|
So the dirty "poor" looking 3468A finally arrived, and it really is not in bad shape at all. Definitively a winner so far.
Nothing loose or rattling inside. Handle broken, only the two side arms remain.
Preliminary test, plugged AC in and power up: self-test OK.
Made a couple of DC volts and 2W ohms measurements, and all readings seem to be quite accurate compared to one of my 974A. An AC volts reading of the 120V power line was off by about 0.5v compared to the 974A reading. Will do some more comparison readings to with the 3478A later this evening or tomorrow.
Lithium battery (black "Eternacell", looks original) prob very marginal at 3.04V. Already placed replacement order at Batteryspecialists.com, $9.95 plus 3.95 shipping.
Main board looks pristine, very clean no dust, corrosion, etc, and nothing seems to have been tampered with or replaced.
Upon pulling main board from lower case shell, turns out the two rear plastic turrets are clean broken off from the cover, and two of the long screws with the small o-ring that hold the case halves together are missing. Turrets should be an easy fix with some MAX1 glue.
LCD front window has a small hairline crack, but LCD screen itself all segments working, although there is? a menacing dark shadow that seems to be growing inwards, specially from the L/R sides. So far so good. Pins of the ribbon cable connector that plugs into the LCD board seem to have a bit of white oxidation on them, but otherwise ok. Seems like humidity got in through the crack in the window? Mating gold plated holes on LCD board seem intact.
Managed to clean up both case halves, all the nasty stickers just peeled off quite easily, only one left a tiny bit of sticky residue. Washed covers with dish soap and soft brush, both came out really nice. Sharpie hand writing on top cover came out quite well with 91% IPO and toothbrush & rag. Only slight ghost shadow of writing remains but you really have to know what you are looking for to see it.
Gallery with follow up pictures up hopefully tonight.
Anyone have an unbroken handle for this model? Seen one on ebay $35 new, but wanted to try here first. Does not have to be new.
A spare LCD would definitively also give me more future-proof peace of mind.
Any comments and suggestions much appreciated. Thanks.
Alex
|
eBay sellers seem to understate the condition of their items to prevent customers complaining, in my experience.? My 3468A was listed as 'fair' but was really in very good condition inside and out.? I've bought 'not working' Fluke gear that was fully working, or functional but needing simple repairs.? Many sellers actually don't know how to test electronics equipment. I have seen front panels for sale but they tend to cost as much as the meter itself, unfortunately.? Just like the handles.? :-(? You might check Test-Equipment-For-Sale-Wanted-or-Exchangeto see if anyone on this site can help...
|
|
|
On Mon, Feb 25, 2019 at 10:19 AM, <gregdunn@...> wrote:
I have seen front panels for sale but they tend to cost as much as the meter itself, unfortunately.? Just like the handles.? :-(? You might check
Test-Equipment-For-Sale-Wanted-or-Exchange
to see if anyone on this site can help...
Thanks for the tip, yes I will send a request in that group as well for the needed/wanted 3468A parts. There is one 3468A front panel on ebay which I was going to buy due to the ominous dark shades my LCD came with, but held off upon suggestion from George to wait and confirm the actual condition once it arrives. It turns out so far it will not need an immediate emergency LCD transplant, this one still works fine, but for how long would be the question. The seller was asking $29 for the whole front panel, and taking best offers. Since I really only wanted the LCD module I offered $20 but he wanted to hustle me back to $25 so it was a no go at least until I knew for sure that I would need it. Now turns out the guy even raised his price to $39. Go figure.
|
Alex,
I can help you and others with these matters but not until later in the year.? ?Most of my inventory for the 3468/3478 is back in Colorado and I have moved to Minnesota.? I will be making it back to Colorado for the parts but not until later in the year? -? possibly fall.
PS? contrary to much of the discussion I have read in these threads, the cal constant backup battery (3V) is not marginal when measuring 3.02v when loaded and drawning current in the 3468. The original lithium batteries are MARKED 3.0V and generally around 3.2v new and I would not begin to get concerned until the voltage drops to 2.7 - 2.8v? when measured loaded in the 3468A.? Many of the batteries are? SAFT? Lithium 3.0? ? p/n? ? LX 1634? ( these are the? "black ones used in later units ).? ?I do? not remember the batteries that were shipped when the instrument was introduced other than they were "white" with black printing -? maybe Panasonic.
?
George
|
My early unit has an Eternacell SDX type G32/13 from Power Conversion, Inc. 3.0v on the nameplate. I found an old listing on AliExpress that says these were 1400mAh cells. The chemistry is lithium thionyl chloride.?
Power Conversion was acquired by Hawker, and at some point, Hawker sold its lithium battery business to Saft.?
To replace mine, I ordered a Panasonic BR-2/3AGE5SP, rated at 1450mAh. The E5SP designates single pin tabs. The tabs have a 34mm spacing, which looks about right for the 3468A board.?
toggle quoted message
Show quoted text
On Feb 26, 2019, at 10:18, " ghnatiuk@..." < ghnatiuk@...> wrote: Alex,
I can help you and others with these matters but not until later in the year.? ?Most of my inventory for the 3468/3478 is back in Colorado and I have moved to Minnesota.? I will be making it back to Colorado for the parts but not until later in the year? -? possibly fall.
PS? contrary to much of the discussion I have read in these threads, the cal constant backup battery (3V) is not marginal when measuring 3.02v when loaded and drawning current in the 3468. The original lithium batteries are MARKED 3.0V and generally around 3.2v new and I would not begin to get concerned until the voltage drops to 2.7 - 2.8v? when measured loaded in the 3468A.? Many of the batteries are? SAFT? Lithium 3.0? ? p/n? ? LX 1634? ( these are the? "black ones used in later units ).? ?I do? not remember the batteries that were shipped when the instrument was introduced other than they were "white" with black printing -? maybe Panasonic.
?
George
|
Glen,
You message jogged my memory.? The units were originally shipped with the Eternacell battery from Power Conversion inc. -- thanks for that.
Again, the battery is? ? NOT? critical.? 3.0v nominal is all that is needed.? Nothing special.? Of course you would like a battery with long life.? ?I do have a large supply of the SAFT batteries but they have been sitting around for over 30 years -- would have to look into the "self life" specs before I would give them to anyone for their 3468A.
George
|
Hi George,
I thought it would just make sense to preemptively replace the old original backup battery, as replacing it now would eliminate the burden of having to check up on the voltage in the months/years to come, not to mention some added peace of mind. Just my thinking. The battery that came with this unit SN 2137A04349 is a black "ETERNACELL", exact same as the one that came with the 3478A. I ordered a Panasonic replacement which will of course be bright yellow. BTW if you happen to know, how does one determine the year of manufacture of a 3468 based on the SN of the equipment? I assume the first 4 digits are the year and week, so if it is in any way similar to the SN of the classic calculators, the "21" needs to be added to some other figure to come up with the actual year. Most IC's on my board seem to have a 82-83 date code.
Any thoughts or suggestions of the corrosion on the LCD board connector pins?
On your generous parts offer, first of all many thanks, and I am certainly in no hurry, so unless something else comes along, waiting for your trip back to CO is more than ok with me.
Oh and BTW I always had a question about the ohms cal procedure of both the '78 and '68. Something I always wanted to ask since I first came across the verbatim in the manual several years ago when I got the 3478A. It states that ohms can only be calibrated in one of the modes (either 2 wire or 4 wire) but Not in BOTH? I am sure my interpretation its probably not entirely correct (it sounds so "un-HP") that the manual seems to suggest that for example one could never get accurate 4W measurements if the instrument was calibrated in 2W mode, or the other way around. But the added comment in the manual that states "Make sure the Multimeter is calibrated? in the [ohms] function in which it will be used" seems only make a stronger case for my interpretation.? Why can it not be calibrated in both ohms modes? Perhaps something to do with the initial 100-parts design limit? If anyone could shed some light on this issue, its certainly would be you!
Thanks.
Alex
|
Mentioned in the 3478A thread but may have gotten buried:? Why does the firmware on the 3468A and 3478A behave differently on the low range VDC display?? A 1 mV input would show "1 mVDC" on the 3478A but "0.001 VDC" on the 3468A.? Was that just a firmware rev which didn't get backdated, or a design choice?
|
Alex,
Yes, that is the way Hp did serial numbers back in the 70s and 80s but do not know if Agilent kept the tradition.
The 4 X 5 serial number of instruments had the following scheme:
? ? ?1234A67890
Year of manufacture =? add 60 to first two digits? ==>? 12 + 60? =? 1972
week of the year of manufacture is the next two digits? =? week 34 of 1972
Letter is country of manufacture:? A = America? ?J = Japan? ?M = Malaysia? etc.
Again you are correct about the Ohms calibration:? 2 Wire or 4 wire but not both.? ?
This was a tough decision to make because of a gotcha.? The gotcha was cost and room for more parts.? We simply ran out of room in storing the firmware and cal constants for? BOTH? ohms functions.? To store more cal constants kicked us up into another chip for memory and COST was a big deal on this project.? Memory was not cheap back then. Besides, there was not enough room on the 3468A motherboard (just one PCB was allowed) for another memory chip without squeezing other areas and compromising analog performance.? I did look at that but just could not make it fly.? There are strict requirements regarding spacings with AC mains and any voltages greater than 42v on a motherboard that also reached out to the outside world (thru the input terminals) that a human could touch.? These spacing requirements eat up a lot of realestate on the MB.? Just was not possible unfortunately.??
Another thing to keep in mind is that the 3478 has an extra range on DCV and Ohms that the 3468A does not.? The 3478A has a 30mV and 30mOhm range that the 3468A does not have.? The 3468A has the capability to implement these ranges ( the switches are there for appropriate signal conditioning and amplifier gains) but the firmware does not allow it.? This was a marketing decision to further differentiate the 3468A and the 3478A from one another besides the 78 being a systems meter with F/R switch and HP-IB interface to justify the cost difference between them.? With appropriate firmware the 3468A could have those lower ranges.? I might have a copy of some special firmware I used that turned on those functions to do some testing on the 3468A but that might take some looking.
George
|
Close - I used to think that, too, but further digging and experience shows that the the prefix is actually the major engineering revision the unit is built to - the actual production date could be years later if there were no major changes.? The number does at least provide the EARLIEST date the instrument could have been manufactured, though.? For a more accurate idea of its 'birthdate', I also look at the date codes on the internal components.? For instance, I have an 8644B synthesized signal generator with a prefix of 3546 (46th week of 1995) - it is an Agilent marked unit, and the date codes inside show it to be late 2000 production - devices have date codes such as 0021, 0006, 0030, 0033, etc.
And stuff from the 60s will have a prefix with three digits rather than four; the second two are still the week and the single digit in those instances is added to 1960 to get the year.?
-Pat
On Tue, Feb 26, 2019 at 02:39 PM, ghnatiuk@... wrote: Alex,
Yes, that is the way Hp did serial numbers back in the 70s and 80s but do not know if Agilent kept the tradition.
The 4 X 5 serial number of instruments had the following scheme:
? ? ?1234A67890
Year of manufacture =? add 60 to first two digits? ==>? 12 + 60? =? 1972
week of the year of manufacture is the next two digits? =? week 34 of 1972
Letter is country of manufacture:? A = America? ?J = Japan? ?M = Malaysia? etc.
Again you are correct about the Ohms calibration:? 2 Wire or 4 wire but not both.? ?
This was a tough decision to make because of a gotcha.? The gotcha was cost and room for more parts.? We simply ran out of room in storing the firmware and cal constants for? BOTH? ohms functions.? To store more cal constants kicked us up into another chip for memory and COST was a big deal on this project.? Memory was not cheap back then. Besides, there was not enough room on the 3468A motherboard (just one PCB was allowed) for another memory chip without squeezing other areas and compromising analog performance.? I did look at that but just could not make it fly.? There are strict requirements regarding spacings with AC mains and any voltages greater than 42v on a motherboard that also reached out to the outside world (thru the input terminals) that a human could touch.? These spacing requirements eat up a lot of realestate on the MB.? Just was not possible unfortunately.??
Another thing to keep in mind is that the 3478 has an extra range on DCV and Ohms that the 3468A does not.? The 3478A has a 30mV and 30mOhm range that the 3468A does not have.? The 3468A has the capability to implement these ranges ( the switches are there for appropriate signal conditioning and amplifier gains) but the firmware does not allow it.? This was a marketing decision to further differentiate the 3468A and the 3478A from one another besides the 78 being a systems meter with F/R switch and HP-IB interface to justify the cost difference between them.? With appropriate firmware the 3468A could have those lower ranges.? I might have a copy of some special firmware I used that turned on those functions to do some testing on the 3468A but that might take some looking.
George
|
On Tue, Feb 26, 2019 at 11:39 AM, ghnatiuk@... wrote:
Again you are correct about the Ohms calibration:? 2 Wire or 4 wire but not both.? ?
This was a tough decision to make because of a gotcha.? The gotcha was cost and room for more parts.? We simply ran out of room in storing the firmware and cal constants for? BOTH? ohms functions.? To store more cal constants kicked us up into another chip for memory and COST was a big deal on this project.? Memory was not cheap back then. Besides, there was not enough room on the 3468A motherboard (just one PCB was allowed) for another memory chip without squeezing other areas and compromising analog performance.? I did look at that but just could not make it fly.? There are strict requirements regarding spacings with AC mains and any voltages greater than 42v on a motherboard that also reached out to the outside world (thru the input terminals) that a human could touch.? These spacing requirements eat up a lot of realestate on the MB.? Just was not possible unfortunately.??
George
George thank you again for another nice explanation right from the original source, its so nice to have you around here, and hope you don't mind all the inquiries and we are not taking up too much of your valuable time. Speaking of, just for curiosity's sake, this brings up just another interesting question: back then when you where going to buy a new 78 or 68, I assume you had to specify beforehand by filling out a request form in what ohms function you wanted to use your instrument the most so it would be custom calibrated? I guess these DMMs where mostly only ordered direct by companies and organizations for internal use, and it was not an instrument that was not so much intended to be available "off the shelf" to individuals. But in any case if anyone wanted to place an order, they had to answer the basic question: 2W or 4W. Wow talk about a tough decision, you order 2W and still you have those two nice additional jacks on your DMM that are not doing much good (or maybe just order one of each!). But in reality, I'm curious, just as a hobby user perspective, how far off would a 2W measurement if the instrument was properly calibrated for 4W? Or the other way around. I am sure this is something that HP or you guys the designers must have spec-ed very well, at least for internal use. Thanks again! Alex
|
Alex,
No problem answering your questions.? It has been a mental exercise for me to reconstruct and remember the design evolution of the "Key" project -- I have been enjoying it.? Actually, I have decided to make a Youtube video or two about designing a laboratory grade DMM and should have that finished by the end of the week.? It will cover the overall design philosophy we used at Hp in making a modern multi-function digital meter and get into the specific analog designs I developed in the 3468/78 to implement each function and what each component part's function is.? That will help all of you understand the instrument should you need to troubleshoot or just want to play around with the circuits.? If you have other questions that might take more time to respond, post them in this thread and I will answer in the video; like why AC Volts does not read zero when the inputs are shorted.
The meters at time of final test/cal were calibrated in 4-wire ohms before being shipped to a customer.? There was no "special" accommodation for a customer to request a 2-wire cal. for there was no way to know beforehand what the customer would be using for test leads.? Such a cal would be test lead specific.?
After a 4-wire cal, the 2-wire accuracy will basically be limited by the resistance of the test leads and only affect the lowest two ranges given that the ohmic resistance of most test leads will be under 0.100 Ohm.? The test leads shipped with the instrument was speced at 100 milli Ohms max. and typically at 95 milliOhm.? ?With 95 milliOhms in the test leads, the meter will read 000.095 on the 300 Ohm range or 95 counts of error out of 300,000 full scale?when shorted together and 9 counts of error on the 3K range -->? ?0.00009 and one count of error on the 30K range -->? ?00.0001? and no error on the higher ranges.? You would need to mentally subtract this offest from subsequent readings in 2-wire Ohms if you wanted best accuracy.??
Unfortunately, we could not implement a NULL function as bad as we wanted to due to limited memory space in RAM for all the firmware code.? You can cheat the instrument to create a NULL offset caused by the test lead resistance by putting the instrument in CALIBRATION mode and perform a ZERO cal in 2-wire ohms without affecting the full scale gain cal.? Then if you want to go back to 4-wire, re-cal the ZERO in 4-wire Ohms.? This is a pain but the only way to get the instrument to null out the offset in 2-wire Ohms if you do not want to mentally subtract the offset.
The best way to handle this is of course is to make a set of Kelvin clip test leads for ohms.? What I do so I do not have 4 clips to attach to a resistance under test, which can be a pain if not impossible when such is a surface mount R etc. is to attach a set of sense lines to the alligator clips or probes at the end of the test leads.? This will give you readings almost as good as a true 4-wire measurement.? The only difference would be due to any contact resistance that may present itself after the point where the sense lines are attached to your main leads.? ?I will make a point to discuss this in the video.
You have good questions that others may also have wondered about.
George
|
Readers,? I have been monitoring only this thread and Greg's thread:
HP 3468A - Working, so far
Any questions directed to me concerning the 3468 and 3478 or other Hp voltmeter products should be posted in these two threads otherwise I may not see them.
Thanks,
George Hnatiuk
|
On 2/26/19 11:03 PM, ghnatiuk@... wrote: Readers,? I have been monitoring only this thread and Greg's thread:
HP 3468A - Working, so far
Any questions directed to me concerning the 3468 and 3478 or
other Hp voltmeter products should be posted in these two
threads otherwise I may not see them. [co-moderator here] A great many of our members use this as a mailing list, not a web forum, FYI. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
Excellent, George!? I look forward to seeing the video.
For those who are interested, it's possible to find some really decent inexpensive Kelvin leads on the Bay.? The ones I have are made from quality co-ax with guard shield and metal/plastic clips which seem sturdy.? They work well with all my HP meters (the plugs are plenty long enough) and my Fluke 8800A.? On the 3468A they read a nice 0.003¦¸ when shorted.
|
Greg,
I have personally examined those Kelvin clips from China and forgot about that source.? You are correct, they really are not too bad performance wise and an excellent buy at the price.? You can even clip then on the end of a SMT resistor.? I would recommend going with that product rather than trying to make your own.? Thanks for point that product out.
George
|
I was quite impressed for the price.? I actually bought an extra set and updated my DER EE DE-5000 LCR meter to have a proper set of Kelvin leads.? Another set stays attached to one of the HP meters.
|
--->Greg: thanks for the tip about the Kelvin cables, just ordered one set. They definitively look quite good, specially for the price.
--->George: yes of course that would make a lot of sense to calibrate the DMM at large for 4W mode. So understanding your very clear explanation, maybe the manual should rather have stated that the best option would be use the 4W cal, and -only- if really maximum accuracy is required on 2W, should that type of cal be favored instead. In any case since I now have one of each (68 and 78) I guess I could cal one for 2W and the other for 4W. Or use the clever zero null procedure as needed. Very interesting.
Come to think of it, I'm going to need a good set of regular test leads as well. Could anyone recommend a decent HP set with exposed banana plugs? It appears PN 11003A might be something to look at, but right now none available on ebay.
Alex
|
My $75 3468A arrived today. ?Looks pristine. ? Opening it up and it was very clean. ? Has a black Saft battery that measured slightly over 3V on my non HP meter. ?
I need to play with it more. ?One thing I noticed was on DC and AC volts with no meter leads in it did not read zero. ?I think it was in the 400 mV. ?Seems to zero nicely with shorted test leads on Ohms. ?
Will test more but looks nice. ?
toggle quoted message
Show quoted text
On Feb 27, 2019, at 5:26 PM, Alex < hpagilentgroup@...> wrote: --->Greg: thanks for the tip about the Kelvin cables, just ordered one set. They definitively look quite good, specially for the price.
--->George: yes of course that would make a lot of sense to calibrate the DMM at large for 4W mode. So understanding your very clear explanation, maybe the manual should rather have stated that the best option would be use the 4W cal, and -only- if really maximum accuracy is required on 2W, should that type of cal be favored instead. In any case since I now have one of each (68 and 78) I guess I could cal one for 2W and the other for 4W. Or use the clever zero null procedure as needed. Very interesting.
Come to think of it, I'm going to need a good set of regular test leads as well. Could anyone recommend a decent HP set with exposed banana plugs? It appears PN 11003A might be something to look at, but right now none available on ebay.
Alex
|
Cubdriver
Dave McGuire