Do GPUs have an open circuit on a single capacitor?

[Adam187a]

I recently acquired a faulty MSI 1060 6GB Armor in an attempt to repair it. The GPU has a fault where it could boot up and work fine, but once crashes refuses to work. Sometimes graphical glitches appear and the computer will crash to a grey display. I changed the thermal pads and paste to no luck, this lead me to check the capacitors.

After checking the capacitors for any obvious damage, I decided to use the multimeter to check for any unusual readings, I used the continuity to find an odd number. Most capacitors were fine, there was this one capacitor closest to the ports that briefly changed numbers before returning with a 1, indicating an open circuit. I'm not an expert at this, so I had gone and asked someone if this was normal, he was no expect either and simply replied that because it's a 1, it's known as an open circuit, apparently this capacitor could be the cause of the GPU problems and to change it would require me to drain the other capacitors of their voltage before fiddling with it. Just want to confirm this with an expert before I continue or decide to get rid of it.

Would like to share a picture but can't seem to find out how to attach it? (Using mobile)

 

[bigpinkdragon286]

You can't accurately test capacitors in circuit, period.

You can however get close.

You can't do this with a multimeter.

With an ESR meter, you can check the ESR value across a capacitor while the capacitor is connected in a circuit, which while it is usually "good enough," does still have problem cases where it isn't always accurate.

If you need to accurately test a capacitor, and you don't have access to an ESR meter, disconnect one leg from the circuit before you test it.

 

[Adam187a]

Thanks for the advice but didn't answer the question, what you said was that my findings were void since a multimeter is simply not good enough to test a GPU circuit, even when it finds an apparent open circuit? I just wanted to know whether or not that capacitor should have returned with a 1 or not. Multimeter may not be accurate, but there's no need to have any accurate readings in this situation, unless the multimeter is completely unreliable when it comes to checking open circuits in a GPU

 

[bigpinkdragon286]

It isn't a matter of your meter being good enough, and I hope that wasn't the impression that came across.

A multimeter is the wrong tool for what you are trying to do, which is test a capacitor while it's connected in a circuit. You aren't measuring the resistance of the capacitor in isolation, but rather the resistance of the capacitor and everything else that is in parallel with it. The complexity or application of the circuit, such as the fact that it's on a graphics card, really has nothing to do with why you can't test the capacitor in-circuit with your multimeter.

Here's an article that will guide you through the methods you can use to test a capacitor using a multimeter. Regardless of which test you use with your multimeter, you must disconnect the capacitor from the circuit in order to perform these tests, otherwise you are not testing the capacitor. You are testing the capacitor plus the rest of the circuit in which it is attached, which usually doesn't tell you anything useful in regards the health of the capacitor.

When the need arises to test a capacitor in-circuit, as you are doing, you need a tool that can perform such a test, and that is known as an ESR meter (Equivalent Series Resistance). Even an ESR meter won't work in all cases however, as if you have multiple capacitors in parallel, you may get incorrect readings from an ESR meter.

ESR meters don't have to be expensive, but can easily range to very expensive for name branded testers, but if you have a need for testing a large number of capacitors often, you will find having one in your toolbox indispensable.

If this is the only time you ever expect to be testing capacitors, it may not make financial sense to invest in an ESR meter, and just test the capacitors out of circuit with your multimeter.

Depending on how the capacitor is attached to the circuit, if you can disconnect even a single leg of the capacitor, that is sufficient to isolate it from the circuit.

 

[Adam187a]

I was talking about in general, from your first post I had picked up as a fact that a multimeter is not sufficient for fault finding in a GPU, but does that mean it's also not reliable when it comes to identifying an open circuit in the GPU? I was just testing the continuity from both legs of a single capacitor, and spotted that one of them returned with 1, this had me thinking whether it was normal or not, hence why I turned to Tom's Hardware.

In regards to testing the capacitor with it being disconnected, is it safe to solder or use a heat gun to remove the capacitor directly from the board of the video card or must I take other measure beforehand as advised by another person? I'm aware of how to measure capacitors using a multimeter, so it'll be straight forward from thereon.

Also want to thank you for the informative descriptions on what an ESR meter is, it's likely I'd end up using one for long term use now that I know what it's intended use is for

 

[bigpinkdragon286]

A multimeter is likely going to tell you that a circuit is open or has infinite resistance when there's no current flowing, but this may be a correct electrical property of the circuit or part under test. A method of determining resistance that multimeters use is to inject a known amount of current and just measure the voltage drop, from which you can use Ohm's law to calculate the unknown resistance.

If the injected current is not sufficient to activate a PN junction such as in a diode or transistor, or charges a capacitor, you're not going to see current flowing, or you'll see it flow briefly and then stop. Once a capacitor charges, unless it's leaky or is discharged in some way, it should stop flowing current. A standard multimeter uses DC when measuring, so in the case of trying to measure a capacitor, it's only going to charge a single time and then should appear as an open. The size of the capacitor and charging source determine how fast this happens. On small capacitors, the charge time can easily be faster than you can witness with a digital multimeter. Nothing wrong with the meters, they just aren't fast enough to measure such rapid changes.

I wouldn't use a heat gun on a PCB if I didn't have to. A simple soldering iron should be sufficient to remove most components that can be safely removed without specialized tools. If you have to spend more than a few seconds per joint, your iron may be of insufficient power, the tip may not be well tinned, or you may even be dealing with a part that in normal usage is expected to dissipate a great deal of heat and happens to be attached to an extended copper plane designed to draw heat away from the part, requiring a different type of iron tip that doesn't lose temperature as rapidly. Some desoldering tape may also come in handy.