EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: edavi on February 15, 2021, 04:37:39 pm
-
Shot in the dark question. I'm somewhat new to electronics. I've been dabbling in graphics card diagnosis and repair, and my cheapo multimeter has done me very well for older architectures. The newer graphics cards have some resistances so low it looks like a short on my current meter. My quest is searching for a meter that can measure at least down in the milliOhm range (microOhm would be nice to have but not necessary), and measure capacitance/impedance is an added bonus. I know of the DE 5000's existence, but I'm looking for something that has all the functions in one. I've found it to be very difficult to search for multimeters that can at least go down to this range! I'm curious to see what options there are on the cheapest side and the nicer side.
Thanks
-
Probe and especially contact resistance sort of spoils that type of resolution.
The best is 4 terminal measurements with a bench meter: DMM6500 has 1uOhm resolution and Keysight 34461A has 100uOhm resolution.
For handheld meters check the EEVBlog 121GW, Fluke 289 and Hioki DT4282 all with 1mOhm resolution.
-
The new BM786 also has a 1mohm resolution.
Nicer side: Gossen 30M 4 wire 100uohm resolution. If you can find one.
-
First, just do the math. An ohmmeter measures resistance by supplying a known current, often 1mA on lower ranges, and then measuring the voltage across the resistance with that current going through it. As the resistances get lower and lower, it has to either supply a higher current or try and accurately measure lower and lower voltages. As you get into measuring lower and lower voltages, effects like thermal voltages, noise, etc start to make it difficult to get an accurate reading.
Second, consider the effect of the leads and connections. Those have resistances too, so that gets added to the resistance measurement. Worse, those resistances may not be completely stable, making it hard to compensate. The solution is to use the 4-wire method and that requires a DMM (typically a bench unit) that has a 4-wire ohms connection.
For examples of how much accuracy you can likely achieve, here's two examples.
My Fluke 289 has a dedicated Lo-Ohms 50R range, it uses about 8mA of test current and has a resolution of 1 milliohm. It has a specified accuracy (in the zero ohm region) of +/- 20 counts, or 20 milliohms, but if I use the REL (zeroing) function and I'm patient, the readings are repeatable to about 5 milliohms or so. That's pretty good for a handheld DMM with a two wire connection.
My Fluke 8846A bench DMM with a 10 R range uses 5mA of test current and has a resolution of 10 microohms. It's specified accuracy would be +/- 3 milliohms, although it seems stable and repeatable down to about 100 microohms or so.
The only way to do a lot better than that would be to use an instrument or homemade setup that uses a higher test current. If you use a bench supply to put 1 ampere through your resistance that you want to measure, all you need is a DMM that can reliably measure 1mV and you can measure 1 milliohm. For you purposes, that may do. Actual calibrated equipment that works down in those ranges is quite expensive.
-
As mentioned in your duplicate post, I wouldn't even bother with trying to measure such low resistances with a handheld DMM. Others have already touched upon the why's, but you will need either a dedicated meter or a very good bench DMM with 4 wire ohms.
At least in the handheld DMM realm I consider meters working in the single digit ohms range continuity checkers and not much more. Some meters might be better than others but I don't know of anything that would do what you are asking.
-
take a look here :
https://www.eevblog.com/forum/testgear/finding-short-on-motherboards-with-a-shorty-(with-display)/msg3446526/#msg3446526 (https://www.eevblog.com/forum/testgear/finding-short-on-motherboards-with-a-shorty-(with-display)/msg3446526/#msg3446526)
it will soon be easy to build one yourself !
the test current is around 60mA
-
8.5 digit DMMs like HP3458, Datron 1271/81 (and their successors by Wavetek/Fluke) have a 100Ohm range, so nominal 1uOhm resolution, but that'll be lost in noise. Historically, low resistances have been measured with Kelvin bridges. Today dedicated instruments or a combination of stable current source and nanovoltmeter are a more convenient option.
(Generally) you'll get more relevant answers by illustrating the problem you're trying to solve. You mention the DE5000 which is an LCR meter. Does that mean you want to measure LSR of capacitors?
-
a DC supply, and a DC current meter, and a handheld i mV mode
is all you need,
how many amps can the resistor handle (the one you want to measure)
then crank up the amps tru it, and measure its voltage drop,
use 2 x 2 wires, so the 4 points dont touch each other, but make it so current go to the resistor
and the voltage you measure, is picked up from the resistor in a point just a mm away from the current input point,
this way you get the correct resistance, at zero money used on new equipment
-
The new BM786 also has a 1mOhm resolution.
They don't have mOhm accuracy though - you might be able to tell the difference between a 0.01 and a 0.02 Ohm resistor but you can't measure an unknown milliohm-value resistor and know it's resistance.
Nicer side: Gossen 30M 4 wire 100uOhm resolution. If you can find one.
4 wire measurement is the only way to do it, ie.
The only way to do a lot better than that would be to use an instrument or homemade setup that uses a higher test current. If you use a bench supply to put 1 ampere through your resistance that you want to measure, all you need is a DMM that can reliably measure 1mV and you can measure 1 milliohm. For you purposes, that may do.
It's unlikely to be practical to pass 1 amp through graphics cards, maybe use a meter with uV range. :)
-
see this picture, the red arrow show the two points you pull current thru
and the two blue are the voltmeter points
if you do it like this over a cross of the miliohm or micro ohm part,
the voltage measures will be most possible accurate.
-
There are specialist meters like the Gossen 27 but it is really designed to measure the skin of helicopters (apparently) :
https://www.gossenmetrawatt.com/resources/tt/hit27/db_gb.pdf (https://www.gossenmetrawatt.com/resources/tt/hit27/db_gb.pdf)
it is also rather pricey - they do come up on ebay from time to time.
-
It's unlikely to be practical to pass 1 amp through graphics cards, maybe use a meter with uV range. :)
Why not? Most of them use more current than toasters or kettles. If the circuit in question can't take an ampere, then we likely don't need to measure it in microohms or even milliohms. Measuring uV or nV is a lab experiment, not a something you casually check with a DMM. A 2 degree (K) temp difference between identical copper alloys will give you a microvolt, the same difference between a stainless steel probe and an ENIG PC trace is probably in the 10s of uV.
This dedicated low ohm instrument uses up to a 1A test current. Some use even more.
https://www.chromausa.com/pdf/16502-E.pdf (https://www.chromausa.com/pdf/16502-E.pdf)
-
It's unlikely to be practical to pass 1 amp through graphics cards, maybe use a meter with uV range. :)
Why not? Most of them use more current than toasters or kettles.
I guess it depends on which part you're measuring. If it's a power rail then you don't need a milliohm measurement you can just plug it in and measure the voltage at various points along the rail.
This raises a question: What exactly is the OP trying to measure on the card?
-
It's unlikely to be practical to pass 1 amp through graphics cards, maybe use a meter with uV range. :)
Why not? Most of them use more current than toasters or kettles.
I guess it depends on which part you're measuring. If it's a power rail then you don't need a milliohm measurement you can just plug it in and measure the voltage at various points along the rail.
This raises a question: What exactly is the OP trying to measure on the card?
Apologies for not clarifying. Typically on the newer cards, the only gray area where it could either be a short or a low ohm area is measuring the vcore to ground. It's my understanding they have some pretty low ohms in that area and I'd like to be able to detect them in the future. I came across a YouTube channel called AHOC where he uses a Keysight U1272A that goes into the milliOhm range but uses two leads. I wanted to pop on here and gather people's opinions/recommendations on multimeters similar to this. I'd love to stay a little under something pricey like that U1272A, but I'm not the best at researching this topic. That being said, I'd love to drop some money on a quality Fluke, but I haven't found one that isn't an exorbitant amount of money that can measure at least in the milliOhm range (doesn't need to be four probe necessarily). Thanks again.
-
I built a low Ohm tester from an LM317 I think. It's four wire and could measure very low resistance. I tore it apart because I already have a few instruments for that.
My HP 3456A is excellent down to very low resistance and great resolution. The GR 1658 has four wire also and is almost as accurate as the HP. The HP 412A is an old unit that measures in a sort of quasi 4 wire. It's 4 wire right to the test lead ends but no farther. It's not super accurate but resolves to a couple percent maybe - it's an analog meter. certainly good enough for most purposes.
I can also make a quick setup with a power supply and voltmeter if that seems appropriate.
-
Apologies for not clarifying. Typically on the newer cards, the only gray area where it could either be a short or a low ohm area is measuring the vcore to ground. It's my understanding they have some pretty low ohms in that area and I'd like to be able to detect them in the future.
The circuit you are trying to measure may not be ohmic (obey Ohms law) and it also is sensitive to voltage so you would want a meter or test device that didn't have a very high open circuit voltage, like less than 1 volt. With some experience (trial and error) on both good and bad boards you might learn to distinguish between them with such a meter, but it certainly isn't a simple process.
-
Agilent 3458A w/ IET SRX-0.1 standard.
Also shown is an old Fluke 8506A I have been working on. Its a relic from the early 80's and was free in non-working condition. Maybe check your local salvage and scrap yards.
-
Ah yes my dream meter the 3458A.
-
The proper way:
Use it as an excuse to spend a grand on a DMM6500, or more on a 3458A (you wanted "something that has all the functions in one" - the DMM6500 does capacitance, 3458A probably not though :P)
The cop-out method:
Stick 1 amp through the circuit, use a DMM with a decent millivolts range (this is a good use for one of the cheap Aneng meters - one of their many ~$10 models has 1uV resolution on the lowest range) to look at the drop across the item of interest. Apply current in both directions if it's likely to be non-ohmic (this is actually good practice even if it is - there may be some thermal EMFs that you want to cancel out). Your choice as to how to interpret the results if they are different depending on current flow direction.
I have certainly used the latter method for measuring the resistance things like PCB planes and brass standoffs, but currently have a DMM6500 on order - the micro-ohm resolution was certainly an attractive feature, but no way I bought it just for that.
If you don't have an alternative LCR measuring device then the DE-5000 seems like a very useful piece of kit in the short time I've had mine. Even if it doesn't do this job then it would be nice to have regardless. Open-circuit test voltage is something like 1.5Vpk-pk though, so maybe no so good in-circuit.
-
The cop-out method may be fine in some cases but some parts may not like an Amp running through them and there's the problem self heating. The resistance / temperature may not even be linear. As always, the choice of method depends.
-
The cop-out method may be fine in some cases but some parts may not like an Amp running through them and there's the problem self heating. The resistance / temperature may not even be linear. As always, the choice of method depends.
The part that the OP was referring to may have hundreds of amperes going through it in normal operation. But the real problem is that the part or circuit in question may not even be a resistive in nature, so self-heating is not so much a problem in measuring the resistance as it is just a normal characteristic of the circuit. A resistance measurement at a very low current by some very sensitive instrument is probably not of any real value.
-
The cop-out method may be fine in some cases but some parts may not like an Amp running through them and there's the problem self heating. The resistance / temperature may not even be linear. As always, the choice of method depends.
The part that the OP was referring to may have hundreds of amperes going through it in normal operation. But the real problem is that the part or circuit in question may not even be a resistive in nature, so self-heating is not so much a problem in measuring the resistance as it is just a normal characteristic of the circuit. A resistance measurement at a very low current by some very sensitive instrument is probably not of any real value.
I have no idea what the OP is attempting to measure or what they are attempting to accomplish. I'm not there looking over their shoulders.
I can only offer that there are going to be cases that require low current measurements. These meters were not designed for the fun of it.
-
I have no idea what the OP is attempting to measure or what they are attempting to accomplish. I'm not there looking over their shoulders.
If you didn't watch the video (I sure didn't) he clarified in his second post that he was looking for short circuits in the Vcore circuit of a graphics card GPU. Nowadays those have voltages on the order of 1 volt and current up to a few hundred amps at full load. I don't know how those would measure in the non-operating state, which is why I suggested that specific experience with specific models would be needed.
I can only offer that there are going to be cases that require low current measurements. These meters were not designed for the fun of it.
Of course. And in other cases those measurements won't be helpful, which is one reason there are meters that use higher currents like one one I posted above. I actually don't know which the OP might need and apparently the person he is referring to does this with a handheld DMM. He didn't say what the readings were that he was getting for good and bad GPUs.
-
I have no idea what the OP is attempting to measure or what they are attempting to accomplish. I'm not there looking over their shoulders.
If you didn't watch the video (I sure didn't) he clarified in his second post that he was looking for short circuits in the Vcore circuit of a graphics card GPU. Nowadays those have voltages on the order of 1 volt and current up to a few hundred amps at full load. I don't know how those would measure in the non-operating state, which is why I suggested that specific experience with specific models would be needed.
I can only offer that there are going to be cases that require low current measurements. These meters were not designed for the fun of it.
Of course. And in other cases those measurements won't be helpful, which is one reason there are meters that use higher currents like one one I posted above. I actually don't know which the OP might need and apparently the person he is referring to does this with a handheld DMM. He didn't say what the readings were that he was getting for good and bad GPUs.
I only saw their reference to some AHOC channel, not a specific video. Per your point, I didn't try to hunt it down. I went to the channel and searched repair. Maybe this one.
https://www.youtube.com/watch?v=0K8hb9RwBrY (https://www.youtube.com/watch?v=0K8hb9RwBrY)
-
I still couldn't sit through it. :-DD
-
I still couldn't sit through it. :-DD
He really needs to switch to decaf and then learn a bit more about DMMs. But I was curious and watched enough to see that he ended up with a 100+ mOhms reading of the GPU core on a known good unit. But that still doesn't mean it behaves like a resistor--a V-I plot would be a more interesting bit of information. Perhaps DMMs should offer that--a 0-10mA sweep graphing a V-I plot.
-
I still couldn't sit through it. :-DD
He really needs to switch to decaf and then learn a bit more about DMMs. But I was curious and watched enough to see that he ended up with a 100+ mOhms reading of the GPU core on a known good unit. But that still doesn't mean it behaves like a resistor--a V-I plot would be a more interesting bit of information. Perhaps DMMs should offer that--a 0-10mA sweep graphing a V-I plot.
Aha, you've just come up with the reason to get into SMU territory - they're even more fun (and $$$) than DMMs.
Just to be clear, I suggested the "cop-out method" ("cop-out" because where is the fun in not using an excuse to acquire more TE?) knowing the limitations, but it sounded like the OP wanted to measure something pretty close to a dead short. I also commented on some of the limitations if you're not dealing with a purely resistive DUT. I probably should have noted that this is for things that you know can take an amp or so without sweating - my examples were big chunks of metal!
-
I probably should have noted that this is for things that you know can take an amp or so without sweating - my examples were big chunks of metal!
I designed a replacement current shunt for a low cost handheld meter. In that case, the only choice was to test it under load as I wanted to see how it drifted.
I have an odd ball Huntron meter that may work for the OP. I've never done anything with it outside of test to make sure it works. The basic idea is you use a good board for a reference.
https://www.youtube.com/watch?v=cusqSx8f264 (https://www.youtube.com/watch?v=cusqSx8f264)
https://www.youtube.com/watch?v=DHoEQivIdQQ (https://www.youtube.com/watch?v=DHoEQivIdQQ)
-
The portable (barely) Fluke 867B has a similar I/V component test, for troubleshooting faulty boards.
-
the huntron will tell you there is a short, but it will not help a lot to localize it precisely.
-
the huntron will tell you there is a short, but it will not help a lot to localize it precisely.
I have used something like the following to locate shorts. There was a company who made them commercially. Sorry, but it's been many years ago.
https://www.edn.com/short-circuit-tracer-finder-uses-low-power/ (https://www.edn.com/short-circuit-tracer-finder-uses-low-power/)