Vertical Mounting for HP3456A - any problems? [SOLVED]

[enut11]

I have limited space on my workbench so I am thinking of taking the radical approach of vertically mounting my HP3456A on my left with LED display at the bottom, terminals at the top. Does anyone know of any problems, say with heat at one end or other, with instrument accuracy in this odd position?

[ManateeMafia]

I have limited space so I picked up a tool cart that doubles as a storage area on the bottom shelf. You can wheel it where you need it . I currently have a Fluke 5100 calibrator and Fluke 5220 stacked on top. Sometimes I add a 3458A on top and it can be easily repositioned.

[enut11]

Thanks. I will also have a look at carts.

[Andreas]

Hello,

you can simply try this. (with a stable voltage reference).

But I wouldn´t do this at least for all LM399 based instruments.
They all have more or less (dependant on individual LM399)
change of output with orientation.

With best regards

Andreas

[enut11]

Thanks Andreas. Over what period of time did the LSD change by 6 units when moving from horizontal to vertical?
enut11

[TiN]

Andreas
I wouldn't be so sure that change cause is LM399 itself. More likely it's change of thermal profile over the board from different orientation.

[Andreas]

Thanks Andreas. Over what period of time did the LSD change by 6 units when moving from horizontal to vertical?
enut11

Hello,

The change is "immediately".
But it needs about 10 minutes to stabilize in the new orientation.

Attached a measurement (different LTZ1000A with 1 minute averaged values).
Orientation:
0 = normal
1 = left side
2 = right side
3 = on top
4 = on back
5 = on front

In HP34401A the LM399 is oriented differently on PCB against K2000
so the maximum effect on HP34401A is on back and front side.

2nd attachment raw (non averaged) values of measurement (first 2000 values with 10 NPLC).

Andreas
I wouldn't be so sure that change cause is LM399 itself. More likely it's change of thermal profile over the board from different orientation.

I have that effect also with bare LM399s (connected with 6K8 pull up resistor to 14V UBat).
So my guess is that the largest amount of the change is through the LM399 itself.
The quick response speaks for a low thermal mass involved. (heater on chip vs. direction of temperature sensor).
And it gets better when you thermally isolate the LM399 from environment.

Of course there are large differences from sample to sample.
I have also seen LM399s with changes of only 1-2 ppm. (CH 6 measurement)

With best regards

Andreas

[enut11]

OK. Looks like I need to locate the HP3456A back to horizontal position. Thanks for all the feedback.
enut11

[e61_phil]

I've done this experiment with one of my 3456A.

The deviation I observed on my unit is much smaller as measured by Andreas.

7,18580 horizontal
7,18579 with display to the bottom

And the meter is specified with 2ppm/hour (transfer) and 8ppm/day.

Take a stable source and give it a try with your meter.

Even with more deviation it shouldn't matter if you use the meter in a fixed position.

[Andreas]

Hello,

According to the manual there are different 3456A references in the wild.
Some with temperature compensated zener and heating resistor, and some with LM399s.
Which one do you have?

The four other possible orientations would be also interesting.

And I hope that the calibration is done always in "normal" position.

With best regards

Andreas

[e61_phil]

It looks like a LM399, but it has a HP part number on it.

I think calibration isn't affected. Even your K2000 example is only -8ppm and +6ppm. This is within the 8ppm/24h specification. And everybody who is calibrating for 24h should know the position of the meter. For normal 1year calibration (23+2ppm) it doesn't really matter. But it is very interesting

[TiN]

I had 2001's (they use selected LM399 as well) running last few days, so I bothered to turn few in positions to give it a whirl 

Sorry, can't be bothered to capture screenshot, here's data live log: click-click, SVG.

First section on left - all meters in normal position.
Then meter 13 rotated 90 CCW (vent side facing bottom) and put on ~10cm spacer (not to block vent holes). ~ -4ppm shift
Same time meter 15 rotated 90 CW (vent side facing top). -3ppm shift

Then meter 15 was reverted to normal position, while meter 13 rotated another 90CCW (making it sit on spacer in dead bug position).

Room temperature : dashed brown line.
Reference value : 2xK2002(reds)+3458(green samples)