5 dollar voltmeter with 5 digit (0.1mv) resolution

[Kalvin]

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! 

According to this video the response time is pretty good, if we are talking about the similar cheap panel meters:


My panel meters are still in mail, so I cannot verify how fast these meters actually are. But according to the video, I am happy with the response time.

[Neverther]

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! 

According to this video the response time is pretty good, if we are talking about the similar cheap panel meters:


My panel meters are still in mail, so I cannot verify how fast these meters actually are. But according to the video, I am happy with the response time.

Hey, thats my video. I forgot I did that.

[paulie]

That appears to be the old version and not the one discussed here. It does have faster refresh (200ms) unlike the new one which is slow as molasses (3 seconds). IMO neither of these are appropriate for a PSU monitor. What's the point of 0.1mv output resolution when accuracy is literally orders of magnitude below that?

I think the little $1 panel meters which have virtually instant response time are better suited. Or those double meters for both current and voltage. At least you only have to cut one hole in the box.

Who needs more than a hundredth of a volt resolution for a power supply? Bling artists and those with money to burn, that's who. LOL.

[Alex30]

Who needs more than a hundredth of a volt resolution for a power supply? Bling artists and those with money to burn, that's who. LOL.

Because it looks cool!!

[paulie]

Touche! I don't have money to burn but must admit to being a bling artist. When I get this new meter calibrated maybe the old one will go into one of my "ghetto" (hi danny) PSU. 3 seconds not useable IMO but 200ms ain't that bad. And who's to know those last couple digits are for show.

Of course it would be really nice to have accuracy of these meters match the resolution. Or as close as possible. I can't wait to see how my friends lab meter readings match up.

[Alex30]

Yeah. If it's as good as I've heard it should be fairly reasonable given the price. I mean $5 for even 1-5mV resolution ain't bad. I just hope they are fairly linear and easy to calibrate so you can make the most out of them.. And yeah 200ms suits me to be honest. Anything less than that I would be wanting a peak voltage meter anyway because I hate trying to pick out numbers quickly.

[Alex30]

So my voltmeters arrived and so far im pretty pleased. I haven't been able to set up a proper test as I don't have a benchtop voltmeter so i'm still looking forward to paulies results. But because I got two I figured I could at least compare one to the other and see what kind of tolerance I got between them. I just quickly hooked em up to batteries and used a voltage divider to get a low voltage for the 0.1mV resolution. I used two batteries so the measurement was isolated from the supply and I noticed a difference of between 0.1-0.2mV which is pretty damned good in my opinion. Refresh rate was also quite good, somewhere between 2-3 per second. I noticed that the readings were a lot more stable when measuring something isolated than measuring from the same supply battery. Quite pleased with this.

[paulie]

Nice colors, yellow and red. Cool. Or should I say HOT. Mine is blue because almost half price. Not as visible but I can live with that.

Well the lab measured references came back and not good news. Both 4040 read significantly out of spec. 0.23% for one and  0.22% for the other, Both high so I'm beginning to think there are issues with my friends meter calibration or some problem with noise or TC. I know he purchased this last year as part of an on-line deal but not sure if he repaired or calibrated or what. I'll have to get in touch to find out more. On the other hand the 5 digit Ebay meter reads well within Nationals specs.

Not likely a $5 Chinese clone could outperform a piece of professional gear that costs thousands but hopefully I'll get to the bottom of this.

[splin]

I tested my .36" 5.5 digit voltmeter (0 - 4.30000 - 33.000V), using an AD584 based reference:

http://www.ebay.com/itm/AD584LH-4-Channel-High-Precision-Voltage-Reference-Module-can-5VDC-charging-NEW-/191342226316?pt=LH_DefaultDomain_2&hash=item2c8ce2ab8c

It has a sticker on it showing the measurements made by the manufacturer using a 6.5 digit HP34401A

I also measured the AD584 reference with my own 34401A (not calibrated for several years) with these  results:

AD584 Calibration sticker (V)	2.49887	5.00149	7.49902	10.00129
My HP34401A, 10V scale (V)	2.4987	5.00144	7.49899	10.00129
Difference (uV)	170	50	30	0
Ebay 5.5 digit meter (V)	2.4984	5.0005	7.498	10
AD584 error (mV)	0.47	0.99	1.02	1.29
Hp34401A error (mV)	0.3	0.94	0.99	1.29
Avg err (mV)	0.385	0.965	1.005	1.29
Avg err (%)	0.014	0.019	0.013	0.013

Well within the .3% spec and not at all bad for the $5.99 I paid last year.

Splin

[splin]

I tested my .36" 5.5 digit voltmeter (0 - 4.30000 - 33.000V), using an AD584 based reference:

http://www.ebay.com/itm/AD584LH-4-Channel-High-Precision-Voltage-Reference-Module-can-5VDC-charging-NEW-/191342226316?pt=LH_DefaultDomain_2&hash=item2c8ce2ab8c

It has a sticker on it showing the measurements made by the manufacturer using a 6.5 digit HP34401A

I also measured the AD584 reference with my own 34401A (not calibrated for several years) with these  results:

Splin

AD584 Calibration sticker (V)	2.49887	5.00149	7.49902	10.00129
My HP34401A, 10V scale (V)	2.4987	5.00144	7.49899	10.00129
Difference (uV)	170	50	30	0
Ebay 5.5 digit meter (V)	2.4984	5.0005	7.498	10
AD584 error (mV)	0.47	0.99	1.02	1.29
Hp34401A error (mV)	0.3	0.94	0.99	1.29
Avg err (mV)	0.385	0.965	1.005	1.29
Avg err (%)	0.014	0.019	0.013	0.013

Well within the .3% spec and not at all bad for the $5.99 I paid last year.

Splin

Some points I forgot to mention:

1) With the input at 0V it reads a solid 0.0000
2) The input resistance is approx 347k ohms
3) The meter only has 3 digits after the decimal point above 4.3V; the 5.0005 in the table reflects the fact that the least significant digit was flipping between 0 and 1 every few seconds.
4) The update rate is quite quick - perhaps 3 or 4 per second?
5) The Power supply is not isolated as some have stated here - the supply is separate from the measured signal but has a common ground.
6) My reference and both meters could all be wildly inaccurate - by the same amount! 

EDIT: 7) The meter uses an STM8S003F3P6 and the ADC is a 6 - pin device marked CAJD
          8. There is no obvious voltage reference - only the regulator marked 7133-1#

[bktemp]

EDIT: 7) The meter uses an STM8S003F3P6 and the ADC is a 6 - pin device marked CAJD
          8. There is no obvious voltage reference - only the regulator marked 7133-1#

Thank you for reading the part numbers.
My guess was right:
CAxx is MCP3421: 18bit I2C ADC with internal 0.05% reference.
This explains why it so accurate.

[electr_peter]

5) The Power supply is not isolated as some have stated here - the supply is separate from the measured signal but has a common ground.

Nobody said that power supply in the panel meter is isolated. The main issue is that power and signal wires are not isolated inside panel meter - this creates issues with powering and isolation.
In posts above it was stated that it is recommended to use isolated external power supply (w.r.t. to your circuit) with this panel meter. Doing it this way allows to achieve better results. Keep in mind that when you connect isolated power supply to this panel meter and connect panel meter to voltage of interest in your circuit, isolated panel supply is not isolated anymore w.r.t. to your circuit.

You can also power these panel meters directly from the same circuit you are making measurements. This has disadvantage of lower precision.

Older type of panel meters could only work with isolated supply - that is a bit annoying when you are dealing with DC circuits. Most lower precision type panel meters only provide non-isolated power and signal wiring for convenience (2 wires - GND and V_power/V_signal or 3 wires - GND, V_power, V_signal).

[paulie]

Thanks for that data splin. It is the most interesting and useful to me so far. Unfortunately I'm still unable to locate a reliable reference for my own unit which I'm hoping is just as accurate. I've modified mine to supply serial output and would love to do some long term stabilty and temperature tests.

Apparently my friends lab meter was a "repair special" and while he did manage to fix it there has been no calibration since. This is probably why his readings did not correlate well with even basic 4040 specs. Maybe somebody here might be willing to lend a helping hand. I just need to get some lab grade (6 digit+) 4040 readings. As you can see in page one photos easy to mail back and forth as letter postage. I can cover all costs. If anybody can help me with this please PM or post.

[paulie]

I've recently added a couple more low cost voltage references, TL431 and LM385, to my collection. Along with the original LM336 , LM4040, and zener diodes this makes 6 types total. Since they are so easy to make and so cheap (as low as 5-6 cents) I assembled a dozen of each type. This will allow me to do some long term drift experiments and also allow me to help some friends get their meters calibrated to accuracies rivaling more expensive units.

Anybody else who might be willing to do some readings for me on some of these please pm. Your help is greatly appreciated.

[paulie]

Someone asked about my setup for measuring voltage references so here's a photo. Black dot is ground, white is +5v or +9v, and last pin is the reference voltage output being measred. I found using batteries for both meter and reference gave most stable readings. If/when I am able to obtain accurate transfer values this will be the same arrangement for calibrating the meter.

[splin]

You can get an AD584 reference module with 2.5V, 5V 7.5V and 10V outputs for as little as $6 off ebay:

http://www.ebay.com/sch/i.html?_odkw=Voltage+Reference+Module&_sop=15&LH_BIN=1&_from=R40&_osacat=0&_from=R40&_trksid=p2045573.m570.l1313.TR0.TRC0.H0.XAD584+Voltage+Reference+Module&_nkw=AD584+Voltage+Reference+Module&_sacat=0

I got a $15 dollar version (those in the acrylic cases) and can recommend them. However, three things to be aware of:

1) They use a 3.6V mobile phone type LI-ON battery - due to shipping restrictions the battery may or may not be included (mine was)

2) Some have a calibration sticker showing the values as measured by the supplier (in my case with a HP34401A, which agreed quite closely with my measurements with my own HP34401A) but others might not. You should check this if you haven't got access to an accurate meter.

3) Don't use long banana plugs as you can short out the circuit underneath the sockets!

[paulie]

Thank you for that information on the acrylic case version. A friend of mine bought the PCB type to calibrate his "repair special" meter but it did not come with the precision readings. Maybe only the expensive units supply that custom data.

I hesitate to spend 4 times more for the reference than for my meter and even then risk not getting accurate values. The AD584 itself has 7mv uncertainty and I'm hoping to get at least an order of magnitude better to take advantage of the meters resolution. My problem may be solved due to the kindness of one stranger willing to provide me with the DIY voltage reference readings.

[Alex30]

Just wanted to drop by to say thanks for your efforts paulie and splin. I have so far been unable to make any progress with getting access to equipment to check out my meters but will let you know if\when i do. Paulie, I was wondering how you plan on creating a calibration circuit for your meters as they do not have any calibration trim as we all know. There are obviously two things to calibrate for, linearity and offset. Surely to do this you need a reference at a high and a low point to ensure linearity? Either that or just an excellent reference at say, 10V and then a really good ohm meter to lower that voltage by a known amount to take your low reading. Thoughts?

[paulie]

Hi Alex. The meter appears to already have an unpopulated calibration circuit onboard. It allowed me to put a 4.7m resistor in parallel with the 330k and a 50k trimpot in series which seems to give a nice 1% adjust. As far as linearity that is not as easy to control so the plan is to live with the MCP3421 specs which ain't half bad as-is. Maybe can be checked with some cheap resistors using flip flop method or even a Hammond network but IDK if it's worth it.

One of my more ambitious mods was to tack on a couple CD4051s and OP07 circuit for monitoring several channels to see how these bandgap dealies really perform in the long run. I'm building a TCO too because of temperature extremes in my shop. Lots of threads on the zener based ones but not so much for the cheapies as splin noticed. Maybe 5 digits is not precise enough unless the drift is significant. Anyway I'm so impressed by this product I plan to buy another before the end of the week that I can dedicate just for these experiments.

[Alex30]

Thanks for the reply. I was just looking at the circuit board (picture attached) and it does indeed have some space there (R7 and R8). The 330k seems to connect to a pad on R8, which then connects to R7 in series, but I can't work out where it connects to after that without probing around a bit. Did you use R7 and R8 for connecting your resistors? Also any idea what the KEY part is for on the left hand side?

Keep us updated on your mods! And yes it is quite an impressive product given it's price, definitely my favorite electronics product I have bought on ebay.

[nikifena]

I found this link with the schematics over the net:
http://eddy-em.livejournal.com/73375.html

It's in Russian.

[electr_peter]

I found this link with the schematics over the net:
http://eddy-em.livejournal.com/73375.html

It's in Russian.

Interesting article. Voltmeter consists of linear reg, ADC (18 bit, or 17 bit for pos. values) and MCU. Save some calibration constants in MCU, read ADC via I²C - and you have a nice voltmeter.

I did not fully understand part about calibration procedure with PD6/UART1_RX. Article implies that calibration values can be written to MCU with UART, but it is very sketchy.

By the way, what does the button/key do? It is unpopulated on the PCB. Is it used for calibration?

[nikifena]

Probably this button footprint exists from old pcb design and it's not used. I have 2 of these voltmeters with different firmwares and they has the same PCB.  Anyway the entire project is very clever with just a couple parts. Pretty nice for DIY projects

[dannyf]

I found this link with the schematics over the net:
http://eddy-em.livejournal.com/73375.html

STM8S03F, the little bugger that can.

It's in Russian.

he was wondering about why they didn't use hardware i2c on that chip, .

[paulie]

That "key" signal and it's debounce cap are just connected to one of the MCU input pins (PD6) so it's hard to figure out what the original purpose was. Maybe related to calibration or seeing how it's on the programming connector possibly something to do with bootloader function. It's lablelled "BTN" which probably stands for "button".

BTW like nikifena says the old 33v model hardware is identical to the new 4.3v. Only firmware is different. Thanks for that Russian link. I scoured the internet (google/bing SUCK!) and couldn't find anything close to that. Strangely they mention the Ebay meter but the drawing is not exactly correct. There were a few minor differences in part numbers, 7 seg display type, and missing components like the "key" button.

I edited it to more closely match my board and attached to the first post.

ps. Yes, Alex30, I tacked on the parallel resistor and series trimpot to R8 and R7.  These meters seem to be very accurate out of the box and I didn't want to risk changing the resistance of the 330k by soldering directly to it.