This a vote proposal to get an EBay HP 3478A and test leads.
Total cost will not exceed $250.
Probably less, as they show up on EBay in the $150 range in various states, but shipping can be nontrivial.
Lorin will help select the actual unit. I can purchase it if he is busy.
yes. I’m an ebay ninja and have no problem playing chase.
If Dave would like to play eBay ninja that is cool with me.
There a a few out there now. I don’t think we are freaked out about calibration but obviously we want a working one. I wouldn’t buy anything until the vote goes through. With the holiday lack of meetings I’m not sure how Elly will handle it.
Let me know how I can help.
Just FYI, we have an HP 3457a multimeter (which is in an order of magnitude more powerful - also huge and heavy). That meter is precise enough to calibrate against IMO.
Sweet, we could use that as a reference.
Lorin: what are your thoughts on calibration? At my work everything is calibrated by an external company so I’m used to having everything trusted.
Would we want to bring in some reference source and verify our stuff? I know we are not making nuclear fuel rods but I would hate to work on something and then find out my tool was off by 10%.
Perhaps a brief tour of the electronics area (when it is completed) could be super useful.
Can do. With these HP/Agilent meters, the calibration does not drift much and most of the error will be seen in the least significant digits. So, on our 3457a, that means the 7th+ digit. On the proposed 3478a, it would be the 5th+. Furthermore, I’ve never seen these meters fall out of cal much. I can guarantee that a used HP will be 100x more precise and accurate than a brand new Chinese meter. It’s all about precision components and tolerances in the machine itself.
12 identical resistors of 0.01% tolerance have [0.01%/(sqrt(12))] total tolerance when in parallel. So, with 12 precision resistors we end up with a tolerance of .0028%. Semiconductors are carefully gain-staged this way as well. With careful feedback, current, voltage adjustment and you can knock it down to the one millionth sort of value needed. This is why it’s a big ol’ box and originally cost thousands.
These days, a handheld DMM is likely based on a linear devices or similar precision 5V source IC. This gives low error for a monolithic IC (0.04% I think), but we’re relying on one IC and not compensating for current, drift, temp, etc elsewhere.
Anyway, there’s a reason that the Agilent/HP designs are referenced as examples in “The Art of Electronics” by Horowitz and Hill…
Note: I meant “cheap handheld”. A good fluke or BK is a different animal.
Thanks for the info. I think when people see electronic digits they automatically assume everything is perfect.
With measuring electrical stuff it is a bit less “physical” since you can’t see/touch the thing your are measuring.
I have a friend that worked on a nuclear submarine and they went into dock once a year and ALL the gauges were calibrated. They basically set a semi trailer on/near the deck and went throughout the whole ship and calibrated everything. He would laugh at our calibration procedures at P&G because people would complain about it.
Sure a steel ruler is probably OK (except that one on our wall of shame missing the 66 digit), but anything with a readout (mechanical or digital) can be wrong.
If anyone wants to check a digital caliper at the Hive, I bought some 1-2-3 blocks that are in the lathe tool box. They are accurate within .0001" for each dimension.
We also have a set of machinist squares in the tool box if you need to check that something is 90 degrees.
Agreed. If we do want to get our electronics bench calibrated I can make that happen at some point (for probably $0 ). I just don’t care to pay for NIST cert…
Totally agree we don’t need “official” certification.
I think having some kind of general verification of the meters is a good idea. Especially if something shows up really off it can be flagged.
I’d be happy to be involved once we get everything together.
Regarding the vote per “meetings in person”.
Technically we will be having a meeting, but I am not putting anything on the agenda and I will likely be the only one there, simply to open, close and take attendance, so that everyone can have a break and recover from enjoy the holidays without worrying about “missing” a meeting or something important.
To put this in a different way, say we had a regular meeting, and NO ONE but the one running the meeting showed up. If a vote didn’t need quorum, it was on the docket and had been on the mailing list for the right time… that vote would still go through that night with the votes cast at that time.
For votes that meet the minimum announcement time, it is really up to the proposer to decide when it gets voted on, not me. I think it is always better to wait and vote in person. (This is why I sent the notification out a bit ago, and why I’ve been mentioning it during every meeting the last month hehe) So it’s up to you, Brad. I would pick and list an official date just to alleviate confusion
Lastly, I will be voting yes.
I vote yes and I was / am planning on being there tomorrow night.
We are not able to vote/close the vote on this tomorrow, just because it wasn’t announced last week, just so everyone is clear on that. This relates to the “meeting break” next week on 12/27.
Tomorrow there is definitely still a regular meeting!