I did 10 years in the cal lab at a Motorola plant, but left the lab for development engineering about the time the 3478A came out (1984). We used them a lot in engineering until the 34401 came out.
I got a "untested" 3478A for about $75 on Ebay. It seems to work just fine and I believe the readings it gives me. I don't know what kind of work requires super accurate measurements, but my daily use meters are the kind you get for $5 from Harbor Freight.
Most of my work is with vacuum tubes and solid state audio, embedded computing, and RF devices. I find that a meter accurate to any fraction of 1% is better than needed.
If I need something more accurate, I have an old Fluke 8000 I believe. The HP is for double checking when I don't trust a reading. It's been turned on a few times in three years. It may see more use when I start connecting up some other IEE488 stuff.
Uh, probably not a Fluke 8000, which was a 3.5 digit dog. 8600 was same, maybe 4.5 digit but autorange. Think you mean one of big systems DMM like 8520? (God Google has become worthless as a search engine!)
Doc
Last edited:
WOOF, WOOF.....This old dog still barks out the right numbers......and it IS an 8000 from the stone age. I did sell my Simpson 260 and Eico VTVM just because someone offered me too much money for them.
The old fluke and the cheapie meter both read 9.63 when connected across a new 9 volt alkaline battery. The 3478 (on a different bench) reads 9.6265 or 9.6266 when connected to the same battery. The $5 Harbor freight meters all read 9.62 or 9.63 on the same battery.
This gives me 100% confidence to believe that the battery is between 9.6 and 9.65 volts, which is good enough for anything I'm doing.......
Beware, most of the cheapie meters are way off in readings if the low battery icon is on. Use fresh batteries, or at least compare all of your meters to a standard (the HP3478) before taking measurements that matter.
I routinely connect up several meters to a tube amp when doing some testing. I will have meters on all supplies, and on 1 or 10 ohm current shunts for each output tube. This takes a lot of meters, so I use the cheapies.
Attachments
Well I've pulled the trigger on a 3478A on ebay. I ended up buying one that is in cal (though probably a 30 year old cal from manufacturing) simply because the cheaper as-is ones were all cosmetically bad. I can fix the electronics, I can calibrate it, but damage and engraving will set my OCD off every time I look at it.
In any case, I'll be verifying the cal against our 8.5 digit monster meter and replacing the lithium battery so it doesn't lose it's cal.
In any case, I'll be verifying the cal against our 8.5 digit monster meter and replacing the lithium battery so it doesn't lose it's cal.
Okay got me. Can't tell you just how many 8000s I repaired.... as I've tried to block such memories. 8000 is probably still listed on many military cal procedures. As long as yors works keep using it. I still own a Triplet 310 with my name engraved on it. I still have and use an HP410BR VTVM (with spare AC tube) that I bought in 1978. Why? Still reads AC to 700MHz with only 1.2pF load. Heck even have a few HP400D. VTVMs. For DC accuracy I fire up my Cimron 6200 with reed relay kelvin varly divider and edge lit Lucite display. (People don't expect a digital voltmeter to make noise when settling.) But still dead on vs my repaired 3458A 10V reference and only meter I have that resolves 0.1uV.
Doc
"Best" is a subjective term relative to application. I've actually changed designation of a Fluke 8600 DMM to specifying a Simpson 260 instead in a calibration procedure. Why? Particular adjustment was to "dip" reading for minimum value. Could clearly see dip on analog movement while the brain just didn't integrate the digital readings near as clearly.Fluke later added bar graphs to their displays for that very reason.
Mirrored scales can be more accurate, when viewing from an incident angle. The mirror cancels paralax.
Highest accuracy I know of on earth is Agilent (Keysite... whatever their name is this week) 3458A. 4ppm DC drift per year drift with high stability option and a mind boggling 100ppm absolute AC accuracy to 10MHz. They pull that off by using a PLL on their A/D with a very high speed jitter clock that moves the digital sample points around. Mechanically the meter sucks. I had to replace quite a few front panel assemblies due to broken input terminal mounts. In contrast I never saw a broken input terminal on the older 3456A era meters. They had cast aluminum frames and solid aluminum front panels. The 3458A has a plastic case and front panel with a single screw into a plastic post for the input terminal mount. Bump the terminals once..... New front panel (many companies don't allow epoxy repairs of such things, only valid new mfg parts.)
Solatrons were okay, but took forever to settle. Loved the Daytron 8200s but a genuine PITA to service. Same for old Fluke 8500 series. (Replace the FETs, check power cord, replace the FETS, check line fuse, replace the FETS, check power switch.....)
For myself I have a first gen Fluke 87, an HP971A handheld DMM, half a dozen HF $3 specials (for loaning and high risk work) and a variety of bench meters I accquired along the way. What I use depends on application. For low ohms I have a custom built high precision wheatstone bridge with solid silver contacts that was given to me by a cal lab, but I get about as good off a simple LM317 based constant current source and the 200mV range on most DMMs configued for 4 wire measurement. My favorite for audio is the Fluke 8050A with relative db scales.
Sorry, rambling again. Point was "best" is relative.
Doc
Yep still listed (as of 4 years ago when I retired) on some Navairs. Never used one our lab had HP/Agilent bench meters, repaired, adjusted and calibrated loads of 3458a's
Mine came from the graveyard..........
Can't think of anything needing this level of accuracy in a hobby environment. I'm in the process of repairing a couple of Tektonix digital scopes and still use my Fluke 187 not the 3478a. I don't even use the 4 wire Ω I have an LCR bridge.
Okay, dumb question time. Now that I’ve bought one and am waiting for it to be delivered, is it going to work with my shrouded Fluke probes, or should I buy some probes with plain banana plugs? I’ve been looking at pictures, and can’t definitively say whether the input jacks on the front are plain 4mm banana or shrouded...
I learned a rather shocking lesson about those plain banana plugs like the one stuck in my Fluke 8000 in post #23. Trim the leads flush and put a piece of heat shrink over the plastic part. The ends of the leads stick through and are connected directly to whatever you have it plugged into.
In my case it was a HP6448B turned up all the way. I tweaked some of the adjustments inside so that it makes 650 volts at 1.7 amps. Yanking a plug out when a circuit blew up will leave burn marks on your fingers and cause loss of vocabulary control.
In my case it was a HP6448B turned up all the way. I tweaked some of the adjustments inside so that it makes 650 volts at 1.7 amps. Yanking a plug out when a circuit blew up will leave burn marks on your fingers and cause loss of vocabulary control.
You can use DMOS as a smart bleeder, but I still use 50K 40W wirewound resistors I bought from Electronic Surplus when they were in Cleveland.
Getting back to topic. Appendix C to this Jim Williams apnote has a useful comparison of various dmm's (including the HP3478 which uses the AD536 log RMS converter):
https://www.analog.com/media/en/technical-documentation/application-notes/an83f.pdf
And here's the chart for an article I wrote in AX on Linear Tech's LTC1966, LTC1967 and LTC1968. (Sadly the LT1088 is obsolete!).
Getting back to topic. Appendix C to this Jim Williams apnote has a useful comparison of various dmm's (including the HP3478 which uses the AD536 log RMS converter):
https://www.analog.com/media/en/technical-documentation/application-notes/an83f.pdf
And here's the chart for an article I wrote in AX on Linear Tech's LTC1966, LTC1967 and LTC1968. (Sadly the LT1088 is obsolete!).
Attachments
I still hold onto the HP3478 I bought calibrated years ago. Though, since picking up two HP34401s, the 3478 has seen very little use.
The 300 V max DC is annoying for tube work, but plenty for everything else. Another interesting quirk is that there is only one set of calibration constants for the resistance meter. That means the 2-wire and 4-wire resistance functions use the same calibration constants. If the meter is calibrated for 2-wire use, the readings on the 4-wire setting will be off.
Tom
The 300 V max DC is annoying for tube work, but plenty for everything else. Another interesting quirk is that there is only one set of calibration constants for the resistance meter. That means the 2-wire and 4-wire resistance functions use the same calibration constants. If the meter is calibrated for 2-wire use, the readings on the 4-wire setting will be off.
Tom
Thanks! I’m separately building a Frex RMS-DC board, which is based on the LT1968. Will be fun to compare that to the AD536 in the HP3478A, vs of course the HP3458A benchmark.
Good thought regarding the stackable banana plugs and high voltages. I really don’t work on anything particularly high at home - say 50-70V is as high as I tend to go. That said butchering a pair of shrouded leads by cutting most of the shroud off is probably still a safer option than using plain banana leads.
I’m assuming HP sold these with leads - alas I haven’t turned any up.
I’m assuming HP sold these with leads - alas I haven’t turned any up.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.