x0xi0 #330-133.5 : Full Build. Next Steps

The x0xi0 mid-build in 2011.

As you can see, this x0x build hit a wall. I followed through the original documentation and built the whole kit in place of doing it in parts. I then had to shelve this project since there were some gaps and errors in those PDFs. I've done numerous other projects since.

A recent x0x repair for someone else inspired me to finally put this project to rest. This has turned out to be advantageous since Brian's documentation is now into its fourth revision and appears to be quite complete. I've been able to do a sweep of updates with little risk and the results have paid off.

2014. It looks sad now but all will be well soon.

So, now that I've reviewed where I left off, here's where it stands. In the interest of extreme experimentation I've socketed most of the transistors and a few key components:

• R47: This extends the low range on the x0xi0
• R61: this controls the ENV MOD setting when Brian's pot is set to 0. 10K is the default in the document.
• R97: as per my original build, this adds a 'natural' boost to the resonance. Some people add a pot for this. I don't see the point. 10K is what's in the document. My last build provided boost at 9.2K. My final decision for this build is pending. What I want is the same behaviour as the the last build. 90% turn in the pot gives my a proper 303 calibration. The extra 10% is the boost and nearly reaches self-oscillation. More than enough for me!
• C21: This add more bass boost and in this case has an effect on VCF Out as well.

The LEDs were also never placed so this task is under way now. here's a dorky photoshop file I've kept on hand to experiment with color combinations:

I'm using the same set of LEDs as the last build but different values appear to be useful here. These will be more intense than the first build but I have time to try out a few combinations.

5mm:
Blue: 5.6K Ohm
Purple: 100 Ohm
White: 2.2K Ohm

3mm - pending delivery of custom LEDs:
Blue: ?
Purple: ?
White:?


There's also loads of wiring to finish up. My pre-assembly calibration succeeded though. VCO, VCF, and ENV all check out.

It's getting there. I want this finished so I can finally have some fun with it.

x0xi0 #330-133.5 : Custom Board VCO

I took a shot at just populating the VCO. This is how the Adafruit documentation suggests one goes about it. As you can see though, this is quite different! The VCO is no longer in a cluster like the original.

So yes - it may make sense to build the x0x outright, then calibrate later (and hopefully not have to troubleshoot). The multiple x0xio headers do allow one to calibrate this though. There are so many jumpers that it must allow access to it. I just need to check.

x0xi0 #330-133.5 : Custom I/O Full

Finally back in the game. A solid draft of the documentation finally made its way to the ten of us using these PCBs so some of us are getting a head start while the manual is being finalized.

This build approach is different since one drops each component in by type first. I'll need to do testing later because the boards are really only populated fully in the in the end rather than being able to test sections incrementally. This concerns me a little bit because it may complicate troubleshooting. The documentation may eventually include test procedures in phases but we'll see.

Note that my own sourced parts, namely the resistors, are also 1/4W instead of 1/8W. I purchased all my parts before getting the mod kit. I hope I don't kick myself later for replacing them with the smaller ones. It appears I have clearance in the I/O but it's also quite obvious that the through holes and spacing on the PCB were design for 1/8W.

x0xi0 #330-133.5 : Top Panel 'Full Kit' Mod

It took about 3.5 hours to pull the top panel together. Like the I/O mod, the steps were relatively simple. Brian has laid the documentation out in a pretty straightforward way. Three components were missing: one 103k capacitor and two 22K resistors. A quick trip to Supremetronic sorted that out.

There isn't really any testing or calibration one can do until it's wired up to a power supply and x0xb0x so there will certainly be more work to follow once the new x0x is built. I'm hopeful that everything is in order here. It looks absolutely amazing. Dropping the top panel on there gives a quick glimpse of the power of this über 303.

x0xb0x SS#01: Finished

Better late than never! The finished x0xb0x, SS#01. I never did take the time to snap that last photo and wrap this up but here it is.

With all the experimenting one can do with a project like this, I have to admit that it was difficult to simply close it up and consider it done. Time and time again I would open it back up, try something new, and tweak it some more. Despite the long lag in posts here it has served my set up really well by being a go-between of all sorts. I've been using it practically daily since my last post. Midi in and thru has been used in full force, with Ableton Live serving as the Master to all the gear and the x0x slaving a Jupiter 6. The CV/ Gate has proven itself worthy controlling a modded MC-202. Finally DinSync, of course, has properly connected the 202 and a TR-606 to the whole rig. So , all in all, this is an extremely useful instrument and tool. The 303 cloning is perfect and the I/O it has on hand bridges the old gear with the new beautifully.

Did I meet my goals?

In one of my first posts I set out a plan for this project:

- Keep the hard costs as close to zero as possible,
- source as many of the parts locally and from within Canada,
- minimize (or at least be aware of) hidden costs,
- document time and cost.

Hard costs:
I'd say I came out ahead in some areas and lost out *a bit* on others. The grand total for *this instrument* came down to about $475.00 CDN when factoring in rare parts end to end (including a $50 BA662A), some custom features like the Korg EMX-1 TB-303 style silver knobs, puchasing from multiple sources, shipping from all of these sources, and duty fees in some cases, and hidden costs. In some cases it was only advantageous to make a larger quantity purchase to justify the shipping. This means building multiple x0xes to spread out the cost. I would say I embellished in some spots to get the sound just right and that increased the cost.

Parts from Canada:
If I only I knew then what I know now! Since the start of this project I have found other sources in Canada that significantly help control cost. I largely used sources based on the BOM that was provided by the x0xb0x community but I now know that many components can be had here for more or less the same price and I can cycle of drive to the store that sells them. My next project, as you'll soon find out, already begins utilizing these sources.

It's interesting to see how wildly prices fluctuate between suppliers. A basic component like a resistor costs $0.09 US each from Mouser. I can get the same carbon resistor locally for $0.02 and a 1% metal film resistor for $0.05. It doesn't seem like much but when you need hundreds of them and factor in shipping and duty this is a significant difference. One can rationalize this purchase by lumping it in with items like pots and encoders and what have you to justify a one supplier purchase, but if cost is important, as it was for me, it's worth watching carefully. My original pruchase from Futurelec helped tremendously. I picked up enough 1% metal film resistors (and other bits and bobs) from them for three units at a cost of $0.02 each.

Minimize hidden costs:
This is where one can bloat and kill a budget. Shipping here and there for this and that, duty (which you can't control), the odd trip for another part, and experiments with another component types. Before you know it you blew another $100. If you click-and-shipped your way through this maybe it's $150-$200. It's so easy to blow that extra cash and then marvel about how you got a 303 for $350. But did you? Nope. Not at all. Not even close. Look at the numbers and it'll tell a different story.

I think I did minimize these hidden costs to a degree. I didn't avoid them but I did minimize it as best as I could given that this was my first crack at it and I was aware that it was part of my goal. As I mentioned above I now know that I can do better in this area but I think I did okay.

Document time and cost:
Grand total: $475.00 for this unit and if I factored in my time at an hourly rate add on another $10,000.00 :D This is a very involved project. If I was in the business of electronics I suppose I could bang this out much faster but cine I was learning it took loads of time and preparation. The caveat with that $475.00 is that I've got enough components to build two more. If I don't pursue that (hint, hint) then the cost is significantly more.

What did I learn?

First of all I learned that I love the 303 more than I did before! This project is well worth going through, not only to have yourself a prefect 303 but to also learn about the inner-workings of a legendary intstrument like this. It's pretty cool to crack open a real 303 and understand what it's doing. For that reason alone I'd recommend it to anyone who's into synths and is prepared to open one up and look at it up close.

Secondly I've learned that I need to be patient. As soon as I got impatient I made mistakes, some of which cost me more money and time. Having said that, I may have been too patient in other areas of this project. A year and half to wrap up the synth and blog is too long. In reality it could have been completed in much less time, real life interruptions notwithstanding.

Finally I learned that I'm capable. I suppose it's like any area of interest, as soon as one gets a little bit knowledge in a new subject a whole world opens up and new possibilities become apparent. Despite the lag on this project's time line, I did make time to work on other synths in between since I had no fear of opening them up and seeing what was going on with them. The more vintage gear one has, the more it becomes a necessity to maintain it all. If you have deep pockets by all means take it to a pro. Quite often though the repair can be fairly simple, so projects like this x0xb0x begin to educate us gear heads.

x0xb0x SS#01: Adjustments & More Tests

Today I set some time aside to address some of the outstanding tasks that were left on my list.

- I removed the BA662A and tested out a BA6110 in its place
- I drilled the power switch hole into the back of x0x
- I cleaned up the knob heights and properly aligned the rotaries. The dremel work needed to be straightened.

What's left:
- Replace the tempo encoder. The shaft has loosened due to my dremel work
- Drop in the heat sink once and for all
- Paint the back, possibly the enclosure
- Wipe down the top panel so it looks fresh again

The BA6110 is a perfectly good alternative in my view. It's nice to have the BA662A in place for the sake of authenticity but frankly it doesn't sound any different. I've left the BA6110 in place for now to put it through several months of jamming.

x0xb0x SS#01: A/B Comparison

I finally got around to doing an A/B comparison of my x0x and a TB. The TB plays through the left channel; the x0x on the right.



I still need to finish up some of the cosmetic elements of the synth but the delay is largely in part to me preferring to jam with it :)

x0xb0x SS#01:Tweaking

One of the things I noticed with my x0x was that it wasn't *quite* a 100% match to the 303 sound I knew. The resonance wasn't quite there and it had a slightly harsher sound to it. The forums cover this in great detail so I won't repeat it all here but I'll touch on a few points that affected my x0x.

The biggest, most immediate change I noticed was swapping Q11 from a 2SC536F to a 2SC945. This noticeably muffled the sound but warmed it up. The harshness everyone describes can be eliminated right away via this transistor. The muffling effect it produced could also be adjusted via TM3, the VCF trimpot, so the range could match an original 303 with the 945s in them. I trimmed this pot after the swap and was able to match it to the real 303.

Secondly, the resonance level needs to match. This is such a signature to the 303 that's it's a must have feature. If the betas are too low in Q8, Q9, and Q10 it doesn't "scream" enough. I have to admit that I found it lacking in mine too. I started by swapping out my 2SA733Ps to 2SA733APs. The APs are found in all original 303s. Unfortunately it didn't have any audible impact for my x0x. As such I dropped the Ps back in and left them. What *did* work was adjusting the resistance in r97. The stock version calls for 10K. I found the sweet spot on this unit at 9.2K. This produced a perfect match with resonance at about 90% and left 10% more for me that exceeded what my original 303 had here. Some people add an extra pot to make this variable but I don't see the need. This produced what I wanted and still kept my 733Ps in use without using "fancy" APs.

During this tweaking I ran A / B comparisons the whole time with matching patterns and I can say that it is SO close to a match now that it's scary. For most ears it's an inaudible difference and in real world use it won't be noticeable. I can hear very, very small differences but it doesn't affect the overall performance and cloning ability of the x0x. This unit is without a doubt a near perfect match to the real TB-303 on my desk here.

So, to round up, it's at least three key steps:

• Q11: swap the transistor from a 2SC536F to a 2SC945
• R97: Reduce the resistor value from 10K to 9.2K (though subsequent units, or your unit, may be a different value)
• Adjust TM3 to match the cutoff setting to your preference. In my case my preference was the TB-303 I had here.

x0xb0x SS#01:Testing & Tuning

All the parts were in place; I was soldering and building the x0x carefully; I certainly spent enough time on it. I couldn't figure out why the x0x wouldn't fire up. I compiled a list of all the tests and went through them once more as a matter of quality control. I completed a battery of tests and got support through the x0x forums and covered everything from my power supply to each IC in the digital section of the pcb. It turns out that my earlier envelope issue was still at the heart of the problem. It was still causing errors and wasn't sending the correct voltage to the amp. After about 2 hours of troubleshooting I was relieved to discover that 2SC536F in Q37 was faulty. I replaced it with another one and low and behold it was fully functional! LIGHTS AND SOUND! Relief & Joy. It wasn't my shoddy set of skills. It was a dead transistor.

Hours of jamming and more tests ensued. I had the x0x drive a TR-606 and vice versa, Ableton and ACID driving the x0x via MIDI, and even had a connection to the c0ntr0l software.

After a good week and a half or so of just enjoying the synth I finally sat down and tuned it as well. It's by no means a complicated step but it does require a bit of patience. The forums suggest setting the tempo to its lowest speed, setting up a low C and middle C, both with long slides and decays and using a tuner to match it all up. This method worked beautifully for me via Wavelab 6 and a free tuner VST.

I must say that I thought the x0x sounded amazing when I got it working but hearing it tuned properly really makes it sing. *Now* we have an instrument here!

x0xb0x SS#01:Sequencer Part 3

The LEDs.

If there's any advice I could give to anyone out there who's about to embark on this step it is the following:

- Install your standoffs, washers, etc., first.
- Set up and just drop in your LEDs onto the board. Do not solder them in yet.
- Set up your top panel so that the height has been correctly set.
- Place easy to remove tape on top of the overlay panel LED holes.
- Set up each LED into their respective holes. Lightly stick them to the tape.
- Solder away!
- Remove the tape.

I suppose if you have the kit's LEDs or your own rounded LEDs this step may be relatively easy. Using flat LEDs is a whole other matter though. It just takes some patience and in my case, a few mistakes to get this right.

Regarding the standoffs:
It seems obvious now but no one wrote it down before so I'll do it now. The two standoff heights you're likely buying are 10mm and 15mm. The 10mm standoffs are used on the four outer corners. The 15mm standoffs are used in the center holes of the board. This prevents the pcb from bowing and keeps the height consistent! Obvious, I know!

Once I had the LEDs locked in I fired it up and .... NOTHING.

x0xb0x SS#01:Sequencer Part 2

I finally got my hands on the missing 16 pin DIP sockets and wrapped up everything in the x0x except the LEDs. Tack on another $3.45 in parts (6 x $0.50 plus Tax) which is not the end of the world but is thievery compared to the $0.07 / unit price from Futurlec. Lesson learned!

The manual isn't kidding about this being a long process and things getting tight in spots. I would have preferred to do all the resistors first (like they always say!), then caps, then dips and switches. It would have made life much easier. In any case the job got finished. Now the only task left is getting the LEDs set up and building the jump wires before we get to start closing this thing up!

As mentioned in my previous post, I opted for these nice flat LEDs I picked up. I diffused a set of them with a dremel for comparison but I'm finding that the clear glass look with really high resistor values is creating a very slick, low blinkyness factor that I rather like. I may regret it because if it doesn't really work visually, I'll be ripping out and replacing 40 resistors. I really hope I don't have to do that. The x0x's LED size is quite large, 5mm, but with this set up the LEDs appear recessed and about 2-3mm in size. It's rather unique looking. We'll see how it goes. The resistors are in so it's too late right now!

These are really high MCD valued LEDs so the resistor values I found that worked for this set up were as follows:

Purple UV - 2000 mcd: 10K
Blue - 5000 mcd: 47K
White - 8000 mcd: 68K

Blue is being used for Tempo and the tools on the left and right side of the panel. White is being used for all the notes and Sequencer steps: 1, 5, 9, and 12. Purple UV is being used for the remaining sequencer steps.

These are pretty high values but it looks elegant. I *think* it'll work. Next up: jump wires and then those LEDs!

x0xb0x SS#01:Sequencer Part 1

Ah, the sequencer. This is where things start to get more creative (in some respects) as opposed to functional. Aside from going through a soldering frenzy - heaps of parts, all pretty much the same - this is the stage where we get to introduce blinky lights. I haven't decided on a final configuration, but I have three flat LEDs on hand: blue, purple UV, and white. I'm considering a combination of all three but it hinges on the enclosure style I implement. In any case, this first phase is simply to get the stock parts out of the way. That means soldering sockets for all the ICs, diodes, caps, and a good handful of resistors.

> Note: I discovered that I didn't order a full set of 16 pin dip sockets. I'm missing six. That's a pain. They were a bargain at $0.07 a pop from Futurlec. Now I have to pick them up locally at probably around $0.25 plus my time and fuel, an inconvenience more than anything. Ebay's certainly an option but with a lengthy delivery time. Either way this is a minor, albeit annoying, setback. It would be nice to see all the ICs in place.

A working x0x is on the horizon! Fingers crossed that the sequencer works! I would love to hear some real sound come out of this.

x0xb0x SS#01: I/O

This was an exciting stage actually. First of all, it's nice to see several of these 'external' components land into place on the I/O board. 1/4", MIDI / DINSync, and CV / gate jacks all make up the familiar territiory of any synth and gear owner. This makes the project feel like it's really starting to come together. Secondly, the MASSIVE leap forward was downloading the Serial <-> USB driver and plugging the I/O into the computer. JOY!! I nailed the USB chip on the first go. Looks like the beer helped after all ;) The computer recognized the card and installed the drivers. As far as WinXP is concerned, this x0x is 'ready to use'. I'm elated!

Of course the x0x is far from complete or running an OS but it sure is encouraging!

x0xb0x SS#01: USB Chip

This was the step I considered to be the 'hump' of the build. I didn't have the I/O card shipped with the USB chip pre-soldered so putting this in place was going to be my big challenge. The 32 points would be my smallest solder work to date. I took my time, about an hour in total (clean the surface, place and solder, and snap pics throughout), and hoped for the best. The beer I had just before doing this probably calmed my nerves but also made me aware that I may very brutalize the whole thing. The true test will occur once the I/O is in place so here's hoping!

x0xb0x SS#01: Headphone

This was another painless step in the process. There's nothing to really test just yet according the documentation. That will follow once the I/O is in place. It's also nice to see the volume pot in place.I couldn't resist snapping a better pic of the BA662 I have too. I'm looking forward to hearing this amplified. The LA4140 also looks ancient despite being new old stock. It just looks vintage alongside some of the other old ICs. Note that I'll also be building the wiring for the jumpers.

Q33's hFE: 200 mfm*

Total time on this: about 2 hours to sort parts, build it, snap pics, and write this.

* My Faulty Meter

x0xb0x SS#01: VCA

The VCA was a reasonably painless step. It was really nice to see the BA662 IC get dropped into place. Seeing it on the pcb starts to give one the sense that the project is taking shape (despite upcoming steps being rather large).

Q31's hFE: 198
(again - based on my faulty meter)

Testing this section gave me some odd results. I was able get a change in sound behaviour but I couldn't say that it sounded like a note. I tested the resistors and got some odd figures here:

r127 (47K) : 39.4 K
r128 (220K): 52.5 K
r129 (2.2K): 20.25 K

The folks on the boards suggest that it's not entirely out of the ordinary and suggested I carry on to the sequencer to trigger notes properly, then troubleshoot again. I'll follow their advice on this one.

The parts for the headphone / mixer are set aside already. I'm looking forward to starting that.

x0xb0x SS#01: Envelope

Today's boost in the VCF results left me with no choice but to drop everything else I was doing and get building. I tackled the envelope section.

It took a short while to get organized, pull the parts and get soldering but all in all the whole process was quite painless. I still socketed all the rare parts too. The two 2SA733P were pulled from my lower measured bags:

Q38: 197 hfe
Q39: 196 hfe

These values are all based on my faulty multimeter though. That's the value I'm getting at this time so I'm keeping it here. I'll document it again properly once I have accurate measurements.

The tests were mixed though. I need to look into that. The first test was successful: 12 volts when connected to D35. D36 does created a pulse but not at 12V so I need to test again more carefully. I suspect it's correct though.

Total time on the ENV today: 3.5 hours including the pics and this blog.

-

x0xb0x SS#01: VCF

Wow. What a frustrating stage this was. I hit a brick wall and the project languished as a result. No point in adding day numbers to my titles anymore!


I put the VCF together shortly after the last blog. I took my time just as I had with the other sections of the x0x and thought I has put it together successfully. Once I began testing it I couldn't get a signal to confirm it was working though. I didn't have scope so I tried to whip up the diy audio probe that's suggested on the x0x build page. I fudged my through it but never really got it working properly or to my satisfaction.

Because this section of the x0x has several rare parts, I started to wonder if one or many of them were faulty. So I spent tons of time checking resistor values, voltages, betas, and the like, all with mixed results. I was sure something was wrong. At the same time, I thought it was high time to get a scope but I couldn't justify hundreds or thousands of dollars on a new one, and frankly, I didn't want yet another 50 pound tool in my arsenal. My hope was to track down a proper soundcard probe and use free software. Between these two challenges it took 6 months. It was really discouraging.

The GOOD news is that I had success with both of these challenges. First things first: I found a probe at a great price by Virtins. $30 US, shipping included to Canada. Can't argue with that. Once I plugged it into my line input on one of my low end sound cards, I wired it all up to the x0x, got testing, and low and behold: the damn thing seems to be working just fine. I just couldn't measure it properly.

I can't begin to express how relieved and elated I am to have past this step. I feel like I can continue and know that I don't have to revisit a problem.

On a side note, the OTHER good news is that I managed to track down 2SA733-APs as well so I'm that much closer to making an authentic 303. The critical components are right in the VCF: Q8, Q9, & Q10. It turns out my hFE reader on my multimeter is also crap so the component log component log isn't accurate either. (I always thought they were unusually low.) I'm going to leave them in and revisit this later. I need time to find a better way to measure these too but this less important than starting to solder again!!

A real Sight for sore eyes:

The VCF test with the cutoff and resonance at 0%.



The VCF test with the cutoff at 100% and resonance at 0%.



The VCF test with the cutoff at 100% and resonance at 100%.

x0xb0x SS#01. Component Log

All components in bold are considered important to optimize the x0x.

-----------------------------------------
2SA733P hFE ratings
-----------------------------------------

• q8: 225
• q9: 220
• q10: 218
  
• q27: 198
  

x0xb0x SS#01: Day 2. VCO

Since the PSU went relatively quickly, I went ahead and started on the VCO as well. Overall this was a much bigger task. While I'm not entirely new to electronics, I also don't have expert ... or even real intermediate skills doing this. A good chunk of my time in this phase involved troubleshooting, trying to get my homemade probe to work with the software scope (unsuccessfully), and trying to follow the steps and instructions really carefully. Total time spent was about 10 hours, four of which involved the above. I also socketed all the ICs and 'rare parts' and made my own cabling as per subatomic's x0xl0g. That also added to my overall time.

Because the scope never did fully work, I had to go the headphone route to test the VCOs. Again - a bit more time was spent on that. Thankfully I got to hear a saw wave before calling it quits for the night. I was quite thankful for that :) I slept better for sure! One unfortunate aspect in this is that the 2SA733Ps I have are all really low hFE. The highest ones I have are 225 so q8 has one of them in there now. I'll certainly be replacing these. It'll be interesting to hear what such low hFE will sound like though. So just as a note:

• q8: 225 hFE
• q27: 198 hFE
  

Day 2 was spent troubleshooting some more and filling out the rest (i.e. majority) of the VCO section. I continued to use the headphones as my 'scope' but I don't entirely trust the results. It really would be nice to see it on the scope just to be sure it's working as it should and that the frequencies are set just so. My multimeter and variable PSU have been quite good but I will need to get the scope up and running to really get the results I want.

One thing that baffled me was that the instructions for the VCO section are also not entirely complete. I had to double check the last round of resistors, caps, and ICs that would have to be added. I learned that I've been relying too heavily on the images at the site. I was a good exercise to have to track things down on my own and place the components correctly. These first two section taught me to have a clearer strategy before embarking on upcoming sections. I can see where I can be more efficient in this process.

All in all it was really nice to hear both the square and saw running. They're tuned quite well so far. Once the scope is up and running I can fine tune them some more.