5 dollar voltmeter with 5 digit (0.1mv) resolution

[paulie]

A while back I purchased one of these:

http://www.ebay.com/itm/310601671625?_trksid=p2055119.m1438.l2649&ssPageName=STRK%3AMEBIDX%3AIT

However the new 0-4.3000v-33.000 version has a 0.1mv spec when used with isolated supply. A fellow amateur radio club member swears he gets close to that and I'm wondering if anybody here also has direct experience with one.

http://www.ebay.com/itm/370966721056?_trksid=p2059210.m2749.l2649&var=640165395235&ssPageName=STRK%3AMEBIDX%3AIT



Update: I ordered one of the 4.3000v meters and it arrived so time to put together a set of voltage references. Included are 0.1% LM4040a-2.5, 6.9v precision zener circuit, LM336-2.5, and 0.1% 4.1v LM4040a (see photo). Out of the box the meter seems to be quite close to expected accuracy. What I would like to eventually do is get some more accurate readings of these and compare those with the meter. Hopefully get the accuracy in the same ballpark as the resolution.

Pinout for the references: black dot=ground, white=Vin (+5v or +9v), third pin=Vout.



Update: Thanks to nikifena we have a schematic from a Russian web site that was very close to the Ebay meter. I made a couple minor corrections and attached below.



Update: In response to an offline request for more info on the calibration circuit mod a photo of that is attached below. A slight improvement over the first version. Basically a 10meg resistor in parallel with the 330k then 9.5k to 10.5k variable resistor (fixed 9.5k + 1k trimpot) in series with the input. These components were attached to the unpopulated pads near the edge so soldering directly to existing resistors was avoided. Didn't want to risk changing original component values by overheating. End result was a small fraction of a percent range which provided very comfortable calibration adjustment.

Ignore the dual header. That is for a multichannel input expander and PC datalog interface and is unrelated to the calibration mod.



Update: Another meter modification. Recently needed to measure geiger tube voltage with more precision than my normal 3 1/2 digit meter so thought it would be nice to take advantage of the 5 digit Ebay unit. Problem is voltages in the area of 650v so decided to build a 100x range expander. Fortunately it's as simple as adding 30 some meg in series which gives up to 3300v measurement capability. It took 3x 10meg, 1.5meg, 1meg, and a pot for fine adjust.

A side benefit is increased input impedance which is always a good thing. Now going from 330k to over 30meg.

One caveat is insuring dangerous HV breakdown won't damage the meter. Spark-gap rating for 3300v is 0.15" so to leave a little margin I kept the hot end at least 0.2" away from everything else. And no 0805 here, only 1/4w or bigger through hole. Naturally great care should be taken with potentials this high so safety measures like enclosure and insulation are required. In this case a little epoxy potting and shrink. However I thought it might be a good idea to show components more clearly before wrapping it up so before doing that some preliminary photos during the test stage.

Works like a charm. Shown below measuring one of the dollar store USB chargers modified for access to rectified mains. Beware, there are dangerous voltage and current levels so take all necessary precautions in your own tests. Good idea to start by checking simple 12v dc or 9v battery which should give some indication of accuracy.

[electr_peter]

There are 0-33.000V and newer 0-4.3000-33.000V versions. Some versions have 3 wires, some 4 (these should be powered by isolated supply).

I had older 0-33V type with 3 wires - worked OK when compared to accurate multimeters. Display digits are slightly smaller than typical 3-4 digit panel meters.
As for 0.1mV - keep in mind that it is resolution only, accuracy is claimed at ~0.3%+2d. I would not trust it too much on low voltages.

These meters are designed to display fixed +4-30V full range, accuracy and resolution at lower end (<1V)is not critical to typical applications anyway. If low range is important, use more sensitive equipment.

[paulie]

Thank you for that input. I was hoping someone had experience with the new hi-res version but it looks like I'll have to order one and evaluate myself.

My goal here might be to calibrate one of these so accuracy can match resolution or as close as possible at room temperature. Unfortunately I no longer have access to a precision lab voltmeter so the plan is to send a home built voltage reference to a friend who does and compare with readings from one of these 4.3000v meters.

What about basing it on something like lm4040a which has 0.1.% initial accuracy spec? Any opinions?

[Alex30]

Thank you for that input. I was hoping someone had experience with the new hi-res version but it looks like I'll have to order one and evaluate myself.

My goal here might be to calibrate one of these so accuracy can match resolution or as close as possible at room temperature. Unfortunately I no longer have access to a precision lab voltmeter so the plan is to send a home built voltage reference to a friend who does and compare with readings from one of these 4.3000v meters.

What about basing it on something like lm4040a which has 0.1.% initial accuracy spec? Any opinions?

Haha I think you may have got that idea from my power supply project. That's exactly what I was going to try and do. Although even if you can calibrate it to the lm4040 you will still be within +/-0 2.5 mV. I will wait for my voltmeters to arrive and let you know how it goes. Should take about a month to arrive and I will post my findings then or two weeks later as I am going on holiday around that time.

[paulie]

Yes, I've been following your thread since the start. It's one of the more interesting ones along with the LM399 and LTZ1001. Like Peter I've had a 33v version for some time now and curious if these new hi-res units are even close to what they claim. It would make for a huge advance in this area if so.

even if you can calibrate it to the lm4040 you will still be within +/-0 2.5 mV.

Or much  better (~0.1mv) if calibrated to a LM4040 that was measured with a high quality lab meter. Not too much beyond that as I don't want to get into spending thousands on standards or worry about sensor orientation or humidity at this stage.

I've just ordered one myself and very interested in how yours works out. Do you own one of those high end lab voltmeters or other means to check things like linearity, stability, or accuracy?

[Alex30]

Yes, I've been following your thread since the start. It's one of the more interesting ones along with the LM399 and LTZ1001. Like Peter I've had a 33v version for some time now and curious if these new hi-res units are even close to what they claim. It would make for a huge advance in this area if so.

even if you can calibrate it to the lm4040 you will still be within +/-0 2.5 mV.

Or much  better (~0.1mv) if calibrated to a LM4040 that was measured with a high quality lab meter. Not too much beyond that as I don't want to get into spending thousands on standards or worry about sensor orientation or humidity at this stage.

I've just ordered one myself and very interested in how yours works out. Do you own one of those high end lab voltmeters or other means to check things like linearity, stability, or accuracy?

No but I was thinking of seeing if I could get friendly with some of the EE staff at my uni (even though I am doing civil eng, which is also one of my other interests) and seeing if I could do some calibrations. For now all I can do is calibrate the voltmeters to the LM4040 alone. Just takes a while for the voltmeters to arrive on the slow boat from china.

[electr_peter]

I did a quick check of 33.000V panel meter accuracy. Measurements were done with known very accurate references and 0.05% DMM. Actual references values are very close (much less than <1mV) to that of DMM.

Reference	DMM 0.05%	Panel meter 33.000
2.5V	2.498	2.497
5V ref1	4.999	4.998
5V ref2	5.001	5.000
7.5V	7.500	7.498
10V	10.003	10.001

Results are surprising - panel meter is off at most by 2mV from accurate reading. I think that this is very good result for a panel meter. 0-4.3000-33.000V meter has intermediate lower range - I expect it to perform similarly well because the panel PCB design seems almost identical.
I plan to use 0-4.3-33V panel meters for PSU - seems like a good choice for metering so far.

[Kalvin]

Thanks electr_peter. I ordered two panel meters for my power supply project and this is excellent news.

[DVDRW]

Yes good stuff. Also using these in my diy PSU.
There is also 5-digit 0-3A ammeter 100uA accuracy
http://www.ebay.com/itm/Digital-Red-Ammeter-DC-0-3-0000A-5-Digit-LED-Current-Meter-High-Precise-/271560868818?pt=UK_BOI_Electrical_Test_Measurement_Equipment_ET&hash=item3f3a4a03d2

[electr_peter]

In case you are measuring DC current of something, that current panel meter is a good choice (when compared to 3 or 4 digit cheap panel meters).

Would it be better to use [two voltage panel meters (1st for current, 2nd for voltage)] instead of [current+voltage panel meters] for a typical lab PSU?
Current signal converted to voltage can be compared to set current, calibrated via op-amp gain+shunt resistor and controlled remotely if needed. Having direct current reading meter does not allow to show set current easily on the same meter (unless you short output terminals).

[paulie]

I'm using one of the new chinese $1 XL6009 buck-boost converters with a discarded laptop brick as basis for my lab supply. 0.9-36v 4amp output. It uses just one panel meter (0-33v ATM) for both current and voltage because personally there's rarely a need for both at the same time. No op amp though, just using PCB traces as a shunt. The meter is sensitive enough for that.

[MrAl]

Hi,

I have one of the 00.000 to 33.000 volt models and i am happy with it.

The model i have reads with only 3 decimal digits to the right, so for example it might read any of:
12.543
10.483
27.123

and for voltages less than 10v it reads for example:
7.123
2.345
1.003
0.001

note that it does not shift down to a higher resolution, so i dont get readings like:
0.0001
1.1234

the readings are limited to 3 digits to the right of the decimal point.

The accuracy is acceptable but even the resolution of 1mv on these models is much better than the resolution on my power supply built in meter which is only 0.1 volt which obviously really blows.  The 0.001 resolution really helps a lot so i use one of these meters.

They can be used as a two wire meter for voltage over 3.3 volts by connecting the power supply lead to the signal lead and using the ground as a common ground.  The current is surprisingly only 10ma and it is bright enough to see in normal room lighting.

I meant to install one into my power supply case as it is much better than the on that came with it, but havent gotten around to doing that yet :-)

Keep in mind that it will not read negative voltages unless you reverse the leads manually.  That's probably the only drawback i have found with these things and for regular power supplies that doesnt hurt really because their output is usually unipolar anyway.
I dont know if they are reverse polarity protected.

[paulie]

Those original models do seem to be quite a good deal for precision meters. The new ones (where IS that slow boat?) are even more impressive. Specially if the low range has similar accuracy and linearity as hinted by Peters tests of the original. Not long ago we had to pay half grand or so to get this kind of performance out of products like Fluke. This wouldn't be the first time cheap chinese "copycats" trump pricey western technology.

ps. I must admit though that for most of my DVM applications the even cheaper 3 digit units fit the bill. I've got several of the $1 Ebay DC converters scattered around with one of those strapped on with rubber band so that I don't have to drag out the DMM every time to adjust the output.

[paulie]

Amazingly the high resolution version of the meter I ordered showed up in one week. Not a record but pretty good for China. So a set of cheap DIY voltage references were thrown together and surprisingly the readings were very close to what might be expected. This is a good sign. To really check this out I need to get more accurate readings for the references from a friend who has access to a lab grade voltmeter. Stay tuned.

[Kalvin]

In case you are measuring DC current of something, that current panel meter is a good choice (when compared to 3 or 4 digit cheap panel meters).

Would it be better to use [two voltage panel meters (1st for current, 2nd for voltage)] instead of [current+voltage panel meters] for a typical lab PSU?
Current signal converted to voltage can be compared to set current, calibrated via op-amp gain+shunt resistor and controlled remotely if needed. Having direct current reading meter does not allow to show set current easily on the same meter (unless you short output terminals).

I thought about this "setting the current limit". You have two displays, one for the output voltage and the other for the output current. Use the output voltage display also for the current  limit setting: For example, you may have a push button which will display the set current limit when the button is being pressed, and normally it will display output voltage.

[Sparc]

Maybe I missed it, but what is the ADC and reference chip on these meters?  Is it just an ADC built into a microcontroller?  A 33000 count meter would need 15-16 bits.  Some micros have a 16 bit ADC.  Or do they do something cheap and oversample a 10-bit ADC?  Just curious, HOW they get 5 digits for $5. 

[coppice]

Maybe I missed it, but what is the ADC and reference chip on these meters?  Is it just an ADC built into a microcontroller?  A 33000 count meter would need 15-16 bits.  Some micros have a 16 bit ADC.  Or do they do something cheap and oversample a 10-bit ADC?  Just curious, HOW they get 5 digits for $5.

Look up "dual slope converter".

[Alex30]

Amazingly the high resolution version of the meter I ordered showed up in one week. Not a record but pretty good for China. So a set of cheap DIY voltage references were thrown together and surprisingly the readings were very close to what might be expected. This is a good sign. To really check this out I need to get more accurate readings for the references from a friend who has access to a lab grade voltmeter. Stay tuned.

Excellent news! Keep us updated! I'm still waiting for mine

[Psi]

Whats the update rate like?
I really want to replace the voltage displays in my gwinstek lab PSU which is like 1Hz 

[Sparc]

Maybe I missed it, but what is the ADC and reference chip on these meters?  Is it just an ADC built into a microcontroller?  A 33000 count meter would need 15-16 bits.  Some micros have a 16 bit ADC.  Or do they do something cheap and oversample a 10-bit ADC?  Just curious, HOW they get 5 digits for $5.

Look up "dual slope converter".

But, what is the part?

[coppice]

Maybe I missed it, but what is the ADC and reference chip on these meters?  Is it just an ADC built into a microcontroller?  A 33000 count meter would need 15-16 bits.  Some micros have a 16 bit ADC.  Or do they do something cheap and oversample a 10-bit ADC?  Just curious, HOW they get 5 digits for $5.

Look up "dual slope converter".

But, what is the part?

Huh? The part is typically a multi-meter chip from one of a couple of suppliers in Taiwan. That doesn't tell you how they measure so cheaply, though. Look up dual slope conversion, look at the schematic for any multimeter, and you'll find out how its done.

[bktemp]

Huh? The part is typically a multi-meter chip from one of a couple of suppliers in Taiwan. That doesn't tell you how they measure so cheaply, though. Look up dual slope conversion, look at the schematic for any multimeter, and you'll find out how its done.

Dual slope needs at least a high quality integration capacitor. I can't see any on the board, except for some MLCCs but those are probably used for the power supply.
Therefore the question for the part number is valid. The board on the first post does not look like a typical dual slope implementation, more like a microcontroller with an ADC. U2 could be something like a MCP3421 or a voltage reference.

[coppice]

Huh? The part is typically a multi-meter chip from one of a couple of suppliers in Taiwan. That doesn't tell you how they measure so cheaply, though. Look up dual slope conversion, look at the schematic for any multimeter, and you'll find out how its done.

Dual slope needs at least a high quality integration capacitor. I can't see any on the board, except for some MLCCs but those are probably used for the power supply.
Therefore the question for the part number is valid. The board on the first post does not look like a typical dual slope implementation, more like a microcontroller with an ADC. U2 could be something like a MCP3421 or a voltage reference.

These devices almost always use dual slope converters, but a recent design might be using an MCU or custom device with a sigma-delta converter. The electrical energy market has driven the development of some very low cost high accuracy wide dynamic range converter modules for MCUs. They are usually only 13 to 15 bit accurate for a single conversion (ENOB), due to AWGN, but average over a number of samples and you can get 19 or 20 bits of linearity from some of them. The DC accuracy will generally be somewhat lower, as its hard to achieve a DC offset which doesn't vary at least a little with temperature.

[Neverther]

I actually have those in my DIY supply, 0-4.3-33 as voltage display and 0-33 as current display with shunt and opamp to output voltage.
Updaterate was more than 1Hz. I can confirm it monday as I'm currently away from my "bench".
Accuracy was good enough, it's just generic supply, not high precision.

[dave_j_fan]

those cheapies are sluggish .. i used one like that in bench supply , i have to wait 1-2 sec to see the new voltages Grrr!