#282

Woah that’s crazy- Probably only 50% slower than a pick-and-place, and $40k less expensive! I’m only ever building one-offs by hand so unfortunately not for me, but this is such a great approach for small-scale manufacturing!

2 Likes
#283

Yeah - he’s also putting these onto solderpaste before going into an oven - I think he just uses a small suction tool to pick up and drop them into place.

#284

The suction tool is another idea that I stole from him. It is an aquarium pump, hacked to invert polarity and fitted with a syringe with a hole (making an on/off control). It works quite well and cost nothing to make. Here is mine:

There are instructions and YouTube videos out there showing how to make this and what pump to buy…

1 Like
#285

For someone(me) with no SMD knowledge, how is the paste applied to the board? And once it is, and the components are placed, is it simply a matter of heating the entire board?

#286

Usually, a ‘mask’ is placed over the board - ordinarily metal, you can get away with plastic - with precise holes above all the pads. You smear solderpaste over the mask, and then remove the mask, leaving the pads covered in paste. You get the mask made at the time you get the PCBs made.

And then you just put it in a temp-controlled oven - ideally, one designed for the task, but many people hack toaster ovens with PIDs.

Of course, you don’t need any of this for SMD - I built my Ornament and Crime with a soldering iron, tweezers and regular solder, and the SMD kits I’m currently designing are intended to be assembled by hand with a regular iron. But this is the more industrial process they’re designed for, and it makes it much easier for doing things in bulk.

2 Likes
#287

Exactly what I’m doing. :slight_smile:

  • I purchased custom stencils from OshStencil (https://www.oshstencils.com).

  • Apply the paste and then drop the components in place using the hacked aquarium pump (above).

  • Bake the boards in a hacked toaster oven (the ControLeo2 - http://whizoo.com/)

The build for the TELEX modules is about the same level of complexity as Ornament + Crime. I’ve successfully done both by hand without any of the additional equipment. The toughest part is the DAC which the TXo and the OC share (DAC8564). That beast is tiny and it is hard to keep the pins from bridging by hand - but not impossible. Many people build Ornament and Crime by hand.

The TELEX have a bit more complexity because, instead of most everything on one side of a larger board, components are on both sides of two boards for each module - with extensive rows of header pins to connect between them and the Teensy processor. I’ve been successfully able to reflow both sides of the boards in my toaster - which is a godsend - but it is a little tricky getting the paste on the backside. That is why I’m 3D printing the jigs (above) that allow me to put boards flat that already have components on the back so they can accept solder paste.

3 Likes
#288

Sorry for the aside, but what’s the story with @c1t1zen building a bunch of O&Cs?

I’m terribly interested in the module, but simply don’t have the time or focus to acquire parts and build.

#289

It is a great module; I have a few myself. Lots of depth in its various crazy modes in the extended firmware.

Magpie Modular is offering pre-built OCs with beautiful panels. @c1t1zen was building units for them. They seem to be sold-out.

http://magpie-modular.myshopify.com/collections/built-modules/products/pre-order-ornament-crime-built-module-white-sandblasted-panel?variant=29967904195

Oscillosaurus also sells a beautiful panel and PCB package - if you can find someone to build it for you (perhaps on MuffWiggler):

http://oscillosaurus.com/shop/ornament-crime-pcb-panel/

blob.jpg332×600 30.1 KB

An even more “standard-looking” PCB/panel package is available at SynthCube:

Hope that helps.

1 Like
#290

Thanks a lot; I appreciate that.

I was aware of most of those options. I just thought/hoped that perhaps another round of builds would become available through a new to me source.

I’ll pick one up if/when one serendipitously crosses my path.

#291

Happy to see others taking the projected parts placement & vacuum pump idea to use in their own projects.
@Larrea every time I do a batch for Magpie Modular they sell out before either of us promote it. I can give you a heads up next batch.

@Galapagoose Thanks! This manual pick and place with projection is really speeding up my batches by hours…I just need to layout my SMD parts better before starting. Think @bpcmusic’s 3D printed dispenser is my next improvement.
I currently use a T-962A IR reflow oven (replace the masking tape with Kapton) and modded the firmware…It’s brilliant once you get proper controls over the thermocouples with the new Swedish firmware. $300+ on Amazon.
I’m also looking to get a TVM-802B pick and place machine for my own modules when I get into production. Lot’s of great reviews on it with people creating Eagle scripts and such…and only $4K instead of $40K

3 Likes
#292

That would be great! Thank you!

#293

TELEX Update - 13 Jan 2017

BITS AND BOBS

Approved the final details from MPC for the panels. They are now officially on their way to production. While I really liked the Lasergist panels (aside from the twisted bur in the corner when the laser does its final cut) - these metal photo panels are going to look AMAZING! They suggested they would be a few weeks out at this point.

  • Final passive components have arrived. All SMD parts are in-hand.

  • The knob order was a little short and a percentage got backordered, but I think I have enough in stock to fill out the reservations.

  • Boxes, packing materials, screws, cables, etc. are all in hand … except for the connector for one side of the Eurorack cable. That is due in a week or two.

  • I’m exploring using a heat shrink gun and wrap to make the i2c cables nice and pretty; will know this evening how it works. (Thanks @tehn for the idea!) I’ll post a pic if I’m excited about it.

  • This weekend I’m going to build a couple of units using the final circuit boards that arrived this week to make sure that they are 100% as I switched to a different manufacturer for the production boards.

EXCITING II NEWS

I have a very strong feeling that the i2c issues are behind us.

  • @tehn and @scanner_darkly (and a host of others testing and thinking through the problem) have updated the Teletype’s i2c-related code and it is super-solid. Thanks to EVERYONE involved in that effort are in order; I sincerely appreciated the response to the call to arms!!

  • Issues around high numbers of devices on the i2c bus and calls to outputs/inputs that aren’t connected, at least as far as the Expanders are concerned, are resolved. They are abated by enabling the internal pull-up resistors on one of the Telex devices. This is only a worry when you have 4+ TELEX devices on the bus (ok, maybe 3+) - but it is dirt-simple to enable in firmware and requires no hardware changes to the TT or modules.

Here is this morning’s torture test. The script is running at a 100ms interval against a Teletype with Ansible, 2 TXi input expanders and 4 TXo output expanders connected.

  • 24 Simultaneous TR pulses
  • 8 TXi Param knobs driving TXo CV values 1-8 (when I turn the knob, the CV outputs on the first two TXo units will change)
  • Random voltages for CV 1-4, which are being read by TXi INs 1-4, which then set as values for TXo outputs 9-12
  • TXi INs 5-8 reading the outputs from TXo 9-12 and driving the outputs TXo 13-16

This means, each clock tick is writing 24 TRs, 24 CVs and reading 16 knobs and input voltages!

Ran this thing for an hour+ with NO ISSUES.

It can totally go faster, too, but the TRs become a bit of a blur which is why I shot it at this speed. At some point you can max out the number of II operations that the Teletype can do per clock and it gets unhappy - but the platform is KILLING it at this point.

Now, I’m onto more production activities. A bit of warning … it will still be a while. I have a lot of manual labor ahead of me.

11 Likes
#294

thank you for the time you dedicated to this, absolutely outstanding! happy to have helped a tiny bit.

3 Likes
#295

Ohhhh yes :slight_smile: is it mean it will work perfectly with the trilogy + ansible ?

1 Like
#296

@mDang - totally should. We’re all still working out the best way to support larger #s of devices - but we do know that it all works together.

FIRMWARE UPDATE

Final experimental TR features were just uploaded to the repository.

HTML Command Reference
Printable Command Reference

Each trigger output has its own pulse divider. Pulse dividing is easy - simply specify the divider and it will only output those pulses (skipping the others).

TO.PULSE.DIV 1 3

Will allow every third pulse. Check it in action:

Sick of putting recurring pulses on the Metro event? Each trigger output now has its own independent metronome that will emit pulses. Here is the same demo as above without using Teletype events:

Easy to use:

TO.TR.M 1-n α	     time for TR.M; α in milliseconds
TO.TR.M.S 1-n α	     time for TR.M; α in seconds
TO.TR.M.M 1-n α	     time for TR.M; α in minutes
TO.TR.M.BPM 1-n α	 time for TR.M; α in Beats Per Minute
TO.TR.M.ACT 1-n α    activates the metronome for the TR output; α [0/1]
TO.TR.M.SYNC 1-n	 synchronizes the metronome on the device #

As above, you can set the Pulse Metro in milliseconds, seconds, minutes, or BPM.

Here is a demo showing off the independence of each pulse metronome; these are only a few ms apart and will phase:

7 Likes
#297

It’s LED Steve Reich in a box! Cool!

1 Like
#298

Wow! I may have to order more TXO PCBs.

1 Like
#299

It feels ungrateful to suggest any features, but I had two ideas:

  1. Add TO.OSC.TIME.S and similar time functions to set oscillator/LFO times in seconds, minutes, etc. This makes them more easily synced to the pulse/metronome outputs.

  2. Add a TO.TR.M.PW pulse width function to the metronome. I have a 4ms QCD + expander that I keep around purely because of the dedicated pulse width knobs. Having variable PW is essential for interesting boolean logic processing.

Reading the documentation is very exciting. I’m loving the quantization functionality. Building an 8-step quantized step sequencer with two TXi expanders is going to be a fun physical alternative to the built-in TT patterns.

2 Likes
#300

H@trickyflemming - not ungrateful at all! Not sure if I’ll have a lot of time to get to suggestions before release, but my goal is for this to be as useful as possible for folks. Also - it is open source, so others can take up the charge as well. :slight_smile:

A couple of questions:

For the OSC, it sounds like you are looking for a cycle length - is that right? Right now we support voltage, note number, frequency (in Hz) and LFO (in millihertz - 10−3 Hz). If it is cycle you desire, how about:

Sample Commands:

TO.OSC.CYC      duration of one cycle in milliseconds
TO.OSC.CYC.S    duration of one cycle in seconds
TO.OSC.CYC.M    duration of one cycle in minutes

Should be easy enough to do.

For the pulse width of the automatic metro pulse, this is directly related to the TO.TR.PULSE.TIME/.S/.M setting. Would you want a way to manipulate that value in a percentage as it relates to the M time - or is this good enough?

For instance if you want a 50% pulse:

TO.TR.M 1 1000
TO.TR.TIME 1 500
TO.TR.M.ACT 1 1

I guess we could add a command to set the TIME relative to the M. (I’ve tried to avoid redundant commands, though.) Like:

TO.TR.M 1 1000
TO.TR.M.WIDTH 1 50
TO.TR.M.ACT 1 1

Which would simply set the TO.TR.TIME for that trigger to 500.

Thoughts?

3 Likes
#301

Great! Yes, the TO.OSC.CYC commands are exactly what I was hoping for. I had only suggested TO.OSC.TIME instead of CYC to be consistent with TO.TR.TIME. CYC is more proper.

TO.TR.M.PULSE would be great. I don’t think it would be terribly redundant. One patch idea that I’m thinking of is to use two TXis and one TXo to create four variable clocks. If I want to change the speed of four clocks but keep a constant pulse width of 50%, I could set TO.TR.TIME to be a DIV of TO.TR.M. However, being able to change the pulse width of each channel would quickly get messy!

It would be especially nice if TO.TR.M.PULSE was maintained even if TO.TR.M changed. That way, I wouldn’t need to use two lines to maintain the pulse width if I’m manipulating the metronome’s speed.

1 Like