Quietly, while we worry about the state of engineering education, the university engineering curriculum is being disrupted before our very eyes.
It began about five years ago when we saw the beginnings of an approach to learning that's now ramping pretty quickly. In 2007, the Mobile Studio project came onto the scene. In short it was a lab-in-a-box approach that put scope capabilities in the hands of engineering students. No need to book lab time, plus you can do your experiments in your dorm room or at Starbucks.
The $99 Digilent Analog Discovery Design Kit and the more advanced $199 Digilent Analog Explorer Design Kit allow students to build and test a wide range of analog and digital circuits using their own PC without the need for any other equipment.
In an interview earlier this year, she, Clint Cole, Digilent's president and founder, and Dave Babicz, ADI's director of global alliances, laid out the financial value proposition: A lab bench costs $5,000-$10,000 to replace; to establish a classroom is at least $100,000 and another $100,000 to man the labs.
"And that wasn't in the cards given the budget cuts we were enduring," Meehan said. "So doing experiments outside the classroom was the only way to establish a circuits lab and get them engaged in the process earlier."
Everyone has their interests in the project of course; the academics manage their budgets better and put something portable and powerful in the hands of students; IC providers get to sell parts and check off corporate-responsibility boxes.
But for ADI, there's something more: "It's not all altruism on our part. We need students to know more about analog circuits," Babicz said.
Disruption in the classroom But what is really interesting (at least to me) about this trend, is how it portends to change the way we educate engineers. We all know about distance-learning and downloadable class materials; Apple and others eye hungrily the e-textbook market.
But things like lab-in-a-box allow educators to reinvent their classrooms.
That was Ken Connor's enthusiastic assessment when I chatted with him during the USA Science and Engineering Festival April 28 in Washington.
"A fundamental issue with engineering education is that universities are incredibly slow to change," said Conner, who teaches at Rensselaer Polytechnic, a key developer of Mobile Studio. "But talk to people in educational research, everybody is abandoning traditional lectures. It's called a flipped classroom. All my lectures are recorded. They watch at home and then they come to class and do something."
Now, with technology like lab in a box, "we can actually give students hardware homework. Send them home to design a filter that does X and they come back in and show it to us."
The hands on, human feedback, peer-reviewed engineering learning then becomes re-cemented in the classroom.
Technology is affording academia a chance at revolution; how quickly (or whether) it accepts the challenges is an open question.
@Flightsfan....thanks for all that. Sounds like a good bit of gear. Even at 105 msps the Analog discovery is only rated at 5 MHz though I reckon it would be OK up to a fair bit more than that. 4 channels is nice on yours. I am on a course at the moment which would get me the discount so I must try and get one or the other.
I own the Electronics Explorer which is a larger/older version of the Analog Discovery. It has a large breadboard space which is really handy as well as having 4 scope channels and 32 digital i/o lines and a whole suite of analog generators, DVM, Spectrum Analyzer, Network Analyzer, and programmable power supplies upto +/-12V.
It is roughly half the acquisition speed of the newer Analog Discovery at only 40MSa so waveforms over 20MHz are pretty much impossible. At 20MHz the resolution is poor but managable. The logic analyzer had no issues keeping up with data flowing across a digital bus at 16MHz. One advantage with the scope is the adjustable buffer size. Really handy when looking at a triggered event. It stored enough data prior to the event that you could look at several thousand samples prior to the event as well as thousands of data points after so you could get a complete picture before and after the event.
Overall I really like it. The Analog Discovery came out after I had already bought the Electronics Explorer. Even with only 2 scope channels (although 2.5x faster) and half the digital lines the Analog Discovery should be as good or better. The Electronics Explorer is $199 for student pricing or $299 for academic pricing. It goes up to $699 for full price so I was glad I was a student. The biggest disadvantage I can think of is the fact that interfacing the board with Labview is not supported. You have to use the Waveforms software which although is decent software, doesn't have the all capabilities that Labview has.
Hmmmm.... The Digilent Lab in a Box has dual channel 14-bit 105 MSPS ADC and can do 2 x 5 MHz inputs. Labview's MyDAQ which I have does dual 12-bit 200KSPS inputs, and is not much use above 20 KHz.
Rats...I've been had!! Would much rather have the LIAB.
Note to self - check if Digilent have an outlet in Aussie....
Simulation is like a short skirt: it's supposed to looks attractive while covering up essential parts.
Having said that, there's always a reason for a difference between the simulation and bench measurement. If it's important, it can be found and accounted for in the simulation. The trick is that such fixes often require being familiar with, ahem, the internal features of the simulation environment.
As Rodney Brooks once said about robot simulators, "Simulations are doomed to success." You can do things in simulations that appear to work very nicely but would not work in the real world. The difference is all the parameters and limitations NOT included in the computer models. This is an old problem.
Bob Pease of National had lots of fun showing that the real world and the SPICE simulation of it were often different. Some designs that worked well on the bench for explainable reasons could not be accurately simulated in SPICE, and vice-versa. The argument is that you have to know what the simulation is doing.
IMO, a big piece of this is learning the "knowledge in the hands" that comes from working with real components in the real world. This knowledge is essential for good designs. If you have this knowledge, SPICE can be valuable for you to increase the precision of your design.
Hands-on knowledge will also let you know when the SPICE is not working, as in it has done the equivalent of taking the square root of your social security number: simultaneously precise and meaningless.
Excellent idea, but can't we get the same affect by using pspice or a cheaper (student) version of many analog simulation programs. There are even how-to books to go along with them. A few years ago, on ebay, I bought an electronic trainer consisting of several pegboard-like hook up slabs with .1" spacing, fixed and variable power supplies, and a waveform generator, all packaged into a neat carrying case. I am not sure who made those, but they should bring them back. So a student could create a circuit using the computer simulator, then actually build it to see how it performs in real life. No million dollar labs reqd. I still do that when my circuit is in a concept stage, before going to prototype.
This is a great idea! Gone are the days when Saturday morning was tied up in a lab (because that was the only time slots open). I can think of a better way to get students engaged then hands on work with the real thing. With the low cost this should be a major help for students and could possibly be in every intro to engineering course.
I am doing a course at the moment and as a consequence of that I managed to get my hands on a student version of Labview with their MyDaq module - $199 the lot. It has 2 analogue inputs and 2 outputs and 12 Digital I/O, and a multimeter. Apart from the Labview software which allows you to program and design custom "Virtual Instruments", it comes with a suite of ready-made ones, including a DMM, scope, function generator, AWG, bode analyser, spectrum analyser.... The one limitation I've found annoying is that the ADC/DAC are 200 KSPS and the associated instruments only go up to 20KHz. But for most basic student work this is not toooo big a deal. I'd be interested to see how Digilent's offerings compare with this. Anyone used both?
Hats off to ADI and Digilent for doing this. Students really benefit from the hands-on learning that can only come from playing with real components, and it does seem that universities have been "virtualizing" lab course work for quite some time now.
Back when I was a student teaching assistant we still had real lab courses with a parts stockroom and lab equipment. I always got a chuckle out of seeing otherwise brilliant students scratch their heads in dismay when they built a circuit and found that the measured voltages & currents were not precisely what their calculations predicted -- or when things changed as components heated up. You can lecture a student about parameter tolerances and temperature coefficients, but nothing beats building a circuit and experiencing those effects first-hand.
Another humorous memory from those days: a two-stage bipolar transistor amplifier design project with feedback. Nearly every student would ask "what value should I use for the beta of these transistors? The data sheet says minimum 60, maximum 300." And of course I would tell them to make sure the amplifier meets all specs for beta = 60 all the way to beta = 300.
The ensuing look of terror on their faces was priceless!
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.