- Mirá que hay formas boludas de ir preso...

- Sí es cierto. Yo quiero que me engayolen por regicidio. Ya me estoy imaginando las placas de Crónica después de atropellar a la reina Isabel con un uniciclo a vapor...

Vuelvo a releer la hoja de datos, los documentos de la libc y mil foros hasta que un mensajito pasa al frente en un instante revelador:

the datasheet for the 8515 says that the USART RX complete handler is named USART_RXC" however in "iom8515.h" it's actually defined as "USART_RX"

Esta me permite sensar la corriente consumida por la carga, la corriente en el inductor y la tensión de salida así como también regular el factor de escala. El driver del fet (un irf540 que tenía dando vueltas por ahí) no es nada revolucionario, en el encendido la compuerta está manejada en forma directa por el circuito externo y se descarga a través de un pnp cuando el terminal de control se deja al aire o a una tensión moderadamente baja. No es súper rápido pero alcanza y sobra a las velocidades que lo uso.

Estoy muy contento tanto con la terminación de la placa y porque funciona de diez. Abajo está probando un circuito basado en el tl494 con límite de corriente y tensión.

Entonces tuve una revelación: estaba usando opamps para fuente simétrica y para colmo el modo común no incluía la tierra (básicamente la fuente es un seguidor realimentado por la caída de tensión en una resistencia muy pequeña, o sea la tensión es muy cercana a 0 volt, unos 10mV en las condiciones que la necesito). Compré un lm358 y todo se arregló.

La referencia de tensión interna no es super estable pero una vez que calentó y equilibró se queda en unos 100uA del punto fijado, por lo menos así lo ve el téster chino no tan barato.

Para después bailar hasta el amanecer

y llegar tres dias después a tu casa sin voz con una sonrisa de oreja a oreja.

Que qué va ahí? Ni idea, el otro chico siempre inventaba algo; total todas las variaciones se parecen.

Total que más da, si no tengo (aún?) el trozo de cerebro para sacar canciones de oído. La próxima vez que vaya a Capital paso por la Biblioteca Nacional y me llevo una copia...

Variación tachada de Canaro en París

Hoy todas las chicas geek están de fiesta, celebrando el día de Ada Lovelace, la primer y gran Geek. Afortunadamente en la sociedad actual hay cada vez mas modelos positivos de mujer geek , mis "ídolas" son Becky Stern y Limor Fried. Soñar no cuesta nada...

Otras biografías interesantes en la web de adafruit: http://www.adafruit.com/blog/category/random/ald/

"aún así entraste en mi casa sin dudarlo"

"es distinto."

Part 1: Process
Part 3: Composition
Part 3: Visuals

Design Fundamentals for Developers: Slides

Random snippets:
The process of design is used to bring order from chaos and randomness.
Sorta like art & has a black shirt.
The alternative to good design is bad design, not no design at all (Douglas Martin).
Authenticity is invaluable; originality non-existant (Jim Jarmusch).
There is no color that is better than black. [...] To me, black is black and red is color. (Massimo Vignelli).
Typography in 8 and 16 bit systems:
Discussion about the fonts used in the most influential systems of yesterday (and the fonts in ttf format available for download).

Typography in 8 bits system fonts
Typography in 16 bits system fonts

The 60s at grain edit:
This is where I go when I need something to inspire me http://grainedit.com/tag/1960s/

Looks nice, eh? I made it with my bare hands.

Know yourself:

We engineers are all heroes capable of amazing feats but be realistic and conservative when estimating delivery dates.  Doing things in a rush never is good and almost always will result in some sort of sloppy work.

Pulling an all nighter is okay from time to time, more even so if you are inclined to the software side but remember to have proper rest. This is an extreme case but I think it serves well as an example: Yale Student Killed in Lathe Accident [MAKE]. My "zone" is normally from 17-24 and I'll happily extend it if  I could avoid waking up early; the point is: plan your schedule so you can work on your most productive time and also have adequate resting.

Invest in tools:

If you are in a hurry or short of money buy whatever you can, but quality tools will make your life a lot nicer.  There are many variations of the same tool, and they can be a very good time saver so while you are shopping look a bit for alternatives instead of buying just what you need that moment. My new favorites are angled nose pliers; with them I can work comfortably in a board without having to put them aside when it's time to switch to another board. That alone saves me about forty minutes in a normal production run.

While we are at this buy an ergonomic chair with lumbar support and good lighting as soon as you can. It doesn't have to be an Aeron but trust me that you will not regret buying it in a couple of years.

Also using both hands is a good skill to acquire; we southpaws have this more or less in our lifestyle but it also helps to relieve/avoid RSI injuries.

Work in batches:

This will improve your usage of time. After a while every repetitive action becomes kind of natural. For the above board I had to bend pins on 240 resistors; if I bent one, then placed it on the board, picking another... and so on I'd still be on that. After the first board I could bend them without even looking. (As soon as I can I'll build a machine to automagically bend the tapes).

In this particular case I mounted first every resistor, then diodes, then optocouplers, terminals and lastly the fets.

Be tidy and pay attention to every possible detail:

This goes both for your bench and your work. I like to have four clear areas in my bench: tools for the current task, required components, boards to be processed and finished boards; it's almost like a fordist line. Everything else is either on another table or stored in shelves. This is a not so very good shot of my in-house workbench:

It could be tidier.

You could be a genius but at the end you'll be seen through your work, so make things neat and tidy as possible. Many oversee little details like orienting all the resistor the same, leveling components and attaining a good finish on the solder side (clean up the flux!). Don't do a sloppy job, it could be the difference between establishing a solid relationship and never working again with that person.

Clean up that flux!

This is much better.

Leave your ego aside:

We like to see our name in things and so does our customer. Refrain from putting your name/logo unless being told so, just use the brand of your customer.

Keep your customer informed and be honest:

Even if it's a one week work give frequent progress reports, send some pictures if you can. It is not nice to work with someone who "disappears" and then magically comes with something.

Sometimes I have to mail-order components because the local shops don't have them or don't have enough stock. In that case I give a tracking number to the customer so she/he can know what's happening and when I'll start working. (if possible do some other part of the job so you don't waste time).

In the (hopefully rare) event that something goes awry tell as soon as you can, be sorry if appropriate and offer means of resolution.

Don't work for free:

It might be tempting to work for very cheap rates when you start but don't lower the bar that much, it will bite you later. It's okay to have low prices but factor in not only the cost of raw materials but the time you take to make it and every ancillary task required. Being cheap involves the risk of a sudden peak of demand with no time nor resources to satisfy it (you don't happen to have a complete and automated production line on your basement, don't you?).

If you cannot build another unit for free with your profit margin and still have earnings(say, because your pet destroyed one) in 99% of cases you are doing it really wrong.

Have a backup:

Don't depend on a single supplier, keep in mind mail delays when working with foreign customers (it is very frustrating to miss a deadline because the transport company had some issues) and buy excess of components if you cannot stock them easily. When making long runs the odds of having failing components have more weight and paying 30$ in postage for a 9$ transistor hurts (and takes time). Also, watch out for duds. This diode felt weird (I mostly place them sensing the bevel and the shorter pin), the casing is reversed:

Led with anode and cathode swapped.

Test!:

This should go without saying but I've seen people deliver boards blindly believing that they work. Just don't.

Design for ease of manufacture and service:

It is not easy to give general guidelines but keep in mind that you are going to assemble whatever you design, so think a while before diving into making stuff. If possible use only stock parts (see below for a home made connector I had to make. Avoid situations like that). If you can go full or mostly surface mount do it, the time saved from drilling and then mounting components is worth every bit of hassle required to learn how to do smd work. Lastly, think about things like ease of  mounting, use and stacking. I like to use proportions based on 2, 4 or 3 to distribute elements but sometimes it is not the best option, despite looking good, as the last picture shows.

IDC connectors, the best thing since sliced tuna.

Last resort connector. Avoid this kind of things at all costs.

Sometimes the center is not really the best place.

Sell yourself:
Word of mouth is a very good way to reach more people but sometimes not enough. There are many places where you can advertise your services (even free ones). It really helps to have some sort of showcase or portfolio available. It's a plus if people can put your name in a search engine and land directly into it (also remember that the net does not forget).

Further reading:
Two influential articles:
Steve Friedl: So you want to be a consultant...?

James Chan: Spare Room Tycoon (Buy it if you can)

Corollary:
Whatever you end up doing, be happy.