Super Klee Sequencer

[Coriolis]

Yes, very nice panel, fonik!
I think I will try to come up with an alternative that uses a bit more space than 3 units. I don't plan on using 19 inch racks, but maybe I should try to stick to those sizes anyway, maybe 5 units for a deluxe version, with 25mm knob/jack spacing. I like those cheap 16mm Alpha pots from futurlec, but should I need to put a 24mm pot in there for some reason, that might just be doable too.

Don't have much time right now though...

C

[fonik]

the final zip isn't released yet and we talk about front panels! on the other hand i always found it good to have a vision or idea of the finished product - it helps.
so lets go the whole hog: does anybody plan to use 19'' front panel? is it desirable to develop a schaeffer front panel at all?
which pots are preferred (the turn angle is of interest: 290/300°)? i think i will go for 16mm alpha pots. and then the rotary: i would suggest a common 12pos rotary switch with adjustable positions like this one: http://www.mouser.com/catalog/specsheets/LN-800005.pdf
that would mean 30° per position marker...
any suggestions or do you think it's too early to talk about that?
perhaps i am too far ahead?
anyway: i posted the fpd file, so anybody can use it as a start, if he wants to and i will update the posted version tomorrow considering the rotary switch.

cheers,
matthias

[Scott Stites]

I don't think it's too early to talk about panels - normally that should have been the first step - design to the panel. But, this one started out kind of as an 'out there' idea, and pretty much stayed there . It certainly helps to see a panel just to get a reality check on overhead, etc. It's really nice to see things organized the way you and Coriolis have done it.

That Mouser switch will certainly work. I'll have to find those I got from Allied and take a pic - much smaller, but the supplied knob is fairly small too. Come to think of it, that may not be that big of an issue as a rotary isn't normally tweaked like a pot is. Very nice action on these switches, though. Might come in handy if space becomes an issue.

Cheers,
Scott

[Scott Stites]

I might add that I like the way you subtly divided the first 8 pots and second 8 pots with that line in the middle - small touches like that make the functions so much more intuitive when looking at it.

Cheers,
Scott

[sneakthief]

fonik - i'm building a 19" version, but i'm going to drill and label my own panel. i had actually started working on a front panel a few days ago, but yours is nicer

in case you're wondering (*lol*), i buy my 19" eurorack cases from ebay for 10 euros and then buy blank panels from reichelt.de for 9,70 euros. cheap and cheerful!

i'm also going to use 16mm alpha pots (thanks futurlec.com!)

[fonik]

FYI: i updated the front panel files (i'm not 100% satisfied with this solution but i have a view months left to think about it before i can start with building the klee).

[Scott Stites]

Looking pretty good!

[v-un-v]

[vtl5c3]

ultimately it makes no difference to me who makes a klee pcb - keeping in mind someone has to agree to do it first. I only mentioned Ken because he's already released some other interesting sequencer and pseudo random designs, klee fitting right in with his lineup.

[Coriolis]

Also, Ray has his own, a bit more conventional, sequencer boards he's selling.
But true enough, much of the activity here is probably due to Ray's designs.
People starting out on the Soundlab forum and going from there and such.
And his boards are very nice (haven't tried Ken's).

But that would be Scott's decision.

C

[v-un-v]

[mosc]

[fonik]

cheers,
matthias

[Scott Stites]

Ah, shucks fellas

Actually, I think of this more as an electro-music project more than anything - this is where it sprung up and was born, out of the interaction on this thread. I haven't even done a web page for it. If I do, it'll probably more or less point to this thread.

As far as Ray or Ken doing a PCB, they'd first have to want to do it. I think Ray prides himself on doing PCBs for stuff he's designed, and it may be the same with Ken, more or less. Either is certainly welcome to do PCBs.

Last night I experimented with different max range voltages - I found intervals of a perfect 5th, minor 6th, major sixth and seventh as the max voltage to be quite useful. This approach lends to more repeatibility of a certain sequence rather than the loose clay of yore, though the loose clay concept is preserved as well.

It would require an extra pot, but one setting on the rotary could be 'manual' setting of max range. Click the range switch to that setting, and the user himself could set the maximum range with the pot.

As for the previous samples, figured out last night I wasn't using the A+B in any of them - I never moved the connection on breadboard to VCO! So, all of those samples are really just the A output, one slewed and other CV control and one not slewed. This is a good example of why having slewed and non-slewed outputs available is a good thing. In fact, all of the older samples were set up in that manner. In this case, the circuitry is already there in the standard output schematic - just tap off of the op amps before the slew circuits through a 100R resistor.

Back to the A+B thing, there is much more variation to be explored other than what I just have examples of. Obviously A+B takes wider swings than A or B, because it's added from those two outputs. It makes for some really interesting combinations. Using it along with a voice controlled by the A output last night, I had a sequence going last night that was reminiscent of the main sequence in Peter Gabriel's "San Jacinto". Well, that's what it made me think of, anyway. Oh, and using A to control a voice and A+B to control the cutoff of the filter it's passing through - very cool, indeed.

Also, one VCO controlled by A and one controlled by B is really fascinating - depending on the settings, in 16X1 mode the effect can be almost fractal-like, because they each have the same pattern controlling them, but the self-same portions of the sequence occur at different times.

Cheers,
Scott

[Uncle Krunkus]

The board I'm doing now, I'm thinking of nicknaming the "nightmare" board. The main push was to get it set up so that it would have 16 stage connections (which come from the C&L + Encoder), 16 step connections (which go to the front panel step led anodes & gate switches), and 16 pot connections (which go to the 50K programming pots).
I wanted them all to be in rows of 8, and in the right order so people could just make up 8 way ribbon cables with headers on the end and never even need to think about getting any of it out of order.
Well, because of the way the pinouts on CD4050s and CD4066s are laid out, doing all that has basically taken up a whole board. I haven't started putting the attenuator circuitry on there, and I don't think there's room for it.
My question is, do you think this pursuit is worthwhile? Or am I just being pedantic and remedial?

[blue hell]

[Uncle Krunkus]

Yeah, I've been thinking the same thing.

[Scott Stites]

Forgive me, Uncle K,

I know it's a bit late in the game, but maybe we should think about averting the nightmare.

The one and only reason I put the CD4050s in there was to 'armor plate' the outputs of the CD4034s.

Each output of the CD4034 has three tasks:

1. Address a CD4066 cell
2. Provide a signal to the gate bus
3. Drive a pattern LED

Now, when I originally breadboarded the Model 2, the CD4050s weren't in there and it worked fine. My fear was that I was asking too much of the outputs of the CD4034s by driving the LEDs from them. I didn't want to go the transistor drive route, because that would involve 16 transistors, plus the supporting resistors.

Well, a check of the CMOS cookbook tells me I really only need 12K to drive LED's at 10V, and we're at 15V. I tried 15K, and there was little change in the brightness of the LED. This is an area where I originally stuck in a value, it worked, and I never thought much more about it.

So - a sea change. LED resistor values will go to 15K. That means each LED will be drawing 1 mA when it's on. Each LED, when it's on, will dissipate 15 mW through the CD4034. Take that times 8 potentially 'on' LED's per package, and that comes to 120 mW. If each input is set to high, that's .68 mA per input times 8 which is 54 mA, power dissipation would be 81 mW. So, with everything high, we're dissipating 201 mW, not counting the clock signal and load stuff. Absolute max power dissipation for the CD4034 at -55C to +100C is rated at 500 mW. Throw in the neglible clock,etc. power, and we're still sitting below half absolute max power dissipation, unless you're working in a very, very hot place (above 100C). Any fluctuation in current will not affect pitch, because pitch is not dependent upon the voltage output of the CD4034.

I've reduced this calculation to the following:

{15K Lload||Stupid 4050pinout^2}/(FEARunnecessary)=screw the 4050s.

IE, let's get rid of those stoopit 4050's - I really don't think they're needed at all. I'll work up schematics without them, and replace the 4K7s with 15Ks in all the LED portions. In the meantime, I'll rework the breadboard, put everything under max load, and see if our CD4034s even begin to get warm. Sorry to put you through that.

Cheers,
Scott

[Scott Stites]

I'm going to revise that resistor value to 6.8K. That'll put it over a little over 300 mW worst case for each CD4034, which is still quite comfortably below the absolute max. Reason is, 1 mA may starve some LEDs while others will work fine (the LEDs I'm using light respectably at 1 mA, but I have no idea what their rating is). I figure I'll spec low current Red LEDs, which typically are at 2 mA. 6.8K will put them there. This way, nobody has any surprises =0).

Cheers,
Scott

[Scott Stites]

Klee is under the stress test:

CD4050s have been removed, LED resistor values are 6.8K, all bit inputs are high, it's being clocked at an audio rate, and reloaded at an audio rate. IE, the CD4034s are working harder than they'll ever really be worked in real life (these conditions don't result in any audible sequence, just a constant pitch because of the bits all being high and it being clocked so fast). I'll let her scream along for the rest of the day, see if she survives or the fat boys get hot.

Cheerio,
Scott

[Scott Stites]

Klee rocked right through the stress test - fat boys stayed nice and cool.

Here's a sample comprising three sections:

First part is a sequence using 1V range. A output only.

Second part is the same sequence at a different range setting - IIRC, I used Minor 6th interval(?) - recorded this the other day and didn't take notes. A output only.

Third part is using the A+B as the 'backing' voice, and A is controlling the 'lead' voice (the one under control of both the Klee and the keyboard). Though both parts are derived from the same pattern and pot settings, they sound quite different.

Note on the first two parts, to recreate the patch with the 'standard' output circuit, one would have to use an external slew - both of those use just output A, one part is not slewed and playing without keyboard control, while the other part is slewed and under keyboard control. If one adds the outputs before the slew, one could patch it up without an external slew. The deluxe output section will have the fixed output and a copy of it that can be modulated, slewed and put under V/Oct control. At least the V/Oct control would be a good idea, because to recreate the effects of these samples, one needs a VCO that has two V/Oct inputs, or an external mixer capable of mixing the Klee and the V/Oct signal.

The panel overhead for that, however, is not inconsiderable, but would facilitate a very easy method to get the most out of the Klee. Alternately, if one could find room for six or seven connectors, with no additional controls, three outputs could be added and three or four V/oct inputs could be added - one each for A, B and A+B. The fourth would be a master V/Oct input that would shift all three in unison. In this case, any modulation could be added externally.

First I have to re-do the schematics to update it to the CD4050-less version with the new value of LED resistors.

Cheers,
Scott

[Uncle Krunkus]

There's no need to be sorry.
It was actually good experience and a great challenge to work it out. I'll keep a copy of it JIC. It's really great to feel involved in this kind of collaborative project, 'cos I get to see the issues which arise and I feel like I'm learning more background electronics theory (at an exponential rate) which I've never had a chance to study formally. (and which actually has immediate and practical relevance to the stuff I'm passionate about)
Still, all credit to you for being an informed, impartial and decisive captain of the SS Super Klee! I just feel honoured to be a part of the crew!

"4050less version on the horizon,... sir!"

BTW We just wiped out 3 twisted little cockroaches! Yeah!

[Scott Stites]

[Scott Stites]

Hey Uncle K,

Well, I'm going to shift gears on you again - part of it will affect the layout you've already done - I hope this isn't too much of a pain:

On the part you've already laid out in stripboard, I'd like to swap the LED and emitter resistor around - IE, tie the 4.7K between +V and the collector, and the LED from emitter to ground on both. I'd also like to change the 56K base resistor on each to 100K.

This doesn't change any functionality, but will serve to make wiring up the front panel quite a bit less intense. I'm not sure if any of this will make the rest of the stripboarding less intense - I have a feeling it might help.

The idea is this: the way those LEDs are wired on the schematic, one has the anode of the LED wired to to +V, then returning back to the board to collector of the transistor. If we make those LED cathodes common to a dirty LED ground, that will standardize things a bit. So the cathode connection of the LEDs would all be to this LED ground, and the anodes would be the individual (not common) connections.

Doesn't sound like much - that's where the rest of the changes come in: the sequence LEDs on the encoder are being changed in the same manner. Each of those, as the current design sits, requires *two* wires apiece - that's 32 wires just to wire those LEDs up. If I swap the cathode resistor around to the anode, and wire the cathode of each of those to ground, that turns it into only 17 wires one has to run (actually, it will reduce the wire count more than that, because all LED cathodes on the front panel can be wired to one common point). The gate bus LEDs are being wired the same, and the reference LED ditto. I believe there are 23 LEDs in all. This method will reduce the LED wiring count to 24 wires in all, with the standard understanding that all of the cathodes go to one point - one doesn't have to remember which LED is connected anode to +V and which is connected cathode to ground. So, for the LED's 23 wires go to the panel, one comes back.

I want to increase the base resistor of the transistor driven LEDs to 100K, because it's more standard than 56K, and works quite well. For the transistor driven LEDs (gate bus LEDs, clock, ref, and load), the current resistor will remain at 4K7 - the LEDs I'm using give the same amount of brightness for those with that value as the sequence LEDs do - the sequence LED current doesn't have to pass through any transistor, so they have more voltage to devote to brightness.

I'm testing the redesigned LED scheme now, and it's working just fine.

One other unrelated note: the gate bus schematic defaults to +5V gates and triggers. That is determined by running the outputs of the LM324s through 3K with 1.5K to ground. This divides the original 15V level to 5V and creates a 1K output impedance, because the 3K and 1.5K are essentially in parallel.

One thing has bugged me about that - not all synths will respond to +5V. Though it's the Electronotes standard, it's not a universal standard. For example, Ray Wilson's EG's will not respond to those levels. This is easily remedied - instead of running the output through 3K with 1.5K to ground, run the output through 1.5K with 3K to ground - IE, swap the two resistors around. The same output impedance is maintained, but now the gates and triggers will be 10V. That's a personal preference decision that can be made at time of build, and I'll note it on the updated schematics, which are coming down the pike.

The new schematics will reflect your IC section swap of the Clock and Load, the resistor/LED swaps and new values, the beheading of the CD4050s, and will include the note about gate bus output level choices. Later the decoder schematic will rev once more - the trimmers will reflect changes in choices of the new set of reference levels (which I'm still choosing - 5th, 6th, m6th and 7th are looking pretty good). Optional output schematics will occur as well (enhanced functionality, but not ultimately necessary stuff).

Hopefully these changes will streamline the Model 2 and put it to bed, once and for all.

Cheers,
Scott

[Scott Stites]

This is Matthew Stites, Scott's lab Igor.

Obviously Scott is under a lot of strain - I should let you know there is no load LED, and the number of LEDs is 22, not 23. Back to the Nintendo...