EEVblog #1017 - Enter The World Of Atto Amps

Dave tears down the Keithley 617 Electrometer, capable of measuring sub-femtoamp (attoamps!) resolution.

Support the EEVblog through Patreon!

Donate With Bitcoin & Other Crypto Currencies!

EEVblog Amazon Store (Dave gets a cut):

[bobcunningham]

I've got war stories from my electronics lab technician job (before and during my time at uni), including one about building a 13-decade logarithmic ammeter that started down at 10 fA (calibrated), that added 3 lower decades to an existing 10-decade system. The PhDs did the circuit physics, the EEs did the circuit design, but I had to build and test the prototype, calibrate it, push it through DFM (make it repeatable and stable), then write the build, test and tech manuals for it.

Counting electrons is totally nuts: They never wind up going where you want them to. I had to enclose my lab bench within a Faraday cage. I had to remove the anti-static mats, clean everything with Freon, alcohol and/or acetone to remove residue traces. Special cables, special solder, special flux; nothing was standard. I had to take many of my measurements remotely.

The signal source was a unique and ultra-sensitive "current chamber" radiation detector driven by high voltage (~12kV) that was located 100m away, meaning it was connected using ultra-low-leakage coax. A nightmare to develop and test in the lab. The cable capacitance alone was horrible to deal with.

Precision log amps are strange circuits. We put a matched Darlington pair in the feedback loop of a Burr-Brown instrumentation amplifier. Simple, right? Not when we had to surround it with bias and thermal corrections to maintain sensitivity and log-linearity. Which added circuit load, which meant I had to start my testing and calibration a full decade lower, at 1 fA. Which meant detecting down to 100 aA to ensure we could reliably measure 1 fA so we could repeatably calibrate from 10 fA. Ugh.

Once the prototype was working, I had to build 10 pre-production units for environmental and accelerated lifetime testing, to ensure the calibration held for the required time under all required conditions. Double-ugh.

My main technical contribution was to thermal stability: I replaced the heat sinks and thermal straps with a machined block of copper, to ensure the instrumentation amp and the Darlingtons were kept within a fraction of a degree of each other, so the thermal compensation circuits would always work as intended and the calibration would be stable. (We almost had to go to a temperature-stabilized oven, but that would have created a cascade of problems that could easily have made things worse overall.)

At the last minute they changed both the PCB conformal coating and the potting compound. Not something you want to do around calibrated low-leakage high-impedance circuits (despite the new compounds being better). To avoid complete retesting, I "handled" it by redesigning the heatsink to become a sealed box, keeping the new materials far away from my calibrated log amp.

While it was a ton of fun, it was this project that convinced me to switch my major from EE to CE (computer engineering), though I did keep an emphasis on sensor/signal processing.

Actually, being an embedded/real-time instrumentation software engineer who is also a fully qualified lab tech has proven to be an ideal career choice for me. I can prove to the EEs where and how they screwed up, but I don't have to fix it myself! Bwa-ha-ha-ha!

[grooeygroo]

Oh my word, I found a Keithley 617 in a skip a year ago! It's in working order and I had no idea it was capable of that kind of range! It will now be treated with even greater respect, I'm glad I rescued it. Cheers Dave.

[gromit]

"Keep your used solder wick" Dangerous business giving engineers more excuse to hoard more 'useful' bits and pieces

[hardwareful]

I made a common-mode joke
but it was filtered

[DaveMcAnulty]

Really like to see a video explaining guards vs ground.

[km]

carefull dave, if you hold it sideways the electrons will fall out and skew your 62, 5 electron count

[nexthack]

Not only you would use an atto-ampere scale to measure electrochemical stuff. But also, you use it for CMOS reliability. It allows you to determine if your gate oxide suffered from SILC (stress induced leakage current), RILC (radiation induced leakage currents), soft breakdown, etc. We used to stress CMOS test structures ("large" capacitors or arrays of MOSFETs or even arrays of floating gate MOSFETs), and then we measured the gate current, to determine if damage was induced on the gate dielectric.

[JeffDumps]

Nice video Dave! I noticed misspelling on femto and Coulomb in the video... I can see why your YouTube proposition is so important! The amount of work and time that goes into production with a highly technical topic is huge! I'm not picking, just mentioning. We all got the point and learned something all while being entertained, well done!

[MrJetra]

Btw. Do you know where the names "femto" and "atto" come from? They are derived from the danish numbers femten (eng. fifteen) and atten (eng. eighteen).

[KouroshMokhtari]

Hi Dave.
The tag or label "Do not touch the circuit board" is because 74CHxxx ICs are actually CMOS ICs and touching the circuit board could potentially zap them due to electro-static dischrges.

[sbalogh]

Nude virgins with grey beards. LOL. I knew a couple old ham radio guys who fit that description.

How do you even make a 250 gig resistor? Teflon windings?

[bytefree]

Even more crazy that this STATE OF THE ART gear is .... HOW DID THEY CALIBRATE THIS THINGS ?. I mean, there most be something even more presice out there .... mind blowing. Now, you can even breath on this can. The stand offs remind me to RF stuff.

[cemx]

At 14:59 the 250gΩ resistor is a lowly 5% tolerance unit. 5% of 250gΩ is 12.5gΩ or 12, 500, 000, 000Ω. Anyone even seen a 1gΩ resistor? Sheesh!

[uzaiyaro]

250, 000.. million, ohms. 250, 000 megohms. what the fuck is this alien shit

[linagee]

Attoamps = a bee sneezing on a wire causing a few electrons to move.

[LutzSchafer]

Nice video Dave. In the earlier days of SLR camera light measurement we developed log amps that had to be precise down to some 10 pA for lowest short circuit current of the photo diode. A good reminder of those days.

[oliverthane]

A video on current guards and very low current measurements would really rock Dave ...its a bit late for me since I have been dealing with pA measurements in photo diodes for the last couple of years now ... but man getting your head around guards, high impedance measurements, and just pcb design and cleaning etc, below 1nA is really a big learning curve ... a fundamentals Friday on this stuff would have been an amazing help for me 3 years ago and hopefully others too !!!

[argonman]

Nice reference to Jim Williams from Linear Technology! He was a master. I worked for Linear for 16 years and Analog Device has now purchased us. Great minds. Cheers!

[BreadboardingDe]

That's the sort of videos I really enjoy whatching. In fact they could be like 2 hours in my opinion. :)

At this point I want to thank you Dave for your great videos, great explanations paired with humor. Your channel has brought me so much knowledge, keep going on!
-> I can't wait any longer for my training to start next month to get even more into this topics. :)

[rutgerhoutdijk]

I love the color scheme of that thing, much more character than those clinical modern things