RoadTest: Keysight 34470A and Texas Instruments DAC8734EVM
Author: 6thimage
Creation date:
Evaluation Type: Independent Products
Did you receive all parts the manufacturer stated would be included in the package?: True
What other parts do you consider comparable to this product?: Keithley DMM7510, HP/Agilent/Keysight 3458A
What were the biggest problems encountered?: n/a
Detailed Review:
Introduction
Keysight’s 34470A is the first 7.5 digit multimeter that they have made (including HP and Agilent) and, alongside the 34465A, is the newest addition to Keysight’s TrueVolt multimeter range. Incidentally, their main competitor, Keithley (in the multimeter market), has also released their new 7.5 digit multimeter at practically the same time.
The first two multimeter’s in the TrueVolt range, the 34460A and the 34461A, were very well received, with their colour 4.3” display immediately setting them apart from the competition. The 34461A was designed as a replacement for the 34401A, adding new features and increasing the measurement accuracy, with the 34460A intending to be a lower cost version, partly for the hobbyist market.
Overall, the 34461A received little criticism, with one of the biggest being that it didn't significantly improve on the specifications of its predecessor and it lacked a capacitance measuring function (which similar bench multimeters were including). However, as a replacement for the 34401A is was a clear improvement. With the release of the 34470A and 34465A, Keysight released a new firmware version for the 34461A which included a capacitance function and some secondary measurements.
The 34465A has been released as a replacement for the 34410A, with its memory and digitising options making it a replacement for the 34411A, both of which were designed as higher performance versions of the 34401A. The 34470A, not being designed as a replacement, fits somewhere between the 34410/11A and the metrology grade 3458A, with it being closer to the 34410/11A than the 3458A.
Blog posts
There will be several blog posts that will be uploaded after this review, which will cover the Texas Instruments DAC, the digitising functionality and sampling model of the 34470A, and the BenchVue remote control software.
Comparison with similar bench multimeters
To help with the comparisons, I have made a spreadsheet which is available at https://docs.google.com/spreadsheets/d/1ZVMTCumXWyNiJ5xm6O5CCFb_HdFU9ntbyYG-e-CKOqQ/.
The 34470A is the cheapest of the four 7.5 digit multimeters that are currently available by over £700 and has the second best DC voltage accuracy, with the Keithley’s DMM7510 being first at 0.0014% versus the 34470A’s 0.0016% (10 V, 1 year). This 2 ppm difference is very small, especially when considering the two fold difference compared with the 34465A and the 34461A (0.0030% and 0.0035% respectively).
The more obvious difference, or lack of, is the current accuracy - the 34470A has exactly the same DC current accuracy (100 mA, 1 year) as the 34465A, 34461A and even the 34460A. In fact, most of the multimeters have a DC current accuracy of 0.05%. The only multimeters that improve on this specification are two by Keithley (including the DMM7510 at 0.015%) and the metrology grade 3458A. Whilst we can ignore the 3458A for most comparisons (due to its high cost and very limited market), it seems to me that Keysight have really missed a trick by not improving the DC current specifications. With the 34470A having the same current accuracy as the 34460A, 34461A and 34465A, it is fairly obvious that the limitation is not the ADC but the load stability of the current sensing resistors. Whilst it is possible to get better stability resistors, with some offering load stabilities that are better than 100 ppm (0.01%), I am a little surprised that Keysight did not include the current sense resistors as part of the autocalibration feature. This would allow the same current sense resistors to be used (which I am assuming are around 500 ppm), but with the autocalibration measuring their resistance value. This would decrease their uncertainty and so significantly increase the current accuracy (by at least an order of magnitude), without increasing cost - as only a couple of relays or multiplexers would be required.
The resistance accuracy has been improved over the 34461A, but, like the DC voltage, the 34470A is the second best 7.5 digit multimeter for accuracy, with Keithley’s new offering bettering it by 10 ppm. With the accuracy being the same as the 34465A, the limiting factor is the multimeter’s current source, not the ADC.
In terms of features, all of the 7.5 and 8.5 digit multimeters are quite similar. The only main difference, other than continuity, frequency range and temperature probe support, is the current ranges, with Keysight’s 34470A offering the largest range, of 1 uA to 10 A. This is followed by Keithley’s DMM7510 (10 uA to 10 A), 2001 (200 uA to 2 A) and 2010/E (10 ma to 3 A).
With regards to usability, for front panel usage the 34470A and Keithley’s DMM7510 are the only two contenders - both offer large colour screens, with Keithley's also being touch sensitive. These colour screens are significantly better than the older VFDs as they allow a user to see the measurement data in a more complete way, through trend graphs and histograms. Whilst I haven't had the opportunity to use a DMM7510, from watching videos, both by Keithley and third parties, the DMM7510’s interface looks a little clunky and sometimes fiddly compared to the 34470A’s. At times it seems that the touch interface hampers the usability, requiring multiple presses to do a simple function. However, the 34470A’s interface is not perfect, with text entry being quite cumbersome and movement of the data cursors sometimes being tricky.
For remote use, the 34470A and the DMM7510 have the advantage of USB and LAN connectivity, with all the 7.5 digit multimeters, except the 34470A, offering GPIB as standard. All of these 7.5 digit multimeters use SCPI, allowing many different computer programs to remotely control them (such as LabView and Matlab). So the remote usability is mainly down to connectivity, of which USB and LAN are significantly easier and cheaper. The only current benefit of GPIB over USB and LAN is its group execution trigger - where multiple instruments can be triggered at the same time (i.e. with a single instruction). For similar functionality with both USB and LAN, the trigger input of the instruments (typically a BNC) needs to be used, which inevitably adds some extra complexity.
Unboxing and initial impressions
The 34470A was packaged quite well, in almost exactly the same way as the 34461A, which is only to be expected. An accessory box, containing a US power lead, a set of probes and a USB device cable sits above the multimeter. Unlike in the 34461A RoadTest, an additional UK power lead was not added by the people at element14. However, this was not a problem for me as I always have some kettle cables lying around and I had already ordered a right angled power lead, as I prefer to use them on the 34461A. Right angle power leads work quite nicely with these multimeters, as they do not interfere with the other connectors on the back and it also means that you can stand the meter on its tail - although, due to the fan, this should not be done on anything but a hard floor, and ideally not for very long (as the multimeter will heat up by a few degrees). As you can see from the photo below, the Texas Instrument's DAC evaluation board comes with a leaflet. The front of this directs you to the TI website for all the information, with the rest of the leaflet detailing TI's terms and conditions for evaluation modules.
The accessory box also contained a leaflet talking about the Agilent to Keysight name change, in case you had been sleeping under a rock for the last couple of years.
The multimeter was snuggly sat at the bottom of the box, being supported by two foam ends. I was glad to see that Keysight have not changed this as it seems to be an excellent way to ensure that the multimeter arrives without even a hint of damage, despite the attempts of your typical courier.
The Texas Instruments DAC evaluation board had been placed down the side of the foam end protecting the front of the unit and was not particularly secure. I believe this was the reason why a couple of the pins on the board were slightly bent on arrival, although this was very minor and easily fixable.
There are some small differences in the appearances of the 34470A and the 34461A, such as the different name and branding and the additional button labels (capacitor and autocal).
One difference that is a bit odd, is that the rear rubber moulding no longer has a ‘no operator serviceable parts’ warning imprinted onto it. Which seems to be a bizarre small change, as there is no obvious reason why they would have removed this warning.
These differences are really small and it can be easy to miss them - if you have multiple models of this series on your bench, then you will more than likely be looking for a way to remember which one is which, as from a quick glance they are identical. When powered on, I like to have the 34470A’s display set to the ‘B’ colour scheme, which is predominantly black, which helps identify it from my 34461A. In a production or rack setup, this is not likely to be a problem as you can set text labels on the display to indicate the measurement the meter is showing.
Additionally, the fan in the 34470A is significantly quieter than the one in my 34461A. My 34461A’s fan is just noticeable in a typical room, where as you can only really hear the 34470A’s fan when everything else in the room is quiet. Both of the fans, according to Keysight’s website, are the same part number (5041-5232), which raises the question - has the fan part been changed to a quieter version, or is there a large variance in the noise the fans make? There are, of course, alternative reasons for the change in fan noise - perhaps the 34470A’s fan driving circuit is different, using a lower fan speed perhaps, or maybe the fans become louder with age (although my 34461A’s fan has very little, if any, dust on it). Either way, both of the fans of the 34461A and the 34470A are largely unnoticeable, and so this difference is really splitting hairs.
One slightly irritating difference I have noticed between the 34470A and its older sibling is the colour temperature of the displays. The 34470A’s display has much colder colours - for example, the background of the on screen buttons is noticeably greyer. This does not affect the readability or usability of the display, but when placed side by side it is quite noticeable. Again, I am not sure of the reason behind this as they both use the same parts, my guess is that the difference is down to the display’s manufacturing tolerances (the display used is the same that is used in an industrial sat nav unit, so precise colour reproduction is not a particularly high concern).
Differences between the 34470A and 3446xA
Overall, there are very few differences internally between the four different models. First of all, the front panels are almost identical, with the same processor and same amount of both RAM and flash memory. The only difference is that the 34470A and the 34465A have a temperature sensor located below the front input terminals and above the input selection switch.
This temperature sensor, which is likely to be a thermistor, is used for cold junction temperature compensation for thermocouples. This is the reason why support for thermocouples has not been added to the 34460A and 34461A with the version 2 firmware update, although it would be possible to add support using virtual / software compensation (where a user sets the cold junction temperature).
With regards to the rest of the instrument, the 34470A and 34465A have extra circuitry around the ADC, which is used for the autocalibration function, and around the current sense resistor, which is used to amplify the measured voltage for the lower current ranges (that is the 1 and 10 uA ranges). Of course, the largest difference between the 34470A and the other three models (34460/61/65A) is the voltage reference. The 34470A uses an LTZ1000, with the rest of the TrueVolt series using the LM399 - although both of these are highly likely to be specially selected, with Keysight more than likely ageing them to improve their performance, rather than them being straight from Linear Technology’s factory. It is from this different voltage reference that the 34470A’s extra stability and precision, over the 34465A, comes from. If you were to swap the 34470A’s reference for an LM399 from a 34465A, then it would perform like a 34465A as this is the only difference between them.
Finally, for completeness, the only difference between the 34461A and the 34460A is that the later does not have either the rear inputs or a fan fitted. I have additionally been told that the 34460A’s lack of a fan is partly the reason why it has a lesser accuracy specification than the 34461A, with the position of the tilting bail affecting the airflow and, consequently, the measurements. Although, I suspect the 34460A uses the voltage references that have been shown to drift more during the ageing process than those in the 34461A and 34465A.
All four TrueVolt models run the same firmware, with any limitations - such as a lack of the trend and histogram display on the 34460A, and a reduced number of secondary measurements on both the 34460A and 34461A - are determined by the multimeter’s stored model and serial number. Resulting in any features in the 34465A and 34470A, besides the improved accuracy, larger ranges and thermocouple support, being arbitrary product lines, rather than fundamental limitations.
It is fairly easy to criticise these arbitrary differences and so I will try to avoid doing so, but the most irritating of them for me are the reduction in secondary measurements and the lack of the data logging and extended memory on the lower models. In particular, I feel that Keysight have neglected all of the existing 34461A owners, who will undoubtedly look at the 34465A, knowing it runs the same firmware, feeling left out that they do not get the peak to peak secondary measurements, an increase in the memory depth and the data logging, whilst knowing that their meter is fully capable of them. I think many 34461A owners would love to have an upgrade path, be it a trade in offer to get a 34465A or a 34470A, or a firmware option to unlock these extra features. However, this is a bit of an aside for a review of the 34470A.
Temperature measurements
The 34470A, and the 34465A, have a new feature of being able to use thermocouples for temperature measurements. This is a very useful feature and one that I have missed on my 34461A. Thermocouples, unlike thermistors and RTDs, produce a voltage across their terminals that changes with temperature, which is caused by the Seebeck effect. Whilst thermocouples are not as accurate as thermistors or RTDs (a couple of °C vs half a °C), they only rely on a welded junction between two wires, making them both cheap and able to withstand large temperature ranges - for example, a K type thermocouple can measure temperatures from -200 °C to 1350 °C (https://en.wikipedia.org/wiki/Thermocouple#Type_K). They also have the added advantage that they are fairly inexpensive.
To use thermocouples, a cold junction is needed as a reference point - this is typically a beaker filled with ice and water, which will happily stay at 0 °C. However, to make life easier, a thermometer can be used to compensate for the cold junction. This is what the 34470A does, with a thermistor being placed next to the input jacks, behind the plastic of the front panel.
From looking inside the 34470A, it appears that there is only one of these thermistors, with there being no additions to the rear inputs. This potentially means that thermocouple measurements taken when using the rear inputs could have additional sources of error. For most uses however, the front and rear inputs will be at very similar temperatures, so the placement of the thermistor is not important. However, if the multimeter has a temperature difference between the front and rear inputs - for example if the multimeter was placed in a rack, with a colder airflow on the front panel - then the measurements would be affected. This is something that is definitely worth knowing about.
Data logging and digitising
The 34470A, and also the 34465A, have two new extra acquisition modes - data logging and digitising. These allow a user to easily configure the multimeter to perform measurements over a long period of time or perform measurements at the multimeter's fastest rate. This provides several advantages, the first is that the data logging can record the measurements to a flash drive (or the internal flash memory), which allows you to easily collect measurements without having to control the multimeter remotely. The digitising mode automatically changes the multimeter’s settings to enable the fastest readings possible (such as disabling auto zeroing and autoranging), which makes it easier to set up the multimeter to perform faster measurements. With these two new acquisition modes, comes a new sampling model, allowing the 34470A to perform measurements in a deterministic way, which is a fantastic new feature. The differences between the sampling modes, and why it is important, are discussed in a blog post (link above).
All of the acquisition modes alter the main state of the multimeter, so if you are performing normal, continuous measurements, say at 10 PLCs with an immediate trigger, and then you change to the digitising mode, with 0.001 PLCs and a level trigger, when you return to the continuous mode, the multimeter will be left at 0.001 PLCs with a level trigger. This can be a little infuriating to begin with, as the digitising mode changes quite a few settings to get the fastest measurement rate. For a new user, these changes will not be obvious and so might cause a little confusion. The easiest thing to do is recall the default settings, or the previous settings if you saved them before entering the digitising mode. However, I feel that this is a step that shouldn't be required, as the multimeter should restore the previous continuous measurement settings when leaving the digitising mode.
Data logging mode
To test this mode, I decided to take measurements of a Vishay 10k VSMP resistor, using 4 wires. The VSMP series have quite low temperature coefficients - typically +/- 0.05 ppm/°C between 0 and 60 °C, with a maximum of +/- 1.8 ppm/°C. For a 10k resistor, this is +/- 0.5 mOhm/°C and +/- 18 mOhm/°C, respectively.
The data logging was for 48 hours, starting in the evening at 23:10.
Setting the data logging up is very simple, when in the acquire menu (from the front panel button), the leftmost soft button (underneath the screen) allows you to enter the data logging mode. You are then presented with the above screen (without the grey bars across), where you can set the duration of the logging (either in samples or time), where the data is logged to and when it should start (either at a certain time or as a delay). The sample interval can be set to either the minimum (where it is governed by the user’s choices of settings - such as autorange and autozero) or to any other value up to 3,600 seconds (1 hour).
When setting the start time, the time will overflow as expected (i.e. incrementing the minutes by one from 59 will increment the hour and set the minutes to 0), with the exception that you cannot increment any further than 23:59 - essentially, the time will not overflow to 00:00. Similarly, it is not possible to decrement the time from 00:00 to 23:59. Whilst this is not an issue, it is worthwhile knowing that it functions this way.
Then by pressing the run/stop button, the data logging begins - or if a delay is chosen, the multimeter waits before it begins (as in the previous picture). Currently, there is a bug in the 2.11 firmware (and previous versions) which means that a delayed data logging when the screen has been intentional turned off (that is through the utility settings menu) does not start when the delay completes, but when the screen is turned back on. However, most users will not come across this as it is uncommon to turn the display off.
When the multimeter starts logging, the screen is updated with the latest measurements, which is very helpful and allows a user to easily check on the instrument’s readings. However, one thing I did find annoying was the two grey bars that tell you the log file and the remaining time. Whilst these are very useful and contain important information, they obscure a lot of the display - in this setup, they completely hide the calculated statistics.
When the data logging completes (or is stopped by a user), the grey bars disappear and the display returns back to a familiar look, where the screen is no longer obscured.
Overall, the process of logging data is really easy to set up, with the interface being very intuitive. However, the hiding of information during data logging by the grey bars, is a little bit of a hinderance and makes the process a little less interactive. Whilst being less interactive when data logging does not initially sound like an issue, te 34470A allows a user to view the logged data whilst it is being captured. For example, the statistics and the information on the cursor positions are both obscured by the grey bars stating the time left and the logging file. This means that any recorded readings of interest cannot be properly investigated until the data logging is stopped or completed.
Both after and during acquisition, the trend graph can be zoomed and panned, which is very useful. The zoom levels can only take certain values, which is a bit of a shame as this means that you can rarely get the data to fully fit the screen. You tend to either get the previous picture, where almost half of the graph is empty, or where you are zoomed into the data and cannot see all of the data (here the beginning and end are chopped off).
Navigating around the graph is quite simple, after pressing the zoom soft button, the left and down arrow buttons zoom out, and the right and top arrow buttons zoom in. The panning works in the same way, with left / down and up / right having the same actions. This is something that I would like to see changed, I think a single zoom and pan option should be available, where the left and right arrows pan, with up and down zooming. This would make it significantly more straightforward to navigate the graph, as currently the user has to switch between either panning or zooming.
Cursors
Alongside the data logging mode, comes cursors to use in the histogram and trend displays. These are very useful for getting quick information about the measurements directly from the front panel, without having to transfer the data to a computer to process.
In the histogram display, two cursors are available, which can be used to find information about particular bins.
Here, I have selected the two bins at the leftmost and rightmost peaks, with the left peak bin containing 1.06% and the right containing 2.13%. I like how the difference between the two cursors is shown on the right hand side (next to span), with the amount of data contained between the cursors (63.1%) also being shown.
However, the information on the widths of the bins is not particularly useful. This is particularly shown for B2, where the bin appears to have no width. When moving the cursors, the bins frequently show this lack of width, as it seems that they are around half a least significant digit wide (i.e. 0.000 000 5 kOhms). Whilst I can understand that showing this digit would be giving more than the quoted 7.5 digits, not showing it is likely to confuse some users - as the resolution is obviously there (to produce the histogram), just not qualified.
In the trend display, the user can use two x-y cursors. In the image below, the cursors are set to track the measured value (y axis). This allows you to measure not only the change in time, but also the change in the readings.
Unlike the histogram view, the data being shown is not processed, so the y-axis values are, and should be, the full 7.5 digits.
Whilst you could always perform data logging by using a computer and one of the remote programming interfaces (i.e. USB, ethernet or GPIB), the built in data logging is very useful as it does not require any external components. More importantly, it does not require a user to spend time understanding which combination of SCPI commands are required to get the best settings. The user can simply use the same interface that they use when making measurements, but instead tell it to log for a set period of time, at whatever time period is needed and with any required start delay. This alone makes this feature practically essential on a modern bench multimeter.
Overall, the new cursors feature is a great addition, that will be useful in many different applications - from quick debugging to high-precision measurements - as the cursors can be used in all three acquisition modes (continuous, data logging and digitising). Although it is worth noting that in continuous mode the cursors do not work in the the trend graph. There are some caveats with the cursors, such as the minor ones above regarding resolution, but the most important to realise is that they are limited to the amount of memory the multimeter has ‘installed’ (or more correctly, enabled by license). When in the continuous acquisition mode, the cursors will automatically disable when the number of readings exceeds either 50,000 (if the MEM option is not installed) or 2 million (with the MEM option). This will also occur in the data logging acquisition mode, even if the output is being written to a file (as the cursors can only use the readings in the volatile storage).
The only features I would like to see added to the trend graph and the cursors are more zoom levels, including an auto fit, and the ability for cursors to jump to points of interest, such as minimums and maximums, both globally and for the current zoom level, as well as next peak and trough functionality. I think these would add a lot to the usability, however, there are only a limited number of soft buttons.
Measurement speed
One of the nice additions, that comes along with the new data logging and digitising acquisition modes and the new sampling model, is the ability for the multimeter to tell you how quickly it can perform measurements. This is accessed via the SCPI command ‘samp:tim? min’, which returns the smallest amount of time that the sample timer can be set to, which is the longest amount of time a measurement can take. It is fairly easy to create a little script to find this out for all of the functions and settings - which I did in Python - and the results are below.
DC voltage, auto zero off, fixed range nplc: 0.001 measurement time: +2.00000000E-05 measurements per second: 50000.00 nplc: 0.002 measurement time: +4.00000000E-05 measurements per second: 25000.00 nplc: 0.006 measurement time: +1.00000000E-04 measurements per second: 10000.00 nplc: 0.02 measurement time: +2.99000000E-04 measurements per second: 3344.48 nplc: 0.06 measurement time: +1.00000000E-03 measurements per second: 1000.00 nplc: 0.2 measurement time: +3.00100000E-03 measurements per second: 333.22 nplc: 1 measurement time: +2.00010000E-02 measurements per second: 50.00 nplc: 10 measurement time: +2.00000000E-01 measurements per second: 5.00 nplc: 100 measurement time: +1.99999900E+00 measurements per second: 0.50 DC voltage, auto zero on, fixed range nplc: 0.001 measurement time: +4.88000000E-04 measurements per second: 2049.18 nplc: 0.002 measurement time: +5.28000000E-04 measurements per second: 1893.94 nplc: 0.006 measurement time: +6.47000000E-04 measurements per second: 1545.60 nplc: 0.02 measurement time: +1.04700000E-03 measurements per second: 955.11 nplc: 0.06 measurement time: +2.64700000E-03 measurements per second: 377.79 nplc: 0.2 measurement time: +6.45000000E-03 measurements per second: 155.04 nplc: 1 measurement time: +4.04500000E-02 measurements per second: 24.72 nplc: 10 measurement time: +4.04496000E-01 measurements per second: 2.47 nplc: 100 measurement time: +4.04496000E+00 measurements per second: 0.25 DC voltage, auto zero off, auto range nplc: 0.001 measurement time: +3.51660000E-02 measurements per second: 28.44 nplc: 0.002 measurement time: +3.52060000E-02 measurements per second: 28.40 nplc: 0.006 measurement time: +3.54260000E-02 measurements per second: 28.23 nplc: 0.02 measurement time: +3.57250000E-02 measurements per second: 27.99 nplc: 0.06 measurement time: +3.72560000E-02 measurements per second: 26.84 nplc: 0.2 measurement time: +4.12580000E-02 measurements per second: 24.24 nplc: 1 measurement time: +7.52880000E-02 measurements per second: 13.28 nplc: 10 measurement time: +2.55088000E-01 measurements per second: 3.92 nplc: 100 measurement time: +2.05508700E+00 measurements per second: 0.49 DC voltage, auto zero on, auto range nplc: 0.001 measurement time: +1.51860000E-02 measurements per second: 65.85 nplc: 0.002 measurement time: +1.52460000E-02 measurements per second: 65.59 nplc: 0.006 measurement time: +1.55250000E-02 measurements per second: 64.41 nplc: 0.02 measurement time: +1.60230000E-02 measurements per second: 62.41 nplc: 0.06 measurement time: +1.82550000E-02 measurements per second: 54.78 nplc: 0.2 measurement time: +2.42580000E-02 measurements per second: 41.22 nplc: 1 measurement time: +7.52880000E-02 measurements per second: 13.28 nplc: 10 measurement time: +4.39135000E-01 measurements per second: 2.28 nplc: 100 measurement time: +4.07959900E+00 measurements per second: 0.25 ######################################################################################### DC current, auto zero off, fixed range nplc: 0.001 measurement time: +2.00000000E-05 measurements per second: 50000.00 nplc: 0.002 measurement time: +4.00000000E-05 measurements per second: 25000.00 nplc: 0.006 measurement time: +1.00000000E-04 measurements per second: 10000.00 nplc: 0.02 measurement time: +2.99000000E-04 measurements per second: 3344.48 nplc: 0.06 measurement time: +1.00000000E-03 measurements per second: 1000.00 nplc: 0.2 measurement time: +3.00100000E-03 measurements per second: 333.22 nplc: 1 measurement time: +2.00010000E-02 measurements per second: 50.00 nplc: 10 measurement time: +2.00000000E-01 measurements per second: 5.00 nplc: 100 measurement time: +1.99999900E+00 measurements per second: 0.50 DC current, auto zero on, fixed range nplc: 0.001 measurement time: +4.88000000E-04 measurements per second: 2049.18 nplc: 0.002 measurement time: +5.28000000E-04 measurements per second: 1893.94 nplc: 0.006 measurement time: +6.47000000E-04 measurements per second: 1545.60 nplc: 0.02 measurement time: +1.44600000E-03 measurements per second: 691.56 nplc: 0.06 measurement time: +2.44700000E-03 measurements per second: 408.66 nplc: 0.2 measurement time: +6.45000000E-03 measurements per second: 155.04 nplc: 1 measurement time: +4.04500000E-02 measurements per second: 24.72 nplc: 10 measurement time: +4.03897000E-01 measurements per second: 2.48 nplc: 100 measurement time: +4.03897400E+00 measurements per second: 0.25 DC current, auto zero off, auto range - fast switch nplc: 0.001 measurement time: +3.60660000E-02 measurements per second: 27.73 nplc: 0.002 measurement time: +3.61060000E-02 measurements per second: 27.70 nplc: 0.006 measurement time: +3.62260000E-02 measurements per second: 27.60 nplc: 0.02 measurement time: +3.71240000E-02 measurements per second: 26.94 nplc: 0.06 measurement time: +3.79260000E-02 measurements per second: 26.37 nplc: 0.2 measurement time: +4.23280000E-02 measurements per second: 23.63 nplc: 1 measurement time: +7.66280000E-02 measurements per second: 13.05 nplc: 10 measurement time: +2.56627000E-01 measurements per second: 3.90 nplc: 100 measurement time: +2.05662600E+00 measurements per second: 0.49 DC current, auto zero on, auto range - fast switch nplc: 0.001 measurement time: +1.60860000E-02 measurements per second: 62.17 nplc: 0.002 measurement time: +1.61460000E-02 measurements per second: 61.93 nplc: 0.006 measurement time: +1.63250000E-02 measurements per second: 61.26 nplc: 0.02 measurement time: +1.74220000E-02 measurements per second: 57.40 nplc: 0.06 measurement time: +1.89250000E-02 measurements per second: 52.84 nplc: 0.2 measurement time: +2.53280000E-02 measurements per second: 39.48 nplc: 1 measurement time: +7.66280000E-02 measurements per second: 13.05 nplc: 10 measurement time: +4.42470000E-01 measurements per second: 2.26 nplc: 100 measurement time: +4.10089200E+00 measurements per second: 0.24 DC current, auto zero off, auto range - continuous switch nplc: 0.001 measurement time: +5.60660000E-02 measurements per second: 17.84 nplc: 0.002 measurement time: +5.61060000E-02 measurements per second: 17.82 nplc: 0.006 measurement time: +5.62260000E-02 measurements per second: 17.79 nplc: 0.02 measurement time: +5.71240000E-02 measurements per second: 17.51 nplc: 0.06 measurement time: +5.79260000E-02 measurements per second: 17.26 nplc: 0.2 measurement time: +6.23280000E-02 measurements per second: 16.04 nplc: 1 measurement time: +9.66280000E-02 measurements per second: 10.35 nplc: 10 measurement time: +2.76627000E-01 measurements per second: 3.61 nplc: 100 measurement time: +2.07662600E+00 measurements per second: 0.48 DC current, auto zero on, auto range - continuous switch nplc: 0.001 measurement time: +3.60860000E-02 measurements per second: 27.71 nplc: 0.002 measurement time: +3.61460000E-02 measurements per second: 27.67 nplc: 0.006 measurement time: +3.63250000E-02 measurements per second: 27.53 nplc: 0.02 measurement time: +3.74220000E-02 measurements per second: 26.72 nplc: 0.06 measurement time: +3.89250000E-02 measurements per second: 25.69 nplc: 0.2 measurement time: +4.53280000E-02 measurements per second: 22.06 nplc: 1 measurement time: +9.66280000E-02 measurements per second: 10.35 nplc: 10 measurement time: +4.62470000E-01 measurements per second: 2.16 nplc: 100 measurement time: +4.12089200E+00 measurements per second: 0.24 ######################################################################################### AC voltage, fixed range bandwidth: 3 measurement time: +2.56430000E+00 measurements per second: 0.39 bandwidth: 20 measurement time: +6.49300000E-01 measurements per second: 1.54 bandwidth: 200 measurement time: +4.33000000E-02 measurements per second: 23.09 AC voltage, auto range bandwidth: 3 measurement time: +2.56430000E+00 measurements per second: 0.39 bandwidth: 20 measurement time: +6.49300000E-01 measurements per second: 1.54 bandwidth: 200 measurement time: +4.33000000E-02 measurements per second: 23.09 ######################################################################################### AC current, fixed range bandwidth: 3 measurement time: +1.74430000E+00 measurements per second: 0.57 bandwidth: 20 measurement time: +2.96800000E-01 measurements per second: 3.37 bandwidth: 200 measurement time: +6.43000000E-02 measurements per second: 15.55 AC current, auto range bandwidth: 3 measurement time: +1.74430000E+00 measurements per second: 0.57 bandwidth: 20 measurement time: +2.96800000E-01 measurements per second: 3.37 bandwidth: 200 measurement time: +6.43000000E-02 measurements per second: 15.55 ######################################################################################### 2W resistance, auto zero off, fixed range, offset compensation off nplc: 0.001 measurement time: +2.00000000E-05 measurements per second: 50000.00 nplc: 0.002 measurement time: +4.00000000E-05 measurements per second: 25000.00 nplc: 0.006 measurement time: +1.00000000E-04 measurements per second: 10000.00 nplc: 0.02 measurement time: +2.99000000E-04 measurements per second: 3344.48 nplc: 0.06 measurement time: +1.00000000E-03 measurements per second: 1000.00 nplc: 0.2 measurement time: +3.00100000E-03 measurements per second: 333.22 nplc: 1 measurement time: +2.00010000E-02 measurements per second: 50.00 nplc: 10 measurement time: +2.00000000E-01 measurements per second: 5.00 nplc: 100 measurement time: +1.99999900E+00 measurements per second: 0.50 2W resistance, auto zero on, fixed range, offset compensation off nplc: 0.001 measurement time: +4.88000000E-04 measurements per second: 2049.18 nplc: 0.002 measurement time: +7.27000000E-04 measurements per second: 1375.52 nplc: 0.006 measurement time: +8.47000000E-04 measurements per second: 1180.64 nplc: 0.02 measurement time: +1.24600000E-03 measurements per second: 802.57 nplc: 0.06 measurement time: +2.44700000E-03 measurements per second: 408.66 nplc: 0.2 measurement time: +6.45000000E-03 measurements per second: 155.04 nplc: 1 measurement time: +4.04500000E-02 measurements per second: 24.72 nplc: 10 measurement time: +4.04496000E-01 measurements per second: 2.47 nplc: 100 measurement time: +4.04496000E+00 measurements per second: 0.25 2W resistance, auto zero off, fixed range, offset compensation on nplc: 0.001 measurement time: +2.85200000E-03 measurements per second: 350.63 nplc: 0.002 measurement time: +3.33000000E-03 measurements per second: 300.30 nplc: 0.006 measurement time: +3.57000000E-03 measurements per second: 280.11 nplc: 0.02 measurement time: +4.36800000E-03 measurements per second: 228.94 nplc: 0.06 measurement time: +6.81000000E-03 measurements per second: 146.84 nplc: 0.2 measurement time: +1.48160000E-02 measurements per second: 67.49 nplc: 1 measurement time: +8.28760000E-02 measurements per second: 12.07 nplc: 10 measurement time: +8.10968000E-01 measurements per second: 1.23 nplc: 100 measurement time: +8.09189600E+00 measurements per second: 0.12 2W resistance, auto zero on, fixed range, offset compensation on nplc: 0.001 measurement time: +2.85200000E-03 measurements per second: 350.63 nplc: 0.002 measurement time: +3.33000000E-03 measurements per second: 300.30 nplc: 0.006 measurement time: +3.57000000E-03 measurements per second: 280.11 nplc: 0.02 measurement time: +4.36800000E-03 measurements per second: 228.94 nplc: 0.06 measurement time: +6.81000000E-03 measurements per second: 146.84 nplc: 0.2 measurement time: +1.48160000E-02 measurements per second: 67.49 nplc: 1 measurement time: +8.28760000E-02 measurements per second: 12.07 nplc: 10 measurement time: +8.10968000E-01 measurements per second: 1.23 nplc: 100 measurement time: +8.09189600E+00 measurements per second: 0.12 2W resistance, auto zero off, auto range, offset compensation off nplc: 0.001 measurement time: +7.47770000E-02 measurements per second: 13.37 nplc: 0.002 measurement time: +7.48170000E-02 measurements per second: 13.37 nplc: 0.006 measurement time: +7.49370000E-02 measurements per second: 13.34 nplc: 0.02 measurement time: +9.55080000E-02 measurements per second: 10.47 nplc: 0.06 measurement time: +9.69090000E-02 measurements per second: 10.32 nplc: 0.2 measurement time: +1.00942000E-01 measurements per second: 9.91 nplc: 1 measurement time: +1.34941000E-01 measurements per second: 7.41 nplc: 10 measurement time: +3.14741000E-01 measurements per second: 3.18 nplc: 100 measurement time: +2.11461000E+00 measurements per second: 0.47 2W resistance, auto zero on, auto range, offset compensation off nplc: 0.001 measurement time: +5.47960000E-02 measurements per second: 18.25 nplc: 0.002 measurement time: +5.48560000E-02 measurements per second: 18.23 nplc: 0.006 measurement time: +5.50360000E-02 measurements per second: 18.17 nplc: 0.02 measurement time: +7.58060000E-02 measurements per second: 13.19 nplc: 0.06 measurement time: +7.79080000E-02 measurements per second: 12.84 nplc: 0.2 measurement time: +8.39420000E-02 measurements per second: 11.91 nplc: 1 measurement time: +1.34941000E-01 measurements per second: 7.41 nplc: 10 measurement time: +4.94741000E-01 measurements per second: 2.02 nplc: 100 measurement time: +4.09460800E+00 measurements per second: 0.24 2W resistance, auto zero off, auto range, offset compensation on nplc: 0.001 measurement time: +1.80623000E-01 measurements per second: 5.54 nplc: 0.002 measurement time: +1.80723000E-01 measurements per second: 5.53 nplc: 0.006 measurement time: +1.81023000E-01 measurements per second: 5.52 nplc: 0.02 measurement time: +2.42497000E-01 measurements per second: 4.12 nplc: 0.06 measurement time: +2.45999000E-01 measurements per second: 4.07 nplc: 0.2 measurement time: +2.56037000E-01 measurements per second: 3.91 nplc: 1 measurement time: +3.41034000E-01 measurements per second: 2.93 nplc: 10 measurement time: +1.06043500E+00 measurements per second: 0.94 nplc: 100 measurement time: +8.26029900E+00 measurements per second: 0.12 2W resistance, auto zero on, auto range, offset compensation on nplc: 0.001 measurement time: +1.20468000E-01 measurements per second: 8.30 nplc: 0.002 measurement time: +1.20568000E-01 measurements per second: 8.29 nplc: 0.006 measurement time: +1.20868000E-01 measurements per second: 8.27 nplc: 0.02 measurement time: +1.62189000E-01 measurements per second: 6.17 nplc: 0.06 measurement time: +1.65692000E-01 measurements per second: 6.04 nplc: 0.2 measurement time: +1.75729000E-01 measurements per second: 5.69 nplc: 1 measurement time: +2.60727000E-01 measurements per second: 3.84 nplc: 10 measurement time: +9.80327000E-01 measurements per second: 1.02 nplc: 100 measurement time: +8.18019100E+00 measurements per second: 0.12 ######################################################################################### 4W resistance, fixed range, offset compensation off nplc: 0.001 measurement time: +6.87000000E-04 measurements per second: 1455.60 nplc: 0.002 measurement time: +7.27000000E-04 measurements per second: 1375.52 nplc: 0.006 measurement time: +8.47000000E-04 measurements per second: 1180.64 nplc: 0.02 measurement time: +1.24600000E-03 measurements per second: 802.57 nplc: 0.06 measurement time: +2.64700000E-03 measurements per second: 377.79 nplc: 0.2 measurement time: +6.64900000E-03 measurements per second: 150.40 nplc: 1 measurement time: +4.06490000E-02 measurements per second: 24.60 nplc: 10 measurement time: +4.04097000E-01 measurements per second: 2.47 nplc: 100 measurement time: +4.04096900E+00 measurements per second: 0.25 4W resistance, fixed range, offset compensation on nplc: 0.001 measurement time: +5.03000000E-03 measurements per second: 198.81 nplc: 0.002 measurement time: +5.11000000E-03 measurements per second: 195.69 nplc: 0.006 measurement time: +5.35000000E-03 measurements per second: 186.92 nplc: 0.02 measurement time: +6.14800000E-03 measurements per second: 162.65 nplc: 0.06 measurement time: +8.95000000E-03 measurements per second: 111.73 nplc: 0.2 measurement time: +1.69540000E-02 measurements per second: 58.98 nplc: 1 measurement time: +8.59540000E-02 measurements per second: 11.63 nplc: 10 measurement time: +8.12850000E-01 measurements per second: 1.23 nplc: 100 measurement time: +8.08659400E+00 measurements per second: 0.12 4W resistance, auto range, offset compensation off nplc: 0.001 measurement time: +7.46690000E-02 measurements per second: 13.39 nplc: 0.002 measurement time: +7.47290000E-02 measurements per second: 13.38 nplc: 0.006 measurement time: +7.49090000E-02 measurements per second: 13.35 nplc: 0.02 measurement time: +7.55070000E-02 measurements per second: 13.24 nplc: 0.06 measurement time: +7.76090000E-02 measurements per second: 12.89 nplc: 0.2 measurement time: +8.36120000E-02 measurements per second: 11.96 nplc: 1 measurement time: +1.35612000E-01 measurements per second: 7.37 nplc: 10 measurement time: +4.95611000E-01 measurements per second: 2.02 nplc: 100 measurement time: +4.09560800E+00 measurements per second: 0.24 4W resistance, auto range, offset compensation on nplc: 0.001 measurement time: +1.60294000E-01 measurements per second: 6.24 nplc: 0.002 measurement time: +1.60394000E-01 measurements per second: 6.23 nplc: 0.006 measurement time: +1.60694000E-01 measurements per second: 6.22 nplc: 0.02 measurement time: +1.61691000E-01 measurements per second: 6.18 nplc: 0.06 measurement time: +1.65194000E-01 measurements per second: 6.05 nplc: 0.2 measurement time: +1.75199000E-01 measurements per second: 5.71 nplc: 1 measurement time: +2.61199000E-01 measurements per second: 3.83 nplc: 10 measurement time: +9.81197000E-01 measurements per second: 1.02 nplc: 100 measurement time: +8.18119100E+00 measurements per second: 0.12 ######################################################################################### Frequency, fixed range aperture: 1 ms measurement time: +1.10000000E-03 measurements per second: 909.09 aperture: 10 ms measurement time: +1.10000000E-02 measurements per second: 90.91 aperture: 100 ms measurement time: +1.10000000E-01 measurements per second: 9.09 aperture: 1 s measurement time: +1.10000000E+00 measurements per second: 0.91 Frequency, auto range aperture: 1 ms measurement time: +6.40400000E-01 measurements per second: 1.56 aperture: 10 ms measurement time: +6.50300000E-01 measurements per second: 1.54 aperture: 100 ms measurement time: +7.49300000E-01 measurements per second: 1.33 aperture: 1 s measurement time: +1.73930000E+00 measurements per second: 0.57 ######################################################################################### Capacitance, fixed range range: 1 nF measurement time: +1.00000000E+00 measurements per second: 1.00 range: 10 nF measurement time: +1.00000000E+00 measurements per second: 1.00 range: 100 nF measurement time: +1.00000000E+00 measurements per second: 1.00 range: 1 uF measurement time: +1.00000000E+00 measurements per second: 1.00 range: 10 uF measurement time: +1.00000000E+00 measurements per second: 1.00 range: 100 uF measurement time: +1.00000000E+00 measurements per second: 1.00 Capacitance, auto range measurement time: +1.00000000E+00 measurements per second: 1.00
From these numbers, it is quite easy to see the effect of having the multimeter switch in different circuits. For example, when autozero is enabled, the multimeter makes two measurements - one with the input and another without. However, this drops the reading rate at 0.001 PLCs from 50,000 a second to 2,049 a second, showing that the time required for the multiplexer to switch the inputs is 448 us. Whilst knowing the exact switching time is not particularly useful, having access to the measurement speeds is very useful to be able to get the best performance out of the multimeter. For example, knowing that autoranging slows the measurement speed down by up to 31 times, means that you are more likely to use a fixed range where possible or the autorange once feature (SCPI command).
Interface usability
All of the TrueVolt series use the same firmware, with certain features only enabled on certain models. This has the large advantage that using the 34470A is very familiar to the 34461A. There are many places where the interface differ where the 34470A has more options or additional features over the 34461A. Such places include the acquisition menu, which has a mode option to support the new data logging and digitising modes, and the entering of PLCs, with the 34470A allowing either the number of PLCs to be entered (from a fixed list) or the integration time to be set manually (in milliseconds). It is a little bit of a shame that the number of PLCs is not listed along the soft buttons, as it is for the 34460/61A, but this change allows more flexibility, as a user can enter an aperture size that is not fixed to PLCs.
Autocalibration
The autocalibration feature is an important part of the 34470A, with its regular use ensuring the multimeter stays within its accuracy specifications. It is particularly important for when the multimeter is placed in a different thermal environment to the one it was calibrated in.
Typically, test and measurement equipment is calibrated at 23 to 25 °C, so when they are used at an elevated temperature, for example inside an equipment rack or a hot lab, there will be additional errors that degrade the accuracy specifications. Whilst these errors are stated as part of the accuracy specification, using autocalibration compensates for the temperature change, resulting in no loss of accuracy.
This decrease in accuracy with temperature is due to the multimeter’s components having temperature coefficients - that is, the value they output changes with temperature. A simple example of this is a resistor, where the resistance is typically proportional to temperature (i.e. the resistance increases with temperature). For a nichrome resistor (which has quite a low temperature coefficient), an increase of 10 °C will result in the resistance increasing by 0.4%. For a precision instrument, this change is very significant.
There is also another source of accuracy degradation that should be familiar to those that have used test equipment - time. As electronic components age, they can also change value, which is sometimes referred to as the components settling. There are many reasons why this happens (such as internal stresses of a component decreasing with time and the effects from the thermal shock of soldering), but we are not interested in why it happens, we just need to realise that it will affect a multimeter’s measurements. It is for this reason, combined with the fact that some components under load will also change value, that manufacturers, like Keysight, provide different accuracy specifications for different amounts of time after calibration.
By far the most important component inside a multimeter is its voltage reference, which is used as a basis for every measurement. To ensure accuracy and precision, this reference has got to be as stable as possible, with all high-end (6.5+ digits) multimeters using ovenised voltage references to keep the temperature as constant as possible (typically around 80 to 90 °C). The ovenisation, unlike that of crystals, typically consists of a heater as part of the chip’s die. However, even these drift with external temperature and, of course, time.
As a little bit of an example, Linear Technology’s LM399 is the go to voltage reference for 6.5 digit multimeters. It has a heater on chip and a plastic thermal shield around its metal can package. With all of this, it is able to have a typical temperature coefficient of 0.3 ppm/°C and a maximum of 2 ppm/°C. This means that for a 10°C temperature change, a measurement of 10 V will be off by 3 to 20 uV just due to the voltage reference - this is 3 to 20 least significant digits on a 6.5 digit display.
The 34470A, like most 7.5 and 8.5 digit multimeters, uses a Linear Technology’s LTZ1000A, which has a typical temperature coefficient of 0.05 ppm/°C. This is an order of magnitude greater, but for this 10°C change, it still results in the least significant digit change by 5. Both of these errors would also be added to the errors of the other components, resulting in an even greater error being shown in the measurements.
To compensate for this, the 34470A contains a very high precision, hermetically sealed resistor that has a very low temperature coefficient - a Vishay VH102Z, which has temperature coefficient of 0.2 ppm/°C, a load life stability of 50 ppm and a shelf life stability of 6 ppm. As far as the multimeter’s concerned, the value of this resistor will not change with temperature or time (particularly as the resistor is only used intermittently with a small load). When the multimeter is calibrated (for example at the factory), the resistance is measured and stored in the multimeter's nonvolatile memory. Then during the autocalibration cycle, the multimeter measures this resistance and then compares it against the calibration value. Any differences between the values can then be used to adjust the multimeter’s readings. This will cancel out the temperature effects of the analogue circuitry and will also cancel out any drift due to ageing.
It is important to realise that whilst autocalibration will cancel out a large proportion of the error, it is not a replacement for a full calibration and adjustment (done by either Keysight or a calibration lab). This is because both the voltage reference and the autocalibration resistor will drift with age, which cannot be compensated for.
As different paths are taken inside the multimeter for different ranges, during autocalibration the multimeter will measure this resistor several times to calculate the adjustments required for each range. Overall this process is pretty quick, taking around 15 seconds, and can be done without removing any of the input connectors (i.e. you do not have to remove your probes).
There is one thing I wish to see as a firmware addition - the accuracy specifications are valid only when the autocalibration has been performed less than 48 hours ago and within +/- 2 °C for the 34465A and +/- 5 °C for the 34470A. So it would be nice if there was an icon that appeared towards the top of the display that indicated if the autocalibration was required to be run again - either due to it being older than 2 days or because it was outside of the accuracy temperature range. Then a user could easily see that they needed to perform the autocalibration again, without having to look at the calibration information, in the utility menu.
Keysight, or HP as it was then called, have designed a multimeter previously that had an autocalibration feature - the 3458A. The 3458A is a very special multimeter, in that the autocalibration is used to calibrate all the ranges and functions from only two references - a 10 kOhm resistor and a 10 V reference. Unfortunately, the 34470A’s autocalibration is not as far reaching as the 3458A’s is, with the autocalibration only adjusting the DC voltage and resistance measurements. A full calibration of a 34470A requires 14 voltages, 13 currents, 7 resistors and a 10 kHz sine wave. Whilst the 34470A has more features, I feel this illustrates quite well the extent of the differences between the autocalibrations.
The reason why the 3458A can calibrate itself from only two references is due to its ADC having a very high linearity, which unfortunately comes with its higher cost, as the analogue circuitry is of a higher quality.
Web interface
All of the TrueVolt range multimeters offer a web interface, which allows you to remotely monitor and control the multimeter.
This is quite useful, even if the interface is a little basic, as it allows pretty much full control over the multimeter. Measurement data, including the histogram data, can be displayed, but there is no export option.
If you want to save the data, you have to highlight it all then press Ctrl-C to copy it (as the text field does not have a right click menu), which is a little bit of a pain.
One of the nice features of this interface is its monitor / observe-only mode - where the multimeter is not placed into remote mode. This allows you to monitor the progress of readings, without having to set them up remotely. Additionally, this works with both the data logging and digitising modes.
This remote interface also offers a nice way to view system information (left) and for you to control and set up different states (right),
including setting a state to be recalled when the multimeter boots. Although this is all accessible by the multimeter’s front panel, this provides a quick way of getting hold of the information. However, the only way to export the system information is via copy and paste, which is not particularly ideal.
An additional feature of the web interface is the ability to send SCPI commands directly to the multimeter - which is equivalent to connecting via sockets or telnet. In the picture below, I performed a standard ID request and also asked the multimeter for its model number.
However, whilst this web interface is very useful, it has one major drawback - it uses Java. Currently, the support of Java web applets by browsers is declining, with both Google Chrome and Microsoft’s Edge browser not supporting them (see https://java.com/en/download/faq/chrome.xml and https://www.java.com/en/download/faq/win10_faq.xml). The only common browsers which support Java applets are Firefox, Internet Explorer and Safari (the screenshots above were taken in Firefox). According to statcounter (http://gs.statcounter.com/#desktop-browser-ww-monthly-201512-201602-bar), this limits the web interface to approximately 36.4% of users (with Chrome accounting for almost 58%), which is a terrible shame. It would be nice, if this Java dependency could be removed, with the web interface re-written into something more compatible. However, web browsers do not allow raw socket connections (from javascript) for security reasons, which means the only way of implementing the same functionality is to wrap the SCPI commands into HTTP requests.
Conclusion
Overall, the 34470A fits quite well in with the other 7.5 digit multimeters - it is the cheapest, whilst still having accuracies that are consistent with its competitors. Keithley’s DMM7510 betters its specification in all areas, has a touchscreen and a 1 MS/s digitiser, but costs £800 more that the 34470A. Whilst I have only seen videos of the DMM7510, its touchscreen interface seems clunky, with the 34470A’s interface appearing to be significantly better designed and easier to use. In addition to this, the 34470A is designed in a very elegant way and has a really good build quality. Whereas the DMM7510 looks like Keithley tried to force all the components into the smallest box they could find - which would make me a little concerned on how it is likely to last and age.
So now to the important part - based on design and usability, I would choose the 34470A hands down. However, I feel Keysight have hampered the 34470A by its supporting components. Whilst a multimeter’s accuracy is always going to be limited by both its voltage reference / DAC and its supporting components (i.e. current source and current sense resistors), it should be a shared limitation. However, with the 34470A this limitation is largely due to the supporting components, with the only advantage the 34470A has over the 34465A being its DC voltage accuracy.
This is not to say that the 34470A is a bad 7.5 digit multimeter - it is typical, but typical with multimeters that were first released in 1993 and 1996 (Keithley’s 2001 and 2010 respectively). To have a new multimeter that does not significantly improve on a 20 year old product is a little disappointing, especially when their competition is.
Whilst the extra digit is very useful, such as when you have to use a fixed range (i.e. measuring current without changing the burden voltage and when digitising), it is difficult for the vast majority of people to justify spending double the cost of the 34465A for an extra digit and a two-fold increase in DC accuracy. Whilst there are certain applications where the increase in accuracy and the extra digit would be required, these are few and far between. In these applications, I think a lot of people would be tempted with the increased accuracy of the DMM7510, even with its large price tag.
Which leaves me in the position of thinking, or perhaps hoping, that Keysight will release a successor to the 34470A. This would still use the same ADC as the rest of the 3446xA series, but include a larger autocalibration that reduces the error in more of the measurements.
Things I would like to see
