Planet Analog DesignLine Blog
Comment
WKetel
I had to design an inexpensive fixture quickly to verify repeatability of a ...
CodeWarrior1241
One of the odder hings I did test-wise was the setup for a college capstone ...
What’s the most unusual tool or fixture you ever made?
Bill Schweber
2/17/2012 12:53 AM EST
Reading the recent column (sorry, blog) from DesignCon about the special type of screwdriver needed to open the latest iPhone (“iFixit blasts Apple for tiny torx screws”) reminded me of a similar incident from my distant past. Many years ago, I was “tasked” with opening up a handheld Nintendo Gameboy.
No big deal, I figured, until I looked closely at the Phillips-like tiny screws used. I say “Phillips-like” because the screw-head had the same conical flare, but with three flutes instead of the four of a conventional Philips screw. I assumed this was done to prevent casual hackers from getting into the box.
What to do? We could have purchased the required screwdriver by mail (I found out later that it was called a TriWing), but that would take a few days to arrive, and we were impatient (and maybe a little cheap, too). Plus, a “real” engineer doesn’t let lack of tools stop him or her, but instead views this as a challenge.
We took a standard Philips screwdriver with the appropriate body diameter, and cut the tip off with a hacksaw. Then we used a bench grinder to form the overall “conical” screwdriver-tip shape. Finally, we used a Dremel hand grinder with an abrasive disk to cut away material, leaving the three flutes (wings) we needed; the tool did the job. The whole process took about two hours, and we were quite satisfied and pleased with ourselves.
And why shouldn’t we have been? We had seen a problem, and improvised a solution: we had done real engineering and tool-making. (Isn’t tool-making one of the factors which distinguishes humans from animals? That’s a topic for another time.)
That final point is what really struck home. Way, way back in the day, engineers and scientists often made their own instruments, tools, jigs, and fixtures. In fact, they also often made the tools needed to make the jigs and fixtures—hard to believe these days, but true.
For example, in the recent excellent book “World in the Balance: The Historic Quest for an Absolute System of Measurement” by Robert Crease, there’s a long section on the vital role which the diffraction grating has played in advances in physics in general, and precision metrology and standards in particular (also see “The quest for ever-better primary standards is a fascinating story”).
One thing which struck me was how the leading researchers, often amateurs working on their own, not only ruled their own gratings, but built the ruling engines needed, or enhanced available engines built by other experimenters to reduce their already tiny imperfections.
Even earlier, John Harrison, solo maker of the legendary clock of the mid-1700s, not only had to cut his own high-precision gears, but he also had to make the gear-cutting machines (see Dava Sobel’s Longitude). Go to any high-end science, industry, and technology museum, such as Science Museum in London, the Galileo Museum of the History of Science in Florence, or the Collection of Historical Scientific Instruments at Harvard University, and you’ll be humbled not only by the instruments themselves, but by the realization that many were not simply bought by their users, but hand-made by them, often along with the tools and calibration tooling.
In addition to tools, engineers often have to build fixtures and jigs for prototype and production test, assessment, and evaluation. When you are pushing the product envelope—whether just slightly or a lot—what you need may not be available or affordable, or is needed right away, so the ability to make that special fixture or jig is just as important to success as a good design and execution of the design.
What’s the most interesting, challenging, or clever tool, fixture, or jig that you have seen, or made yourself? Was there a simple, clever one that stands out in your recollection? ◊
|
|
|
|
|
antedeluvian
2/17/2012 10:01 AM EST
We make a series of products that measure AC currents (1A-100A). In order to test them one could use a variac and a resitive load, but that would prove inaccurate, and possibly dangerous as a result of the voltage and/or the heat.
The solution is to create a toroidal transformer with a low number of output turns so that there is low voltage and high current. We also controlled the transformer input by a micro and monitored the output closing the loop with some PID control. The jig is still in use.
The whole project was written up in a Circuit Centre Online article "Developing an AC Current Generator" in June 2000. Unfortunatley nothing lasts forever on the Internet. If interested, I can provide a copy.
Sign in to Reply
antedeluvian
2/17/2012 10:28 AM EST
Following in the theme of current generation, we often have to run an electrical conductor through a Current Transformer and the pass a current to test the unit. To make a high speed connection the best parts I have found are made by Hypertronix called the Y series. These are single pins and sockets with very low insertion/removal force and high current capability. Unfortunately their web site does not appear to detail these at all. You can seen them on page 1/2 of their catalog
http://www.hypertronics.com/Files/Full%20Catalog%20Download%20(Low%20Res).pdf
I got better information from their distributors Kensington Electronics who have a pdf document with more detailed ordering information.
Sign in to Reply
antedeluvian
2/17/2012 10:53 AM EST
Forgive a further story on AC current generation (we actually do much more than AC current measurement)- this is the most bizarre equipment I have ever seen.
We asked our Chinese subcontractors to make a current generator for about 5A. They steadfastly refused to use the transformer technique or even a variac into a fixed resistor.
Instead they used two power resistors mounted in a large plastic enclosure box. The resistors consist of an exposed resistive track spiralling around a cylinder of about 9" length. The space between "loops" is about 1/2".
Above each resistor they mounted a threaded bar supported through the walls of the enclosure, and on each bar they mounted a carbon motor brush that touched onto the exposed element of the resistor making two rheostats. They are connected in parallel to reduce the heating effect. To adjust the current, the bars are twisted allowing the brush to move along the resistor and they are supposed to be adjusted more-or-less in tandem. Added to this improbable form of adjustment, the brush width is less than the 1/2 inch spacing between spirals and so there are points where one or even both brushes do not make contact and so the current goes through steps with huge variations. This is also compounded by the fact that the threaded bars are not (or at least do not stay) completely straight and contact resistance varies.
Seeing is believing! I would love to show the pictures here.
Sign in to Reply
Mike_at_icmanage
2/17/2012 12:57 PM EST
This one is fairly simple, but some time travel needed...In my off-hours quest to convert 60's era organ vacuum tube amps into workable guitar amps, I was really scratching my head about why the frequency response of my modified amp didn't seem to line up with the SPICE results (yes - you can get SPICE models for tubes!).
So - how to measure without expensive test equipment?
Connect an iPod touch to the amp input with frequency and white noise generator apps and an iPhone with a frequency analysis app to measure the speaker output response.
An interesting juxtaposition of 50 years of technology evolution and a few billion transistors needed to measure what's going on with circuit designed with slide rules and consisting of a handful of tubes!
Sign in to Reply
no clever name
2/17/2012 2:15 PM EST
Due to poor chinese quality control we needed to test automotive alternator stators. Conventional testers were very expensive, and not 100% accurate when be bought one on approval, which we declined.
So, I had THE idea, if you drive the stationary rotor with 60 HZ ac, magnetically it looks to the stator as if it is spinning. You then turn it slowly to align with each of the three phases, and look for balance. This tester was cheap, and 100% accurate. Just so happened that 110 VAC there was enough impedance to limit the current to about 2 amps, so I got to use a inexpensive isolation transformer. In intermittent use, the rotor never overheated.
Rod
Sign in to Reply
ccalvin
2/17/2012 2:36 PM EST
Years ago I made a tri-wing from a nail to repair an iron. I now know that you can by a box of "security" bits at Harbor Freight for less than $15. Most everything you will ever need except the Apple 5 point and perhaps really tiny sizes.
Sign in to Reply
Bob T
2/17/2012 3:59 PM EST
We built an ESD test setup with a box (about the same volume as a person) covered in aluminum foil suspended with fishing wire. it was charged from the high voltage circuit on a monitor.
The second one was an FCC test station using a black and white TV with open loop AGC. The system passed the FCC related tests on the first try much to the astonishment of the tester.
Sign in to Reply
zeeglen
2/17/2012 4:13 PM EST
Two of many electronics fixtures stand out, one was a block of stainless steel for testing a Peltier-cooled EDFA pump laser and a power transistor, both bolted to the steel. Controlling the transistor power dissipation heated the steel and provided the hot ambient to test the Peltier cooling effectiveness.
Another was the CanLan, a 250 Mb BER tester in tin can shield housings. A photo is in this article:
http://www.eetimes.com/electronics-blogs/pop-blog/4212155/Opportunity-knocks--management-fails-to-answer
As for mechanical tools the easiest was for screws that had a small blob of metal in the center of the slot. Took all of 3 minutes to file a matching notch in a flat blade screwdriver.
The hardest was a jig for carving the plastic walls of a Z scale model locomotive shell. The manufacturer made 2 versions, a "powered A unit" with a cab and a motor, and a "unpowered B unit" with no cab and no motor to trail behind the A unit. The axle friction was atrocious and severely limited the pulling power of the A unit. The strange thing is the manufacturer for whatever reason made the plastic B shell walls too thick to fit around the motor brush holders without bulging outwards. The jig is a wood sliding carriage in a wood guide attached to the base of a Dremel drill press. To shave the walls in the right area to the right thickness I just place the shell in the carriage and slide it back and forth while slowly lowering the Dremel side-cutting bit. Now the B shell fits the motarized mechanism without bulging, I've done this for many people.
Sign in to Reply
EDA360 Insider
2/17/2012 4:26 PM EST
I built a teaching tool at HP back in the mid 1970s, when DIP switches were a new thing. I'd designed two DIP switches (one with eight SPST switches and one with two DPDT switches) into a serial-I/O card and needed to train the field on how to set these for various configurations. I built two giant DIP switch models measuring about a foot by a foot and a half from foamcore and appropriately cut Styrofoam disks and then made a video tape (on half-inch reel-to-reel videotape!) as a training guide. Pushing those giant DIP switch rockers was funny, and made the training memorable.
Sign in to Reply
David Ashton
2/17/2012 5:58 PM EST
I've always made a lot of my own test gear, mostly for cost reasons. Probably the first was a car light bulb soldered to a PL-259 connector when I was about 19 and a radio technician in the police in Rhodesia. We used 10W 40 MHz AM gear and the light bulb was a great dummy-load-come-wattmeter. Being AM it brightened when you modulated - built in modulation meter as well. I used the technque for 80 and 150 MHz TXs as well but it needs a trimmer across the bulb to null out the inductance at those frequencies.
Later while running my own business supplying dumb terminals to travel agents, I had problems with terminal controllers whose RS232 chips kept getting blown by nearby lightning in the summer. I made up an "emulator" for the Z80-SIO which lugged in in place of the IC and used a 555 to put out 1Hz on all the output lines. I then had a plug on the RS-232 socket with LEDs to show if the +/-12V were getting through. This also had a loopback swithc that routed the signals back to the input pins and back to the SIO tester, which had LEDs to show the input signals. It was then an easy matter to identify which RS232 chip (the usual 1488/1489) were faulty.
These terminals also had keyboards with an 8-bit parallel interface (much like a centronics printer port). Not all the keyboard functions could be tested on an offline terminal, so I made up a test box that indicated with 8 LEDS the state of the output lines. Dead keys or dead output lines were then easy to identify. I actually converted the box to a centronics port monitor later, I think I still have it somewhere.
I wrote this up in more detail some time ago here:
http://www.eetimes.com/electronics-blogs/other/4211470/How-to-have-fun-with-klunky-old-terminals
Sign in to Reply
David Ashton
2/17/2012 6:12 PM EST
I also have made up a lot of my own Breadboard systems, obviously with power supply built in, but also with switches, DIP Switches, speakers. power transistors and signals available on contacts beside the board. I take off the 100Hz from full wave rectified 50Hz before smoothing, square it up and then divide it with a 74LS390 down to 0.5 Hz, with all the intermediate taps available. Not much use if you're doing RF stuff, but I don't do a lot of that.
I won one of the excellent Xminilab boards in the recent 555 contest (no, just a lucky door prize, but a nice one...) and want to build that into my next breadboard - its a scope and sig gen in one, more info here:
http://www.gabotronics.com/
Sign in to Reply
hm
2/18/2012 7:53 PM EST
I spent few months to make different tools and setup to make my first fibre optic current and voltage sensor work. That was very satisfying.
Sign in to Reply
Joe.Sleator
2/19/2012 5:16 AM EST
Bought a Chinese manufactured commercial 5 bladed helicopter head for a 450 sized electric helicopter. The price was very nice, less than $30, without the carbon-fibre blades, but the quality control was non-existent. The central bore-hole was eccentric. Not only not centred in the part, but also not parallel to the centreline of the part. Thought I might TIG up the hole with fresh aluminium and re-bore it, but later decided to machine a new one. There's no video of the new one, as it'd be very boring, with no machining errors. Here's the video of the very eccentric original:
http://www.youtube.com/watch?v=Mu5honUI3jc
Sign in to Reply
Haven Wildcat
2/21/2012 5:12 PM EST
We designed a solid state contactor that makes and breaks 85 Amps at 28 VDC for de-icing a horizontal stabilizer on a business jet. For a product to be deemed airworthy by the FAA, it must pass a series of qualification tests outlined in a document named RTCA/DO-160. One of the tests simulates a run-away generator by subjecting the UUT to an 80 V surge for 100 mSec with no specified source impedance. It would be tested both ways, but when our switch was ON when the surge occurred, the load on the test setup would be roughly 250 Amps. Even though the test lab is very familiar with the test, I had a suspicion they wouldn't have anything able to make and break 250 Amps at 80 V. I built my own surge switch out of eight big honkin' obsolete MOSFETs and three 9V batteries to power one of our timers adjusted for 100 mSec. I offered it to the test lab at the start of the testing, but they assured me they had some equipment that could handle it. After they fried 2 different multithousand dollar pieces of test equipment, we shipped them mine overnight so they could stay on schedule.
Sign in to Reply
dleske
2/22/2012 9:15 AM EST
I had to test industrial computer boards, intended to be sold rated for -20C~+70C, so I built a thermal test chamber.
I got a large cardboard box (~1m^3), lined all 6 faces inside with 100mm Styrofoam, then covered that with corrugated cardboard. (Styrofoam is too easily damaged or melted.)
Inside I placed a VME card file, loaded with test cards, supported on bricks.
Under it I placed a domestic fan heater - modified by disabling the protective thermostat, adding separate supply cords for the fan motor and heating element, connected to an external industrial thermostat controller, with the PTC sensor mounted inside the box. [This provided a controlled air temperature, not blowing directly on the test items.]
I connected multiple type-K thermocouple sensors to an 8-channel PC-interfaced acquisition unit. Sensors placed on key points, eg: frame (~ambient) temperature, CPU heatsink, PSU electrolytic caps, PSU switching transistor, Fibre-optic transceiver.
With power & data cables connected, and the lid closed, I could monitor system function and log internal temperatures.
I happily ran it up to Ta=40C, then 50C, then 60C, 70C, 80C (for a week) and 90C (for a few hours) - and was very pleased to see my boards function perfectly. They will work OK in the western Saudi desert (Ta=50C)!
Then for low temperature, the heater was replaced with a load of dry ice (CO2). The concern was for crystal start-up at low temperatures, so the system was Off during cool-down, except for brief function tests with every -5C drop.
Everything worked fine, until suddenly the unit went dead at -25C. We found the crowbar zener on the 5V rail was burnt out! Vz reduces at low temp, while regulator output increased. Re-spec the zener with thermal tolerances allowed for, and success down to -40C.
A couple weeks very enjoyable work, and I was very pleased. It gave the company confidence to expand the rated operating temperature range of the product, to meet customer demand.
Sign in to Reply
SherpaDoug
2/22/2012 9:35 AM EST
My first job out of college was on a team building a capacitive discharge impulse source of 4kV @ 40kA for geologic soundings. While others worked on the ignitron switch I was to work on a voltage monitoring system. Without a working switch how could I test my part of the system? We built a "whammode" consisting of a couple of tungsten contacts held apart with a fiberglass spring. When we hit it with a fiberglass handled rubber mallet the contacts closed and the system fired.
Sign in to Reply
Don J
2/22/2012 10:48 AM EST
Recently we had a development vehicle with a custom Front End Accessory Belt Drive. As we were preparing for a Demo drive we started having belt squeak issues. We didn't have time to tear things down and find the source of the problem. The belt supplier suggested applying bar soap to the belt to stop the squeak. We went to the local store and bought some soap. I took a piece of wood from my garage and duct taped a piece of the soap to the stick. I was able to carefully insert the "Soap on a Stick" into the front end of the engine, brace it against a bracket and then lightly touch the soap to the running belt to apply the soap. This worked great and got us through several demos without any belt squeak.
I still have the "Soap on a Stick" in my drawer, and once in a while I get a whif of the soap perfume scent to remind me it's still there.
Sign in to Reply
D J Liu
2/22/2012 1:51 PM EST
My 8 year old son built an inverter using parts in a Snap Circuit set: a battery pack, a DC motor and a transformer, then used a pair of LED to prove what he got was really AC.
Sign in to Reply
Streetrodder
2/22/2012 4:14 PM EST
My most interesting test jig was an XRay phantom to simulate inserting a catheter into the human body. We needed to test the image quality of a angiography room. Turned out to be incredibly easy. Clear plexiglass is optically almost identical to the human body. I bought some 1 inch thick by 1 foot square pieces and some 1 inch thick 2 inch rounds for spacers. From there we could build the needed 'body' and catheter depth to test and adjust the image quality.
Total cost $120. Cost from phantom vendors for something that didn't work as well - $15,000.
Sign in to Reply
jimfordbroadcom
2/29/2012 9:08 PM EST
Like dleske, I've worked with an improvised temperature chamber, although I myself didn't build it. Back in the mid to late '90's I was at a now-defunct startup company, and since we didn't have much of a budget, we had to use a wooden box for a temp chamber. Ours only went hot, and the boards we were testing generated plenty of their own heat. VME boards, just like dleske. With a plexiglas window and a stick-on digital thermometer, we could monitor the temperature and the status LED's of the boards. IIRC, we had to keep some chassis slots unpopulated to avoid overheating the boards.
At another company, I put together several 250 W Dale power wirewound resistors to stress test an ultrasonic amplifier card built around a hybrid power op amp. It put out ~30 Vp-p and 20 Ap-p IIRC and we were able to stress it enough with pulses of several hundred milliseconds to the point where we could see thermal pulse droop (scope at ms/div) and waveform distortion (scope at microseconds/div). I named this load the "Mother Load"!
Late last year at my present employer I was checking out some problems with FM radio measurements showing too much variation from bench to bench, and I needed a way to verify that the filters on the test boards were operating properly against conducted EMI. I couldn't probe on the point I wanted to directly, so I built up a circuit just like the series ferrite bead and shunt cap that was on the boards. I first measured the unfiltered spectrum with a piece of semi-rigid coax (a pigtail) with a series cap soldered on the center conductor connected to a spectrum analyzer. Then I added a series ferrite bead and a shunt cap from the center conductor to the outer ground conductor and probed at the same unfiltered point. Voila, the peaks on the spectran dropped by at least 60 dB, so I knew that the conducted EMI was obliterated. We ended up cutting the radiated EMI by performing bench tests in sealed RF test chambers. Variations solved.
Sign in to Reply
CodeWarrior1241
3/1/2012 4:49 PM EST
One of the odder hings I did test-wise was the setup for a college capstone project. We had worked out a system that would examine an EKG plot, identify the presence of several common arrhythmia variants, and deliver a drug through a trans-dermal patch into the skin. The patches were available off-the-shelf, and the microcontroller/FFT stuff I tested separately using canned datasets that are openly available online.
What we didn't know how to test was the trans-dermal patch... The model we chose was electrically activated with a low-voltage pulse, and contained a small reservoir for the substance to be injected through skin. Artificial skin (for lab work) exists, but it's more expensive than college students are prepared to pay for. So, to demonstrate the effectiveness of the delivery system, we filled the patch fluid reservoir with blue food coloring, and cut up and stretched out single a lambskin condom over a clear plastic can filled with water. When the system turned on, we could see faint blue color in the water inside the can, proving that the "drug" was successfully delivered. We could also examine the amount of fluid in the reservoir of the patch itself, confirming that we delivered only as much as was necessary.
Sign in to Reply
WKetel
3/2/2012 10:42 PM EST
I had to design an inexpensive fixture quickly to verify repeatability of a developmental contact sensor that was used to detect crash impacts. The final iteration used a motor driven cam to lift and drop a small hammer, with the impacting part being a hardened steel ball, which hit a ground surface steel plate that the sensors were clamped to. The prior version used a pneumatic gripper designed for robotics to drop a steel ball onto the plate. That one worked well, but it was inconvenient and had to be manually reloaded each time. The motorized cam and hammer had a UHMW cam follower to reduce cam wear. THat fixture was fun to build.
Sign in to Reply