penrose build problems

a773

I have finished my penrose, and I think somethings not right:

1) Without anything in trigger in, shouldn't the gate out send trigger on every new note detected? The only thing I can get from gate out is a copy of trigger in (regardless of note change) if I connect something here. 

2) If I enable only C on penrose, I'd expect inputs of C#, D up to about F to quantize DOWN to C, however they all quantize UP to next octave, so not nearest note, next note up. 

3) All notes enabled on the penrose, playing C on my BSP (same with CVPal) in 7 octaves. First three octaves (0V, 1V and 2V ) gives me C on penrose as I'd expect. However octave 4-6 (3V, 4V and 5V) is detected as C# and ovtave 7 and 8 (6V and 7V) is detected as D. I did calibrate and the output of a penrose generated C is ok. The problem is the calibration of the INPUT of the penrose, it seems like it's stretched, but there's no calibration to be done here. 

Here are a few videos that show how my penrose behavcs:

video 1

video 2

video 3

n0rd

1) "GATE" should output a gate on a note change when there's no cable plugged into "TRIG". (The LED on Penrose should show when a gate is outputted). When there's a cable plugged into "TRIG", the "GATE" will only output when there's a note change and a trigger into "TRIG". (Again, the LED will show when a gate is outputted).

2) My Penrose does the same. I guess that's how it's implemented in code.

3) Sounds like either you need to re-calibrate Penrose or re-calibrate your input signal (ie re-calibrate output from BSP/CVPal).

Steve

a773

Thanks the the reply!

1) so somethings wrong here

2) ok, strange but I'll accept that

3) the BSP and cvpal checks out ok according to my multi meter. No calibration possible here. Seems like Penrose input is wrong.

So any ideas where to start looking for mistakes in my build? Could the two issues be related?

n0rd

1) On your vids, the LED on Penrose appears to be working correctly, indicating when you should be getting a gate. What version PCB do you have? Check out this thread: Altering the Penrose Gate voltage.

3) Re-calibrate Penrose using BSP as input.

a773

I'm away from my modular ATM, but still:

1) I was informed that the R21 mod was unnecessary on PCB 0.4 (which I have). Also the problem is not that the gates are not strong enough, but that they appear at wrong places. On video 3 in my first post, you'll see the penrose gate driving a ladik env (red LED) at 0:47 it's clear that the penrose fires gate out on EVERY trigger it receives, not just the ones where the 1v/oct changes. Then at 0:59 (when the trigger in is pulled from penrose) the gate out no longer fires (red LED stops flashing).

3) If I turn on all notes on the penrose, 0V in is recognized as C (the C button is red), 8V is recognized as D (the D button is red). Using the calibration knob I could make that D output 8V, but it would still be lighting up the D button. So while the calibration is rubberband-stretching the OUTPUT, my problem seems to be calibration is needed on the INPUT before it is seen by the ADC.

I'll go crazy with my meter when I get home, later today, but I'm pretty sure (all due respect) that you're not really seeing the problem as I see it.

n0rd

1) If you are getting gates at different times to which the LED on Penrose shows then something isn't right.

3) Sending 8V to Penrose and it quantising to D8 (and not C8) suggests that Penrose needs re-calibration. It suggests that each Penrose step is close to 1/12th of a Volt but slightly higher. So for the first couple of octaves this small amount extra doesn't matter but as you increase the number of octaves (and thus the total number of steps) this small amount has an cumulative effect. This is why you said that 3V, 4V and 5V quantises to C# and higher volts quantises to D etc.

a773

Ok, back with my meter, somethings definitely wrong. But maybe the problem is somewhere else. Here are the voltages of BSP in 8 octaves and the corresponding voltages from penrose out (only C enabled):

BSP   PENROSE

0.00   0.00

0.99   1.00

1.99   2.00

2.99   3.00

3.98   5.00  Oops, penrose jumps 2 octaves, somethings wrong!

4.98   5.99

5.98   6.99

6.98   8.00

I've reflown most of PCB1, especially the ICs, but to no avail. Can't see anything wrong when visually inspecting the board, so I could really need some pointers regarding which section of the build could be the problem.

NB. regarding 1) Yes, the LED flashes just as I expect, but the gate out doesn't follow the LED, so somethings rotten here too.

n0rd

Those numbers appear to be related to what you originally stated in 2) and that Penrose step is not calibrated close enough to 1/12. (ie it's slightly higher). So after a few octaves, the cumulative effect of the slightly larger step means it quantises incorrectly up to the next octave.

As for the gate issue look at the schematic and check corresponding components on PCB. Given the LED seems to work correctly maybe it's the 3.5mm jack. Check that they're installed in the correct orientation. (Left and right jacks have a different orientation - Check position of jacks here).

a773

Gate issue fixed, R21 was soldered wrong. Output scaling problem persists...

a773

Still struggling with my penrose. Here are voltages in + out of penrose, all notes enabled:

    BSP   Penrose

C   0.00  0.00

C#  0.08  0.07

D   0.16  0.15

D#  0.25  0.24

E   0.33  0.34

F   0.41  0.41

F#  0.50  0.48

G   0.58  0.57

G#  0.66  0.67

A   0.74  0.74

A#  0.83  0.81

B   0.91  0.90

C   0.99  1.00

C#  1.08  1.07

D   1.16  1.14

D#  1.24  1.24

E   1.33  1.34

F   1.41  1.41

F#  1.49  1.48

G   1.58  1.57

G#  1.66  1.67

A   1.74  1.74

A#  1.83  1.81

B   1.91  2.00  C

C   1.99  2.00

C#  2.07  2.07

D   2.16  2.24  D#

D#  2.24  2.33  E

E   2.32  2.33

F   2.41  2.47  F#

F#  2.49  2.57  G

G   2.57  2.67  G#

G#  2.66  2.74  A

A   2.74  2.81  A#

A#  2.82  2.90  B

B   2.91  3.00  C

C   2.99  3.07  C#

C#  3.07  3.14  D

D   3.16  3.24  D#

D#  3.24  3.33  E

E   3.32  3.41  F

F   3.41  3.48  F#

F#  3.49  3.57  G

G   3.57  3.67  G#

G#  3.66  3.74  A

A   3.74  3.81  A#

A#  3.82  3.91  B

B   3.90  4.00  C

C   3.99  4.08  C#

C#  4.07  4.14  D

D   4.15  4.24  D#

D#  4.25  4.33  E

E   4.32  4.41  F

F   4.40  4.48  F#

F#  4.49  4.57  G

G   4.57  4.66  G#

G#  4.65  4.74  A

A   4.74  4.81  A#

A#  4.82  4.90  B

B   4.90  4.99  C

C   4.99  5.09  D LED lights up

C#  5.07  5.15  D LED flashes

D   5.15  5.27  D# LED lights up

a773

I should note that

1) the voltage regulator get's pretty hot and

2) I accidently soldered R21 in the smallish hole under the resistor, I don't know if this could have blown something up that's causing the current problem

Regarding nOrd's suggesting that I calibrated wrong: At least on my penrose the calibration only affects the output CV, not what notes penrose recognises input voltages as meaning what buttons on keyboard lit up. In other words: I can turn that calibration pot all the way, and my penrose, still put the notes in the higher octaves in "wrong boxes" either 1/2 or 1/1 note off.

Frankly I'm beginning to think (in desperation) all penrose behave this way, I suppose most people send CV from a sequencer with continuous CV knob or from a random source (noise, nano rand or touring machine) to the penrose in which case they won't notice this problem/phenonomen. So I'd be extremely happy if someone with a midi2cv interface or a quantizing sequencer would confirm that with all notes enabled on penrose, a chromatic scale through 8 octaves indeed makes the correct knobs go red on the penrose.

Enough speculations, back to supposing my build is wrong: My logic tells me something is slightly off between the input and the atmega, as if they should be calibrated. Could it be a resistor having an unprecise value?

Looking at the schematic, I checked R6, R7 and C16. I noticed it says 10V->5V below R6, could anyone elaborate on what that means, what's the role of R6? It seems to have a more specific value than the other resistors. Also is it ok for the voltage regulator to get hot?

a773

Ok, being a sofware-kind-of-guy, I tried scaling the input in software, by inserting 

input = input * 0.98;

in line121 of quantizer.c (first thing in quantizeValue). Seems to work perfectly.

But before I go on with my life, I'd like to get a comment on:

1. Would it be safe to use the module with this software fix (esp is it ok for the voltage regulator to get hot)?
2. Would you agree that the problem might be caused by an unfortunate combination of actual resistor values, esp R6 being off?

Basically a happy camper here :-)

FadingCanvas

I had my voltage regulator quite hot too, don't know if it's normal or harmful.

a773

Ok, thanks, maybe it ok then. I left it singing fof a couple of hours today, no magic smoke...

ArguZ

Hello there,

I would like to chime in and confirm that i have the same offset in incoming pitch.

The extremely accurate BSP outputs , i measured around 0,002 V fluctuation with a good voltmeter turn into 1% offset which results in 2+-,05 notes offset on the high C.

I really like a773s idea of just changing the firmware to accustom ones personal RI offset.

Alternatively a 2K poti would be a nice option to calibrate the input gain.

a773, could you please post a link with your firmware, i'd like to try in on mine.

I am no software-dev guy, so i would not know how to bake my own  

Oh, and is the original FW available somewhere as a fallback solution ? 

Cheers

Sascha

a773

My hacked firmware is here:

https://dl.dropboxusercontent.com/u/4343030/penrose.wav

If your hardware is not off by the same as mine, you have to modify the quantizer.c, but you can try with my values first, maybe it's just right :-)

To update, you need to play it (loud) to the trigger input as described here http://www.sonic-potions.com/penroseAssembly

Update procedure

Connect your audio player to the trigger input jack

Hold the first note select button down and turn the power on

The red LED of the button will start to flash red quite fast, indicating the module is now in firmware update mode.

Start the audio playback of the firmware file

The LED should switch its color to green and continue to flash

If the LED lights up red and stops flashing an error occured -> see troubleshooting.

Once the update is finished the quantizer will start the new firmware automatically

Not sure if the original firmware is available as wav, but I can make on for you, if mine doesn't fix your problem.

Hope it helps!

ArguZ

Sweet, thanks for the speedy reply

I am also having the no gate issue, but someone here of the tech department will inspect my unit later today.

I tripple checked my components and have tried the reflow on both boards with no change at all.

So there is a tiny chance the input scaling might be introduced by another problem down the line..

We will see.

Nether less i am having a lot of fun using it behind my Turing machine...they go well together.

a773

Ok, maybe you (like me) soldered R21 in the small hole that's supposed to be under the resistor instead of the correct larger hole. That's why my trigger out wasn't working, at least (if I remember correctly)...

ArguZ

haha, kind of..but more creative

I turned R2 and R9 90 degrees around their Y axis.

So they are in the right place, just poking into each others hole.

And of course someone else just saw the flaw right away.

We looked for hours and could not see it.

The effect was that the LED started to flash per gate output, but dimmed away, including the output being almost zero on the Gate, just not all zero.

ArguZ

Hi, so i fixed my gate out problem now and flashed it with your firmware (RYO AmpMix made it happen)

And guess what, its now off by ONE note

So, if you find the time and could do your magic with even less scaling i would be very happy to run that FW.

I am in contact with Julian regarding that matter, we might see an official FW in the future.

a773

It might be better to do your own adjustments, it might take some attempts to get it right. I'm pretty stressed out until after the weekend, maybe then I can show you how to roll your own. If I forget to get back, feel free to ask again 

ArguZ

Aight..better in the long run