-
I promised in the TEA thread to share what I had here... Over on the groups.io HPAK forum I ran into the engineer who was responsible for the analog design of these meters, and... well... pretty soon I found a copy of each and compelled them to join me on the workbench.
One of the interesting things I learned was that even though these two have slightly different specs, the HP engineers considered them pretty much equivalent in terms of long term accuracy. I don't have anything good enough to do proper analysis of them, so that's not something I'm going to worry about. They both present an interesting contrast to the Flukes with their bright LEDs and mechanical front panel switches.
The 3468A, listed as "fair" by the seller, was in fact nearly perfect barring some grime which cleaned up quickly. I don't care about the odd nick and scrape as long as it doesn't impair usability or make the display ugly. The handle was the only thing which had any cosmetic issues anyway.
The 3478A, shown with some "cosmetic" issues, looked as though someone had dropped a boat anchor (figuratively or literally) on it. Nothing was obviously broken, and the dent in the case, though serious, clearly wasn't going to crush anything inside so at the price I took the chance. The only thing I was worried about was a crushed corner of the front bezel which potentially could have damaged the display board.
When I got the 3478A, I tested it immediately and it seemed to work just fine on all modes and ranges. I managed to lever the case off without further mangling anything, and freed the bezel as well (you pretty much have to disassemble the instrument to get the bezel off). The case is beyond my modest metalworking capabilities, so a friend with a shop is going to take a look at it for me. It's not all that bad, but I'd like it squared up and the dent removed so that it will at least slide on/off without too much scraping. As I anticipated, it didn't hit anything inside - there's a lot of free space above the PC board in these units. The inside was very clean.
The bezel: apparently the early units have a metal bezel but mine was the later plastic model. It looked like someone had dropped it (or possibly the boat anchor which landed on it bounced...) but the design prevented any of the damage from even reaching the front panel, let along propagating the force to other parts. It was surprisingly easy to heat the plastic with a paint stripper set on low and gently prod it back into shape. A little tan epoxy and some judicious sanding, and it looks OK. Not perfect, but at a glance it doesn't immediately get your attention.
The 3468A battery is 3.4V, which for an ancient 3V battery is pretty impressive. The 3478A battery is 'only' at 3.2V so it's a little closer to the tipping point. I don't have any way to get data out of the 3468A (in fact, I'm not sure there is a method to talk to an HP-IL meter and retrieve cal data?) but I could get it from the 3478A via HP-IB, I suppose. I'm really not sure I can justify building an interface just to extract a few bytes of data - or pay twice what I spent on the meter for a HPAK clone interface. I may just settle for doing a live swap on the battery.
Overall consensus: I really like these meters. They're fast, quiet, and even the LCD is easy to read in the lighting of my workshop. Unfortunately, now the HP meters are challenging the Flukes for bench space.
-
before playing with them there is a software thread here that will backup the calibration data for the 3478a
https://www.eevblog.com/forum/testgear/free-hp3478a-multimeter-control-program/
you need a prologix gpib adapter or a clone
https://www.eevblog.com/forum/projects/ar488-arduino-based-gpib-adapter/
https://www.eevblog.com/forum/testgear/$5-usb-gpib-adapter-for-ezgpib/ -
Nice job on that bezel, Greg! I was quite worried about the extent of the creasing in the corner. You revived it really well. How soft did you have to get the plastic to successfully work it?
In this case, it's actually beneficial that the bezel was plastic instead of aluminum. The metal probably would've cracked from fatigue. -
@coromonadalix: Thanks, I knew there was a 3478A option. Maybe I should try to build that $5 board and see what I can do with it. For the 3468A it seems there's no option other than using a transfer standard or calibrating the instrument from a traceable reference...
@bitseeker: it's still a bit uneven, but so much better than what I started with that I'll count it a win!
I used a paint stripper set on low power, holding it about 3-4 inches from the plastic and trying to spread the heat a little by moving it around. It took maybe 15 seconds for the plastic to become just soft enough that I could reshape it without melting it. I was very nervous about that so I erred on the low side. After a few trials it became more obvious just how much heat and cooldown it could tolerate. I found that the plastic had developed enough memory after being deformed, that I couldn't just press and leave it; I had to keep a little pressure on it and blow on the part to cool it down more rapidly; then it stayed. Altogether it took about 5 minutes of repeated working cycles before I got the shape I wanted.
The plastic did separate a little (some small crevices opened) which may have been from overheating it a bit or just from fatigue. I happen to have some tan epoxy which matched the color well, and so I spread a little on the surface and then sanded it after it hardened. It enabled me to get a smooth surface and fill those tiny gaps at the same time. If I was really anal-retentive about it I could probably have gotten it perfect but the rest of the bezel is not pristine either, so I moved on to other stuff.
Incidentally, to get the bezel off without breaking it you have to pretty much disassemble the meter - that's a good 5-10 minutes depending on whether you've done it before. There are a number of components mounted to the chassis and connected to the board.
-
Thanks for the additional details on your process, Greg. Good to know for future restorations.
-
Someone asked me about the following, but I can't find the appropriate message to follow up.
So I'm posting the answers here because (1) they're relevant; (2) I can point them over here when I do find the message; and (3) perhaps they'll see it here anyway. These are from the engineer who did the analog design of both meters.
The 3468A and 3478A have different numbers of ranges and a different display for DC mV - the 3468A shows "0.xxxxx V" whereas the 3478A shows "0.xxxxx mV" on the lowest scale. This was purely part of a marketing decision to justify higher cost for the 3478A. Likewise the tighter specs for the 3478A even though the basic accuracy of both meters is the same, and the addition of HP-IB. Apparently a firmware change could add the lower ranges to the 3468A with no loss of performance. We're hoping to hear more on this.
You can only calibrate 2W ohms or 4W ohms and not both; this is because they ran out of cal memory and the budget wouldn't stretch to fitting a larger RAM.
The battery spec is 3.0VDC - any voltage above this is entirely satisfactory and it's not a problem till the voltage drops below 2.7-2.8V. There is some misinformation about the battery voltage being 3.6V and this is not the HP spec. It may have to do with someone finding a substitute battery in their meter, then spreading the info.
Before acquiring the 3490A and these 2 units, I was not familiar with most of the HP meters, especially in the home lab. After using them for a while, I'm really impressed with the level of engineering that went into them - just like most of the HP gear. Must... not... buy... more... -
It wouldn't be so hard except that they're still available at stupidly low prices most of the time. Interestingly, after some discussion with the designer, several people went out and bought the least expensive examples on the Bay, and now prices have risen accordingly. No doubt they're worth the $100 or so, but at $30-40 they are an utter steal. I suspect prices will drop again unless a bunch of EEVBlog members fuel the fire.
-
The only major issue is the surgery required to replace the dying calibration memory's battery. Most people probably aren't inclined to do it, so it's less desirable and prices fall accordingly. That's certainly a win for those who are handy with an isolated-tip soldering iron.
-
I think the lcd is the same as the 3458 main frames
-
You mean 3457A, correct? 3458A is a VFD type.
The 3457A was made a couple of years after the 3468/78 series, so it might well have had the same LCD. According to the designer, the 3488A switch controller used the same display driver as the 3468/78 but a different glass (because of the annunciators). -
All the HP meters snuggled together on the bench.
-
You mean 3457A, correct? 3458A is a VFD type.
The 3457A was made a couple of years after the 3468/78 series, so it might well have had the same LCD. According to the designer, the 3488A switch controller used the same display driver as the 3468/78 but a different glass (because of the annunciators).
It is not only the display is similar to the 3457. AFAIR the user interface and the programming and even some HW (hybrid for input switching) is similar / same. A nice thing about the 3478 is that it has a 30 mV range. -
there is some hp mainframes who seems to have the same lcd display, the 3488a seems the same module, haven't seen the service manual
I've seen them as low as 35$ usd
Even thoses at 24$
https://www.ebay.com/itm/QTY-1-HP-Agilent-3488A-Data-Acquisition-Control-Mainframe-Unit-HPIB-no-modules/291374750766?epid=1500240266&hash=item43d749dc2e:g:dSQAAOSwMFFbwfvZ:rk:179:pf:0 -
All the HP meters snuggled together on the bench.
That looks like a happy family. I guess you need some VFD versions for your collection.
-
Yes the 3488A is very inexpensive these days; HP sold thousands. But while the driver will work and the numbers will display correctly, the display annunciators don't match the 3468A/3478A and so it's not a 1-to-1 swap.
-
Damn sorry did not check it carefully nice pair o meters, glad for you