is obtained which corresponds to a 15dB gain.
1730 mVpp signal at the first amplifier stage output with low bias  (gain=5.8, 15dB) (q1d)
The figure below shows the signal observed at the output of the first amplifier stage (Q1D) with mid bias. A 3060mVpp distorted sinewave is obtained which corresponds to a 20dB gain.
3060 mVpp signal at the first amplifier stage output with mid bias  (gain=10.2, 20dB) (q1d)
The figure below shows the signal observed at the output of the first amplifier stage (Q1D) with high bias. A 700mVpp distorted sinewave is obtained which corresponds to a 7.4dB gain.
700 mVpp signal at the first amplifier stage output with high bias  (gain=2.3, 7.4dB) (q1d)
The figure below shows the signal observed at the output of the second amplifier stage (Q2D) with low bias. A 970mVpp distorted sinewave is obtained which corresponds to a 10.2dB gain.
970 mVpp signal at the second amplifier stage output with low bias  (gain=3.23, 10.2dB) (q2d)
The figure below shows the signal observed at the output of the second amplifier stage (Q2D) with mid bias. A 3220mVpp distorted sinewave is obtained which corresponds to a 20.6dB gain.
3220 mVpp signal at the second amplifier stage output with mid bias  (gain=10.7, 20.6dB) (q2d)
The figure below shows the signal observed at the output of the second amplifier stage (Q2D) with high bias. A 940mVpp distorted sinewave is obtained which corresponds to a 10dB gain.
940 mVpp signal at the second amplifier stage output with high bias  (gain=3.1, 10dB) (q2d)
The figure below shows the signal observed at the final buffer output (FXOUT1) with low bias. A 630mVpp distorted sinewave is obtained which corresponds to a 6.4dB gain.
630 mVpp signal at the final buffer output with low bias  (gain=2.1, 6.4dB) (fxout1)
The figure below shows the signal observed at the final buffer output (FXOUT1) with mid bias. A 2170mVpp distorted sinewave is obtained which corresponds to a 17.2dB gain.
2170 mVpp signal at the final buffer output with mid bias  (gain=7.2, 17.2dB) (fxout1)
The figure below shows the signal observed at the final buffer output (FXOUT1) with high bias. A 330mVpp distorted sinewave is obtained which corresponds to a 0.83dB gain
330 mVpp signal at the final buffer output with high bias  (gain=1.1, 0.8dB) (fxout1)

Assembly

DLS guitar pedal has been mounted on a 1550B enclosure
The figure below shows a bottom view of the Hammond 1550B enclosure open without lid showing the DLS guitar pedal with audio jacks and DC jack, the 3PDT footswitch, the 9V battery


The figure below shows a top view of the Hammond 1550B enclosure showing audio jacks, DC jack, potentiometers, LED, toggle switch and the 3PDT footswitch: