Geller Labs
Geller Labs


The HP 3458A DMM

Be sure to hit refresh for the latest version.

Just a few notes on the Agilent 3458A Digital Multimeter:

The new hp 3458A was described in the April 1989, HP Journal. [1]

Serial Numbers: Note that while a general rule of thumb with hp instrumentation is that the S/N prefix plus 60 gives the year of manufacture, the prefix appears to be more of "series time frame" for many models. For example, the hp 3400A prefix series 2415A which presumably began in 1984 (24+60) includes meters manufactured at least from 1986 and 1990.

A particularly good example of the exception to the rule is the hp 3458A, where instead of the traditional four digit prefix, on at least some units, "US280" was the prefix (Not, for example "2803"), with the numbers that followed, the specific units S/N. Serial numbers were over 32,000 by the year 2005. Close inspection of the S/N tags on some units shows that the US280 was separately colored or highlighted on the label. Earlier serial numbers appear to conform to the hp standard numbering system, e.g. 2823A 0nnnn for 1988 or a 1988 series. 2823A 19100 is another example from the mid to late '90s (so the "2823A" series, presumably beginning in 1988, ran through the 90's). The change over might have been sometime in 2004 (after the Agilent name change about 2000). 2823A ~22820 has the hp logo, while ~ US280 27740 (sold ~2004) has the Agilent logo. It is possible that the serial numbers have continued sequentially despite the 2823A to US280 prefix change. ~ US280 32890 was sold about 2005. SG450 40233 to SG450 40350 were from about 2008, as were MY450 43514 to MY450 4413. An October 2013 eBay listing for a new Ag 3458A opt 001/002 shows the latest S/N seen to date, MY450 49479.

Earlier S/Ns can be identified in part by the fine meshed cylindrical dust filter over the fan opening on the back of the unit. The oldest units, perhaps below about 2823A 00200 have paper looking white background serial number tags. By about 2823A 00900, the tags were similar to the modern silver metal foil tags. Later units had a simple grill and sheet of filter material. Earlier serial numbers appear to conform to the hp standard numbering system, e.g. 2823A 02nnn. The earliest serial number seen in recent sales was just above 2823A 00100.

eBay prices vary widely from $200 for parts units to $6k to $8k asking prices, with actual sales in the range of $2k to <$5k. $3k seems to be a recent average sale price as of about May 2013. Here is a snap shot of eBay sales as of May 2013 which also shows some of the serial number progression PDF. Note that many of the lower priced units were inoperative, not all that surprising for a unit packed with relays, switches, battery ROM, and some with passives and actives dating back to the late 80's. The range between the red marks is the range apparently impacted by Service Note 3458A-18A (see footnote 2 below), units that might need a replacement A3 card. There are many just plain broken 3458A on the used market, not all drift problems should be related to the 18A service note. Some serial numbers sell over and over again.

We might start to correlate serial numbers to manufacture year if we get enough data. If you have opened your unit and know the mfgr date from component date codes, we would appreciate any input data. On request we can hold your exact serial number, or just report it to the nearest ten or so (e.g. round 30876 to 30870) and tell us you rounded.

AutoCal: In our experience, many hp 3458A users do not realize how important the auto-cal operation is. Auto Cal is not a "calibration event". In normal use, Auto Cal is performed regularly, at least every 24 hours, and preferably for every change of ambient temperature > 1 degree c (for highest measurement accuracy, yet another reason why metrology labs are typically kept at a relatively constant room temperature). If you are only working in DCV, our understanding is that it is okay to just do AutoCal DCV (instead of the more time consuming AutoCal "ALL"). If you find you need to use AutoCal more often, you might have a bad A3 board in need of replacement.[2]

Note that the specs for the 3458A DMM temperature coefficient without use of ACAL for every 1c change, are pretty comparable to modern high quality 6.5 digit DMMs. The 3458A page has a link to the data sheet PDF. For example, on page 10 of the data sheet, the temperature coefficient for the 10 V scale without use of ACAL for a >1 c change is (0.5 ppm of the reading plus 0.01 ppm of the range) per degree c. With ACAL, the temperature coefficient for the 10 V scale is (0.15 ppm of the reading plus 0.01 ppm of the range) per degree c. Some 3458A might perform much better than spec.

Service Notes: Be sure to review the hp 3458A Service Notes!

The amazing! hp / Agilent 3458A (we have had the good fortune to work with many over the years since about 1990), can measure AC volts to a very respectable noise floor in the SetRNDM AC mode. With the input terminals shorted, we typically found noise floor values from about 14 uV to 25 uV. If you get a chance to check yours, please let us know! (joegeller -at- gellerlabs -dot- com)

If you are moving up to a hp 3458A from a hp 3456A, note that it will take some time to get to used to the new command set (esp. via front panel button operations). The command set was very IEEE-488 (GPIB) centric and in our opinion, not very user friendly. However, given time, and good use of the custom programmable front panel function buttons (F1, etc.), new 3458A users will see what a remarkable instrument the hp 3458A is. It is well worth the time to learn it by a combination of studying the manuals and lab practice.

Undoubtedly, some day there will be a 34***A series meter to replace the 3458A from ~1989, however for now, in our opinion, it still stands alone as the greatest DMM ever made in production quantities! So how long will the 3458A be around? The prices (not adjusted for inflation) of three known relatively long lived instruments were plotted versus hp catalog year PDF. The hp 5359A Time Synthesizer (an ultra high accuracy time delay/pulse width "pulse generator") shows a rapid change in price increases towards its last years, no doubt related to the production design style as well as demand. Closer to home (DMMs), the hp 3456A 6 1/2 digit DMM shows a similar increase in slope of the price curve as it came to an end. It is difficult to draw conclusions based on the graph, especially since we are looking back 31 years, past the current generation of managers. Yet, if the slope of the price curve can be used as an indication of product run lifetime, the 3458A might be around for some years to come (assuming no significant competition or another production problem(?) such as indicated by Service Note 3458A-18A [2]).

It should be noted that many used hp or Agilent 3458A units are broken, or somewhat defective (certainly not all). The 3458A really shines in metrology applications, (when properly working and calibrated) at constant room temperature. To maintain specified accuracy (well under 8.5 digits absolute in most cases) ACAL needs to be run when the temperature changes by more than 1 degree C. Of course, the venerable 3458A has also been used in thousands of other non-metrology applications ranging from academic research and R&D at national laboratories to all sorts of other industrial applications.

Probably many small labs would be far better served by buying one or two brand new Agilent 34410A (not the older 34401A) at around $1,300 each than a used 3458A of unknown origin. Another point is that most of the 3458A high level DC performance verification is done using a calibrated FLUKE 732B transfer standard. While used FLUKE 732B come with some risk of not operating properly, many labs looking for absolute calibration might benefit from having one or two 34410A along with a working 732B. Using offset to bring the 34410A measured value just below 10.000 00 V on the 10 V scale reveals 7.5 digit with second display statistical averaging (e.g. 9.999 999 V). It is unclear how "accurate" the last digit is, however once at 7.5 digits in applications where averaging is acceptable, the difference between the 3458A and 34410A is far more grey than black and white. Also, for 10 V applications with a FLUKE 732B, null measurements on a 34410A 100 mV scale are more accurate than a direct measurement on an Agilent 3458A 10 V scale.


1. 3458A Digital Multimeter, HP Journal, April, 1989.

2. After we sold our late model 3458A (badly missed, but temporary financial realities prevailed), while trying to understand the 3458A serial number system, I came across the 3458A-18A Time Drift issues Service note from 2007-02-13. Mostly I used our 3458A in recent years in conjunction with the FLUKE 732B after long times of power off, so I did not look into its short term (24 hour stability). However, apparently something went wrong on a series of the late model units (which is remedied by return and replacement of the A3 board). Apparently for a time (inferred from the service note), there was some problem (a manufacturing problem and/or part failure or substitution?) with A3 card production .

Serial numbers to watch out for are Serial US280 31400 / US280 32927. From the note, "Some recently produced 3458As may be “out of DCV specifications” due to time drift issues. One issue is long-term drift of the internal reference that may cause the instrument to be out of DCV specification in time intervals less than 1 year. The second issue is short-term drift that may cause some instruments to be out of specifications within 24 hours following an ACAL." Parts Required: 03458-69503 PCA A/D & Inguard Logic (rebuilt A3 assembly). My understanding is that this was an A3 board problem unrelated to the reference (not a reference problem).

Generally, when asked for advice, I recommend purchase of later S/Ns since so many of these fine DMMs were run 24/7 with open input terminals causing high rates of relay state changes (e.g. from autoranging relays changing state with the hi Z input and open terminals).

However, with some later S/Ns as per Service Note 3458A-18A, if the drift rate is greater than 0.43 ppm per day, you might be looking at an additional A3 board swap out. hmm, sometimes newer might not be better. (Also, like the 3456A, occasionally there could be some trouble with the mechanical front-rear terminals mechanical switch, particularly where it was never moved in twenty years. It might be possible to spray some contact cleaner (e.g. Deoxit Gold or D5 then P5) into the switch, not sure.)

Note that the service note specifies a FLUKE 732 A/B type transfer standard to check its DCV stability.



hp manuals and related catalog pages are reproduced with Permission, Courtesy of Agilent Technologies, Inc.




Tech Notes
About Geller Labs