Turns out these $5 999ma Ammeters are low-side only

[cellularmitosis]

So I was tempted by the "too good to be true" price ($5 shipped) of these "0-999ma" ammeters from China on ebay.  I snatched one up to use in a simple LM317 project I'm working on.

example: http://www.ebay.com/itm/0-36-Blue-LED-Digital-DC-Ammeter-AMP-Mini-Current-Panel-Meter-DC-0-999mA-/321241201719?pt=LH_DefaultDomain_0&hash=item4acb77bc37

There are many listings for what appear to be the exact same ammeter, and some of them show them only in a low-side configuration, but one of them showed it in a high-side current sensing position.

Well, it turns out they are low-side only

I wired it up in this schematic: https://github.com/pepaslabs/DualLM317BenchSupply/raw/master/DualLM317BenchSupply_schematic.pdf

and stupidly hooked it up directly to a wall wart, instead of a current limited bench supply.  It let out just a little bit of magic smoke before I unplugged it, and displayed "999".

Amazingly, it still worked when I rewired it into a low-side configuration.

So, buyer beware, these are low-side only.

[FrankBuss]

±0.1% measurement accuracy sounds good, can you verify it for different currents? Or maybe Dave could buy one and expose the real accuracy I guess it could have such a good precision, but for this accuracy I would expect an expensive resistor which alone would cost more (at Digikey there are 5 mOhm/0.1% resistors for $7, even for larger quantities), or it needs calibration. And the resistor looks strange: 050 looks more like 50 mOhm, not 5 mOhm.

Could you post a picture of the rest of the circuit (I assume on the back) and what parts they are using? I hope it's not the usual black epoxy blob so that we can learn a bit from it how it works.

[Monkeh]

050 looks more like 50 mOhm, not 5 mOhm.

050 ought to be 5R..

[sleemanj]

050 ought to be 5R..

No you you are thinking of 5R0.  Where shown the R substitutes the decimal point and there is no multiplier.

R050 is 50 mOhm

Edit: sorry just noticed what you did there, yeah, the R is shown on the image in the eBay but you must have only looked at the post previous to yours which omitted it.

[Monkeh]

050 ought to be 5R..

No you you are thinking of 5R0.  Where shown the R substitutes the decimal point and there is no multiplier.

R050 is 50 mOhm

I hadn't actually looked at the photo. '050' without a preceeding R would, I would imagine, indicate 5R.

[FrankBuss]

Thanks, I didn't know the "R" is important. But with the R it looks like it is indeed 50 mOhm:

http://www.digikey.com/product-detail/en/LVK12R050FER/LVK12R050FERTR-ND/1632729

But I bought one from eBay, too, just for fun. Found one in Germany, no shipping costs, even a bit cheaper, nearly the same product photo:

http://cgi.ebay.de/ws/eBayISAPI.dll?ViewItem&item=141208282939

In this auction the description doesn't say 0.1% accuracy But even 5% would be still useful to add to some projects. There are versions up to 100A for the same price.

[Harvs]

You could just add a $2 SIP isolated DC-DC converter to get over the low side issue.  Then you can stick it wherever it's needed.

In fact I'd go as far as staying that would make it quite useful.

edit:  This is the sort of thing I was refering to:
http://www.ebay.com.au/itm/260988729339

I always buy them by the x10 because they're so useful.

[FrankBuss]

I hadn't actually looked at the photo. '050' without a preceeding R would, I would imagine, indicate 5R.

Right, looks like 050 without a "R" would be 5 ohm:

http://www.marsport.org.uk/smd/res.htm

First two digits are the value, third digit is the multiplier. If there is a "R", it is the position of the decimal point. If there are a letter at the third position, it could be the EIA-96 marking method (I've never seen this).

[sleemanj]

There are versions up to 100A for the same price.

Careful there, I think you'll find that the higher current devices require an external current shunt, they won't have an onboard one.

That's of course obvious from looking at the size, a pcb trace to take 100A of current would be rather large

[rdl]

Somebody please explain this "low side high side". The meter is measuring a voltage drop across a 5 mOhm resistor in a current loop. What difference does it make here? Seems like there current and voltage drop would be the same at any point in the circuit being metered.

[sleemanj]

Somebody please explain this "low side high side".

The device has 4 wires
1. VCC
2. GND
3. Current In
4. Current Out

Current Out is I expect connected to GND on the PCB by it's design (they are not differentially sensing, just sensing the voltage at the top of the shunt and referencing that to GND).

So if you put it high side, you have supply current going into Current In, which now shorts straight to GND though the PCB bypassing the load which you connected to Current Out,

[FrankBuss]

There are versions up to 100A for the same price.

Careful there, I think you'll find that the higher current devices require an external current shunt, they won't have an onboard one.

That's of course obvious from looking at the size, a pcb trace to take 100A of current would be rather large

This 100A amperemeter looks exactly the same as tha 1A version, but without the SMD resistor. Maybe it uses the PCB traces. But of course, the wires and traces would vaporize, if used with 100A.

[cellularmitosis]

Frank, mine is about 10% off, but that's after letting a little bit of magic smoke out, so I hope that's not representative.

I think I'll order another and sacrifice this one for a tear down (the circuit is hidden under the 7 seg display)

[sleemanj]

Maybe it uses the PCB traces. But of course, the wires and traces would vaporize, if used with 100A.

Here's one that shows the back of the board, there's no trace there between the SMD pads.  I'm pretty sure that the only way their photo showing it working is real is if they have an external shunt that the "load" wires (now only sense wires) are connected to off-camera and the load is going through that external shunt.  That would explain why
1. it isn't a ball of flame
2. why the on board resistor is removed (because the shunt is offboard)

Maybe they include an external shunt in the package (but surely they'd show it), but as pictured, yeah, that can't work.

[iloveelectronics]

Maybe it uses the PCB traces. But of course, the wires and traces would vaporize, if used with 100A.

Here's one that shows the back of the board, there's no trace there between the SMD pads.  I'm pretty sure that the only way their photo showing it working is real is if they have an external shunt that the "load" wires (now only sense wires) are connected to off-camera and the load is going through that external shunt.  That would explain why
1. it isn't a ball of flame
2. why the on board resistor is removed (because the shunt is offboard)

Maybe they include an external shunt in the package (but surely they'd show it), but as pictured, yeah, that can't work.

All these Chinese ammeters require an external current shunt for their 5A or above variants, and I'm pretty sure it's always a separate purchase.

[cellularmitosis]

Ok, got a replacement ammeter in from china.  The results aren't much better.  It is mostly about 6% off.

I soldered the ammeter into an LM317 supply I've been working on, and used an adjustable dummy load to dial in various loads, and measured the load using three meters: an HP 3478A, a UNI-T UT61E, and the $5 eBay ammeter.

Here's a chart of the absolute values, HP vs the other two:



Here's the same data, but formatted as % error vs the HP 3478A:



I'll tear one of these down and see what the circuit looks like.  Too bad I don't see a trim pot, it looks like the error is just a scaling issue.

[Bored@Work]

I'll tear one of these down and see what the circuit looks like.

Why? If this is what you indeed bought

example: http://www.ebay.com/itm/0-36-Blue-LED-Digital-DC-Ammeter-AMP-Mini-Current-Panel-Meter-DC-0-999mA-/321241201719?pt=LH_DefaultDomain_0&hash=item4acb77bc37

then the shunt is right in front of your eyes:

http://i.ebayimg.com/00/s/MTIwMFgxMjAw/z/DhwAAOxy3NBSdKgg/$%28KGrHqV,!mEFJtkgC7VuBSdKgeONew~~60_57.JPG

Your readings are 6% too high, so add a 1 Ohm resistor and a 10 Ohm trimpot in parallel and adjust to reduce the overall shunt resistance by ~6%.

[cellularmitosis]

Your readings are 6% too high, so add a high resistance trimpot in parallel to the shunt to reduce the overall shunt resistance by ~6%.

Good point!  Thanks

[cellularmitosis]

Teardown photos:

[cellularmitosis]

Following Bored's advice, I corrected it with four 1 ohm resistors (1 ohm in parallel with 3 ohms).

Here's the results:



And here's a link to the spreadsheet with all of the data: https://docs.google.com/spreadsheet/ccc?key=0Aps_jTQV0tKndEdDQjlsdnpEQUxyNGZqcVBrY1VfTnc&usp=sharing

[sleemanj]

Neat photo capturing a segment half-lit :-)

[FrankBuss]

That's interesting, more chips than expected under the LED. This is the microcontroller:

http://www.digikey.com/product-detail/en/STM8S003F3P6/497-11946-5-ND/2765682

Not a cheap 1-bit CPU epoxy blob.

I guess the 7130H IC is a voltage regulator. And maybe the 3rd 5-pin IC would be an OpAmp.

[cellularmitosis]

As a follow up to this, using a 10 ohm mutli-turn trimmer is the perfect thing for dialing in these ammeters.

[sleemanj]

Bit of a follow up to this, thread dredge I know, but anyway.

I picked up some "0-10A" versions of these, they are identical to the one posted in this thread, except for using an on board shunt link rather than the smd resistor.



Unfortunately, a brief play around with them reveals they have some interesting properties.

If you power it from it's own supply separate to the supply powering the load (of course remember that the negatives will be connected on board), everything is hunky dory, the accuracy seems reasonable.



If you power the meter from the same supply as is powering the load, and leave the negative supply of the meter disconnected (remembering that the bottom of the shunt is connected to the meter's supply negative on the PCB), then it seems to read consistently about 80mA low - not a percentage of the load just always seems to be 80mA low.



If you then connect the negative supply for the meter to the power supply, remembering mind you, that they are already connected together on the PCB, things get really crazy.



There is supposedly a "calibration" procedure, but I've only ever seen it mentioned as existing, not what it actually is.

The unpopulated 6 pin header on the left side of the board (see previous images in this thread) I have buzzed out as (numbered from bottom to top)...

1 = Pin 20 of the mcu, "PD1(HS)/SWIM"
2 = Pin 2 of the mcu, UART1_TX/AIN5/(HS) PD5 (this is marked +++ on the board)
3 = Pin 3 of the mcu, UART1_RX/AIN6/(HS PD6 (this is marked ---)
4 = Pin 4 of the mcu, "NRST"
5 = VCC
6 = GND

If I get enthusiastic I might see if there is anything useful to be had by way of serial.

[sleemanj]

With experimentation I've stumbled upon some sort of calibration for these.

Put a constant 1 Amp through it, then touch the pad marked +++ to ground it, should show 1.00 (at least for this 10A one I have), touch it again I think to save, or the one below it.
The bottom most pin when grounded goes into some other mode which could possibly be some sort of adjustment, I haven't made sense of it.
The third down from the top is reset.

This seems to work with the "supply" negative (small black wire, as opposed to load negative large black wire) disconnected if you are powering from the same supply.  If you connect the supply negative, it goes bonkers.

The pin marked --- seems work the same way for calibration to 5A, but I can't try it out at the moment.

[cellularmitosis]

Fantastic!  I wonder what current the 999mA version should be calibrated with?