Keysight 34470A and Texas Instruments DAC8734EVM - Review

Table of contents

RoadTest: Keysight 34470A and Texas Instruments DAC8734EVM


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?: Honestly, there is not much out there to compete with the 34470A except for the Keithley 7.5-digit DMM.

What were the biggest problems encountered?: The DAC8734EVM could have shipped with a USB-to-SPI bridge cable [they are available] and some demo software to go along with that, but it's a great product for the price.

Detailed Review:

-------------- 34470A ------------------


I will start with the 34470A first.  When it first arrived and I took it out of the box, I was stunned by how small it was in comparison to my 6.5-digit meters [HP, Keithley, Fluke].  It also was quite heavy for it's size, and gave me the impression of the extreme high quality of the old HP instruments, when Bill and Dave still ran the company.  This instrument simply oozes "quality", and I am quite impressed that they did not go the way that everyone else is going-- to make cheap plastic toys.  This DMM really has the look and feel of a pro-quality DMM.


The LCD is large, and the screen is very readable even in a bright room from extreme angles.  It is not a touch-screen, which is fine with me-- I don't like touch screens as they end up getting fingerprints on them, and it is just one more thing to fail down the road.


The menuing system is very well thought out, and I have yet to actually read the manual for operational advice.  This DMM has nice feature though that if you press and continue to hold a button, a context-sensitive help screen will pop up to describe how you use that button.  The help screen goes away when you release the button [but the function is not activated until you press the button normally].  This allows you to explore the buttons and the menu without committing any actions until you want to.  This help feature also works with the shift-key functions too.


I ran the DMM for several days [in graphing mode] while it was connected to my Fluke 732B, and even though my lab varied about +/-5C during this time, the indicated voltage never varied more than about +/-0.5ppm.  This was without autocal.  I suspect with autocal, the variation on the 10V scale would be minimal.  There were no popcorn noise artifacts like you might see with a DMM that has an LM399 reference.  With this kind of performance, you can absolutely rely on 1ppm accuracy for periods of at least a month without even using the autocal feature.  Very impressive.


I also checked the linearity against my 3458A, and the 34470A is quite impressive [less than 0.5ppm]-- much better than specifications [1.5ppm].  This makes the "last digit" only a little bit better than a "marketing digit", but remember this is INL, which applies if you are in the same range, but the two readings are far apart.  The DNL, is about 0.2ppm [or maybe better, but I could not tell due to noise limitations]-- which applies if you are looking at two voltages that are nearly the same.  It seems that Keysight is being very [if not overly] conservative on these specs.  Maybe they are waiting for some time in the field, and then they can re-visit the specs to see if they can change the documentation.


I did not check if autocal incorrectly modifies the 100V and 1000V scales as Dr. Frank found.  I suspect that this is caused by the autocal not applying 100V or 1000V to the divider, and is instead using a lower voltage, which does not duplicate the change in resistance [or higher dissipation] in the main divider resistors with applied voltage.  This is something that can be corrected for in a future release of the firmware, but for now at least these ranges are still within spec after an autocal.


My unit came with all of the options turned on, so there was no way to test if the data can be streamed to the USB flash drive without the memory upgrade [but I suspect you can't].  Maybe the manual can clear that up.  If someone else has one of these DMMs without the memory upgrade, maybe they can post a test of data streaming speed to the USB drive.


One thing that I'm a little disappointed with is that the maximum AC input voltage is 700Vrms [providing a peak of 1000V with a sinewave].  Even my Tektronix DMM4050 [a re-badged Fluke 8846A] has a 1000Vac input.  My 3458A also has this same limitation [without a special option to allow higher AC voltage], so they match in their limitations.  In this new design, they could have gone to some extra effort on the AC voltage measurement capability I think-- but they took the "me too" route of just using an analog RMS converter like almost everyone else is using.  For really serious work in AC voltage, one needs to use a thermal converter anyway, so perhaps this is a bit harsh of a criticism.


I also noticed that when I have the statistics screen up, and the converter is in the middle of a conversion, when I press the "clear measurement" key to clear out the statistical values-- the partial conversion then becomes part of the next statistical calculation.  This is not a good thing if you know the input is changing at that moment.  I think the "clear statistics" key should throw away any measurement that is in progress-- [go ahead and complete the cycle, but just throw the value away, and start the statistics on the next measurement].


One of the things that I really like is that I can upgrade the firmware myself without having to send the unit into a service center.  You just put the firmware upgrade file on a USB flash drive, insert it into the DMM, and reboot the unit, and the firmware will automatically be updated.  My Tek DMM4050 must be sent back to the service center for a firmware upgrade, and it is not cheap.


I've tested pretty much every function on the meter, and everything works pretty much as you would expect, except that you get so much more capability with the new LCD screen and user interface.  This meter is approaching something that you could use in metrology, but without the high price tag.  I think it fills a niche that was lacking in the Keysight lineup, and Keysight engineers have answered the call with a very elegant design.  Good job Keysight!


All in all, I'm very impressed with this DMM [which is saying a great deal coming from me], and I would definitely recommend this DMM if you are looking for something better than your average 6.5-digit meter, but do not need the super high linearity [or cost] of a 3458A.  I will keep this meter until I retire I expect, as it is very useable on the bench.  I will probably end up selling my DMM4050, as it is now redundant.


--------------  DAC8734EVM  -----------------


The T.I. DAC8734EVM is a *great* way to go if you need 4 analog outputs that have 16-bit resolution.  In fact, there is no reason at all that you could not use the evaluation module in an actual product.  It is very high quality, and the reference design does not degrade the INL or DNL of the DACs.  I tested the module for transient response, and it's output conformed closely with the DAC's datasheet specs.


My only complaint is that it did not come with an easy way to drive the DAC "SPI" interface.  I have some ARM-Cortex-M3 based boards that I hacked up to drive the board, but that might be too much work for someone that just wants to evaluate the DAC.  I think this module should come with [or T.I. should make available] some kind of USB-to-SPI bridge cable, and maybe with some already done demonstration software.  That would make this module usable by far more people.


This board is very nice though-- 4 independent outputs, and very fast updates [to the limit of the SPI interface].  I found that I was more limited speed-wise by the cabling and interconnections between boards than I was with the SPI interface on either the DAC board or the ARM board.


I would definitely recommend this board [or the DAC chip that it is based on] if you a lot of high-resolution analog outputs.