(Update 23/08/2015: Source files on Github)
I am a big fan of shoegaze / noise rock bands and one of my favourites bands is Jesus & Mary Chain and their album Psychocandy. That album is well known for its unique and characteristic dense wall of sound with that white noise (switch-on vacuum cleaner, washing machine and all appliances and let them couple to the guitar amp) that served as inspiration for a coming shoegaze generation of bands: Ride, My Bloody Valentine, Cocteau Twins, Lush, Spacemen 3, Loop... and even today shoegaze bands like A Place To Bury Strangers, Skywave, Ceremony, Screen Vinyl Image or 93MillionMilesFromTheSun, The KVB, The Lost Rivers, The Soft Moon...
It seems that one of the keys of that sound is the use of a Shin Ei Fuzz-Wah (8 transistors) pedal. This is a two/three effects into one pedal, a 6 transistor fuzz with octaver (high octave) plus a 2 transistor Wah effect at the end. JAMC did not make lots of use of the Wah pedal as it is normally used, but they used it to reinforce the gain of a band in the mid tones to couple the sound, so they set the pedal at one position and didn't move it.
These are the schematics of the original Shin Ei Fuzz-Wah pedal:
The fuzz with octaver is also used on the Univox/Unicord Super Fuzz (6 transistors) pedal. See schematics here:
Find here a demo video of the Shin Ei Fuzz-Wah pedal:
With all that in mind I made a first version of the Super Fuzz with Octaver pedal with scrapped through-hole conventional components on a Vero board using 2N2219A transistors
The first version on the Super Fuzz pedal can be seen here on the left of the picture:
The result was quite deceiving, the circuit has lots of gain and it was too noisy, too much, even for playing a PsychoCandy cover, specially when powered at 9V. A higher DC power of 15V reduced a bit the noise but it was still an undesirable result.
I decided to restart the work using surface mount devices (SMD), a professional PCB and include the Wah sections as well as several improvements and modifications to the original circuit.
These are the schematics of the Super Fuzz-Wah design:
I used BC847C high gain (hfe = 520typ) SMD SOT-23 NPN transistors.
The improvements on the circuit are the following:
Replacing the two clipping germanium diodes by one dual BAT54S Schottky diode SOT-23 in series with 100 ohms (D1, R22).
Replacing the Wah bulky and expensive inductor by a gyrator circuit based on a transistor (Q9) plus capacitor and biasing resistors.
Doubling the 9V DC power supply, with a charge pump circuit based on Intersil ICL7660SCBA
The charge pump regulator ICL7660SCBA allows doubling the power supply from 9V to 18V, however this charge pump regulator uses an internal oscillator with a quite low and audible switching frequency of 10 kHz. If it is used in its default oscillation mode this frequency is seen as a ripple frequency on the power supply, since this pedal has lots of gain in its transistor circuits, this ripple noise can be amplified and heard as a very nasty and unpleasant high pitch noise. For that reason, it is absolutely required to short pin 1 (boost frequency pin) with pin 8 (V+) in order to get a higher switching frequency of 35 kHz, out of the audible spectrum.
Wah circuit simulation with LTSpice
(Update 23/08/2015: Source files on Github)
The following schematics shows both Wah circuits: to the right the original wah circuit based on an inductor, and to the left the circuit based on a gyrator circuit, where the inductor has been replaced by a transistor plus capacitor and biasing resistors. Basically the circuit framed by the square on the right side is replaced by the circuit framed by the square on the left.
This is the frequency response of the original circuit when varying the wah potentiometer:
It is a tuned band-pass filter with its peak moving from 240 Hz to 1.28 kHz, increasing the gain from 12.5 to 23.5 dB.
And this is the frequency response of the gyrator circuit when varying the potentiometer:
Again it is a tuned band-pass filter with its peak moving from 260Hz to 1.28 kHz, slightly increasing the gain from 16.5 dB to 19.2dB.
So we obtain a quite similar result with even a more stable gain peak.
PCB Layout
(Update 23/08/2015: Source files on Github)
The PCB was made on two layers with dimensions of 80 mm x 50 mm:
Schematics and PCB layout was designed using Eagle CAD. The PCB layout can be uploaded into Eurocircuits website, and a quote can be obtained immediately, I paid 70 € for 2 boards prototypes in a 7 day turnaround, the quality is really good.
I purchased the components at Mouser website. You can find here the BoM, total cost of components for one prototype was 26 €.
To solder the components I used a low temperature (138°C ) CR11 solder paste Sn42Bi58 and a hot air soldering station, the results are very professional and clean. This is the finished PCB with all components (except external switches, jacks and potentiometers) mounted.
All the cables, jacks, connectors, switches and potentiometers soldered and ready for debug and verification
I wanted to do a sound check of the TubeSim v1.0 VOX AC30 valve amplifier emulator and a comparative with other VOX AC30 amp emulators. This is it.
First I found on the web a sound file with a clean guitar riff:
Then I fed this sound file from the computer Line output into the TubeSim Line input. The TubeSim 25W 8 Ω output was connected into the Rezzonics v12 DIY Cabinet speaker with a Celestion Vintage 30 12''. The output sound was recorded with Audacity in the computer using an AKG D40S dynamic condenser microphone (it's a vocal microphone but that's the best I could get for the sound check)
This is the TubeSim v1.0 guitar riff sound with gain at maximum, no bass, no treble, tone cut off:
I think the sound has a nice crispy distortion, with lots of mids and lots of sustain.
The best would have been to compare it with a real VOX AC30 valve amplifier, but unfortunately I don't own one.
I decided to compare with several amp emulators.
First the open source amp emulator Poulin HyBrit Head v1.1 in the same conditions, maximum gain, no equalization:
The Poulin HyBrit Head v1.1 has a lot of gain too, it's more saturated quite brittle with more treble than the TubeSim v1.0
Now the GuitarRig 2 ACBox
GuitarRig 2 ACBox resembles more the TubeSim, less saturated than the Poulin with maybe more bass than the TubeSim v1.0
Amplitube 3 offers two different AC30-like amp emulators:
Amplitube 3 Hot AC30:
The Amplitube 3 Hot AC30 is less crispy, lots of bass, not much saturation, it's a bit muddy and more somber, less sustain. In my opinion is the worst sounding of all the amp emulators.
Amplitube 3 Hot AC30 Copper:
Amplitube 2 Hot AC30 Copper is a bit more saturated than the previous one, not much sustain, less muddy, more mids, similar to the Guitar Rig 2 ACBox but with less treble, less crispy than the Guitar Rig 2 and the TubeSim.
Then I coneccted the Line Output of the Tube Sim directly into the Line-In of the computer, first with the Speaker Simulator bypassed:
And now with the Speaker Simulator active:
The Speaker Simulator reinforces the middle tones as a guitar speaker would do.
Find here the playlist to listen to all the different sounds used in the comparison:
Which one is your favourite? Why?
In order to test the TubeSim 25W 8Ω output I decided to make a DIY 1x12'' cabinet speaker with a classic Celestion Vintage30 8Ω speaker installed.
I was inspired by the series of 4 magnificent and instructive videos made by thefirstbab. The first one of the series can be seen here:
I changed the dimensions of this cabinet based on a 1912 Marshall to provide nicer proportions to the cabinet. To do so, I used (approximately) the golden ratio 0.618 and came up with the following external dimensions: 600 x 400 x 300 mm(W x H x D) for a total internal volume of approximately 50 liters.
I used 18 mm (3/4'' approx.) thick plywood (from exotic okoumé wood) already precut acquired in a DIY store. The wood cleats were cut from two 22 mm x 2400 mm beech squares
The panels are fixed with wood glue and 3.5 x 30 mm wood screws. Wood filler paste was used to fill holes and imperfections. Find here below the cabinet mechanical drawings (AutoCAD was used but open source LibreCAD can also be used)
The first step is to glue and screw the bottom panel to both sides. To ease screwing, 2 mm holes are drilled. In order to hide the head screws, shorter 6 mm holes are drilled too, once the panels have been glued and screwed together, wood filler paste is applied to completely hide holes and head screws. Then the top panel is glued and screwed to both sides.
Four 22 mm x 564 mm thick wood cleats are placed on the internal top, bottom, front and back sides. Note that front cleats are fixed aligned to the front borders, while rear cleats are fixed leaving 20 mm space to the rear borders to allow adding external rear cover panels. Picture below shows the bottom wood cleats glued.
Two 22 mm x 320 mm wood cleats are glued and screwed to the front left and right sides. Shorter and beveled wood cleats 22 x 270/242.5 are fixed on the rear left and right sides. This shape allows introducing the front panel that will hold the speaker. The picture below shows the wood cleats glued, ready to be screwed. The 2 short beveled cleats resulting from cutting must be reserved to finish the rear frame once the front panel is installed.
Two beveled cleats 22 mm x 210/189 mm are fixed on the bottom sides, while shorter beveled cleats 22 mm x 107/85 mm are fixed on the top sides. This allows sliding the front cover from the rear top of the case, place the bottom side of the front cover panel onto the internal bottom front side and then pivot the front cover panel to place the top side on the internal top front side of the case as shown by the red traces of the mechanical drawing, see the two pictures below. To allow this pivotal movement, top side reinforcement beveled cleats must be shorter.
Three circles of Ø283 mm, Ø297 mm, Ø309 mm respectively are drawn in the center of the front cover as recommended by the speaker manufacturer Celestion for the Vintage 30 model. The position of the four speaker fixing screws are also drawn, they are 210 mm apart from closest screw, horizontally and vertically.
The external circle of Ø309 mm represents the maximum dimensions of the speaker.
The middle circle of Ø297 mm represents the placement of the four screws. Logically the relationship between the diameter of the circle and the sides of the square is respected: sqrt(2*210*210) = 297 mm.
The internal circle of Ø283 mm diameter represents the hole that must be left for the speaker sound to come out of the box. An electrical jigsaw has been used to make this hole and then the borders slightly softened with a wood file and sandpaper.
All external sides and corners must be rounded. I don't have a router, so I had to use a wood file. The radius of this rounding must correspond to the internal diameter of the corners used. Note that Marshall corners used in the first video have a larger non standard radius 1'' (25 mm) and rear sides are not rounded. Check this article Choosing amp corners for further info.
I purchased all cabinet and amp accessories in this great website: Tube-Town. I bought a set of 8 standard black corners 1/2'' (12.5 mm internal radius). In this case, all corners are identical and all the sides of the cabinet must be rounded, rear sides too.
I traced lines parallel to the sides at 6.3 mm and 12.5 mm (they can be seen in the first picture), with the wood file I made 45 degrees between the lines at 6.3 mm and then I finished the rounding up to the 12.5 mm lines, I cut a 12.5 mm semicircle in a thin plywood piece to be able to properly control the radius while filing, I used one of the corners to be sure that it adjusted properly to the corners of the cabinet.
I then filled in any irregularities with wood paste and sanded all the surfaces, see pictures below:
Next step is spray painting in matte black all the internal sides of the cabinet. Several layers will be needed for a good result:
Front cover is painted on both sides. Before painting it file the borders leaving at least 2 mm margin on each side and make sure that once the grill cloth is fixed on it there is enough place inside the cabinet to put the front panel in place. Speaker hole screws were previously drilled and hole borders were lowered in order to hide countersink screw heads. Head screws will have to be painted in black too, so they cannot be seen through the grill cloth. Screws will have to be put in place before fixing the grill cloth.
I wanted a vintage look for the Vox AC30 TubeSim cabinet with Celestion Vintage 30, so I opted for a golden vintage grill cloth, two meters of Red basket VOX type tolex and two meters of white medium piping to enhance and frame the grill cloth.
Before fixing the grill cloth, place the front panel in place and make the holes corresponding to the fixing screws. To fix the grill cloth to the front cover I used a cheap upholstery stapler. Place the four speaker screws, before fixing the grill cloth. Use a hairdryer to heat the grill cloth before stapling it so that when cold is as stiffened as a drum batter head. It's important to select the proper staples, if they are too short or too thick they may not be pushed to the end. This actually happened to me when I fixed the white piping to the beech cleats frame, since solid beech it's a very hard material and I had to finish with some hammer taps.
Use a soldering iron to melt the grill cloth over the holes that will be used to fix the front panel with screws to the cabinet, otherwise grill cloth can be damaged when screwing the front panel to the cabinet.
Gluing tolex is probably the most tricky part of this project, at least for me, since it was the first time. I recommend watching thefirstbab video series I mentioned at the beginning of this post and also the videos from Uncle Doug:
A neoprene glue spray is recommended for gluing the tolex to the cabinet. I drew the shape of the cabinet in the back of the tolex (see picture below), marking the place where the cabinet borders had to coincide and leaving 1 cm margins. The tolex is only one piece 2m long and the joint should be placed at the center bottom of the cabinet. Corners are not yet cut, this will be made afterwards to compensate for tolerance or fabric deformation. The tolex piece is rolled inside out to ease manipulation while gluing, exposing the first 300 mm section. Glue is sprayed on the central square of the tolex section and half of the bottom cabinet side and let dry for some seconds, then applied from the center of the bottom side. With a clean cloth, press the tolex fabric from inside to outside to expel any bubbles. Glue is sprayed on one side of the cabinet and the center square of the second section, special care must be taken on the border, press the tolex on the border carefully to avoid forming bubbles. Process follows on the top side. Careful must be taken to be sure that the tolex is properly aligned to the cabinet while unrolling the tolex piece over the cabinet and the drawn lines follow the border of the cabinet.
When both tolex extremes meet on the bottom side of the cabinet, they must overlap and with the help of a ruler and a cutter they are both cut in the middle of the overlapping section, remove the top strip just cut, detach the top fabric extreme and remove the strip left on the bottom fabric, so that a perfect joint is left between the two fabric sections.
Now it is time to glue the tolex on the front and rear sides of the cabinet, this is even trickier. Follow the videos I posted before. Fold symmetrically the fabric at the corners and make a cut in the middle, overlap both sides of the fabric and make a 45 degree cut from the outer corner to the inner corner, but no longer!!, otherwise you will leave the internal borders without material, see on the drawing above how the 45 degree angles only apply to the first section of the border exposed to the front or the rear, but it does not apply to the internal sides. Any excess of glue on the tolex can be cleaned with white spirit.
Leave to dry the neoprene glue for at least 24 hours.
Cut and staple or nail down the piping to the rear side of the front frame. Put in place the front panel with the grill cloth and speaker screws and fix it with screws to the cabinet from inside.
Rubber feet and handle can be fixed now doing some small cuts on the tolex at the place corresponding to screw holes. Corners can be fixed with two screws per corner, make sure that they are properly placed with screw ears on the cabinet inside.
See below the front view of the finished speaker cabinet.
I decided to do a rear open cabinet. Given the volume of air inside the cabinet, a closed cabinet could damp the bass sound pressure waves.
From the rear panel, 2 pieces of 115 mm wide are cut for the top and bottom covers, middle cover is unused. A Ø24 mm hole is made with a wood drill at the center of the top cover to place the speaker jack. Tolex is applied to both pieces and 2 mm holes are equally spaced all around to fix both covers with screws and countersunk flange finishing washers to avoid damaging the tolex with the screws and obtain a better finishing. Radial cuts are practiced with a cutter over the tolex at the connector hole on the top cover and connector is fixed with two screws.
The speaker connector is an hybrid SpeakON/ 1/4'' jack Neutrik connector. Speaker is fixed into the front panel using washers and bolts.
A speaker cable 2 x 1.5 mm2 is soldered between the speaker connector and the connector taking care of + and - connections.
Top and bottom rear panels are fixed with abundant screws and countersunk flange type washers.
See below the rear view of the finished speaker cabinet.
The 1x12'' cabinet speaker with the 25W TubeSim amplifier.
I decided to personalize the cabinet speaker adding a Rezzonics logo at the top front of the cabinet. I named the cabinet speaker Rezzonics v12 for the 1x12'' speaker installed.
I used Inkscape for the logo design, I tried several fonts: Magneto, Aeroplane Flies High, Mottel, Norton and Puma Oblique. The Magneto font is the one I most liked.
Several layers of extruded extended text are overlapped to provide a 3D volume effect, with a silver grey in the front layer, darker grey in the middle layer and black on the top layer:
I used some recycled materials to make the logo and I changed the colors: Bottom layer aluminum grey, middle layer in black and top layer in silver. A piece of 2.5 mm plywood coated with aluminium was used as the base or bottom layer. Another piece of 2 mm plywood painted in black was used for the middle layer, and a 1 mm soft plastic lined with aluminum paper for the top layer.
Printed paper was glued to the plywood pieces and then the text form sawed with a scroll saw. For the thin plastic layer I used scissors. The three layers are centered and glued and then the whole logo is lacquered.
This is the final result.
Rear view of the cabinet speaker with Celestion Vintage 30 speaker installed:
Front view of the finished cabinet with logo and 25W TubeSim amplifier.
Now it's time for the sound check and comparative with AC30 emulators!!
For the tube amp debugging I used a digital multi-meter to verify transistors and diode polarization, as well as DC and bias voltages, a signal generator and an oscilloscope. I have a Velleman PCSGU250 Oscilloscope ($200), it has a signal generator output and a two oscilloscope channels, it connects to a PC via USB and has a simple and easy to use graphical interface, that offers frequency spectrum analysis.
Fig 1. Velleman PCSGU250 Oscilloscope ($200)
I have a vintage digital multimeter very similar to the first Fluke digital multimeter models, mine is a KT 605 (probably a Fluke copy):
I had many issues with cold soldering and oxidation due to the use of a soldering paste not suitable for electronics, this caused many DC regulators breakdowns. I had to redo many of the soldering joints.
Another issue was related to excessive noise in some of the amps stages, I had to replace some of the amps by followers to reduce the amplification of noise floor.
Film capacitors are very sensitive to high temperature and some of them had to be replaced since they where damaged while soldering. It would be better to use low temperature solder paste (Sn Bi) and hot gun to avoid heating too much capacitor pads.
25W class-D amplifier gain was set with input pins PA_GAIN[1:0]=11 to get a maximum gain of 36dB.
Verification:
Signal generator and an electric guitar were used as input of the tube amp, oscilloscope signal capture at different test points on the board are shown below.
Fig 2. shows the time signal of a G note at the output of the guitar and input of the tube amp, with approximately 0.5V peak-to-peak
Fig 2. G note (open 4th string) signal at the tube amp input
Fig 3. shows the frequency spectrum of the same signal by using the FFT function of the scope.
As it can be seen fundamental frequency of the signal is 196Hz which corresponds to a G note, open 4th string, harmonics at multiples of the fundamental frequency can be observed
Fig 3. G note frequency spectrum signal
Fig 4. shows the signal at the output of the preamp first double operational amplifier stage (TP6) (refer to schematics on the Tube Amp Schematics post), signal is amplified by almost 10, more than 4V peak-to-peak, with no apparent distortion.
Fig 4. A G note (open 4th string) at the output of preamp 1st stage (TP6)
Fig 5. shows a G note at the output of the preamp second double operational amplifier stage (TP8). Some distortion appears, with more gain on the positive cycle and some soft clipping. Signal range is around 4V peak-to-peak.
Fig 5. G note at the output of the preamp second stage (TP8)
Fig 6. shows a G note at the output of the equalizer (TP21) with bass and treble at 0. Signal range is 1.6V peak-to-peak.
Fig 6. G note at the output of equalizer (TP21). Treble = 0. Bass = 0
Fig 7. shows a G note at the output of the equalizer (TP21) with bass at 0 and treble at 10. Signal range is 1.5V peak-to-peak.
Fig 7. G note at the output of the equalizer (T21). Treble = 10, Bass = 0
Fig 8. shows a G note at the output of the equalizer (TP21) with bass at 10 and treble at 0. Signal range is 0.4V peak-to-peak.
Fig 8. G note at the output of the equalizer (T21). Treble = 0, Bass = 10
Fig 9. shows a G note at the output of the output amp first section (TP20). Distortion appears on both cycles with some hard clipping on both cycles. Signal range is around 0.8V peak-to-peak.
Fig 9. G note at the output of the output amp first section (TP20)
Fig 10. shows a G note at the output of the output amp second section (TP18). Distortion appears on both cycles with harder clipping in the negative cycles. Signal range is around 0.8V peak-to-peak.
Fig 10. G note at the output of the output amp second section (TP18)
Fig 11. shows a G note at the output of the main volume potentiometer (TP22). Signal range is around 0.6V peak-to-peak. Volume = 10
Fig 11. G note at the output of the main volume potentiometer (TP22). Volume = 10
Fig 12. shows a G note at the output of the Speaker Emulator filter first opamp (TP 25). Signal range is around 1.5V peak-to-peak. Mid frequencies are boosted.
Fig 12. G note at the output of the Speaker Emulator filter first opamp (TP25)
Fig 13. shows a G note at the output of the Speaker Emulator filter first opamp (TP 29). Signal range is around 1.5V peak-to-peak. Mid frequencies are boosted.
Fig 13. G note at the output of the Speaker Emulator filter second opamp (TP29)
Fig 14. shows a G note at the output of the Speaker Emulator filter third opamp (TP 24). Signal range is around 1.75V peak-to-peak. Mid frequencies are boosted.
Fig 14. G note at the output of the Speaker Emulator filter third opamp (TP24)
Fig 15. shows a G note at the output of the Speaker Emulator filter fourth opamp (TP 27). Signal range is around 1.6V peak-to-peak. Mid frequencies are boosted.
Fig 15. G note at the output of the Speaker Emulator filter fourth opamp (TP27)
The real signals correspond quite accurately to LTSpice simulation results.
Now is time for the sound tests, but first of all I needed a Speaker cabinet to connect to the 8ohms power amplifier output.
