Super Klee Sequencer

[Scott Stites]

Thanks, guys. Sometimes when the Klee goes off on a tangent, I just want to stand in front of the breadboard with my lighter held up....

[Coriolis]

Man, I would love to see a quick example gif of a frontpanel for this thing, or for differences between Klee 2 and 3.

I mean; I get that it's probably going to be 16 pots and three outputs (A, B, A+B), but beyond that...? Oh, and some cv-inputs, an external clock input.

Maybe just a quick list of ins, outs, pots, switches, functions?

Maybe I should just let you finish this thing!

C

[Scott Stites]

Coriolis,

First of all, there are four rows of '16'

16 LEDs
16 Voltage Pots
16 Pattern Switches
16 Bus Switches

LEDs probably should be right above the pattern switches, with the pots right below the pattern switches, and the gate bus switches below the pots.

Control section will have a momentary load switch, a momentary step switch, and a toggle Bus 1 reload switch. In addition to that, there is a toggle 8X2/16X1 switch, a toggle Random/pattern switch, a random reference pot and a random source level pot. I haven't included it on the schematics, but a simple toggle switch to prevent the clock from shifting the sequence would be handy - if one has a sequence acquired through random methods, and wants to use that same sequence at a later time, one stops the sequence and flips the pattern switches to match the LEDs. Right now the only way to stop the pattern from shifting is to disconnect the clock source. This switch could be wired entirely on the panel.

Control inputs are clock, load, and random signal.

Control indicators are clock LED and random reference LED.

The gate bus control section will have three 'merge' toggle switches. Outputs are master gate, master trigger, gate 1, trigger 1, gate 2, trigger 2, gate 3 and trigger 3. Four LEDs will serve as gate bus indicators - master, bus 1, bus 2, and bus 3. If one places these LEDs close to the clock LED indicator, the master gate LED could be eliminated - it will always be synchronous with the clock. If they're spaced apart from it, though, it's handy to have the Master LED next to the rest of the bus LEDs just so one can observe the relationship of the master to the three busses.

Output section will be A, B, and A+B. Each of these outputs will have a lag control.

It's up in the air about the offset and modulation inputs and controls.

It looks like a lot of controls, but if you peruse through any other analog step sequencer, it's about on a par. Check out the ARP 1601, the Milton, or the Modcan sequencer and count the controls/connectors on those. I've sort of perused a 4U rack panel, and it could fit on that without too much of a squeeze.

I plan for Model 3 skip, stop/start, gated/triggered operation and sychronous functions (IE, you send it a reload pulse, and it waits for a clock pulse before actually loading, things like that). There will be an additional gate bus function that fixes the gate time at the length the stage is high (the current merge function will do that if there are no adjacent active steps). That's also the one that will have additional sampled and held outputs, under control of the gate bus. I'm also considering two control sections for each 8 bit register, so that when in 8X1, two different tempos can be employed.

Model 4 will have the features of 2 and 3, but there will be an additional, standard single bit up/down sequencer built into it, which means another row of LEDs and another row of pots. That is the one that will reflect the diagram in the first post of this thread. With that, either row of pots can be selected to be controlled by the Klee section, the up/down one bit section, or both. The gate bus source will be selectable in the same manner. It will also have the 'interlace' function, which will allow it to produce 32 step sequences from the same Klee pattern, up/down sequencer, or both.

That one will be big.

Cheers,
Scott

[blue hell]

[Coriolis]

Thank you so much for breaking it down a bit, Scott!
Now I have something to refer to, when I read through this thread, to try and understand what's going on in this circuit.

I whipped up a quick layout of the Model 2 in Autocad.
Not pretty, but for reference, it works.
Everything is spaced 25mm apart, this works nicely for 0,5 inch knobs on 16mm pots, and 4mm banana jacks, which I use.
There is slightly more than 25mm between the top four rows and the two control sections.
Looks like a model 2 fills a 5-unit rack panel the way I have it set up.

-White crosses are toggle switches
-Yellow circles are pots
-Green crosses with little green circles are in/out jacks
-Red circles/crosses are LED's

Hope it's useful to someone else...

C

EDIT: Dang! Those letters are kinda small to read...

---

Klee2.png
Description:	A quick and ugly panel-layout of the Super Klee Model 2...
Filesize:	7.57 KB
Viewed:	481 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[v-un-v]

[Coriolis]

Those look nice!
I'm using Autodesk ABS for my schoolwork, but if I had to have something for hobby use, I'd rather eat than pay license fees for it, so it's good to see there are cheap alternatives!

C

[v-un-v]

[Scott Stites]

[Uncle Krunkus]

I just had a look at the decoder schem, and realised that I still don't understand exactly what's going on!
Doesn't matter, I can still get it all on stripboard and work it out later.

I've decided to put the 4050 buffers on the decoder board. There's not much else on there, one side of every buffer goes to a stage controller (4066) anyway, so they are multiple board interconnections which would have to happen anyway. This way I can sort out the 4050 connections into neat little sequential lines. That makes it easy to connect with ribbon and 8 way headers. No one needs to worry about getting out of order. (except for me!) Why they couldn't put all the 4050 inputs down one side of the chip, outputs down the other, power back where it should be and an extra buffer across the NC pins, God alone knows!

So the first board will be served up shortly! Just have to get the buffers off it and tidy up.

[Scott Stites]

Those buffers are hideous. Agreed on the pinout - why in the world......

Can't wait to see the board. To be honest, I'll have to review where the saga left off - decoder, I believe. Time for another chapter. In short, the decoder has a step attenuator made up of the resistor/trimpot pairs for the different voltages. Notice the op amp with the 10/20K resistor divider - this buffers 10V, which each resistor/trimmer divides into the approprate voltage. Each of those voltages is sent to the input of a 4051.

A rotary switch sends +V to the diode matrix, which selects which voltage the 4051 will put out. That voltage is buffered and sent to each CD4066 section (4 CD4066's, 4 sections per CD4066, 16 sections total). A CD4066 section will pass that voltage when one of the the fat boys sends a bit to its control input. The output of each CD4066 section feeds one of the step pots. In this manner, the selected voltage range is presented to the top of the pot. Pots 1 through 8 feed 8 resistors which are summed together to be sent to the A voltage summer. Pots 9 through 16 feed 8 resistors which are summed together to be sent to the B voltage summer.

Forgot to mention in the last post, but matched resistors are pretty much a necessity on the mixer board and decoder, if one wants to have accurate ranges. Not that big of a deal - I just took a bag of 100K's and sorted them out with an Ohmmeter. The most of the same value I had was 99.8K, so I put those into the decoder, and the op amps that sum the voltages - BIG difference in tunability. I also put in offset trimmers to the first two op amps. The way it works, one op amp sums A, one op amp sums B, then the output of those two op amps go to the A+B mixer. Well, if those two have an offset, they'll add up in A+B, which will offset the output of A+B even more. Not a whole lot, a few millivolts each, but they do add up.

I should also mention that the range trimmers should be multi-turn pots. That makes it easy as fishin' to cal the voltages in.

Had a rough time posting that last post - dial up, PeoplePC, don't ask...but in the time it took to post it, while it was uploading the file for the third time, I went downstairs and tweaked another pattern that is just blowing me away. It's one of those that, no matter what key I hit on the keyboard, it just seems to fit.....sort of an arpeggiator that just won't let you hit the wrong key. I did it using the 1V range and setting one - just one!! - pot dead center and shaking up the pattern a bit then switching the gate bus around. Holy cow....

I'm not getting anything done.....again. Last thing to do is test that Gate Bus parts reduction. Must...get.....it.....done.

Cheerios,
Scott

[Scott Stites]

Hmmm....thought I should mention it - I've made no mention of it on the schematic, and it's not normal protocol to mark it as such, but you'll notice some of the ground symbols have a little 'A' next to them. This is 'analog ground'. IE, they should share their own trace/jumpers that go back to the source only at one point. Also, there is +VA and -VA, which have their own special little filter caps. Just insurance to keep any jitters out of our triocular exotic dancer.

Cheers again,
Scott

[Scott Stites]

The shiny bauble I mentioned a post or two ago....

1V range, 16X1 mode, all pots either hard left or hard right, except for one.....

Cheers,
Scott

[Uncle Krunkus]

Yeah, that's cool, I'm hip to the whole "analogue ground" idea. I decided to incorporate it wherever appropriate from now on. My new power supply (when I get around to building it!) will have 4 way connectors to support DGround & AGround.
Multiturns on the range trimmers - check.
What do they reference from? - might not matter if they're all set at the same time?

[vtl5c3]

not as raunchy as toadskin teddy, but I like it. Imagine if they had built a superklee into grampa's wurlitzer.

[Scott Stites]

[Uncle Krunkus]

Just while I remember Scott,
I've swapped the comparators on the encoder schem.
I'm using pins 1, 2 and 3 (6&7 shorted, 5 to ground)
Almost done.

[Scott Stites]

Noted.

I didn't get through the whole Gate Bus reduced parts count experiment, but I ensured eliminating the first two inverters on each gate bus output (right before the NANDs) had no ill effects. One can Mickey Mouse a NAND leg as well as an Inverter (why does that sound vaguely obscene?).

I'm pretty sure I can get rid of those last two inverters on the Master Clock output as well. The one thing after that to check is to make sure the quad 324s will work fine instead of the dual 358s. I gotta find some breadboard space somewhere to try that out.

I'm wondering if a Zener on the voltage reference might not be a bad idea. I haven't had any problem referencing straight from the rail, but who knows what kind of power supply people will be using. I don't know if I have any of the value I'd want (9.1 to 10V) in any event, it wouldn't harm the stripboard a jot - the 20K would just be replaced by a zener. Shouldn't affect the resistor/trimmer values either.

Cheers,
Scott

[Uncle Krunkus]

Super Klee Stripboard (Klee-B1)
All of the Clock & Load, and most of the Encoder.
Double checked, as yet unbuilt.
All the parts from the encoder schem have 200 added to their designator.
With no buffers, I've been able to make a little bit more room. There's space to use proper headers for the data lines, (if people want to).
There are 84 links all up.
16 of which are under chips. I've shown them on the copperside pic (in white) for clarity but they should really go on the parts side before the IC sockets go in.
If there are any queries or problems, let me know. I'll edit this post to keep it the only accurate one for this board.

Edit (9-Oct-06) Updated layout.

---

Klee-B1.jpg
Description:
Filesize:	254.5 KB
Viewed:	336 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

---

Klee-B1CS.jpg
Description:
Filesize:	305.86 KB
Viewed:	294 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[Scott Stites]

Sweet!! Excellent work - quite amazing to get all of that on one stripboard. Nice apportionment of tasks - it makes more sense to put the CD4050s with the gate bus, since their main task is to drive the LEDs and provide the signals to the gate bus. Other than those blasted CD4050s, the gate bus will probably be rather stripboard friendly. Especially a reduced part version of it - gates aren't scattered out quite as haphazardly.

Not sure if the rest will fit on one board (doubtful) but it makes it obvious that the decoder and final mixer could share a board - it's the only section that uses +/-V and all that analog ground crap - it would make it easy to keep the grounds and power separated.


I didn't number the ref des's consecutively as a whole project, it made it easy to work out each section. I was going to go back and do something quite similar, so that'll work out - I'll do that with the schematics. Stay tuned for 'Showdown at the Decoder Corral', coming up next.....

Cheerios,
Scott

[Scott Stites]

Episode 3: Showdown at the Decoder Corral

I've uploaded the current version of the Decoder schematic. Actually, it's the "Scaler/Decoder" because it also scales the maximum value each of the sequence pots is capable of providing.

Moving from left to right, we first see the 10V reference. R43 and R45 form a voltage divider that derives a 10V signal from the +15V rail. This 10V is fed to U5B, which is configured as a voltage buffer. The output of the buffer provides a low impedance 10V source that is fed to 8 sets of resistor/trimmer pairs. These resistor/trimmer pairs, SW1, and U6 form the stepped attenuator which is used to set the output range of the Klee sequence.

Each of these resistor/trimmer pairs is responsible for generating the range voltages, which are 1/12 V (0.0833V), 1/6V (0.166V), 1/4V (.025V), 1/2V (0.50V), 1V, 2V, 4V and 8V. Each of these voltages is fed to a pin of U6, a CD4051 1 of 8 multiplexer.

SW1, an 8 position rotary switch, R33 through R35, and D1 through D12 provide the means to control which of these range voltages becomes present on pin 3 of the CD4051. D1 through D12 form a diode matrix switch which converts the positive voltage selected by SW1 to a 3 bit binary address, which is used to determine which of the input pins of the CD4051 is connected to the output pin, pin 3. R33 through R35 serve as pull-down resistors so that when an address pin does not have a positive voltage on it, it will be reliably pulled to 0V.

The output of U6 is buffered by U5A and sent to the 'stage activation' section of the decoder board. This voltage is applied to the input of 16 total CD4066 sections, each of which acts as an On/Off switch. The control input of each CD4066 section is connected individually to one of the output pins of the CD4034 'fat boy' shift register ICs in the encoder section. These are what the labels "Stage N" ('Stage 1', 'Stage 2', etc) signify.

The output of each CD4066 section is connected to the top of a sequence programming pot. These are the pots on the front panel of the Klee that the operator uses to program the sequence voltages.

When the control input of a CD4066 is brought high by one of the shift register ICs, it passes the voltage applied to it from the stepped attenuator section to the top of its dedicated control pot. The pot then forms a voltage divider, variable by the operator, to set the output voltage for that particular stage. When the CD4066 control input is low, the output of the pot wiper will be 0V.

Each of the pots feeds a 100K resistor. You'll notice that the 16 resistors are grouped into two sets of eight resistors. The first group of eight resistors are combined to put out Step Mix A and the second group of eight resistors are combined to put out Step Mix B. This allows the Klee to operate in either 8X2 or 16X1 modes. Step Mix A and Step Mix B each go to the as yet unseen voltage mixer section of the Klee, the final piece of the puzzle. The 16 resistors forming Step Mix A and Step Mix B must be matched to within 0.1% - this allows a fairly precise mix of voltages to reach the voltage mixers when each pot is cranked wide open. This allows that all pots cranked wide open will supply an accurate mix of the voltage provided by the stepped attenuator.

Matching the resistors is a piece of cake. I just took a bag of them, measured them on my DMM, and sorted them out in piles according to what they read on the DMM. I had more 100K resistors with a value of 99.8K than I had of any other, so that's what I used for the breadboard. When measuring them, I made sure they read *solidly* 99.8K - if it looked like the DMM wanted to flicker down to 99.7K or up to 99.9K, I rejected those resistors from the pool. Make sure you have around 25 of them that match up exactly (you'll need a few more down the road in the voltage mixing section).

I used 50K pots for the programming pots - this allows the pot to not be loaded down by the output resistor, and allows for a smoother range of adjustment in the pot.

Notice the voltages have an 'A' by them, and the ground also has an 'A' - this signifies that these paths must go back to the source directly, and not be mixed with all that nasty digital crap that's going on elsewhere in the Klee.

And that's it, in all it's anticlimatic glory......


Cheerio,

Scott

Edit: Schematic has been revised since this post. Zip file with the latest revision is downstream in this thread. Check first post of thread to find exact whereabouts.

---

Klee_decoder_1.JPG
Description:	Klee Decoder Rev 1.00
Filesize:	252.67 KB
Viewed:	648 Time(s)
This image has been reduced to fit the page. Click on it to enlarge.

---

[v-un-v]

These mp3's are beautiful!

So is the stripboard

Thank you so much

[Scott Stites]

Yes, a mighty thank you to Uncle K for going where no stripboarder has gone before.....

Cheers,
Scott

[Uncle Krunkus]

Hey Scott,
you probably already explained this, but why the 4051? Why not just switch 1 of 8 trimmers through to the buffer? Is it the wire runs up to the front panel and back? Yeah, that's probably it. Thanks for the quick reply!

As for the layout, thanks. I think it looks pretty good too! The question is, does it work? Who's gonna test them out? Hey, that can be your job Tom! I'm sure you can take a month off from your studies to build a Super Klee can't you?

[Scott Stites]