[BUILD] Bytes Turing Machine expander

From circuitbending to homebrew stompboxes & synths, keep the DIY spirit alive!

Moderators: Kent, Joe., luketeaford, lisa

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

[BUILD] Bytes Turing Machine expander

Post by windspirit » Wed Aug 20, 2014 1:21 am

Ok here is the link to all of the relevant information on the Bytes expander. I know synthcube has PCBs in stock, Grayscale panels are on their way, and thonk will have PCBs soon.

http://www.circuitshaman.com/2014/09/bu ... ander.html

Image

Image

Image

Here is the trace cut, you think you can handle it?

Image

******************TROUBLE SHOOTING**********************


SYMPTOM: If you had working locking on your un-modified turing machine but no locking when you added Bytes:


One builder accidentally used a 4051 instead of the second 4015 on the expander. Double check that this is not your problem.

I recently found a bug on 2 of my units, it's tough to say what the cause is but it seems like the extra draw of the expander was causing some noise on the original turing machine board. This was causing a fluctuation in the comparator reference voltage to compare the noise value to, causing tiny crackles on the data input of the shift register. The solution that I found was to increase R6 to 220k, increasing the gain of the summing amp which drives the noise comparator. This resistor is all the way to the left of the TL072 at the bottom of the board, just above the clock input jack.

SYMPTOM: You're unit's buffer clears for long periods of time. This happens especially in a short loop length (4 or less) with the prob. knob set to around 1 oclock.

This is actually a problem with the original Turing Machine design that the Bytes Expander brings out at certain settings. Basically the transistor which generates noise is supposed to be chosen out of a batch of them for preferable noise characteristics, but most people dont realize that and so they place any old 2n3904 as their noise transistor. In my own builds I have found a success rate of roughly 1 in 10 transistors, this means that there are a whole bunch of builds which have poor noise transistors in them! If you're transistor allows your unit to lock properly you probably haven't noticed any strange behavior, but the transistor's characteristics can also be responsible for the amount of randomness in the middle of the probability range. Therefore if you have a bad noise transistor, you will get a lot of repetition even if your prob. knob is close to noon, making it so that when the buffer is clear it will stay clear for a long time. This means that if you have a short loop length you have a higher likelihood of getting a clear buffer and then your buffer will stay clear for awhile.

Anyways if you just skimmed that don't worry, just replace your noise transistor if you experience this behavior. It's the transistor closest to the TL074 towards the bottom of the board. I have heard that 2n2222s work better than 2n3904s but I can not verify this. You can either socket your transistor or you can leave it unsoldered while you test it and bend the legs to make good contact but not contact any of the other pins (there is enough room for this, it's how I usually do it.)

What I do to test is:

1) After the transistor is situated and you're rack is fired up, give the turing a really fast UNIPOLAR clock. Then put the turing in locking mode (prob. knob all the way clockwise) with a loop length of 8, then fill the buffer with 1s and leave your TM alone for a good couple of minutes (I will usually test for at least 15 to be absolutely sure). If you come back and your Turing's buffer is still filled completely with 1s then you're transistor will lock correctly.

2) Do the same thing except with anti-lock (aka inverted locking, prob. knob all the way couter-clockwise.) Generally if one works so will the other but I have come across a few transistors which worked for lock but not antilock. If you clear the buffer it will fill with a checkerboard pattern of alternating 1s and 0s, with each lasting for 8 cycles. If you leave it locked like this with a fast clock running and come back 5-10 minutes later, it should still have the same checkerboard pattern. If it does you're anti-lock works.

3) Now this is to test for randomness. With a small loop length (3-4) and still a fast clock, put the prob. knob around 1-oclock (just right of noon). Now clear the buffer several times. If you have a good noise transistor the buffer should fill up right away. If not it will take several seconds or longer to fill up. You can also usually tell if your noise transistor is not random enough by putting your Turing Machine in a length of 8 and a slow clock with the prob. knob around 1 oclock. If it is no good you will notice that the pattern does not change very much, at this setting you are supposed to have a whole bunch of random!

Anyways I will probably upload a video demonstrating my calibration method for transistors.

BOM:

Harware:

3 x 100k linear right angle potentiometer ( Alpha # RD901F-20-15R1-Value or equivalent )

1 x on-off-on toggle switch

1 x 100k trimmer potentiometer, bournes vertical adjust multi-turn with legs in a straight line

ICs and silicon:

1 x TL072

1 x TL074

2 x ADC0820

2 x CD4051

1 x 78L09 voltage regulator

6 x 1n5817 schottky diodes (do not use another kind of diode)

Sockets and headers:

3 x 16 pin IC sockets

1 x 8 pin IC socket

2 x 20 pin IC sockets

1 x 14 pin IC socket

1 x 40 pin breakaway header (or equivalent 16 pin IDC headers for expansion ports)

2 x single breakaway sockets (optional - for jumper sockets)

Caps and resistors:

6 x .01uF ceramic capacitors

10 x 470k resistors

2 x 100k resistors

5 x 2.2k resistors

Fuses:

.12A Resettable fuse (trip current = .24A)
Last edited by windspirit on Wed Dec 10, 2014 2:44 am, edited 2 times in total.

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Sun Sep 21, 2014 2:17 am

Updated the info, here we go!

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Wed Oct 01, 2014 7:28 pm

To all of you who have already bought the BOM there is a slight update, you need to buy a new fuse for your backpacks. This is only for those of you who are adding all of the expanders to your turing machine as the old Fuse trips when all of the expanders warm up.

Here is the new fuse:

http://www.mouser.com/Search/ProductDet ... 76-250R120

The BOM has been updated on the website.

User avatar
Zaibach
Common Wiggler
Posts: 202
Joined: Sat Jul 26, 2014 2:13 pm
Location: Germany

Post by Zaibach » Sun Oct 26, 2014 8:04 am

Hey guys,

First of all, thank you for this module! I received mine yesterday and can't stop wiggling!

But, there seems to be an issue with the panel. I got my bytes expander from Thonk with the Spiro panel. At first I was totally confused when I wanted to adjust the lenght for a 8-step sequence and nothing happened when I wiggled the upper pot, but I could adjust the lenght from 1 to 8 steps using the lower pot, which was marked with the numbers 9-16. After a few hours double and triple-checking my build I noticed that there are apparently panels with different layouts concerning the lenght pots and the switch. I attached a picture, from left to right:

1. Circuit shaman website
2. Bytes user manual
3. Grayscale's website
4. Thonk spiro
5. PCB - note the markings for long and short for the pots

Image
I think the picture from the Bytes user manual is a preliminary version, but when you compare the Thonk and Greyscale panels, the marking for the short and long pot are different.

My problem was solved when I wired the pots so their function matched the panel. Not pretty, but now It works as advertised :nana:

Image

User avatar
Thonk Support
aka Monobass
Posts: 8818
Joined: Fri May 29, 2009 5:39 pm
Location: Brighton, UK

Post by Thonk Support » Tue Oct 28, 2014 7:39 am

ack, I had no idea about that, I used the grayscale panels as reference... and obviously ended up somehow using two as reference!

I'll get new panels made up.
Thonk - CLICK HERE - Modular Synth DIY + Eurorack Accessories Store

AS of October 2020 this account which used to go by the name of Monobass is now manned by various staff at Thonk.

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Wed Oct 29, 2014 2:44 am

Ouch sorry about that. The other panel was a first draft, we decided to switch the numbers around so that they would reflect the action of the loop length switch (aka long is up and short is down.) I guess Steve somehow got an old version of the panel for reference. Sorry about that Steve!

Anyways I'm glad to see that someone else has built the expander and that it is working for them. I think you may be the first one :party:

User avatar
Zaibach
Common Wiggler
Posts: 202
Joined: Sat Jul 26, 2014 2:13 pm
Location: Germany

Post by Zaibach » Fri Oct 31, 2014 11:57 am

windspirit wrote:Anyways I'm glad to see that someone else has built the expander and that it is working for them. I think you may be the first one :party:
Well, I don't know if I am really the first one :razz:

On a related note, maybe useful for those who are about to use the stylish acylic panels from thonk. I used the recommended jack from the build doc but it turns out that the thread is too short to reach through the thicker acrylic panel. I don't know if there are that many manufacturers for this kind of jack, but i couldn't fix mine on the panel with a nut.

Image

I just used a normal jack and wired it, no biggie...

Image

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Fri Oct 31, 2014 8:33 pm

Both good ideas, glad this stuff is ending up on the first page.

User avatar
woodster
Super Deluxe Wiggler
Posts: 1348
Joined: Mon Mar 05, 2012 9:36 am
Location: UK

Post by woodster » Sat Nov 01, 2014 10:29 am

I recently built one of the Bytes expanders and think it's brilliant.
There is so much fun to be had with multiple sequencers set to different step lengths.
I'd like to offer windspirit some :yay: for the documentation.
It all made the whole build pretty straightforward.
This Turing I've just upgraded had Black PCB's, I'd built the Bytes on a day of terrible weather, so had quite poor light conditions.
I couldn't really see the trace clearly, so the pictures really helped.
The Fuse swap was fine, as the Fuse is in the most accessible place you could hope for (if component swapping is your aim).

Regarding the Jack Socket and the Thonk Acrylic panel, I did manage to secure mine, granted the nut isn't quite fully threaded, but it works for me...

Image

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Sat Nov 01, 2014 10:32 am

Thank you :tu:

If you are using the suggested jack it honestly shouldn't matter if it isnt bolted to the panel since everything else should be keeping it pretty snug.

MatMillion
Common Wiggler
Posts: 51
Joined: Fri Sep 13, 2013 2:55 pm

Post by MatMillion » Wed Nov 05, 2014 4:23 pm

I just got a PCB and panel from thonk. (also got one with the wrong layout.)

Unfortunately the mouser cart is currently not available. Did anyone save it and wants to share? :love:

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Wed Nov 05, 2014 5:27 pm

FARK. I have had my last THREE attempts to save a mouser project dissappear! Does anyone else know wtf is going on?

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Wed Nov 05, 2014 5:29 pm

BTW I will have a new cart uploaded by tonight.

MatMillion
Common Wiggler
Posts: 51
Joined: Fri Sep 13, 2013 2:55 pm

Post by MatMillion » Thu Nov 06, 2014 5:09 am

thank you!

User avatar
Thonk Support
aka Monobass
Posts: 8818
Joined: Fri May 29, 2009 5:39 pm
Location: Brighton, UK

Post by Thonk Support » Thu Nov 06, 2014 6:31 am

ok I have new panels and will be contacting everyone who has purchased one tomorrow to offer a replacement, so hold tight :)

The text is fixed and the jack hole is countersunk so you can properly get the nut on.

Image
Thonk - CLICK HERE - Modular Synth DIY + Eurorack Accessories Store

AS of October 2020 this account which used to go by the name of Monobass is now manned by various staff at Thonk.

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Thu Nov 06, 2014 7:46 am

Aw yeah get that nut on!

(You set me up)

User avatar
jules
Ultra Wiggler
Posts: 841
Joined: Wed Jun 23, 2010 6:27 am

Post by jules » Thu Nov 06, 2014 9:39 am

Hello,

Which ADC0820 are we supposed to get?
There's one from TI with 1 channel, and another form Maxim with 2 channels..?

EDIT: by the way, there's a problem on your website with the link to the mouser card... "error=true".

EDIT 2: there's a picture on the build doc showing that you're using the maxim.
Just have too look really hard.

EDIT3: the bom should mention 3x 4051 (not 2).
Dieter: "who needs all this??!!??"
tak poo-poom

User avatar
woodster
Super Deluxe Wiggler
Posts: 1348
Joined: Mon Mar 05, 2012 9:36 am
Location: UK

Post by woodster » Thu Nov 06, 2014 11:34 am

jules wrote:
EDIT3: the bom should mention 3x 4051 (not 2).
I'm sure the 3rd 4051 relates to the need to remove one from the Turing Backpack.

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Thu Nov 06, 2014 5:00 pm

Yes mouser has been in the habit of deleting my saved carts for some reason. Ill have a new BOM up soon.

Actually there are only two 4051s, the other chip is a 4015 that you grab from the turing main board. I think you got mixed up :-).

Ill double check but I think I have used both adc0820s and they both worked.

User avatar
clive.grace
Common Wiggler
Posts: 100
Joined: Tue Oct 28, 2014 4:40 am
Location: 255.255.255.256

Post by clive.grace » Thu Nov 06, 2014 6:17 pm

Did you ever manage to get that BOM up as I am about to start a build. I think I have all the parts but I want to be sure.... especially the toggle switch and silly little mechanical stuff.

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Thu Nov 06, 2014 6:59 pm

Sorry, I promise I will have the BOM up by tonight. If you want to double check you can look at the smallbear links that I provided. For whatever reason Mouser provides no pictures of their alpha pots and the data sheets are always for a collection of pots, so I am never really sure that the pot I want is the one that I am going to get. Ill get back to you with the right number later.

As for the switch you need a on-off-on (latching non momentary) spst switch. Usually momentary is denoted with parenthesis, aka (on)-off-(on).

markfrancombe
Common Wiggler
Posts: 192
Joined: Mon Apr 21, 2014 2:40 pm

Post by markfrancombe » Thu Nov 06, 2014 8:22 pm

Just finished my BYTES.
After Initial wiggling Im a little sad.
Was looking forward to differing loop lengths, but didn't realise that so much NORMAL other functionality would be affected.

For example, the long length switch really cant be used in a locked sequence at all, it just either gets stuck on one note forever, or all notes disappear, depending on which lock you have.

Maybe Im missing something, so is there a setting so that the Turing machine, and related expanders, works exactly as before? Because I loved having a 8 note sequence and then flipping over to 16 note, for variation. (I was looking forward to having a 5 note sequence and flipping over to 10)

I appreciate that random is fun, but, the nice thing about Turing was its ability to find something cool for you so you could lock it, This seems to have gone (at least from long version, cos locking it goes into the aforementioned whine (or silence).

Theres some fun stuff to be had for sure, but after 1/2 an hour listening to Coltrane, frankly I want Kraftwerk back....

Mark

User avatar
windspirit
Super Deluxe Wiggler
Posts: 3065
Joined: Thu Sep 20, 2012 11:43 pm
Location: krellifornia

Post by windspirit » Thu Nov 06, 2014 8:48 pm

Huh, that definitely does not sound like typical behaviour. Are you modulating the length of the loop when you have a locked sequence? That will definitely change the sequence of bits.

If both length knobs are all the way CW the turing should behave exactly like a regular unmodified turing (with loop lengths of 8 and 16) with locking still intact.

Besides that it is possible that you built something wrong. How much experience do you have with debugging? It's hard to give you suggestions for where to look if I dont know what is misbehaving in your unit.

The test that I always use with locking is I fill the buffer with ones and then leave it on with a fast clock for awhile. If after you have left it for 5-10 minutes it is still all ones then locking should be working.

Did you take the original 4015 out of the turing and put it into the bytes?

I understand your frustration, hopefully we can get you your kraftwerk back ;)

User avatar
clive.grace
Common Wiggler
Posts: 100
Joined: Tue Oct 28, 2014 4:40 am
Location: 255.255.255.256

Post by clive.grace » Fri Nov 07, 2014 3:14 am

You are not alone in your BOM woes, they lost three of my BOMs last night, I feel a stiff letter to their customer services department is in order, these take time to get right.

markfrancombe
Common Wiggler
Posts: 192
Joined: Mon Apr 21, 2014 2:40 pm

Post by markfrancombe » Fri Nov 07, 2014 3:17 am

Thanks for quick reply...
windspirit wrote:Huh, that definitely does not sound like typical behaviour. Are you modulating the length of the loop when you have a locked sequence? That will definitely change the sequence of bits.
no... I have tried it of course, but no...
windspirit wrote: If both length knobs are all the way CW the turing should behave exactly like a regular unmodified turing (with loop lengths of 8 and 16) with locking still intact.
Yes, if I do this, it works as expected on the short loop setting (notes grabbed/filled during UN-locked remain there and do not disappear).

However, on the 16 note long setting, the sequence will play once then, depending on WHICH lock you are on it will either. fill with single notes (lock knob CCW) or empty of notes and become silent (lock knob CW)
windspirit wrote: Besides that it is possible that you built something wrong. How much experience do you have with debugging?
I can use a Multi, to check if ICs are getting power, or follow instructions for testing, but I dont understand how components work, or how to read a schematic really, so.. answer.. not much or none...
windspirit wrote: It's hard to give you suggestions for where to look if I dont know what is misbehaving in your unit.
Well I have told you what is misbehaving, functionality-wise, can you be more specific as to what kind of info you mean, I can check and report back.. for example "check for 5v at pin blah blah...???) Mind you it will be tricky, this sequencer is in five pieces with all the expanders (I gonna HAVE to get me an all in one greyscale panel) So its hard to get at one piece to test...
windspirit wrote: The test that I always use with locking is I fill the buffer with ones and then leave it on with a fast clock for awhile. If after you have left it for 5-10 minutes it is still all ones then locking should be working.
Not working with long 16 note, see above. Think AOK on 8 note.
windspirit wrote: Did you take the original 4015 out of the turing and put it into the bytes?
I DID remove it yes, bit NO if didn't use it in BYTES; had new ones. I bought via Mouser Bom, just to be sure I got correct parts...
windspirit wrote: I understand your frustration, hopefully we can get you your kraftwerk back ;)
I will of course pull the whole module and double check solder joints, and look for bad ones, apart from that I can mention 2 things that I might be unsure of.
Firstly I THINK my trace cut was good, thru a magbifiyingglass I could see the shiny 2 ends of trace with board in between, and the other thing was that I used one half of a Double pole switch (Just snipped the legs of the other side so they wouldnt encroach on the board. But that was the only variation from the Bom, and that part should not affect normal operation anyway, only the CV stuff.

Any hints appreciated...

Dont get me wrong, I can see some very cool uses for this expander, but I must get it to work, or else Im gonna have to revert to nrmal operation cos that was AWESOME; now its gone a bit "meh!"

PS: I read that you perform this installation for people, so, in a disaster, how much do you charge for taking a look? :) I guess THAT might be my last resort...


Mark

Post Reply

Return to “Music Tech DIY”