Sunday, September 28, 2008

cagibi exposition

in support of the pop montreal circuit bending workshop there is going to be a small gallery of diy electronics all week at cagibi. i decided to contribute these.
everything makes anything sound like lasers. and the sewing machine pedal is a gain/slight overdrive. none are that recent but they are all quite interesting.

probably more on this later. the vernissage is this wednesday.

Wednesday, August 27, 2008

Bass Cab

I made this bass cab with a "sammi sound" speaker i purchased at songbird music in toronto last year. My friend Celidh helped with the sewing.

Because i am paro about the speaker getting punched i put a metal grill behind the mesh. The grill is just held on with velcro. The box is made out of 3/4 inch particle board. I ported the back and it sounds quite nice. It is much tighter and clearer than the other 15" cab i made with an eminence delta.

The upholstering is far from professional but maybe the next one will look nicer now that i sort of have an idea.

Mostly i have been spending time on PIC programming and a soundlab with some extras that will hopefully appear soon.

Friday, July 4, 2008

Sunn 200S

I did some work i've been talking up for a while on my 200S. Basically two things, implementation of solid state power rectification and replacement of all electrolytic capacitors (from 1967).

The original power cap is one of those jumbo 4 in one caps which i am not interested in sourcing so i replaced it with several separate caps. The 30uf was replaced with 2 x 68uf in series (34 uf @ 800v), the 20uf's feeding A and B were replaced with pairs of 47uf's in series (23.5 uf @ 700v) and the last 20uf with a 22uf @ 450v. Overall this increases the total capacitance of the supply a bit resulting in a decreased power factor. Also to insure that the voltage is shared nicely between the capacitors in series i added dividers using 100k resistors. This will cause a bit more current to be drawn through the power transformer but since it is extremely large i don't think it will cause problems. (2.9mA on the B+, 2.85mA on 570 tap and 2.17 mA on the 433 tap = about 4 watts of power dissipated over the new voltage dividers). The 15k resistor was bumped up to 17k. Ideally the dividers should have been made with 220k and the 15k replaced with a 20k but its what i had on hand and it will work alright. With a 20k the voltages on the preamp taps would remain very close to the original design. with a 17k they are just a bit higher.

The caps on the negative bias voltage supply were also swapped out and replaced with the same type of 47uf mentioned above.

Here is the original supply schematic for reference:

Below is a schematic of the new power supply capacitor section. The voltages are the end result measured after changing to solid state rectification. Unlabeled resistors are 100k 3 watt.

Here are some pictures of the physical layout, first the pairs of 68uf and 47uf.

Here you can see all the electrolytics replaced.

The diode rectification is very simple to convert to. I have seen it suggested 10pf caps be put in parallel with each diode but i do not have any rated at the appropriate voltage so for now i have omitted them. Mostly you just have to make sure that the PIV rating of the diode is about 3 times greater than the rectified DC. In this case the resulting voltage is about 595 volts. And the PIV of a 1n4007 is 1000 volts. 3 X 595 = 1785. So if we use 2 diodes on each half of the rectifier we should be fine.

Here is a picture of the final result with the diode rectifier.

The amp was re biased for the new B+ and so far so good. It seems clearer sounding and punchier. It also seems to have a bit more gain. Possibly from the higher voltages in the preamp or possibly my imagination.

Saturday, March 1, 2008

PCB Etching

This is the second series of etchings i have done. The main difference this time around was the heat transfer method. A problem that keeps coming up is uneven heating of the board when transferring the toner which results in many broken or missing traces.

The previous method involved ironing the copper side of the board with the acetate in between the iron and the board. Pressure is applied on the iron and reasonable results can be had but it not very consistent.

This time the heat is applied to the fiberglass side of the board. This method provides much more even heating to the copper. The transfer is then placed on teh heated copper and a flat surface is smoothed over the surface during heating to put some pressure on the transfer.

Above is the basic setup. The iron is inverted and held in place by a vise. The timer is just to indicate roughly how long the board has been on. The transfer takes about 2-4 minutes depending on the size of the board.

Heres the setup in action. An oven mit is used because it gets real hot on the hands other wise. The aluminum is just a random scrap piece. Anything with a flat surface will do.

The rest of the process is trivial. I used ammonium persulphate to etch. The board is teh "Tube sound fuzz" from tonepad with a few modifications. (Mainly a millennium bypass and a switch for high/low gain.

Above, a sketch of what the extra traces are for/ the modified PCB.

Below, some other pictures from the process.

Tuesday, February 12, 2008

YBA-1A Re-Cap

At some point during rehearsal my amp started blowing fuses. So i decided to use it as an excuse to replace the 40 year old caps that were inside and learn a bit about tube amps.

One thing with this amp is that there is no bias pot and the bias voltage is set to -46. This seems to have possibly caused a problem when the capacitors started to dry out. When electrolytics dry out they begin to have resistive quanlities as well as a capacitance. A less negative bias results in a greater current flowing through the tubes. Which could burn them out prematurely, or overload the plate and cause it to glow bright red (seriously shortening the tubes lifespan if not ending it right there).

This is the schematic for my amp. Mine is either a later or earlier revision and has a few less parts plus runs a higher voltage to the preamp tubes.

Basically the main B++ power supply caps were replaced as well as a few others. A bias pot was added as well for a bit more versatility.

Heres a close up of the power supply section. C15, C17, C18, C19 and C20 were all swapped out for 100uF 450V nichicon electrolytics.

But one thing at a time. Going back to the bias problem. When I measured the bias had sagged to -41v.

Looking back at the schematic, if C14 had even a somewhat large resistance, it would be in parallel with R31, effectively lowering its resistance and lowering the voltage drop across it. since the -46v comes from the voltage divider setup by R32 and R31, C14 developing a resistance would cause the bias voltage to become less negative. So i replaced both C14 and C15. C14 with a 10uf and then changed the circuit a bit to have a bias pot that can vary from about -44v to - 55v. Essentially R32 is replaced by a 5k resistor in series with a 10k pot. The wiper of the pot is connected to a 47uf cap and is considered the output voltage. The 3rd leg of the pot goes to R31 which remains unchanged.

Here is a sketch of the old and new bias sections

Here is some math showing the effect of the resistive qualities of a bad cap on the bias voltage

Heres the bias section modified with the pot. Also you can see the replacements for C15 and C19.

Heres C20, C16, C17 replaced. The holes are where the can caps used to be.

I also replaced a few resistors that were floating out of spec. And I changed R16 from 10k to 11k. The reason is because on my version the voltage at c20 is exactly 450 volts (if everything is ok) which is the caps rated limit. Although this is not likely a problem I bumped R16 up to 11k to reduce that voltage just slightly to around 442 volts.

And finally I also replaced C21 with a 10uf 450V xicon electrolytic. While doing this I melted a bit of one of the filter caps being careless with the iron. So I replaced it with a metal film cap of the same value (to the left of the yellow capacitor in the picture below).

Saturday, February 2, 2008

Scrambler Clone

This is a clone of the ampeg scrambler pedal that you may or may not have heard of. Its a pretty high gain sounding fuzz box that can also do an octave up buzz. One knob controls the wave shape and one blends the new sound with the clean sound.

The PCB was made from the tonepad layout. However better bypass circuitry was added.

I perf boarded this pedal about 4 years ago and it worked but the output was so low. I asked about this on a forum and the situation was left unresolved.

the thread can be seen here.

Anyways years later here we are making another one and guess what? same deal. The output is about .6 volts. When you blend to the clean side its going to be at a volume level equal to whatever is coming in. Which in the case of any instrument I have is at least 2 volts and in terms of guitar more like 3 (courtesy of a dimarzio X2N). This works out to around a 12dB drop when the effect is switched on.

Quick searches of the tonepad comments and the DIYstompbox forum revealed that other people have noticed as well.

here and here

If you had low output pickups you might not notice. square waves sound louder than sine waves so the distorted sound will sound louder then a clean guitar signal with the same amplitude. (Rms of a sine wave is approx .707 * Amplitude, Rms of a perfect square wave = Amplitude) .

A lot of people are mentioning building a booster in the same box to bring the volume up. This is poor solution for a few reasons. This circuit calls for 3 high gain darlington transistors and has 2 gain stages and a buffer section. I shouldn't need to add anything to bring the level up.

The problem can be resolved by modifying the final output stage comprised of Q3 and Q4. They are setup as a differential amplifier with a single ended output. Without doing to much math to calculate A (gain) = Rc / re'. Where Rc is the 1K resistor coming from +9v to the collector of Q4.

To calculate re' you must find the DC emitter current and so on. This is not necessary at this point as we can see from the equation in the last paragraph that Rc and A are directly proportional so increasing Rc will increase the gain. It will also however decrease the current through the transistor but this is of no consequence for our purposes.

I put mine up to 39k which gave me about 2.0 - 2.4 v out depending on the wave shape from a 3 v sine. (.707 * 3v = 2.12V so it should "sound" like about the same level). Since the clean part of the blend does not pass through this final differential gain stage its volume is unaffected. It would seem the problem is solved. I would suggest for anyone doing this to use a trim pot to find quickly they're preferred output level, 50k or 100k should be fine.

A millennium bypass was implemented after the board was made so thee are no traces for it.

Heres a video of the effect on a sine wave. The top trace is the output and the bottom is the input measured straight off the function generator. It gives a good idea of how the octave up is generated. First the effect is off. Its switched on when you see the square wave. Then the tone knob is twirled, then blend, then a bit of one and the other.

Sunday, January 20, 2008

Variable (+/-) DC Supply

I started designing thing with +/- 15V mostly instead of the usual 9 volts I was working with before. I don't have a split supply so i built this simple one. It is variable from about +/- 2 to 18 volts DC.

Its based on the LM337 and 317 of course. The transformer is overkill but the point was to build it out of parts I already had.

The circuit is simple enough to perfboard. It is straight from the application notes (using a 1k pot and an 82 ohm resistor) with 1N4004 diodes protecting the regulator as suggested. The large caps off the transformer are 4700 uF.

There is no display of the output voltage. I monitor it with a multimeter.

Probably not the best supply in the world but more than enough for what I need.

Saturday, January 12, 2008

DIY Bass Repair

At some point I bought this bass online:

"you've got mail!"

Ive never seen something get quite so mangled in the mail. Anyways my money was refunded by the seller and they never filed the claim with usps so i still had the pieces of the bass. I asked some local luthiers about a cost of repair but the cheapest i could find was 80 dollars just for the neck. I contacted Dean and they were willing to sell me a new replacement neck for about the same price. so I decided i didn't have much to lose by attempting the repair myself.

I used wood glue and a bunch of clamps to glue both the neck and the body back together. I just used normal wood glue from the RONA near me and it worked perfectly fine. The hardest part was probably clamping the headstock area but in the end it turned out nicely.

this is the part of the fretboard that was originally cracked

the body "repaired"

various neck shots

Also the pickup was smashed so I replaced it with a seymour duncan "basslines" musicman pickup. It required some sort of preamp so I designed a very simple one with the option of some slight asymmetrical clipping using a RC4580 op-amp from texas instruments.

The preamp has 3 controls. A gain knob, a master volume knob and a switch which places 2 leds (used as clipping diodes) into the circuit. Below is the schematic i came up with (note that NE5532 and RC4580 are pin for pin compatible and both high performance low noise op-amps). Also the "volume" pot is Audio tapered not linear taper as stated in the schematic.

Without the preamp the output of the pickup is unbearably low.

The bass is now quite usable. I may repaint the body so that you cannot see the crack
but im not sure if I really care that much.