EEVblog 1382 - Keysight EDU34450A 5.5 digit Bench Multimeter TEARDOWN

[tszaboo]

Another question is do student labs need 5.5 digit multimeter? We used venerable Mastech M830 in our physics labs, and it worked a charm. Typical labs don't need anything better. I'm not advocating for M830, of course, but if you are spending more than $100, you are probably just wasting money.

I can see there may be a need for improved performance in a research lab, but why bother with crippled EDU versions then? Just buy real equipment.

The labs we had were equipped with HO 54000 series scopes, 34401A, their test gear in general. Some of the labs were supported by companies they had newer gear. It does makes a difference, because after I got out, I could use the newer Agilent stuff, everything was where I thought it will be. It also set a standard in my mind, so when I saw a Tektronix TDS 2000 series, I saw how much hot trash it was.
If you have new high end gear in the lab, you can see how great it i s to have good tools.
If you have old high end gear in the lab, you can see how dependable and reliable these tools are.
If you have low end gear in the lab... That sends the wrong message.

[ataradov]

That's all great from a company perspective. What do I care as a university if you like some vendor gear after you graduate?

Obviously of Keysight bundles that thing for free, it is fine. I just don't see the point of considering buying this outside of the bundled deal.

[maxwell3e10]

Regarding 50/60 Hz rejection from the spec sheets of the meters:

EDU34450 NMR (normal mode rejection)  For 60 Hz (50 Hz) ± 0.1%
Slow mode 5½ digits, medium mode: 4½ digits, 60 dB
Fast mode: 4½ digits, 0 dB

34450 NMR (Normal mode rejection) For 60 Hz (50 Hz) ± 0.1%
Slow mode 5½ digits 90 dB
Medium mode 4½ digits 55 dB
Fast mode 4½ digits 0 dB

34465
Integration time    Normal mode rejection 1
≥ to 1 PLC                    60 dB 2
< 1 PLC                         0 db

So, according to spec sheet EDU34450 should have decent suppression of 50/60 Hz noise on medium or slow reading speed. Perhaps it wasn't properly set to 50 Hz?

But I agree that "fast" "medium" "slow" are annoying specifications, one should be more specific and, if used for education, teach engineering students about NPLC.

[wizard69]

Another question is do student labs need 5.5 digit multimeter?

Personally I think the answer is no but a qualified no.  Students need understand when precision and resolution matters which is something not covered well in many schools.    In my opinion putting a 5.5 digit meter on every students bench is a mistake.   On the flip side students need exposure to an array of instruments. 

We used venerable Mastech M830 in our physics labs, and it worked a charm. Typical labs don't need anything better. I'm not advocating for M830, of course, but if you are spending more than $100, you are probably just wasting money.

Well we could discus the optimal price forever however a 4.5 digit  meter is more than enough to get students off to a good start. 

I can see there may be a need for improved performance in a research lab, but why bother with crippled EDU versions then? Just buy real equipment.

Exactly!   

What EDU needs is a low cost meter that exposes the students to as many potential measurement techniques as possible. 

[floobydust]

I'd do a test and see if the mains (normal-mode) hum rejection is actually working. Should be a zero-cross detector on the board so the DMM can figure out line freq. as its predecessors did.
The DMM seems to spaz out way too much? I also don't find these Keysight bench multimeters (34461a, 65a) a good drive. The readings are all over the place, so much flicker it's like a slot machine in a casino to read them even with a dead quiet source like a battery. It's like it's picking up tons of strange noise that actually isn't external to the DMM.
Another test I would do is look at the common-mode noise the DMM has at the (-) input, with a scope or spectrum analyzer. You can a put a 10nF from (-) input to earth-ground to stop HF CM noise, assuming it doesn't already have that as a Y-cap from the floating A/D to PE.
I was disappointed to see Keysight copied ANENG by using Hongfa relays  and a 10k resistor across the coils seems weird.

A problem is the DMM looks expensive to manufacture - PCB with SMT on both sides, through-hole and flying leads, tons of manual assembly with the AC mains wiring, fuse etc. and a kilogram of shields over shields.
Instead of lowering cost by using cheaper components, a cost improvement project would first lessen manufacturing labour costs instead of relying on china for assembly.
Recycling the old enclosure and sheet metal saves development costs (NRE) but in the end it might add too much fat because it's not ideal - the power transformer +wiring are sources of hum but poorly shielded from everything and not away from the analogue board.

[gamalot]

I was disappointed to see Keysight copied ANENG by using Hongfa relays  and a 10k resistor across the coils seems weird.

Hongfa is not as bad as you think. They obtained UL certification in 1987 and acquired Hella's relay business at the end of 2019.

https://www.hella.com/hella-com/en/press/Company-12-11-2019-18428.html

[tszaboo]

That's all great from a company perspective. What do I care as a university if you like some vendor gear after you graduate?

Obviously of Keysight bundles that thing for free, it is fine. I just don't see the point of considering buying this outside of the bundled deal.

They wanted to send students away with usable knowledge. It wasn't just theory. There is also a large pressure from automotive companies to teach communication protocol, like CAN and Flexray. It reduces the amount of training a graduate needs before being useful.
The class "Measurement technology 1" for example started by explaining how a scope works. And the underlying mathematics of error propagation and statistics.

[madires]

30 years ago the T&M lab of my university was equipped with typical stuff you'd have also found in commercial EE labs. And the communications lab had some fancy stuff like ISDN protocol analyzers. No special educational devices, all standard T&M. Not the latest and greatest, but devices which get the job done. And if you want to get students "ready for job" you should train them with standard T&M anyway. I don't see any point for overpriced educational T&M. A "my very first DMM" for a kid would be fine, but not for EE students. For hobbyists/makers a standard hand-held DMM is much better suited.

[Kleinstein]

For most part there is no special need for extra educational series. At least the 34450EDU is not especiall overpriced - it is some $200 less than the old version. There are a few cases (.g. RF gear) were extra robust instruments for educational purpose may be helpfull.

Other users may find those versions useful too - the difference is not that large and the form-factor may also fit a commercial bench.

There are companies to sell special instruments for schools. This is extra robust, extra safe (for use by the students) or with an extra large display, so that students can see it from a distance. These instruments may look overpriced (and sometimes underperforming) compared to normal T&M.

 

But I agree that "fast" "medium" "slow" are annoying specifications, one should be more specific and, if used for education, teach engineering students about NPLC.

A more accurate manual would definitely help, so one would know how the measurement actually operates. Most users would not mind, but it sometimes makes a difference.  It does not hurt to tell what the actual integration time and measurement speed is.

[rsjsouza]

I personally don't dislike the form factor. This can potentially reduce the depth of benches which, at least in places located in prime realstate, is significant to pack more seats per m². Also, it may well be a way for Keysight to test the waters to accept a new form factor - the circuitry of such devices does not demand the amount of bench space of its older brethren and therefore it may set a trend at least for a segment of the T&M market.

The internals are alright as well, although the reference is not the best and I hope the binding posts have some back support from the plastic enclosure - otherwise the PCB will be broken in no time.

The large display is very handy, but I didn't spot a "help" or "tutorial" button à lá Tektronix educational line. This could be handy for instructors and not a bad idea given the size of the display.

Being a 5-1/2 meter can help the instructor explain the various differences between accuracy, precision, resolution all in the same equipment. I see the value in this.

I obviously dislike the price for a single seat, although compounding it to an entire lab has significant savings when compared to the more professional tools from Keysight (Rigol/Siglent is another story, though). The lack of customizations and built-in trend/analysis tools, as well as the SW being a trial version, are bad ideas.

[jusaca]

but I didn't spot a "help" or "tutorial" button à lá Tektronix educational line. This could be handy for instructors and not a bad idea given the size of the display.

This is actually one of the things Keysight is doing really well - you can press and hold any of the buttons and you will get a popup on the screen explaining everything around the corresponding mode / parameter / whatever the button is doing.
See for example the photo I attached: hold the button "HighRes" for something like 2s to get the popup. As far as I know, all their newer gear behaves like that.

[3roomlab]

Being a 5-1/2 meter can help the instructor explain the various differences between accuracy, precision, resolution all in the same equipment. I see the value in this.

I obviously dislike the price for a single seat, although compounding it to an entire lab has significant savings when compared to the more professional tools from Keysight (Rigol/Siglent is another story, though). The lack of customizations and built-in trend/analysis tools, as well as the SW being a trial version, are bad ideas.

i guess after adobe and altium
keysight also wants more subscription products

[chickenHeadKnob]

The internals are alright as well, although the reference is not the best and I hope the binding posts have some back support from the plastic enclosure - otherwise the PCB will be broken in no time.

The binding posts bothered me as well. I don't think the PCB's will break routinely. Instead if the board is allowed to flex I see the failure point happening at the solder fillet of the post, resulting in  dicky continuity and even more range-relay hunting Klackery.

[Kleinstein]

The mechanical load to the jacks / tubes could be a problem - actually split jack could have been the better choice as they are a bit more flexible. Still it would be mainly the solder joints that may turn bad - not sure how strong the lead free solder is. It least it would be relairable.


For the non populated BNC foot print where the ground lead is soldered, it looks like this was meant for some trigger input / ouput (e.g. measurement complete), but probably removed to save a few cents or not yet ready in time. The rear metalwork looks like different from the sope, as it does not habe the hole(s) for the fan.

I am a bit confused with the input bias current: with some 1.8 mV with 10 M input resistance this would be some 180 pA, so a relatively high input current.  I had expected better.

The large THT resistors for the AC input are a bit odd - I would more like expect a string of SMD ones. It could be just a take over from the older 34450 version.

From the specs sheet, the medium speed mode seems to be 1 PLC for DCV and DCI and 2 Wire Ohms.  Still odd to get less than 49 readings per second via LAN. The µC they have on ethe GND based side soes not look that slow.
Compared to the old 34450 they seem to have doen some cost cutting for the current ranges. Otherise it still looks very similar - so the numbering part is not that confusing.

[Keysight DanielBogdanoff]

My thoughts on this DMM. Disclaimer: I have the luxury of not needing to worry about price points or +- a couple few hundred bucks - my main personal experience is more about usability & capabilities.

I really like the big bright screen. I turned off auto-dimming almost immediately, but understand it could be nice for some folks since the screen is bigger and brighter than a typical DMM.

I've been able to pop it on my bench and use it without any issues and with a bit less of the usual 1st-time-user woes. I worked with Dave & our R&D on the relay thing and the beta firmware was much worse in that respect. For the NPLC stuff, the higher end DMMs have specific circuitry to detect and deal with that, this one doesn't for cost reasons. However, pulling it out of "fast" mode works wonders. As much as I like "fast" it's probably faster than I really need from my benchtop DMM, traditionally the super fast measurement frequency is more of a manufacturing line concern and I sort of prefer the longer averaging cycle (especially to get a higher NPLC).

Anyone think the BOM on that thing exceeds $100?

Handily, I've seen it.

[scopeman]

When I first saw this I thought I must have overslept and it was April 1st already.
From my point of view it would make a lot more sense for Keysight to offer near cost discounts on the real gear for university and even senior high school labs than bringing out such strange crippled gear for the educational market.

After all when you graduate and get into industry would you really buy the EDU stuff? I think not. When I was in college we used the real stuff
HP/TEK/Fluke/GR etc. When I got out of school I knew what I wanted on my EE bench.

Sam
W3OHM

[Kleinstein]

The fast mode for DCV / DCI is of limited use. Some 110 SPS can no longer suppress mains hum, but it cannot really resolve it either. So it is prone to hard to handle / detect errors from mains hum. The 190 SPS of the original 34450 is not that much faster, but about where a faster mode starts to make sense.

Fast readings make limited sense on the screen, but it can be useful for some data logging. So it would really make sense to write the data at some 50/60 SPS to memory or a USB drive. Just 1 SPS is rather slow for some experiments, though a good choice to get easy scaling.

Most younger users should not have too much trouble handling a few settings in the menu-system - so no real need to make it too simple / limited. It is still nice to have a dummy mode, but also nice to have control.

The EDU34450 is not that much different from the non EDU version: AFAIK less current ranges and lower grade shunts and slightl slower maximum speed. The performance range may be just the point were the chips for handheld DMMs reach there limit, and using seprate switches, amplifiers and ADC chip adds to the circuit and BOM. At least for the normal 5.5 digit meter it may be time for an upgrade.

[Neganur]

Despite liking the display size and formfactor in general, I am having a hard time not spending the 300 EUR more to get the 34460A instead (thinking for home desk use). Or maybe even two of Dave's EEVblog handheld meters.

Isn't the DCA range a bit weird too?

I don't know what education market this aims at, not seeing many DMM at all in universities here in Finland. And if, then more the handheld type than the desktop ones simply because they are easier to lock away after use... Additionally, for proper scientific measurements you'd need the proper tools anyway. Maybe colleges are different? And does the education sector not get pretty substantial discounts anyway?

[floobydust]

For student use, jack-detect to prevent a zillion blown fuses is obviously a good idea that was missed.
The auto-ranging algo sucks, have you not heard of successive-approximation? Why start at the bottom and work up, oh yeah to be the slowest in town. Hint: look at the middle middle middle.
Does diode-test display more than 3 sig digs on a 5-1/2 product? Why throw away resolution that was paid for, it's used for diode temperature use.
Can you make the font smaller and sillier on the (-) button please. And the beeper even more quiet. I like the big display, wish I could change font or colour.

edit: where's the Kensington lock

[free_electron]

this thing is cool but disqualified for one and one reason only : you can't stack em or rack mount em. game over.

[EEVblog]

this thing is cool but disqualified for one and one reason only : you can't stack em or rack mount em. game over.

Their target market doesn't stack'em and rack'em though.

[CChin254]

Design Wins:

Predicted Block Diagram:

IC Identification:

[Kleinstein]

KS offers some additional kit to stack them. It would likely still be limited (e.g. only 2 units and no power supply on top).

Isn't the DCA range a bit weird too?

I don't know what education market this aims at, not seeing many DMM at all in universities here in Finland. And if, then more the handheld type than the desktop ones simply because they are easier to lock away after use... Additionally, for proper scientific measurements you'd need the proper tools anyway. Maybe colleges are different? And does the education sector not get pretty substantial discounts anyway?

The fuse is not that sensitive - this is possible by using only 2 diodes (instead of a bridge) for the protection. On the downside this may limit the crest factor for AC amps and the usability for low currents. However the meter is far from providing fine current resolution, so the low current use is limited anyway.

The current ranges are a bit odd, but in parts due to cost savings: they use only 2 shunts and it looks like a seprate relative low noise amplifier (1/2 the AD8629). A low noise amplifier can use a single shunt and relay over a larger range - kind of makes absolute sense, as a low noise OP is cheap compared to extra shunts and relays.  I am a bit surprized to have the shunts with 0.1 and 1 Ohms. With the extra low noise amplifier I would have choosen more like 0.1 and 10 Ohms (or even 100 Ohms) to cover a larger range at not extra costs.  A 10 Ohms shunt could even be lower power.
A limited number of current ranges is common for cost savings and even some  digit meters like 34401 or Keithley2000 have this, though with more sensible values.

The large diodes make absolute sense, as they need to stand the current to worst case blow the fuse.

The physics / EE hands-on experiments at the university use quite a few DMMs. Here some bench meters could make sense. Because of costs there are still many handhelds. Bench top versions may be where higher performance is needed or where available (e.g. more like surplus or older units). Many cases could get away with a even lower grade version. Still the resolution is not so much driving the costs.

I would not consider the new meter exclusive for educational use.  Others can be happy with this meter too.

[Fungus]

Isn't the DCA range a bit weird too?

You mean the "DCI" range? 

[Neganur]

woops, Freudian slip