Bench "Standard" Check Up

[MED6753]

The hp 3456A is my bench voltage/resistance bench "standard" so it's time to check it's accuracy against some known references. First up is DC volts. Primary reference is an AD584-M that was checked at the factory with an Agilent 34401A at 21 C ambient. Both this reference and the 3456A were powered up for 6 hours at 22 C ambient. Results as follows:

AD584-M.............3456A

2.50066V.............2.50061V
4.99794V.............4.99791V
7.50011V.............7.50010V
9.99691V.............9.99690V

Next up is the secondary AD584-M reference. That was checked at the factory with an Agilent 34401A at 25 C ambient. The 3456A was powered up for 9 hours and the AD584-M powered up for 3 hours. Results as follows:

AD584-M.............3456A
2.49937V.............2.49933V
5.00030V.............5.00026V
7.50042V.............7.50039V
10.00096V...........10.00094V

Based upon these results the 3456A has experienced basically zero drift on DCV since the last time it was checked.




I don't have a high accuracy ACV reference other than a 1.00VAC which I wouldn't trust beyond 2 decimal places. When applied to the 3456A it reads 1.018286V. Good enough for my needs.

Next up is resistance. Leeds and Northrup 4747. This resistance box is showing some wear and tear so I had to do a lot of cleaning and futzing with it. The 2 binding posts are corroded and initially had a contact resistance of about 0.0450 ohms. With cleaning I got it down to 0.0148 ohms. But future I'll be replacing those binding posts with high quality pieces. Also on the 100 and 1000 ohm ranges some additional contact resistance is present. The switch decks show considerable wear and corrosion that deoxit won't touch. I'm considering using some metal polish. Future project. The 3456A had been powered up for nearly 12 hours. So the results are as follows:

4747...............3456A

1.0.................1.0159
10.0...............10.0191
100.0.............100.0629
1000.0............1000.317

The 1 and 10 ohm are OK while the 100 and 1000 are a little suspect but I think it's somewhat safe to conclude that the 3456A is OK.



Tomorrow will check my 10MHz "standard".

[Specmaster]

Next up is resistance. Leeds and Northrup 4747. This resistance box is showing some wear and tear so I had to do a lot of cleaning and futzing with it. The 2 binding posts are corroded and initially had a contact resistance of about 0.0450 ohms. With cleaning I got it down to 0.0148 ohms. But future I'll be replacing those binding posts with high quality pieces. Also on the 100 and 1000 ohm ranges some additional contact resistance is present. The switch decks show considerable wear and corrosion that deoxit won't touch. I'm considering using some metal polish. Future project. The 3456A had been powered up for nearly 12 hours. So the results are as follows:

4747...............3456A

1.0.................1.0159
10.0...............10.0191
100.0.............100.0629
1000.0............1000.317

The 1 and 10 ohm are OK while the 100 and 1000 are a little suspect but I think it's somewhat safe to conclude that the 3456A is OK.



Tomorrow will check my 10MHz "standard".

I think you need to double-check your calculations, I would say that the most accurate resistances you have are the 1000 and 100 as both are giving you readings of 99.96831% and 99.9371395% respectively while the 10 and 1 are providing you with 99.8093641% and 98.4348853% respectively.

[MED6753]

I think you need to double-check your calculations, I would say that the most accurate resistances you have are the 1000 and 100 as both are giving you readings of 99.96831% and 99.9371395% respectively while the 10 and 1 are providing you with 99.8093641% and 98.4348853% respectively.

You're failing to consider the binding post contact resistance of approx 0.0148 ohms which accounts for the apparent greater error on the 1 ohm and 10 ohm. If you subtract the binding post contact resistance you get 1.0011 ohm and 10.0043 ohms.

[Specmaster]

I think you need to double-check your calculations, I would say that the most accurate resistances you have are the 1000 and 100 as both are giving you readings of 99.96831% and 99.9371395% respectively while the 10 and 1 are providing you with 99.8093641% and 98.4348853% respectively.

You're failing to consider the binding post contact resistance of approx 0.0148 ohms which accounts for the apparent greater error on the 1 ohm and 10 ohm. If you subtract the binding post contact resistance you get 1.0011 ohm and 10.0043 ohms.

Oops That would make a difference, either way the results are still pretty good.

[MED6753]

Continuing with "Bench standards". 10MHz reference. Mine is a stand alone unit built up from a reference given to me by nixiefreqq back in 2019. I enclosed it in it's own case with power supply. It's only powered up when I need it so obviously it has to be on for several hours for best accuracy.




After several hours of up time on both the reference and monitoring frequency counters this is the result:




After running overnight the Star Trek Shuttle Fluke 7260A was showing +2 to +3Hz high. So time to break out the big guns and see what it says.



So after about 6 hours up time on the 5248L every other sample was either nominal or +1Hz.




Conclusion: This 10MHz reference is adequate for my needs and worst case is perhaps +3Hz high although the 7260A might be at fault here and need an adjustment. But I'm not going to sweat it and leave it as is.

[nixiefreqq]

for those youngsters new to direct counters (not you med):


"The value of the LSD Displayed often is the same as the quantization
error which represents the smallest non-zero change which may be
observed on the display. Because of this, resolution and accuracy
statements often specify quantization error as ± LSD Displayed."

"Although we say ±1 count, we do not mean that a particular measurement
can vary by both + and – one count. The measurement can vary
by one count at most. The reason that you have to say ±1 count is that
from a single measurement, you don’t know a priori which way the next
measurement will jump or if it will jump at all. So the specification has
to include both possibilities."

consider the 10000001 displayed to be a feature, not a fault......i still love direct count nixie counters and view reciprocal counters as cheating.

the 5248 is a thing of beauty until the tunnel diodes in the trigger go tits up. hmmmmm. i may have some trigger boards in my massive stash of 5245/48 parts.

[MED6753]

the 5248 is a thing of beauty until the tunnel diodes in the trigger go tits up. hmmmmm. i may have some trigger boards in my massive stash of 5245/48 parts.

Oh wonderful. So hp gear uses unobtainium tunnel diodes like older Tek gear? Didn't know that. Knock on wood (taps side of head) I haven't (yet) had one go bad.

[Zenith]

Having looked into it slightly, I don't get the impression that very many tunnel diodes are exactly made of unobtainium. There's some NOS as well as pulls. There's new production from AmericanMicrosemiconductor and others. The price? Well don't look if you have any history of a heart condition.

Then there are the Russkie ones still available on ebay from Ukraine and places such as Bulgaria. They seem reasonably priced apart from the postage and import duty. I've no idea what the Western equivalents are. I suppose someone has looked into it and there's a web page or two on them. I'm slightly tempted to get some to make a tunnel diode pulser.

[nixiefreqq]

the 5248 is a thing of beauty until the tunnel diodes in the trigger go tits up. hmmmmm. i may have some trigger boards in my massive stash of 5245/48 parts.

Oh wonderful. So hp gear uses unobtainium tunnel diodes like older Tek gear? Didn't know that. Knock on wood (taps side of head) I haven't (yet) had one go bad.

maybe it is a possibility.......maybe not.

thought about it some more and i seem to remember that only the early 5248's used tunnel diodes on the A17 and A21 boards.

but both of mine have dip ecl chips on those boards and the parts list does not show any tunnel diodes. but some parts lists definitely show tunnel diodes.

no idea what is in your 5248.

[MED6753]

but both of mine have dip ecl chips on those boards and the parts list does not show any tunnel diodes. but some parts lists definitely show tunnel diodes.

no idea what is in your 5248.

I have no idea either and I really don't want to go looking for trouble. It's working, leave it alone. But I do wish there was an easy solution for that god awful noisy fan.

[nixiefreqq]

but both of mine have dip ecl chips on those boards and the parts list does not show any tunnel diodes. but some parts lists definitely show tunnel diodes.

no idea what is in your 5248.

I have no idea either and I really don't want to go looking for trouble. It's working, leave it alone. But I do wish there was an easy solution for that god awful noisy fan.

no doubt you already tried all the obvious fixes for the fan. (a little 3 in one oil, or just clean the dust off). in my experience sometimes the blades have gotten bent out of alignment and a little careful tweaking gets them back in balance.

cub might have a fix for the noisy fan problem. (but he will probably just suggest that you get a cat).

[mnementh]

Naaahhh... he just needs a redhead to take his mind off those parts per million.

Red-ette.png

mnem