Super Klee Sequencer

[Scott Stites]

Oh, that's easy for *YOU* to say - how would you like to play 22 rounds of 'operation' - swapping LEDs and resistors in a component encrusted, wire clustered nightmare of a breadboard archipelago?? Not as much fun as Nintendo, huh? Why I oughta....what? This is online?

Oh (harumph), ummm.....

Like I was saying, there are 22 LEDs - the 16 pattern LEDs, the 4 gate bus LEDs, the clock LED and the random reference LED. All of these can share a common dirty ground connection. This dirty ground connection can also be used for the 16 pattern switches and any other switches that switch ground in and out (merge switches, step and load switches, etc.). That will relieve a lot of wiring and hopefully stripboarding overhead.

The random, clock and step inputs, and gate outputs should share their own ground.

The CV outputs and programming pots should have their own ground - analog ground.

The rotary switch will actually have +V going to its pole.

So, for flying wire stuff - that's three ground connections and a +V connection. Then, of course all the other stuff.

Cheers,
Scott

[Uncle Krunkus]

I assumed that the sequence LEDs would have the 4K7 hanging off them at the front panel, before being commoned to DG. I know it's not good practice to have components "air sculptured" off the back of the panel, but I thought "no ones gonna want to run an extra 16 wires if they don't have to".
This change you've made makes a lot more sense, and is more robust and therefore more reliable. Makes very little difference to the stripboard. Same for the transistor drivers. No sweat.
I'll see if I can get these changes done tonight.

I think I'll soon be taking some new shots of the Sorcerer. It's looking very nice. Can't wait to hear it make a sound!

[Scott Stites]

[blue hell]

Great plot

Especially Matthew kicking in, pwew

Was wondering about one detail that sort of still dangles, the 10 V reference - could that be a 78L10 or something ?

[Scott Stites]

Hi Jan,

Yes, I'd nearly forgotten that - the remaining seed of doubt. I'm using a method Buchla used to derive a stable reference, but I'm not all that trusting of power supplies. Since a power supply isn't part of this design, that is going to be a variable that could affect the stability - if there's any variance in the power rail, that reference is going to move. and all of one's calibration will be for naught.

A 78L10 should work there - power rail will be +15V. Is dropping +5V too much to ask from those - I don't imagine so, but I've never tried it.

An alternative to that is to leave the circuit as is, but replace the lower resistor with a 10V zener (may have to adjust the 10K resistor down). What's your thought on that - in your opinion/experience, is a zener going to be stable enough to rely on?

Another alternative is a precision 10V reference such as TI makes. I voted that out (A) because I didn't want to get too exotic with the parts and (B) I can't for the LIFE of me find my stash of them. They disappeared during the inventory that started this thread.

If the 78L10 is used, the buffer could certainly be dropped, but I'd still want to hang onto the one on the other end (the one that's feeding the CD4066s). Wish I had a 78L10 handy - I'll have to place an order - should have it by the end of the week and know for sure. Hmmm....if the voltage drop is too much for a 78L10, a 78L12 would surely work. I *might* have some of those around.

Thoughts, Jan, anyone?

Cheers,
Scott

[Uncle Krunkus]

AFAIK, you can hang a 7812 off 30V, it will still put out 12V. (yeah, I just checked, LM7810 max input is 35V)

[blue hell]

When you're flexible for the output voltage ,,, a 5V1 zener is best regarding temperature stability and zener action, IIRC.

Without checking data sheets ... 5V drop won't be a problem for a regulator that needs 3 for good regulation I'd guess, it would disappoint me otherwise. Current drawn under 15 mA, results into < 75 mW dissipation, seems doable for a transistor size package.

Using a precission reference should probaly be an option, if used at all. I guess most VCO's out there wouldn't be good enough to justify for it.

Ok, out of lazy mode a bit, just checked http://pdf1.alldatasheet.com/datasheet-pdf/view/143138/ETC/GM78L10-S8.html , should be OK, should be better than the zener as well.

[Scott Stites]

OK - it's a done deal. 5.1V won't deliver quite want the Klee wants, so I'll definitely go for the 78L10.

+10V reference IC would be more or less polishing a turd, unless I really go all out and start going for expensive low offset op amps and such, I'd think. The TL072s seem to be holding up with the cal quite well on breadboard, which is usually worst case in situations like this. I'm sure you've noticed on the schematic that they get an offset trim before they're mixed together for A+B - it keeps any accumulated offset from terribly affecting the A+B ouput, so it stays in line with the intitial voltages of A and B. There is always a slight bit of offset downstream of those in the inverting stages before the lag circuits, but it's negligble and no more than any other like device I've used before. That kind of thing is usually adjusted with a fine adjustment on a VCO, which is going to get tuned anyway when setting up the patch. I was really surprised how much of an effect matching the resistors had when relying on max range to set intervals - that was really nice. Of all the intervals I've tried so far, 1V really works out quite nice, though it's better if one or two pots are set for some other interval than full rail.

Downstairs, the Klee is rattling through this really neat techno-spaghetti-western riff (is techno-spaghetti-western an established genre yet?). I should record it, because my 'pattern switch' wires got hopelessly jumbled around today during my resistor/transistor shell game, and I don't know which one is switch 3 or which one is switch 14. It'll be nice to actually be able to flip a *real* switch when all of this is done.......

Long suffering Uncle K - you might hold off on the stripboarding of the decoder til we give this a shot, or I can also work up a schematic of how I *think* it will go.

Cheers,
Scott

[fonik]

hi all,

your kind feedback encouraged me to post this 2U version of a possible klee2 front panel. yes, it's very condensed, podgy fingers not recommended. the "gate grid" boosted the price but can easily omitted to reduce the costs.

---

klee2_2U.jpg
Description:	schaeffer front panel 2U
Filesize:	240.37 KB
Viewed:	353 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

---

klee2_2U.zip
Description:	schaeffer front panel designer file	Download (listen)
Filename:	klee2_2U.zip
Filesize:	3.83 KB
Downloaded:	290 Time(s)

---

[Uncle Krunkus]

I reckon a compact version like that is fine, as long as it's fairly easy to find components that will fit. I don't think it matters so much about the switches being close to the top and bottom of the pot knobs, but I would get as much width as possible so there's room for a finger on either side. BTW the layout looks very professional, considering how hard it is to fit everything in. Personally, if I make one, (which is becoming more and more a "when" rather than "if") I'll go for 3U or more. It would just be to frustrating to go to the hassle of building something like that and then finding it unfriendly to use.

[fonik]

[blue hell]

[Uncle Krunkus]

I just updated the posted pics of board 1.
I'll get into the next one soon, should be heaps easier now. As I finish each board, should I send you a higher res copy Scott, so you can include it in the main .zip?

[Scott Stites]

Yes, hi res shots of the stripboard would be great to include in the final zip file.

I didn't think it would be possible to shrink this thing into 2U - far beyond my brain to imagine even doing it. Looks really great, but tweaking it may or may not be kind of a chore - depends on how large the switches are.

I wrote down my first Klee patch last night - I'm going to design a patch sheet for it so those that build a Klee can share settings.

Take care,
Scott

[Scott Stites]

Thanks for bringing that up, Jan. It would be a real drag if someone built the thing only to have ripple on the output.

Cheers,
Scott

[Scott Stites]

Hey Fonik,

I really like how you've accentuated every fourth bus switch - that helps one at a glance to find his bearings more easily and also sticks to the spirit of the original Klee, which was programmed by four pots at the same bit increment (the 4th, 8th, 12th and 16th, but it's all the same when things are moving in a circle ). I find myself, on the breadboard, counting pots to find out where I'm at all the time. Even if the stages are numbered, this coding still makes it much more intuitive, at least for me.

You've really put a lot of thought into that. It really tempts me to splurge with Schaeffer.....

Cheers,
Scott

[fonik]

FYI, i updated the 3U frontpanel...
many thanks for your kind feedback.

[Scott Stites]

Fonik - looks great!

Jan - 78L10s arrived today - works like a charm! Thanks for the suggestion, I honestly would never have thought of using one of those. Feels good to get any dependence on the power supply for voltage stability out of the picture.

Uncle K - I'll update the schematics for the decoder with the 78L10 - I plan to put the voltage value changes in as well.

Fonik (again) - I'll send you the updated sch files when I get those changes in.

Cheers,
Scott

[blue hell]

[Scott Stites]

I hit it pretty hard with a hair dryer - it got pretty hot, much warmer than one would expect in any reasonable environment (I don't have a temp probe, couldn't tell you how hot it was).

DMM I'm using at 10V range only has 10 mV of resolution - room temp it's on the edge between 10.1 and 10.2. Warmest I got it, it measured 10.2 without flickering; I'd estimate it moving perhaps 5 or so millivolts. It's pretty solid, I'd say - no worse of a tempco than any other of the components in use.

Don't have any cold spray, so I couldn't say how it would work at a December outdoor concert in, say, Oulu

I don't think temperature is going to be an issue with this component.

Cheers,
Scott

[Uncle Krunkus]

Great work!
This is shaping up to be a project we can all be proud of!
Second stripboard is coming along nicely. I'm only a bee's dick off firing up the Sorcerer for the first time, so forgive me if it's a little longer to wait than the first one.

[blue hell]

[jksuperstar]

Something I just found, might alter you plans for the front panel:

http://electro-music.com/forum/post-93038.html#93038

damn that's cool. & cheap.

[Scott Stites]

Those are certainly some cool keypads - colorful, too!

I'm hoping to finish up the new decoder schematics today with the 78L10 and new max range values - it's the decision making thing for which voltages to choose that takes the time.

Here's a Klee tune about Iris. It's a song of her longing to go back to the island of her birth, again to play upon the windswept beaches of Trinotaur with her long lost pet pole squid, Delia. Such an innocent time that was!

One oscillator synced with another, and the DSC2000 reacting atonally (as usual) to one of the outputs. 16X1 mode, 8V range with the original Klee programming of the 4th, 8th, 12th and 16th pot only. The master output is passed through an analog delay. One take/track, as usual.

Cheers,
Scott

[Scott Stites]

I've mentioned it a couple of times, but the more I think of it, it certainly would make things even more versatile.

I'm talking about a position on the rotary that would allow one to program the max voltage with an additional pot. As I go through and test different intervals, I've really been doing it with a multi-turn trimpot. Major 3rd, perfect 4th, perfect 5th and minor 6th and one octave all seem to be the most conducive to programming interesting patterns easily by treating the pots more as switches rather than pots - either full on or off. Usually just setting one or two pots to some interval less than the max range provides more variation to the pattern, but generally it's much more easy to program patterns in this manner than the original Klee 'loose clay' approach. I'd still like to preserve that mode of operation, which is why pots are used in the first place, but very nice patterns can be created just by treating the pots as switches as well.

Having said that, there are other intervals that are quite intersting as well - intervals above an octave, but below two octaves produce fine Klee patterns. I've attached a quick sample of a pattern using one octave plus one step -using just a pattern I randomly set up while testing. Settings like this can produce nice quick patterns or drones, still highly variable by pattern, which pots are 'on' and which are 'off', outputs used and gate bus settings. For this sample, I made no effort to program any specific pattern into it (it was acquired randomly), but it still produced interesting results. I'm not after creating a device that takes no effort, but rather something that can provide results without having to master the whole 'loose clay' philosophy of the original Klee, which, though rewarding, can be an arduous task and may be enough to make the prospective builder wonder why the hell he went to all the effort.....

Anyway, if one were to have a ten turn pot in the 2 V setting, one could program one's own interval in addition to the 3rd, 4th, 5th, minor 6th and octave positions. 4V range and 8V range would be above that setting.

But, the pot adds one more front panel control, and ten turn panel pots aren't generally tiny. The thought returns to how the max voltage is set up. Taking up just as many controls as two pots would be two rotary switches, for example. The first rotary would select half steps through just under one octave (twelve positions - 0 through 11 half steps), the second rotary could select octaves( 9 positions, 0 through 8 ). Using two rotaries would allow any interval to be set as the max range. One could use these intervals to program patterns, yet still make 'free form' original-style Klee patterns as well.

In that case, the CD4051 would have to be dispensed with and put more reliance on the rotary switch(es). By the same token, though, calibration may be made easier by using a resistor string with tap points along the string rather than individual trimpots. One pot would calibrate the range.

OK, it may be too late in the development of this thing to make such a sea change. I don't think this is really a case of feature creep, but sort of something that gets to the heart of the whole concept of the sequencer. I could be wrong, though.

Any opinions out there?

Cheers,
Scott

---

octave+2nd.mp3
Description:	One octave plus one step interval.	Download (listen)
Filename:	octave+2nd.mp3
Filesize:	856.87 KB
Downloaded:	1701 Time(s)

---