Hioki DM7276-01 Precision DC Voltmeter (9 PPM)

[HighVoltage]

This HIOKI DM7276-01 has been added to my lab.

It is advertised as a 9ppm/year 7 1/2 digit DC Voltmeter with touch screen a very small footprint and lightweight.

Some specs:
- 100 mV ( ±120.000 00 mV) to 1000 V ( ±1000.000 0 V ), 5 ranges
- High-accuracy model with 1-year 9ppm Accuracy
- Capacitance contact check (using built-in C-monitor)
- Supports global production with built-in variable power supply
- Built-in EXT I/O, LAN, and USB
- Outstanding Long-term stability and Temperature characteristics
- Excellent noise performance approaching an 8-½ digit DMM
- Specifications ideal for everything from R&D to production lines at ¼ the conventional costs

I am just starting using this instrument and the first thing that is noticeable is its very fast warmup time from around 3 min, maybe less.

Stay tuned, first measurements should be interesting.

Anyone here having experience with this instrument?

[MegaVolt]

A very interesting find !!
How much did it cost?

Can you share the documentation that is only available upon registration? "Instruction Manual", "Communication Command Instruction Manual".
I am already myself )

[MegaVolt]

Teardown? ))))))))))))))

[HighVoltage]

In regards to the documentation, here are the direct download links

HIOKI DM7275, DM7276 - datasheet english
Hioki product overview - Test equipment for batteries english

Hioki DM7275, DM7276 - Manual english

Application notes:
Effect of the measurement target output resistance and voltmeter input resistance on measured values
Measurement of the package potential of laminated lithium-ion batteries

[HighVoltage]

Here I tested the warmup time of the DM7276.

My LTZ1000 reference was warmed up well, so no influence from that.
This little peak shortly before 2 min seems normal for this instrument.
It seems within 2 min this DM7276 is warmed up to get reliable values.
And this is even faster than the Keithley DMM7510.

[HighVoltage]

In this picture the Hioki DM7276 was hooked up to my LTZ1000A reference over night
Settings:
- 100 NPLC
- Smoothing filter ON with 5 average readings

The nominal value of my LTZ1000 reference is 7.065,90xV
For comparison I had the Keithley DMM7510 running in parallel to the Hioki.

[MegaVolt]

It is twice as noisy as the 7510...

[Kleinstein]

It is twice as noisy as the 7510...

The factor is visible from the example data, but AFAIK the DMM7510 is very mixed perfomance when it comes to noise. For the data I have see, it is very good in the 100 mV range and also good with short integration like 1 PLC. However with 100 PLC the performance is not that great and a bit disappointing - at least with a lower applied voltage. With the test at 10 V the references could be a main factor.

Both meters are not really made for ultimate precision at 100 PLC. The main use is more like relatively fast production tests at 10 PLC or 1 PLC, often with a scanner.

[Kleinstein]

The Hoki meter is a voltmeter only - so no shunt and not current sources for resistance. This allows for a lower price.
Chances are the accurcy specs for the 10 V range are to a large part limited by reference drift. The external temperature regulation for the LTFLU in the DMM7510 has some disadvantage in this respect. The specs are also only estimates about future performance - the actual performance is usually better, but unknown upfront. Some estimates are more conservative than others.

[guenthert]

     It's even a DC-only voltmeter, which further simplifies the input stage and eases timing requirements (while the 3458A is often praised here for its precision and linearity, it was really its speed which puts it ahead, more so in '89).

     I was even wondering why they bother with a multi-range DC-only voltmeter.  Why not make it 10V only, simplifying the input stage even further (making it cheaper or more precise) and selling it as digitizer?  Seems they are eyeing the currently hot battery market.   

[aronake]

Do anyone have more experiences to share on this multimeter?

Maybe some teardown pictures?

Anyone know what voltage reference it uses?

Linearity?

With the collapse of the Japanese yen is is getting relatively inexpensive from Japan.

https://store.shopping.yahoo.co.jp/firstfactory/soku-hioki-dm7276-02.html

373K JPY is around 2.7K USD. Some freight etc added of course.

[aronake]

Similar price in Europe apparently.

https://www.calplus.de/hioki-dm7276-01.html

[HighVoltage]

While Keysight 34470A and keithley DMM7510 had a dramatic price increased in the last few years, this Hioki Voltmeter has been stable.

Since the warranty has expired on my DM7276-01, and if there is any interest, I could open it up and take some pictures from the inside.

[MegaVolt]

Since the warranty has expired on my DM7276-01, and if there is any interest, I could open it up and take some pictures from the inside.

It is very interesting

[HighVoltage]

It has an LTZ1000ACH on a separate small board with all SMD components.

First set of pictures

[HighVoltage]

Second set of pictures

[HighVoltage]

Last set of pictures

[HighVoltage]

Back together and still working.

[maxwell3e10]

In this picture the Hioki DM7276 was hooked up to my LTZ1000A reference over night
Settings:
- 100 NPLC
- Smoothing filter ON with 5 average readings
The nominal value of my LTZ1000 reference is 7.065,90xV
For comparison I had the Keithley DMM7510 running in parallel to the Hioki.

Thanks for the tests and teardown. Something seems wrong with Hioki displayed statistics. P-P cannot be less than twice S_n. P_P would be equal to twice  S_n if the distribution is purely bimodal, otherwise P_P should be more than twice S_n.

[laichh]

Here are the inside of a Hioki DM7275, I don't see any different compared against a DM7276, seems identical to me.

[laichh]

more Hioki DM7275 pictures.

[HighVoltage]

Thanks for the tests and teardown. Something seems wrong with Hioki displayed statistics. P-P cannot be less than twice S_n. P_P would be equal to twice  S_n if the distribution is purely bimodal, otherwise P_P should be more than twice S_n.

This was a quick and dirty setup, definitely not reliable.
Let me do a proper hookup and testing.

[Kleinstein]

The inconsistancy between the p-p and standard deviation values has nothing to do with the input signal it is a pure math thing.  For a given P-P value one gets the largest RMS value / standard deviation with a kind of square wave and this gets 50% of the p-p value. It is only the SD estimator for a small sample size that adds the extra factor of square root of (N/(N-1)) that can make it slightly larger.
The test description indicates more like a longer test.

[HighVoltage]

Here are the results of 100 measurements with 100 PLC and smoothing filter ON
The source is a stable LTZ1000A

[HighVoltage]

1.5 hours later and 1394 measurements later
Same setting

[maxwell3e10]

It is an impressive stability, only 1-2 uV. But the statistics is still wrong, maybe the smoothing filter messes it up. It also would be interesting to download the data and plot, see how many digits are used internally.

[iMo]

It does smoothing with "moving average" - how many samples do you average?

PS: What is the architecture of the ADC in that nice meter? We can see the AD7175 ADC there, as well as parts used in the Mutlislope (even though no fpga or mcu in the inguard?)..

[Kleinstein]

They say that they use a sigma delta ADC and chances are this would be the AD7175 and not an extra seprate build one. The AD7175 is a slightly off choice, more made for high speed, but still OK. The relatively fast ADC may be used to get still relatively fast trigger / start of a measurement. A slower SD ADC may have too much delay at the start.
Noise wise the AD7175 should be good enough, even with a background check of the ADC and ref. divder gain.

Chances are there is a µC for the ingurad part, as there is a 2nd crystal or clock generator besides the one at the ADC. Chances are the 16 pin chip next to it and close to the signal isolators would be something like a µC.

[iMo]

Unbelievably complex inguard board, however.. My first thought had been it is a kind of multislope with 7175 as the rundown something.. 

[Kleinstein]

The circuit indeed looks quite complicated for just a simple single channel DC voltmeter.
As an extra function there is the contact checking mode via capacitance, but I would not expect that much circuit for this either.

There some relatively large ceramic capacitors connected to the four ADG1211 switches. This looks a lot like switching capacitors for voltage division.
So this gives quite a few chips to replace some recision resistors.

It gets a little more complicated from having some kind of ACAL - though still no reason for that much circuit (e.g. 1 x ADG1208, 1-2 relays and some resistors should be enough).

[HighVoltage]

Here the measurements have been running over night with the same settings.
The smoothing filter is set to Average over 10 measurements

 

[iMo]

FYI noise with RAND() and 1uV resolution.

[DavidKo]

While Keysight 34470A and keithley DMM7510 had a dramatic price increased in the last few years, this Hioki Voltmeter has been stable.

Since the warranty has expired on my DM7276-01, and if there is any interest, I could open it up and take some pictures from the inside.

I like Hioki products, but for home user their products are not too much interesting. When the products are out of the support, they do not calibrate the device nor send the calibration/service manual (same experience with Japan and Czech support). Compared to mentioned Keysight or Keithley, where the calibration adjustment procedure is available.

[HighVoltage]

While Keysight 34470A and keithley DMM7510 had a dramatic price increased in the last few years, this Hioki Voltmeter has been stable.

Since the warranty has expired on my DM7276-01, and if there is any interest, I could open it up and take some pictures from the inside.

I like Hioki products, but for home user their products are not too much interesting. When the products are out of the support, they do not calibrate the device nor send the calibration/service manual (same experience with Japan and Czech support). Compared to mentioned Keysight or Keithley, where the calibration adjustment procedure is available.

That is very true.
Japanese companies in general do not like to give out service manuals or calibration procedures or schematics.
However their products are usually very good.

Keithley and Keysight are going more and more in the same direction and I would not be surprised, if they also would not supply calibration procedures in the future.
 

[ap]

Unless its broken, and given the fact that this Hioki is DCV only, one can get away with verification. This way the drift history is not covered either by re-adjustment.

[aronake]

The voltage reference board read BT4560 in the corner, indicating that it is the same voltage reference board as in BATTERY IMPEDANCE METER BT4560:

https://www.hioki.com/in-en/products/resistance-meters/battery/id_5897

There are sellers on Taobao who claim to have ADR1000 for sale.

https://item.taobao.com/item.htm?abbucket=5&id=761537983785&ns=1&spm=a21n57.1.0.0.5256523cWu7jic

They are on very similar voltage reference boards with an ADR1000 but read DM7275. So it looks like ADR1000 may be used in newer models, and as has been indicated that Linear is now selling it for companies using it in products.

[gamalot]

They chose OPA2140 on their LTZ1000 board.

[gamalot]

Sorry to go off topic again, are these resistor values 1.58K? I have never seen such top marking before. 

[Lajt]

If anyone is interested, I've made a DIY temperature probe for Hioki DM7275, DM7276 & RM3544.
The build is documented on https://lajtronix.eu/2024/02/21/diy-hioki-dm7275-76-temp-probe-z2001/
It has been tested and is working great.

Br, Lajt

[laichh]

Sorry to go off topic again, are these resistor values 1.58K? I have never seen such top marking before. 

1.58 kohm is correct, these resistors are likely made by KOA.

[gamalot]

Sorry to go off topic again, are these resistor values 1.58K? I have never seen such top marking before. 

1.58 kohm is correct, these resistors are likely made by KOA.

I have a lot of small light blue KOA Speer chip resistors and most of them don't have top markings, even the 0603 size ones.

[gamalot]

If anyone is interested, I've made a DIY temperature probe for Hioki DM7275, DM7276 & RM3544.
The build is documented on https://lajtronix.eu/2024/02/21/diy-hioki-dm7275-76-temp-probe-z2001/
It has been tested and is working great.

Br, Lajt

I happen to have a Z2001 sensor that came with my RM3545, maybe I can do some simple measurements on it.

[Alex Nikitin]

Well, I've purchased the DM7275-03 for my home lab. So far I am quite happy with what this small unit does. I like that it is completely quiet, seems to be very stable and as arrived did differ about -3ppm from my usual standards (which I count as about 10ppm accuracy in any case). The input current measures about -20pA and seems to be stable and not noisy. The temperature measurement feature is very handy and useful. I am already getting some good data on the low frequency noise of my HP3245A and other sources, with the passive filter the baseline noise is about 600nV p-p in 0.1-10Hz range at 1NPLC.

I have a question though - does anybody knows a calibration company in the UK which would calibrate this unit with traceability, data and uncertainties (to my surprise, the unit came without any calibration data from the manufacturer).

My next home project now could be to build a universal shunt box for the Hioki and the HP3456A, to measure DC currents from about 100uA FS to 1A FS. I have good 10K and 1R0 references but it would be much more convenient to have a box with proper range switching and fuses.

Cheers

Alex

[alm]

I have a question though - does anybody knows a calibration company in the UK which would calibrate this unit with traceability, data and uncertainties (to my surprise, the unit came without any calibration data from the manufacturer).

I have no suggestion for a lab, but you'll need a lab with sufficiently low uncertainties (check their scope of calibration) that doesn't mind doing something they may not have a procedure for. The calibration procedure in the manual (see attachment) should be straight-forward for any cal lab with the right equipment. So I don't see why it would be difficult for them to produce a traceable certificate with data and uncertainties. Adjustment may be more tricky, but do you really need adjustment? Can you not live with just knowing the offsets?

I'm a bit surprised there is no linearity check like a negative voltage or say 50% of range in the calibration procedure. I didn't study the design, but is the linearity guaranteed by design?

[Alex Nikitin]

I did some checks on zero volt, range and polarity change, and in general errors are not more than 1LSB (which for the range change could be 1ppm of value, measuring +/-1V at 10V range). The zero drift on 100mV range with a shorted input is below 50nV for 10min, the offset noise on Slow speed is ~350nV p-p (and that is without offset correction). Which means this small thing might be as good a Null-Meter as the HP3458A (and might be even better as it stays cool unlike most accurate bench meters).

On calibration I am not worried too much especially as the difference between the meter at 10V and my own references is slowly getting smaller in time and now below 2ppm. However it might be nice to have a way of adjusting the unit in the future, and that should increase the value, as HIOKI policy on second hand units is simple - they would deal only with the original owner, as far as I understand.

Cheers

Alex

[Alex Nikitin]

I am already thinking about building an "extender" for the Hioki, a box with relays and resistors controlled from the I/O connector of Hioki and providing DC current and resistance measurement options. It is not that Hioki can not measure DC current on its own (see the photo below  )

Cheers

Alex

[Kleinstein]

I'm a bit surprised there is no linearity check like a negative voltage or say 50% of range in the calibration procedure. I didn't study the design, but is the linearity guaranteed by design?

It is indeed odd that the calibration points are only 0 and full scale. Checking the negative side would be good and not that difficult. It's not a strickt INL test, more a crude check the amplifier / supply is not broken.

The ADC is an AD7175 high end SD ADC chip. So the linearity can be pretty good and normally no separate adjustment needed for the positive and negative side. If the design is OK it is unlikely the the meter would develope high INL over time.
At the expected linearity level it is anyway a bit tricky to check, as the specs for the usual calibrators are limited (the actual performance may be better).

[Alex Nikitin]

DM7275/6 has an interesting feature, called scaling, where it is possible to use a linear scaling factor and offset, plus rename the units if required. The scaling coefficients A and B (for A*x +B expression) can be entered from the front panel or remotely with 7 digits precision and later the combination can be saved as a setup and recalled when needed. In essence it makes the unit very convenient to use with current shunts, as long as the shunt is stable enough, the tolerance doesn't matter much as it can be corrected out. The unit also offers a (linear) parametric correction from temperature sensor readings, up to +/-1000 ppm/C. I did make a nice 1 Ohm reference recently using a precision hermetic high-current JRL 1 Ohm resistor, the same resistor as evaluated on xDevs. The resistor I've used measured about + 560 ppm and I just found almost exact correction resistor value. However with the HIOKI one can use one of these JRL resistors as is and just calibrate the error out by scaling when use it as a shunt to measure current.

Actually the scaling can be used as a user calibration option to correct for the reference drift if needed one day .

Cheers

Alex

[Alex Nikitin]

I did some measurements on the capacitive coupling of my bench meters to the mains ground (i.e. to the meter case in all three units) for another thread on the forum here. The newcomer came on top by not a small margin, to my surprise:

Hioki DM7275-03

Positive input to mains ground - 10V range and below <30pF, 100V and 1000V <10pF
Negative input to mains ground  <30pF
Between input terminals  ~1500pF for 10V range and below*, <10pF for 100V and 1000V ranges.

Here I can add that I also measured the DC leakage to ground from measuring terminals with my Keithley 617 electrometer: for +100V above the ground the leakage is 2.5pA, for -100V it is -6pA.

Cheers

Alex

* - possibly something to do with the contact checking feature, even if it is OFF. The cure is simple (if the input C is a problem) - to put a 10K resistor in series with the positive lead and use Hi-Z, on 10V and below the meter has ~20pA input current or ~0.2uV added error with the resistor.

[Alex Nikitin]

Six days continuously powered on, the HIOKI is settlings down nicely, the difference with my reference meters and Fluke 731B is now essentially gone, the initial difference of about -3ppm has disappeared in about 3 days time and the meter appears to be stable - however I will continue to monitor its stability. Here just as an illustration the HIOKI measuring the Fluke 731B for 60 min when the temperature in my home lab was reasonably constant (the TC of 731B is about -2ppm/C, so the 0.2ppm rise over 60 min means the temperature has dropped by about 0.1C ). The red line is 30s running average, vertical scale is 1uV = 0.1ppm /div. It is good to see two low noise references (LTZ1000A and RefAmp) working well together.

Cheers

Alex

P.S. - I've corrected the graph's title, the measurements were made at NPLC10, 1s intervals, not NPLC1 .