homelist of postsdocs about & FAQ

People in science and engineeing - Part 1

and the never ending rivalry

Lately the topic about science and engineering has become quite hot in our group. It is probably the fact that we always have lunch in the neighbouring buildings hosting the theoretical physics and mathematics departments. Being almost the only engineers having lunch in this canteen, otherwise packed with sciencey' theoretical physics people, I started noting differences between them and us. I am sharing some of my observations as an engineer. I do not aim to put more gasoline into the fire here, but instead note some of the aliasing artifacts between "them" and "us" in a friendly manner.

A fact, people who are into pure theoretical sciences are usually very sharp and confident in their fields. This however allures them to think that they know everything. Well, they don't. It isn't possible to learn and know everything of course, and in general, the more you know the more you learn that you know nothing. And because theoretical/natural scientists often study and work on very nieche problems they kind of lose the whole picture. Don't get me wrong, I am not saying they are stupid, which is a typical illusion they get by seaking to "us". People within pure natural sciences are very eager to have discussions within their comfort zone (whatever their field is) however when we step-over to other fields they often misbehave. Because it often happens that they know nothing about the other fields, often, their first reaction is to become arrogant and try to defend themselves. My experience shows that their defense is usually expressed in insults and explanations on how easy it is to solve "this" or "that" engineering or whatever problem. Let me give you some brief examples:

Not a long time ago I had a discussion with a friend (mathematician) with whom I somehow ignited a discussion about the MP3 compression format. I politely asked her if she knows the basic principle behind MP3 encoding wanting to share some thoughts on types of music and compression density. An immediate quite confident reaction was followed with the words "Yes, fourier transform!", with that the topic was exhausted. Well, not quite, she feels very comfortable with Fourier transformations and is probably quite good at it, however this by far does not stretch over the basic principle. When I tried giving a brief explanation, instead of listening carefully to my "intuitive engineering explanation" on the psychoacoustic model of the brain and f-domain truncations, she simply ignored and did not listen to any of my words. Head was up and pride rockets in the sky - it is Fourier transformation, too smart to listen to "intuitive stuff".

Enough talking about competition, there are more interesting facts to cover on people's stereotypes, based on their fields of work, no matter if these are pure or applied sciences or engineering. Stereotypes contain enough truth to be humorous but also quite objective. We all do some sort of preliminary judjements based on stereotypes, probably quite a bit more often than we are brave to admit. Stereotype models give us fast and efficient cognitive shortcuts and save us a lot of energy and time. I will soon try to elaborate on the stereotypes of people in science and engineering in Part 2.

Date:Sun Jul 05 15:36:14 CET 2015

Applied chaos theory

I was browsing around some cool circuits recently and I discovered Chua's circuit. How come I haven't come across chaotic circuits before?

I want to do a quick analysis on this non-linear circuit, and this post will ironically follow a chaotic fashion and begin upside down. Let's start with Chua's circuit non-linear behaviour. This is a non-linear circuit having a current-voltage relation resembling something like this:

Chua's classic circuit transfer function

So that you don't consider that I am describing a black hole, here is the classic chaotic Chua circuit diagram:

Chua's non-linear diode circuit diagram

Taking an initial look at Chua's circuit we can identify an LC resonance tank and an active non-linear element, which adds-up energy in the resonance loop and forming a special "energy injector" which injects current in the system depeding on the voltage's magnitude following a non-linear function as show on the first figure. The active non-linear element is effectively a negative impedance converter with a twist. The opamps in the Chua configuration act as adding or subtracting a varying voltage in series to the voltage drop across an equivalent positive impedance. The voltage polarity of the current direction of the element is reversed and thus a phase shift of 180 degrees between the voltage and the current is introduced.

Having a single negative impedance converter would simply keep the LC tank in resonance and would not introduce any form of chaos in the system. For this reason, Chua's circuit in the from as shown in the figure would normally contain two negative impedance converters in parallel with a slight slope offset.

I will now leave you with some pictures of the circuit in operation. The circuit was not very easy to tune and get into chaos with my setup but the wobbling sinewaves were quite fun to play with.

The spider's web
Plot X vs Y, circuit is out of chaos here
A semi-chaotic behavior starts to appear.
We're in chaos now
We're in chaos now
We're in chaos now

I started to wonder how would that sound as the frequencies generated are in the audible spectrum. It isn't hard to attach a speaker to it and listen to some chaos. Not very surprising or exciting sound, but still, here is how I heard it:

Too many pictures, too little explanations? Have a look at www.chuacircuits.com.

Date:Tue Jul 04 18:00:23 BST 2015

A precision DIY Scanning Tunneling Microscope

Last weekend I paid a visit to Kris at the University of Bath. She is currently finishing-up with her doctoral studies and invited me to pay a visit to her lab and the materials group's DIY scanning tunneling microscope. During that weekend we scanned a few Silicon 111 wafer samples and injected Toulene molecules on the wafer.

I was quite impressed that the tunneling current noise floor of their microscope was in the order of a couple of hundred femto amperes, which is quite impressive for a measurement system operating at room temperature. It seems like they were using a commercial low-noise programmable gain amplifier produced by the German company Femto. It was quite tempting to open the amplifier box and see what's inside. I doubt that there is a custom OPA in it, but I really wonder which commercial chips did they use and what was this low-noise PGA feedback configuration used. Anyway, here are some fancy pictures of my visit and this has now inspired me to run an investigation on the OP and low-noise needle amplifiers used.

Outer look of the STM, the long pipes are manipulating shafts used inside the chamber; the turbopump as well as the piston pumps can be seen on the left side of the machine
Different angle, the turbopump as well as the head chamber are seen in the centre of the image
Here is the scanning head, together with a sample attached. The head is using piezo elements for micropositioning and scanning in both ways
The low-noise programmable gain amplifier manufactured by Femto Gmbh, note that the gain is changed by turning the rotary switch and not electronically
An STM image of the Si 111 sample, we can see the individual atoms as well as some impurities. Seems like the Czochralski method is indeed a brilliant casting technique

Besides the immense pressure/vaccuum required for this machine to work, the electronics behind it is fascinating. A PLL and all other sorts of regulating loops to precisely control the needle. The piezo elements have a natural vibrating frquency for which it should be corrected. A needle with passing fA currents though the sample, very low-noise programmable gain amplifier with filters and a 16-bit analog to digital converter. Moreover some digital image acquisition and needle vibration correction algorithms are used to reconstruct the images. Simply a symphony of virtuoso circuits!

MORE IMAGES IN THE GALLERY HERE

Date:Tue May 09 15:00:23 CET 2015

A few useful functions for Virtuoso layout

I am waiting for a simulation to finish and am a bit bored having done layout whole day. Let me post some very simple but useful SKILL finctions for the layout editor of Cadence's tools.

The first one I use regularly adds a specific layout instance when invoked. I have binded it to various keys and thus can add cells from my own common library, e.g. VIA_M1_M2_min, VIA_M2_M3_min, M1_PO_min, M1_PO_minx2, M1_AA_min etc...

Now you might argue that there are normally provided p-cells for thus purpose, which is true. However, for the image sensor processes I use, this "feature" normally is not provided by the foundries. Or at least, the ones I've used so far haven't got any p-cells. A quick solution is thus having own cell library and adding layout views from it.

I also kind of like that freestyle layout drawing as it gives you the freedom to do "virtuoso stuff", which you normally can't do with p-cells, or it is tedious to set. Besides, p-cells rely on SKILL and Virtuoso's ROD engine, if that setup somehow fails, then the whole design can not be "compiled" / "viewed". Anyway, here is how I instantiate my cells:

;A function to add specified layout instances

procedure(placeInstance(lib cell view "t")
leSetEnv("instLibName" lib)
leSetEnv("instCellName" cell)
leSetEnv("instViewName" view)
hiRegTimer("hiToggleEnterForm()" 1)
leHiCreateInst()
)

;Binding it to a key:
hiSetBindKey("Layout" "Ctrl Alt1" "placeInstance(\"commonLib\" \"M1_M2_min\" \"layout\")")


You can set hiRegTimer("hiToggleEnterForm()" 1) to 0 if you want the enter form to appear and not be auto-minimized, or on the other way around '1' would auto-minimize the window and you are good to go.

Another quick and handy layout bindkey function is the toggle layout: Here is how we can toggle layer visibility, I know there's tons of resources on that, but still may be useful to have a yet another source.

;Toggle layer selectability procedure for metal layers

procedure( toggleMetal(num "x")
let(((metlay list(sprintf(nil "M%d" num) "drawing")) )
if(leIsLayerSelectable(metlay) then
leSetLayerSelectable(metlay nil)
else
leSetLayerSelectable(metlay t)
)
hiRedraw()
)
)


We can set it to an arbitraty layer defined in our techlib by minor corrections. We can even choose to toggle schematic editor layers.

Alternatively, we can also toggle visibility which is one of my favorite functions:

;Toggle layer visibility procedure

procedure( CCStoggleNW(num "x")
let(((metlay list(sprintf(nil "NW" num) "drawing")) )
if(leIsLayerVisible(metlay) then
leSetLayerVisible(metlay nil)
else
leSetLayerVisible(metlay t)
)
hiRedraw()
)
)


I guess the most important stress in this post, and in doing fast and efficient layout (bar good floorplanning and thinking in advance), is the good set of bindkey functions. Try to use the mouse as less frequent as possible, i.e. only for drawing or zooming-in-out.

Date:Tue Apr 28 17:45:49 CET 2015

"Alchymia" in circuit design

I stumbled upon an interesting book - The Collected Papers of Lewis Fry Richardson. Richardson was an English scientist of "everything" - maths, physics, psychology, meteorology... actually a pioneer of modern day mathematical techniques of weather forecasting.

Now all that is great, but what got my attention is the following figure: (I hope reprinting figures from Richardson's papers from 1930s is not a severe copyright violation)

This all looks to me like some crazy alchemistry in circuit design. No matter how many times I tried to read this paragraph it all simply does not make any sense at all. It really is fascinating how great scientists in the past tried all sorts of crazy ideas (including the addition of an electric eel to an RLC circuit) to see "if something great would happen".

But wait, there is even more:

It all smells some deep alchemistry happenning around. But hey, if I can, why not?

Date:Tue Apr 19 14:07:40 CET 2015