Keysight's new 34465A (6.5 digit) and 34470A (7.5 digit) bench multimeters

[dmm2018]

Hello...

Received my 34470A one month plus ago but only started to use it 1~2 week ago.

I have just done a test. A Fluke 884X Short was installed on the front panel terminals. The meter has been powered up nearly 90 mins before the statistical data recording (image below) was initiated.

The calibration menu shows that the meter was calibrated in July 2018.

Is the reading is OK? Since "Auto Zero" is ON, I thought the reading should be (very close) to 000.00000.

Note: When the test was carried out, the (4m x 4m) room's aircond temperature setting was set to 18 degC, while the temperature shown on the calibration menu (besides Uptime) is around 24.1 degC.

[TiN]

To get best accuracy you should use meter with same ambient temperature as what used during calibration (typically +23 °C).
I think your meter is fine, but to know for sure you need much more than just a short.

[HighVoltage]

My 34470A shows almost 0 for this quick test
Here is a picture of a comparison on my instrument.

You should test again at about the same temperature as per calibration certificate

[dmm2018]

To get best accuracy you should use meter with same ambient temperature as what used during calibration (typically +23 °C).
I think your meter is fine, but to know for sure you need much more than just a short.

My 34470A shows almost 0 for this quick test
Here is a picture of a comparison on my instrument.

You should test again at about the same temperature as per calibration certificate

Thanks TiN and HighVoltage.

I do not have a thermometer to measure the room/ambient temperature now. I will get one soon. By the way, what is the temperature (the one keeps changing 24.0~24.2 degC) shown on the calibration menu of the meter? Is it the internal temperature of the meter?

[HKJ]

With any warmup time or ACAL:

[TiN]

I would expect it to be just constant value stored in meter (typed in by calibration procedure step?) during the calibration, for user's convenience.
Sure manual should cover this detail with correct explanation. I don't have / never played with 34461A/65A/70A to know shenanigans related to these meters, sorry.

[dmm2018]

I would expect it to be just constant value stored in meter (typed in by calibration procedure step?) during the calibration, for user's convenience.
Sure manual should cover this detail with correct explanation. I don't have / never played with 34461A/65A/70A to know shenanigans related to these meters, sorry.

I switched off the room's aircond and after nearly 10 mins the temperature shown on calibration menu changed from 24.2 to 25.2 degC. The new statistical data (after switching off aircond) is shown below.

[dmm2018]

The temperature (shown on the calibration menu) which I referred to in my previous posts was taken from the menu in image below (copied from manual pp. 148). The temperature is not fixed but changing. After switching off the room aircond for one hour plus, the temperature is 27 degC now. The latest statistical data recorded in given below, now it's approaching 000.00003...

[Dr. Frank]

Hello...

Received my 34470A one month plus ago but only started to use it 1~2 week ago.

I have just done a test. A Fluke 884X Short was installed on the front panel terminals. The meter has been powered up nearly 90 mins before the statistical data recording (image below) was initiated.

The calibration menu shows that the meter was calibrated in July 2018.

Is the reading is OK? Since "Auto Zero" is ON, I thought the reading should be (very close) to 000.00000.

Note: When the test was carried out, the (4m x 4m) room's aircond temperature setting was set to 18 degC, while the temperature shown on the calibration menu (besides Uptime) is around 24.1 degC.

Hello and Welcome to the forum!

At first, this deep cooling in buildings in Malaysia and also in Singapore is always a torture for non - residents (like me), but also for analogue precision instruments and metrology processes. 

The calibration process is done at about 23°C, even @ Keysight in Malaysia, where your instrument might have been calibrated (see your cal certificate).
Highest uncertainty is achieved only at this calibration temperature. Otherwise, correction calculations using T.C.s apply.

If you have a look into the manual, chapter about calibration, first thing calibrated are the offsets of all modes, especially DCV, DCI, Ohm.
They use a low e.m.f., low ohm short to mitigate thermal voltages at the input jacks, but this calibration involves all (solder) junctions inside the instrument as well. See also this lengthy thread somewhere here, about DIY shorts.

e.m.f. voltages of several µV can occur all the time, even at 'normal' lab temperatures, e.g. by using lab grade cables, but that can be mitigated by using the NULL function, before making the real measurement. That does not affect the calibration of the instrument, but increases the precision of the measurements.

The gain constants (Full Scale) are also calibrated at ~23°C, and might out of specification, if you use the instrument @ these usual Malaysian temperatures.
Even the '470A has a T.C. specification including its internal LTZ1000A reference, of about 2ppm/°C...
18°C room temperature is just at the edge of the usual lab temperature, maybe additional 10 ppm gain error.

The 34465/470 have an ACAL function, which reduces this gain error at other temperatures to some amount (see manual), but it can't correct for these internal e.m.f. errors.. again that's what the NULL function is for..

In summary, these offsets you see @ 18°C are perfectly fine, and no sign of defect or mis-calibration.

I suggest to prepare a special lab room at 23 +/- 2 °C.

Frank

[bson]

Somewhat tangential, but I often wish I could null my 34465A to the statistical average of a run, rather than take a new measurement.

[djnz]

Again a bit of a tangent, but firmware version 3.0 now includes the digitizing and advanced trigger (former option DIG) as standard.

[HighVoltage]

I had my 34465A running over night with a short block on the input.
Here is the histogram for comparison.

[dmm2018]

I had my 34465A running over night with a short block on the input.
Here is the histogram for comparison.

Hi HighVoltage,

The histogram of your 34465A is really nice. Before starting collecting the samples for the histogram, how long has the 34465A been powered (warmed) up?

Thanks

[HighVoltage]

Hello dmm2018

This 34465A was turned ON for about 24h, before I started the data collection for the histogram

[dmm2018]

Hello dmm2018

This 34465A was turned ON for about 24h, before I started the data collection for the histogram

Thanks HighVoltage...

I tried to produce a similar histogram plot for my 34470A. I found it's not easy to obtain such a nice bell-shape curve for my unit. I noted that the different orientation (rotated 10-30 degree) of 34470A will affect and shift the mean of the histogram plot quite obviously.

[HighVoltage]

If you have your 34470A ON for a few hours and the temperature will not change much, you should get a very similar bell curve.

OK, I will do the test again with one of my 34470A over night and see the results tomorrow.

[HighVoltage]

These data were taken over about 24 h on one of my 34470A with a shortage block across all 4 input connectors.

[splin]

The 34470A noise is double that of the 34465A which is a bit surprising - were they using the same NPLC/filtering/AZ settings? What were the settings?

How come the 34470A is only showing 6 1/2 Digits?

If they were using the same settings then I wonder if the noise difference is normal variance between instruments - I expect the 34465A and 34470A will have the same front end and ADC,  the main difference being the reference (and the price!)

[HighVoltage]

Both, the 34470A and the 34465A were set to ...
Aperture: 10 PLC
Auto Zero: ON
Input Z: 10 MOhm
I use the same short block on both

I will repeat the test and do both instruments at the same time and also record the room temperature
May be I will add a 34461A for comparison as well.

[HKJ]

I also tried letting my 34465A/34470A warm up, do a ACAL and then histogram over 24 hours. Except for histogram settings all parameters are default on the meters.



Large size: http://lygte-info.dk/pic/cpf4/DSC_5321a.jpg

[splin]

From the 34465A and 34470A specs the noise adder for 1NPLC is 1ppm of range + .5uV = 6ppm for the 100mV range = 600nV

At 10NPLC the expected RMS noise would be 600/sqrt(10) = 190nV. That compares to:

34470A	dmm2018	136nV
34465A	HKJ	180nV
34470A	HKJ	200nV
34465A	HighVoltage	80nV
34470A	HighVoltage	190nV

So most are pretty close to the spec except HighVoltage's '65A which is surprisingly good.

[EDIT] FWIW The 3458A equivalent spec is 54nV - the difference  being mostly due to the ADC.

I estimate the 34470A's input amplifier is approx 4X noisier and the the ADC is around 14X noiser.

[eplpwr]

How come the 34470A is only showing 6 1/2 Digits?

On the 100 mV range, 34470A defaults to 6 1/2 digits. You can manually change this to 7 1/2 digits with "Digit Mask" button in DCV mode, which also results in an additional digit of resolution in the statistics window (Avg, Stddev et al).

[HighVoltage]

OK, repeated the histogram noise test on 4 instruments at the same time over night.
The temperature span was about 1 degree.
Since I only have one good short block, I used a single short cable on the voltage input of all instruments.

It seems the test results from earlier are repeatable.
I am not sure why my 34465A has such little noise, compared to the two 34470A's

All standard settings, 10 NPLC

1. Room temperature measurement with 5k thermistor on a 34461A
2. Noise 34461A    = 130 nV StdDev
3. Noise 34465A    = 100 nV StdDev
4. Noise 34470A-1 = 160 nV StdDev
5. Noise 34470A-2 = 130 nV StdDev

[dmm2018]

The calibration process is done at about 23°C, even @ Keysight in Malaysia, where your instrument might have been calibrated (see your cal certificate).
Highest uncertainty is achieved only at this calibration temperature. Otherwise, correction calculations using T.C.s apply.

Frank

Today installed BenchVue and tried retrieving the "temperature of last calibration" (as described on pp. 242 of 34460-90901 Operating and Service Guide) using the following SCPI command:

CALibration:TEMPerature?
Returns the temperature of the last calibration in °C.

The returned temperature of last calibration of my Keysight 34470A (L) is 25.41 °C.

According to the Certificate Of Calibration, I think the meter was calibrated @ Keysight (Malaysia). Is it normal that the Keysight 34470A was calibrated at 25.41 °C?

[e61_phil]

25.41°C would be the internal temperature and not the lab temperature