Milliohm Meter Kit from Ascel Electronic Review

[Martin72]

Hi,

Last month I´ve ordered this kit here:

AE20218.

I´ve choosen all order options, so it comes with the suitable housing, a usb interface, kelvin clamps and power supply.
All smd devices are soldered before they claimed - well, this means exactly 3 components...
The rest is to mounting and soldering by yourself, by following the instruction guide (component placement).
This takes about an hour(pic1, pic2).
The correct mounting of the 4mm banana jacks was a time waste…
You have to mount them in the frontpanel FIRST, then solder it...Should be remarked in the manual but it doesn´t.
Fun fact :
Everything is diy, even the powersupply- you have to solder the 3.5mm plug on the chord...

Soon I´m finished it (pic3) and try some measurements(pic4, pic5), to see if it´s working.
The final adjustments according to the manual I´ve done at work - You have to trim the current to 100mA as precisely as possible at 24 range, 10mA at the 240 range.
This I had done with our yearly calibrated keysight benchmultimeter.
After this, after warming it up, I´ve done several measures, it´s really precisely.
I think, for it´s money it´s a really good meter.
Further measurings (long time and under different ambient temperatures) will follow.
Conclusion:

Pro:

Cheap because it´s a kit.
Precise measuring (so far).
USB connection, software included for logging/remote operation, works proper.
Con:

Components are not specified by position, you have to look at the value and then to the component placement ( Example 10K would not be marked as R10, R11 on the bag, even the value is not marked on it, you have to use the colour code for identify).
Except the measuring resistors (0.1%) all other resistor are carbon type.
Grid dimensions for the diodes are really small…
Kit is therefore not for beginners


Martin

edit 07.10. : Tolerance of the measure resistor corrected to 0.1%

[HKJ]

A milliohm meter is many different things, I made one for checking mains and other cables:

https://lygte-info.dk/project/CableLowResistance%20UK.html

It is designed to be practical (i.e. show if a cable is within specs or not), not to have a lot of significant digits.

[Martin72]

Not bad 

A thing I´ve forgotten is this:

One of the screws for the lcd-display is close to another circuitpath, this could cause a short circuit when the screw is tighten too hard…

[Electro Fan]

Nice project and product - Nice work and Thanks for sharing.

[Martin72]

Thank you, I thought it was worth a thread meanwhile some cheap ready-to-use units are on the market like this.

[nctnico]

Did you try measuring 20mOhm resistors with it? There is no accuracy spec and it seems it uses a rather simple ADC (can't read the part number from your pictures) to measure the voltage across the resistor while assuming the current is adjusted to precisely 100mA using a single turn pot. This doesn't instill much confidence in me.  For about the same price you can buy an XJW01 LCR meter which can also measure capacitance and inductance with good accuracy (resistance with 0.1 milli Ohm resolution at the lowest range). I have the Ascel LC meter (if I haven't thrown it away yet) but I have to conclude it was a waste of money; the measurement principle is just too simple and from the looks of it this resistance meter has the same problem.

[Martin72]

Hi,

Did you try measuring 20mOhm resistors with it

50mOhm Shunt on heat sink, but I couldn´t find out it´s tolerance.
Will repeat it in the next week, hopefully I find the specs for it.
First measuring was 52mOhm +/- 100µOhm, between 20min duration time.

[nctnico]

Hi,

Did you try measuring 20mOhm resistors with it

50mOhm Shunt on heat sink, but I couldn´t find out it´s tolerance.
Will repeat it in the next week, hopefully I find the specs for it.
First measuring was 52mOhm +/- 100µOhm, between 20min duration time.

A 1% current shunt (and I'd assume a current shunt is 1% or better) should measure between 49.5 and 50.5 milli-Ohm.
My XJW01 measures a 10 milli-Ohm 1% resistor as 10.2 milli-Ohm (where .2 is the last digit).

[Martin72]

My XJW01 measures a 10 milli-Ohm 1% resistor as 10.2 milli-Ohm

Fine. 

it uses a rather simple ADC

Yepp, a 22bit one, MCP3551.

[Kleinstein]

It is odd to see 2 trim-pots instead of calibration in software.
I see no CMOS switches or similar to switch the inputs / reference resistors on a regular basis. So I would assume they use just a simple divider from the ADCs reference voltage with a range setting resistors and the DUT. So the wire- and relay- resistance is only partially compensated.
With large resistance there is the obvious problem that the diodes used for protection can start to conduct. One may be able to fix this if for one of the diodes 2 in series are used.

Not so perfect 4 wire measurement would be noticeable in the low range with a relatively large resistor. Some extra resistance in the drive wires would than change the result, in proportion to the reference resistance. With an assumed  200 Ohms as a reference values 1 ohms wire / contact resistance would result in 0.5% of error - noticeable, but not dramatic, given the limited accuracy.

At the 0.1 mOhms resolution and maybe up to 10 mA of test current, it would need to resolve some 1 µV at the DUT. At this level, there can be thermal EMF causing problems. For a much higher test current the reference resistors look like rather low power rating.

Does the software alow to use the sense input as a low voltage voltmeter, e.g. to measure thermocouples ? Chances are the Hardware could allow this.

[Martin72]

Hi,

Looks like not, but maybe it could be implemented in the next update.

It is odd to see 2 trim-pots instead of calibration in software.

The current source is independed from "everything", the µC switches the relay k1 between the 2 values, so you have to trim it with an external currentmeter for each range.
The incoming voltage will be converted by the 22bit ADC and serial data goes to the µC.
With the given hardware, a calibration in software would not be possible, I guess.

[Kleinstein]

The resistance values is calculated from the ADC reading and calibration could work with the factors used in the software instead of adjusting the resistors. So quite sure the same HW could support software calibration, just a question of the software.

It could still be beginners project - especially if the software or at least some documentation is available so one could write ones own software for it.

With an independent current source (worst case even using something like an LM317 / LM 1117 regulator would avoid a problem with the wire resistance, but have the ratio of the 2 voltage reference and thus not so good long time and temperature stability.

Anyway the hardware side looks rather crude compared to the  XJW01 RCL meter and resistance resolution seems to be not much better, though measured in a different mode.

[Martin72]

Current source is realized by LT3092 and LT1634, see pic below.

They recommend to trim the current appx every 6 months or by using a reference resistor.

The XJW01 was interesting too, for L and C I need a meter.

[Kleinstein]

The test current is quite high - so quite some heat to the current setting resistors. This could lead to some drift after connecting the DUT.

If there is no extra capacitance, there is some danger to see oscillation if the DUT is quite inductive.

The trimmers also have quite some effect. Normally the trimmer should have less effect as trimmers tend to drift over time and have a high TC.

With the given current source and the diodes for protection it is not a surprise that it does not work well with more than some 100 Ohms. 

[Martin72]

With the given current source and the diodes for protection it is not a surprise that it does not work well with more than some 100 Ohms.

Therefore it´s limited to 240ohms, I guess.
I´ll test some different resistors in the next week under several conditions, comparing to the keysight.

[ogden]

They recommend to trim the current appx every 6 months or by using a reference resistor.

Precision current source? - It's brain surgery through **shole  There is ratiometric and bridge resistance measurements that requires just shunt with good TC (already there) and differential ADC with external reference voltage input (there as well, MCP3551).

[Martin72]

Here too.

[HighVoltage]

I have a client that needs to read 1 Ohm values at 100mA, this might be a solution. So I will order one and see how stable it is.
Thanks for sharing

[ogden]

Here too.

What you are talking about? Yes, THIS is where I was pointing to. - That "Milliohm Meter Kit" contains all the parts needed for ratiometric resistance measurement that does not require precision current, precision voltage sources that would require trim every 6 months.

[Martin72]

Yes, THIS is where I was pointing to

So I must have misunderstood it, my failure.

[Kleinstein]

It is an odd construction. With less parts (skip the references and trimmers) they could have reached better performance. For better long time stability they might still need better reference resistors. So spend the money saved on the references on a better resistor.

Maybe wait for a better version and see this one as a chance to learn from mistakes.

[Martin72]

We will see how bad it is, I´ve ordered some measuring shunts from isabellenhuette (100,50,10mOhm 0.5%) and will compare the measurings with a calibrated keysight meter.

[Martin72]

Arrived, tomorrow or friday I´ll make the measurements

[Martin72]

Not bad so far….

[ogden]

Not bad so far….

Good to hear that. Inefficient or overengineered instrument does not necessarily mean bad. It still may measure well and be fit for purpose

[Martin72]

Of Course, they are differents to the last digit about +/- 1, but all in all I´m surprised in a positive way.
Next step will be a long time measurement as a "colleague" of mine suggest it.
( He is like me a administrator of our forum, at work a group leader of R&D in electronics)

That means mounting the shunts on a heatsink, measure their values for min. an hour at ambient temperature.
Then increase it to say plus 10 Degrees more and measure again.
Nevertheless, the possibility to adjust the current at every time is a benefit what precisely measurings concerns.

[Kleinstein]

With the low value shunts there is not that much resolution left. With 100 mOhms it is just 3 digits. So some small moderate time drift would likely no be visible. It would need a slightly larger resistor (e.g. 1 Ohms) to see drifting.

From the circuit design and pictures I would expect some drifting in the first 20 seconds or so after connecting the DUT: the reference resistors get quite some power and would thus show quite some self heating effect. 15 Ohms and some 1.25 V gives some 100 mW to the small resistor. This is within specs, not to damage the resistor, but the resistor would heat up considerably and thus show quite some change from the TC. I would expect something like 10 ppm/K * 100 K = 1000 ppm of resistor change as a reasonable order of magnitude.

The instrument could still meet the rather modest specs.

[Martin72]

Today I took a 1R21 metal film resistor (1%).
After say 10sec the displayed value is rockstable until I switched it off ( after 30min).
Damn, didn´t pay attention to the range, will repeat it on monday.

I would expect some drifting in the first 20 seconds or so after connecting the DUT

The drift is about 5..10sec.

[Martin72]

After long time searching for it, a distributor of isabellenhuette was so kind and send me a free sample of an PBV 1R0. 
With this one I´ll do "long time" measuring.
Because, with an "ordinary" metalfilm resistor 1R0, the values were instable, maybe due to the "high" current.
This one I can mount on a heatsink.
Measure results will be posted here when done.

[Martin72]

Today I did a 2 hours long measuring using the PBV 1R0.



Logging the values with the milliohm meter software



Two negative points about the software:

The time interval couldn´t be changed (always one second) and the output format is csv - but the content (value, range, time) is all in a single column 




So it takes a while to create a diagram (for me, I´m a totally noob in excel)….

At the end one positive thing:
The tolerance of the PBV resistor is 0.5% - And the measured values are always in this range - Nice...

[Kleinstein]

The photo show how not to connect the Kelvin probes:
There are 2 good ways:
1) Use only 2 of the contacts of the resistor and then the probes in both sides. This measures the 2 wire resistance of the shunt.
2) connect in a way that die current and voltage contacts go to different pins of the resistor. This measures the 4 wire resistance.

For the CVS file one can usually choose the separator character on importing to excel or similar. One may also have to choose the right decimal separator (. or ,).

[Martin72]

First I got the clips connected on the pins outside, but it was mechanical unstable so I put them like on the photo.
There was no difference between it what the values concerns but I could repeat the measure, trying to find a way for better fixing the clips on the pins.

For the CVS file one can usually choose the separator character on importing to excel or similar.

Ah, I remember - Thank You !

[HighVoltage]

Looks really stable.

Can you test it with stable 1 Ohm resistor?

I ordered one unit but it has not arrived.

[Martin72]

Hi,

Can you test it with stable 1 Ohm resistor?

This test was made with a Isabellenhuette PBV 1Ohm/ 0.5%  resistor.

[HighVoltage]

I was typing too fast ...

What I wanted to say is:

Can you make some longer measurements with true 4 wire connection from the Isabellenhütte resistor 4 wires to the instruments 4 wires, to see if there is any drift.

[Martin72]

Without kelvin clamps,  four wires directly on the pins soldered ?
Sure, why not...
On monday, I guess - Yesterday I took everything back to home.
Except the resistor...

Edit: Thinking about it, this is the better method, you can see any movement of the clamps as a disturbance due more or less contact on these little pins

[nctnico]

Looks really stable.

You can't say over such a short period. Better leave it on for a couple of days. You'll likely see it acts as a good thermometer.

[Martin72]

Better leave it on for a couple of days

That make sense, I´m also doing this at work (testing a resistor or something else for days) and everytime I got trouble with my boss because he thinks I´m too slow, what a stupid..

[Martin72]

Can you make some longer measurements with true 4 wire connection from the Isabellenhütte resistor 4 wires to the instruments 4 wires, to see if there is any drift.

Prepare the resistor today, measure follows in the next days, want to acclimate DUT und and tester to my room temperatures (at home, all tests before done at work).
Nevertheless I´ve test it for a short while at the evening without logging.
In the first ...say 2,3 minutes the measured value drifts a little bit before it becomes almost stable - but in every case it´s within the tolerance range of the resistor.
More to come in the next days.
Nice thing:
I wrote to Ascel, moaning about the "useless" csv export, they promptly doing an update and give me a link to this - nearly close to sunday midnight...
Of course, the next measuring will be with this version.

[andy3055]

Not bad 

A thing I´ve forgotten is this:

One of the screws for the lcd-display is close to another circuitpath, this could cause a short circuit when the screw is tighten too hard…

Martin,

You can use a fiber washer under that screw.

Where did you get that "holder" you are using to hold the PCB in place?

[Martin72]

Hi,

Thankyou for the hint, I´ve used a Isolation washer from work for this.

My pcb-holder I´ve bought from here:

Pollin Electronic

But also :

ELV Electronic

Amazon

Martin

[andy3055]

Martin,

Thanks for the links.

[Martin72]

You´re welcome.
For it´s price it´s really good.

[Martin72]

Of course, the next measuring will be with this version.

I´ve done an "one hour quickshot" today, with the resistor directly connected to the Ascel and using the new software.
Pro Thing:

Measuring the resistor in such manner will result in "rock-stable" values - in the whole hour I´ve done it, the value varies only about 100µOhm - Really nice.

Con Thing:

The updated software is still somekind of crap, when you export the logfile.
Now the values are departed in different columns - good.
But still you can´t create a diagram directly from it.

My conclusion:
For its price the Ascel milliohm-meter is a good choice.
The precision lies on your own hands when you calibrate/adjust the measuring currents.
This could be/is an advantage, when you want to measure everytime as precisely as possible - Just adjust the currents before.
The pots could be reached on the front without disassembling the unit.
Although you don´t need to do it - Ascel recommend to adjust the currents every half year, my adjustment is more than 2 months ago and still it measure pretty good.

What the logging concerns, they should/must work on it.
But to be honest, who really need this…
You grab your DUT and want to measure it here and now, after say a minute or half you got a stable result - And that is all you want.

Nevertheless I will keep this thread alive by posting eventually sw-updates their results.

[HighVoltage]

Today I received the kit and assembled it.
Worked right away and seems to be stable.
I just did not like the cheap and supplied 4mm posts and used a different kind.

Here is my first picture.
I am testing here a stable 2.2 Ohm resistor and the value shown sits at 2.200,5 Ohm in the 24 Ohm range.

[Martin72]

Looks good 

Today I did another measuring about two hours.
Stabilizing time is about 2 minutes from the start - but even in this time it´s already in the tolerance value of the resistor.

[andy3055]

Nice job!

[nctnico]

You really need to run this for a day (preferably even longer) with temperature variances. The way this device has been designed is not good and subjecting it to different temperatures will show that. Running it for 2 hours proves nothing at all from a metrology point of view.

[Martin72]

None test a resistor longer as a few seconds - If you know your device needs some warming up, you test a little bit longer, say a couple of minutes.
The Ascel needs some warming up as stated before and after this, the displayed value vary only in µOhms.

The used resistor got a tolerance band of 0.5%, 995 mOhm to 1.005 Ohm.
The Ascel measures 999.4 mOhm.
I think it´s trustworthy enough.

Sure I can take it back to work and do some climatic tests with it, but usual I´m testing at normal room temperature at home or at work, so how it will be measuring when ambient temperature is for example 40°C....for me it´s not important.
Nevertheless, if I get the chance for climatic tests, I´ll do it and post it here.

[nctnico]

None test a resistor longer as a few seconds - If you know your device needs some warming up, you test a little bit longer, say a couple of minutes.

You are missing my point: the Ascel milli-ohm meter is not constructed in a way that it will give you a guaranteed accurate reading after a year AND / OR when the temperature has changed. A properly constructed resistance meter (DMM) will be able to do that. It is all about making sure the reference is stable and cancelling aging / temperature effects. The Ascel meter has none of these (as others and me already pointed out). Doing a measurement over a day will show the effects of the temperature as an example.

[Martin72]

You can calibrate the measuring current every time when you think it will be needed (therefore the holes in the frontpanel, to get everytime access to the trim-pots), so it´s not important if it could measure with the same precision as a year before - just trim the current as precisely you can to 100ma and that´s it.
You´re right, DMMs should deliever this because they´re a "black boxes", the customer shouldn´t (and haven´t) play on it´s factory adjustments.
And again yes, the principle of this unit is simple.
But it works, regardless of it´s cheap price.
Comparing to our yearly calibrated benchtop multimeters from keysight, it doesn´t doing anything worser.
Only a few more than 100 bucks, but it works - Sad but true.
But to repeat it, if I get the chance, I´ll test it under different ambient temperatures.

[HighVoltage]

Over night I had a 34461A hooked up to the 10mA current output.
Here is the result.

I think for a small cheap unit, this is not bad.
For a client project I needed something like this to measure 0.50 Ohm and this could work perfectly.