EEVblog 1442 - DON'T DO THIS! (PCB Vibration Testing)

An innocent TO-220 regulator gets tortured using a mystery device from the bunker.
The art of environmental vibration PCB testing.

00:00 - What is this thing?
01:00 - I used to use these all the time
02:40 - Electrodynamic Shaker
03:56 - How PCB's are vibration tested
07:00 - Teardown
08:10 - Accelerometers
12:27 - The datasheet
12:57 - Does it work?
15:13 - Let's torture this poor TO-220
20:00 - Slow-mo looks cool!
20:45 - It's going to come-a-gutsa!

Buy anything through that link and Dave gets a commission at no cost to you.

Donate With Bitcoin & Other Crypto Currencies!

#ElectronicsCreators #Vibration #Testing

[flymypg]

When I was developing aircraft instruments, we used two shake tables. One was inside a temperature/humidity environmental chamber, and the other was at the center of a 3m magnetic field cube. The one in the environmental chamber was dual axis (XY), and the other was single axis (Z). We had separate operational vibration profiles we had to put the instruments through to meet environmental specifications for propeller aircraft, jets and helicopters. Doing a full vibe test run would take nearly a week, assuming everything went well.

We learned so much stuff on the shake tables. We used to pot our instruments, until a shake table test showed the potted instrument actually suffered damage easier than the unpotted one. So we went with only conformal coating instead. We also found that sheet metal changes from switching suppliers could greatly affect instrument resonances, despite the metal stock from each source meeting the same specifications. Which was fixed by pressing ribs into key areas, meaning we could use a wider variety of sheet stock. We used ultracaps for temporary backup power in some instruments. All of which failed vibe testing. We tested every ultracap Digikey carried. We had to get some made with custom mechanical changes (Maxwell was local to us), then support them in a metal cage that tailored the vibe spectrum the ultracaps encountered. That project alone occupied our shake tables for two months.

Most shake tables can't deliver high amplitude impulses. When I was developing a super-rugged ultra-high-speed digital video camera we cared more about high level impulse response than vibrational modes. So we got to go a bit caveman: We mounted the camera to a large steel plate, attached a 3-axis reference accelerometer to it, then whacked the back of the plate with a sledgehammer, repeated with the camera mounted in 26 different orientations (6 faces, 12 edges, 8 corners). The camera was used inside cars during crash tests, and within 10 meters of missile warhead tests. It was rated for a lifetime of 100g impacts.

[WizardTim]

For those on a tight budget or time contained project the "Womens Health & Personal Care" category on Amazon has somewhat equivalent products for a fraction of the cost, they also often come with excellent ingress protection so you can do small product ingress testing under vibrational modes, but good luck explaining your need to the purchasing department.

Please don't shadow ban this comment YouTube.

[Arek_R.]

Watching this channel for like 8 years now and hearing Dave complain about to220 "flapping around in the breeze" countless amount of times and now finally he has proven it, we're now fulfilled and can die in peace...

[johnyoungquist]

A strobe light is a key accessory to vibe testing. I built a circuit to drive the strobe at a fixed offset frequency from the vibe drive frequency so the unit under test would display its motion at a fixed alias frequency like 5 Hz or so. I could tune it over about 1 - 10 Hz. It was great for frequency sweeps. It was confusing and fascinating at the same time. You would hear the drive frequency sweep while the motion rate was constant. Great fun and very convenient. You could spot critical resonance modes quickly and not miss any of them. I have exactly the same driver you showed and some larger ones like it. We use a 700W amp to drive them.

[AirCrash]

The first company I worked for after leaving college made these, the size of houses, to test aerospace parts. They used 50Kw audio amps to drive them, like they used to use in football stadiums. Which consisted of hundreds of transistors bolted to water pipes in parallel. The most amazing thing about them was that everything and I mean everything would hit its resonant frequency when we did the test frequency sweeps. Even your vocal cords, if you kept your mouth wide open you would make sounds that you had no control over. Also the tube ceiling lights made from fine glass would flex by a couple of inches without breaking.

[sefarkas]

As a young engineer I had a frequency counter based on RTL logic. one day the pass transistor flapping around in the breeze leaned over and applied unregulated DC to the logic and poof. A while later I asked the a Fairchild field engineer that was visiting my company if he could get me the obsolete logic.
About a month later I found a jewel box on my desk with 40 or so RTL chips. I tracked him down to say thanks. he said "just thanks?" he told me to take a look at the devices again. They had date codes of a week earlier and had been hand packaged in ceramic packages with extra epoxy oozing out on all sides. he said it probably cost Fairchild $10, 000 to make these for me. i mounted the regulator and put a crowbar on the VCC rail.

[donvito]

Hey Dave.. Please repeat this with some new TO-220s, one flapping in the breeze, one down the the shoulders of the leads, one staggered pin arrangement, and one where "somebody had fun" as you often quote, gunked down to the PCB. It'd be good to get an appreciation of how much more securely the different mounting options work.. Not much point in trying a bolted to the PCB/heatsink version.
I was soooo expecting a timelapse/ I'll leave it overnight video waiting for the failure, this happened much more quickly than I expected.

[markc]

You just demonstrated why the Air Force teaches the "High Reliability Soldering Techniques" course that I took 40 years ago. It stressed making solid physical connection and not using solder to bridge gaps because of vibration and thermal movement. I spent the last 30+ years as a civilian shaking my head at the bad soldering techniques in all the equipment I've repaired.

[gordonwelcher]

I used to use a larger version of this to calibrate electromagnetic vibration sensors for testing Rolls Royce turbine engines.
It was driven by a vacuum tube amplifier with 4 807's in push-pull parallel.
Driven by a good old HP 200 signal generator.
It used an electromagnet controlled by a Variac.
A Bruel & Kjaer reel to reel was used to record the engine tests.
A vintage Nicolet 446 spectrum analyzer did the analysis. The boards inside were all
wire-wrapped TTL logic
This setup ran for years with no problems except sometimes you had to whack the Nicolet to make it work.
This was in the 1980's.
There were no airplanes crashing or natural gas pumping stations blowing up so I guess it worked. At least the vibration tests were accurate.

[soundcheck]

Small shakers are also good for getting bubbles out of mixed epoxy resin.

[codingbloke]

I remember using these with PCBs and strobe lighting to actually see how the loaded PCB distorts (perversely testing PCBs for Frequency Response Analysers) . The tall electrolytic caps would appear to sway about like skyscrapers in an earthquake. We wouldn't just use frequency sweeps though, we had recordings of road vibrations as felt in the boot (trunk) of a car or in the cargo hold of an airplane. Fun times!

[robbennett]

Built a shake table with Aura Bass-Shakers and a basic stereo amp and IPhone app for sig source. Had to convince my boss that a 1k pot would be a better choice rather than a 10k pot. Even though the trimmer pots were staked, the 10k trimmer pot changed value enough to throw off the calibration. Tested phase matched cable sets for 1.5 GHz and sometimes they would also change enough to no longer be matched or would fail completely. Did not take much power (10 to 20 watts)at 20 Hz to 1KHz to get the table shaking. Different wave forms were cool to watch with a cup of water on the table. Cost about $200 in parts and pieces to build and was EZ PZ to build. Another good vid Dave!

[gwapod]

We used to use them to simulate human breathing into an oxygen cannula. Our products would sense the patient’s breathing via the pressure change in the nasal cannula and deliver a pulse of oxygen at the beginning of inhalation.
We would use this in production to calibrate the trigger threshold. At least that’s what the modification could be used for.
Great video. Brings back some old memories 😁

[brynnrogers]

A lot of people call them 'voice coils'. I was doing the vibration test on a product at a test house that had a 'car' sized shaker . one of their engineers hooked his bass guitar to it's 10, 000W amp for 'testing'

[mdeslaur]

This very thing happened with transistors on Stern pinball boards mounted underneath the playfield. They had a kit they sent out to owners that contained plastic separators and tie-wraps to fasten the transistor so it wouldn't fall off from the vibrations.

[actionjackrecycles]

I have made many repairs of car amps that were mounted to the subwoofer. Many broken parts from vibration.

[thomasives]

Car audio amplifiers often go down to 2 Ohms and below, which might be ideal for this transducer. Great content and experiment, love the channel. Cheers!

[dougfraser]

If AvE was reviewing this it would be a double-entendre fiesta

[BradRaedel]

The ring that keeps the armature centered is called a 'Flexture'. On larger shakers, they use optical sensors and air bags that dynamically inflate/deflate depending on mass and keep the armature centered vertically in the stroke. They can also be configured for single-shot where you start at the bottom of stroke. We have also used 2 shakers and controlled the phase relationship. For serious vibration testing, a MAST (Multi-Axis Shaker Table) is used where you can control 6 degrees of freedom.

[wbfranks]

Another vibration sensitivity you will see with analog circuits is offsets induced due to flexing of the die inside the package. Once had to rotate a 14 pin dip thermocouple amp 90 degrees and add some mounting screws to reduce its sensitivity to the pcb flexing.