Why are Gechos out of stock (often)

While the truth about plumbuses is now out, it may still be a mystery how Gechos are made. Let me try to illustrate what is involved and why it takes so long to make them.

First, the electronics is made from panelized PCB boards assembled in the factory in Poland. After they arrive to us, we unpack them from ESD protective bags, carefully check every single one under the microscope for bad solder joints (not every element just those that are known to be problematic) and fix where necessary, either with solder iron or hot air. Then the boards are programmed, which tests powering and reset circuit, boot modes, USB communication and SD card interface.

Enclosure panels are checked for defects and filed at edges which come to contact with sliding door. Then battery holder wires are soldered and the unit is assembled. Next step is testing all functionality - powering from batteries, clear sound from microphones, reaction of sensors & LEDs, signal from two HP out connectors, line out switch, line input, MIDI I/O and the related switch.

No two screws are created equal

When it comes to manufacturing, there is no consistency in anything. Pretty much every batch of material that arrives is a surprise. Not all screws are the same. Not all standoff colums are precisely the same length. The quality of panels is varying, sometimes I need to throw away 50% of them. Blue panels that use lead-free HASL surface finish tend to have holes covered by excessive solder. Black panels use solder mask that often has cosmetic defects, which is expected as it does not matter at all with normal intended use (this is the same material that printed circuits are made from). It is better with side and bottom panels as thanks to the smaller surface area, chance of a defect appearing there is lower. Orange battery door panel slightly varies in thickness, which requires extra time matching every unit with a suitable panel, or filing the side panels further.

Black screws (that look the same on pictures) may come in 3-4 different variations: diameter of the head, depth in which the (x) is embossed, surface finish, and some do not work because of damaged thread. Of course it is very important to pick four of them that look identical. Steel screws which go to the bottom side have variable diameter, they can be thinner or thicker, slightly bent (as the head is quite large to hold well with the magnet), and sometimes the top falls off completely when the screw is tightened.

Battery holders are a whole new level of madness. They all look great on pictures (which the sellers probably steal from each other), but some are as bad as completely unusable. Batteries are hard to put in, then jump out at a slightest knock. Even the texture of plastic varies and not every double sided tape sticks well enough to every holder.

Wooden boxes take time to make

They are hand-made by our partner in Slovakia. Except for photos on their instagram, I do not have a video footage but the process is similar to enclosures for Glo whale, apart from the curvy tail. Christmas and Easter is the busiest time of their year as they specialize in making wine cases, wooden toy animals and a lot of other cute things that are sought as presents.

After receiving boxes, we glue magnets in them at a precisely defined position to give the best grip for the screws and let them hold the unit centered - surprisingly, this is not achieved by placing them exactly under the screws as that's not how magnets work against a flat piece of metal. We use this contraption to help with the task:

Finally, every unit has to be paired with a box. Because the wood is warping very slightly and the screw heads are not of equal thickness (which actually helps as they can be swapped around), it takes a few attempts and often a screwdriver too, to find a box for each unit where the insides do not wobble with a clanking sound every time you press a button.

Each of these operations add up to hours and hours of work. And that's not the end, because there is packing and shipping; postal docket forms & customs declarations, stickers, queueing at the post office, writing down the tracking numbers from stickers and putting them into the system, sending notifications, answering customer support emails, attending to the forum...

At the moment there is as many boxes left as required to fulfil the orders waiting since last year. We should receive more boxes in the spring but until then, and also before the remaining material arrives (mainly panels, and they are of suitable quality) I am not comfortable accepting more orders.

Conclusion & vision for the future

The past year was very demanding. I would like to take the time to rethink the project, optimize the build process, decide how many complete units can I afford to make before they need to be replaced with a plastic enclosure... But mainly, do things at my own pace rather than being under a constant pressure from hundreds of orders that were waiting for months.

Also, it is important to regain the balance between building & packing and working on the software side of things, as the firmware still lacks some nice features that the hardware is capable of supporting. This was meant to be a "magic music box" (before I knew that pocket synths existed) but over time, thanks to the feedback from many happy owners, features were added that make it somewhat a synth: MIDI, CV I/O, accelerometer... At the moment MIDI is implementated at a very basic level, CV hardware works but not doing anything useful and accelerometer is only available in some of the channels.

Certain things may be expected to work out of the box when you look at the hardware features, but the firmware needs time to catch up with that. What sounds trivial may actually not be simple at all, for example syncing the tempo to MIDI clock becomes very tricky with some channels. Deriving the tempo from Arturia Keystep's MIDI clock or from Pocket Operator's sync signal is easy enough and already working. But making it to do something useful in soundscape channels is tricky; the sound processing there isn't quite sequential (as if it was generated from bits and pieces and you can send a chunk of sound on demand, according to the clock). When adjusting the delay buffer length up and down, glitches will appear. It makes more sense to detect tempo in the beginning and then adjust the delays and timing by it. Although commands like start/stop/continue would be nice to have too. In some channels it is possible to react to the changing clock information in the usual manner, e.g. in drum sequencer or a FM synth (Dekrispator).

While this is an universal platform that could become a "swiss knife" of sound processing (as much as the DSP resolution and CPU power allows), personally I do not feel like adding functions that are already performed well by other devices, as is often suggested (however as there is SD card with plenty of storage now, it would be nice to have a 4-track looper :) The goal of this project is to invent new weird sound effects that are fun to play with, rather than adding precisely defined and fully controllable sound effects as known in other gear. The source code is freely available and anyone who wants to implement their own ideas, is very welcome to do so. Porting the code over from other open source projects is possible too, and I've done it multiple times. On the other hand, copying functionality of other devices doesn't make much sense, unless unique controls or features of this platform add an extra "twist" to it.

Last but not least, the website needs to be reorganized. It does not explain the project well, and the shopping experience isn't ideal. It is a result of how it evolved when things (models, choices, accessories) were added and there was no time to rebuild it all, so you do not have one place to pick all that you may want. Also, I am getting too many "I still haven't received it" enquiries despite "my order" page showing position in queue, the culprit is that the emails containing access links and automatic reminders for picking the colour scheme often end up in spam so if you don't read everything carefully, you may not even know about that page existing. Ideally, you should gain access to your order right after paying for it, however with recent changes in credit card processing, what used to work smoothly, now requires some amount of manual interaction at my side, or a complete rebuild of the payment gateway.

Thank you all for your support and patience, wishing you Happy New Year and all the best for 2020!

Mario