The HP 3458A is still the best multimeter ever built
Designed in 1989, still the reference standard in metrology labs, still not really beaten. One came through our bench. Here is why it matters.
An HP 3458A came through in a lot of test equipment. It is on the bench now.
If you do not work in test and measurement, that means nothing to you. If you do, you already stopped reading this sentence and scrolled down to see what condition it is in.
Why this instrument is different
The 3458A is an 8.5 digit multimeter. Hewlett-Packard designed it in 1989. It is still in production, under the Keysight name, at a price that will make your eyes water. And it is still, thirty-odd years later, the reference instrument that national metrology laboratories use to calibrate other instruments.
Think about that for a moment. In a field defined by relentless improvement, a design from 1989 is still the thing you check other things against. There is no other corner of electronics where that is remotely true.
What 8.5 digits means
A decent bench multimeter is 6.5 digits. That is already good: it will resolve a microvolt on a one volt range and it is far beyond what most electronics work needs.
8.5 digits means the 3458A can resolve about one part in three hundred million. On a ten volt measurement, that is resolution down into the tens of nanovolts.
At that level you stop measuring the thing you pointed it at and start measuring physics. The thermal voltage generated where two dissimilar metals touch in your test leads is larger than your resolution. The temperature of the room matters. The instrument is so sensitive that using it correctly is a discipline in itself.
That is what a reference standard has to be: an instrument whose own error is so far below the thing you are checking that it can be treated as truth.
Why it never got beaten
The honest answer is that the market for instruments this good is tiny, and HP built one so far ahead of its time that nobody has found a commercial reason to try to beat it.
The engineering is remarkable. A custom multi-slope integrating analog to digital converter. An internal reference so stable it is the reason the instrument holds its calibration for a year. A design where the hard problems were solved in the analog domain by people who understood that no amount of digital cleverness fixes a noisy front end.
It is also a genuinely beautiful piece of instrument design, in the way that a tool built by people who cared is beautiful. It has a reputation among engineers that borders on reverence, and the reputation is earned.
Where ours stands
It is untested. It is on the bench, it is in the queue, and it has not been powered up and evaluated yet.
I am not going to tell you what it is worth or what condition it is in until we have actually looked at it, because on an instrument like this the difference between “works” and “holds calibration” is the entire value, and I do not know yet which one we have.
What I can tell you is that we know what it is, which is not something you can assume of whoever listed it as part of a pile of bench instruments.
The general point
This is why we read the boring lots.
A crate labelled “misc bench test instruments” is worth exactly nothing to most bidders, because most bidders will not spend twenty minutes working out what is in the photograph. The value in surplus is almost never in the lots that announce themselves. It is in the ones where somebody has to already know what they are looking at.
We will publish what we find when it goes under power.