About "How high is too high": usually fine cap has ESR several times lower than stated in the datasheet (2-3 times lower, correspondingly to my concrete ESR-meter). I remember (+-) those numbers from my experience, but I don't hesitate to look at the datasheet too. It is a question of how well you have used to your meter (you understand). (My ESR meter has 0.01 Ohm resolution - this is fine. It is harder to work with a 0.1 Ohm resolution meter).
If I see measurements (ESR) close to a datasheet value (for example, 0.05 R is a rated value and I see 0.04, instead of 0.02-0.03 Ohm) - then the cap is tired, it is somewhere at the end of his life. (It wants to be replaced). If I see 0.02-0.03 with 0.05 Ohm rated (about half of a datasheet value or less) - the cap is healthy and will live long.
But this small information is 80-95% enough. Most parameters deteriorate simultaneously. If I do a service or repair I definitely don't want to know EVERYTHING about all caps.
An oscilloscope is for the next, more serious level. When a first fast step (using simple meters - DMM/ESR) was not sucessfull.
Last edited:
You are right in general (but you don't have to worry about my exact case: I have a master's degree in electric engineering, so I definitely understand (almost) all assumptions. ( A repair is always assumptions.)
I think we don't have to disconnect parts at the first repair/service stage unless there is a reason. That is for some next repair stage when the first stage was not successful.
About leakage and dielectric condition: I have a leakage current in my mind, but it is a very rare reason in my experience (with my 'area' of devices). Possibly you work with high-voltage (> 100V) devices more than I so you have to worry about leakage more.
And, when I measure the DC voltage mode of the device (with only DMM), I have some info about possible leakages too.
Last edited:
Hi Vovk Z,
Okay, I strongly disagree with you.
No problem, but for young technicians and hobbyists, I think they should understand what they are doing. A repair is all about known quantities. Your methods are not acceptable and I would have either trained you properly, or dismissed you as an employee had you ever worked for me.
To use any meter, you absolutely must disconnect one lead. This isn't an option, it is the only proper way to do this. My early training after university was with an Austrian technician, I had to prove I was worth training. In the proper light, my university training is valuable, but I had to become humble and understand that very good service technicians knew a great deal more than I did and my assumptions about my engineering skills did not hold water. They taught me to break things down to what is actually going on and understand what I was doing. Of course everything seemed so simple before learning that many things matter that I discounted much as you are now. Once I was able to let go of my "superior training" I learned that even an old TV service technician knew things I did not, and that what he knew was based on what is really going on. So while they may only know a few things, they were well worth listening to. This allowed me to accept information and reach a very high level of service ability, and even prepared me for servicing higher quality electronic test equipment commercially. Later training in Metrology further squashed any thoughts of being superior to others (already humble by this time) and pounded home the truth that you must look at everything and accept that there is no room to make gross assumptions. In fact, calibrating instruments showed me that it never pays to take a short cut.
My very worst customers? Electronic engineers, computer technicians and programmers. The worst service I see? Electronic engineers and computer programmers (plus DJs, they take the cake). They all share the same trait. They assume they know more than everyone else. They also assume they know how things work.
Do me a favor and do not train anyone. You are not going to do them any favors.
-Chris
Okay, I strongly disagree with you.
No problem, but for young technicians and hobbyists, I think they should understand what they are doing. A repair is all about known quantities. Your methods are not acceptable and I would have either trained you properly, or dismissed you as an employee had you ever worked for me.
To use any meter, you absolutely must disconnect one lead. This isn't an option, it is the only proper way to do this. My early training after university was with an Austrian technician, I had to prove I was worth training. In the proper light, my university training is valuable, but I had to become humble and understand that very good service technicians knew a great deal more than I did and my assumptions about my engineering skills did not hold water. They taught me to break things down to what is actually going on and understand what I was doing. Of course everything seemed so simple before learning that many things matter that I discounted much as you are now. Once I was able to let go of my "superior training" I learned that even an old TV service technician knew things I did not, and that what he knew was based on what is really going on. So while they may only know a few things, they were well worth listening to. This allowed me to accept information and reach a very high level of service ability, and even prepared me for servicing higher quality electronic test equipment commercially. Later training in Metrology further squashed any thoughts of being superior to others (already humble by this time) and pounded home the truth that you must look at everything and accept that there is no room to make gross assumptions. In fact, calibrating instruments showed me that it never pays to take a short cut.
My very worst customers? Electronic engineers, computer technicians and programmers. The worst service I see? Electronic engineers and computer programmers (plus DJs, they take the cake). They all share the same trait. They assume they know more than everyone else. They also assume they know how things work.
Do me a favor and do not train anyone. You are not going to do them any favors.
-Chris
Chris that takes me back to one of my first engineering lectures (engineering didn't take but the lesson was memorable).
One of the very first things the professor told the class was that out in the plant the people who ran the equipment had an intimate, valuable knowledge to which it was well worth paying close attention.
One of the very first things the professor told the class was that out in the plant the people who ran the equipment had an intimate, valuable knowledge to which it was well worth paying close attention.
How does the ESR meter seperate the rest of the circuit from the cap? Like the other capspacitance (including parasitic) in parallel with the cap your testing?
Of cause, it can't separate one cap from another, if they are in parallel. But usually, this is not a problem but rather an inconvenience. We have this situation only in some cases where we really have several caps in parallel. E.g. motherboard power supply, or ATX power supply. In most other cases the answer to your question is very easy - the cap under test is usually the lowest impedance in that part of a circuit (if it is healthy). (take into consideration the ESR-meter typical test frequency 70-140 kHz).
Last edited:
I don't know why you don't understand me. I don't say we don't have to unsolder parts at all. I just say that we do a fast and simple check first, and then, on this or next step - we do smth more time-consuming, like unsoldering.
Of cause, you have to unsolder parts when you work with a typical RLC-meter because it uses quite large voltage levels (several one-tenth of a volt or more) and it needs that part was unconnected from the rest of a circuit. But the beauty of ESR metering is that a good (or a typical good) ESR-meter works with quite low voltage levels (somewhere near 0.1 VAC). This voltage level doesn't open p-n junctions. And a capacitor under test is usually the lowest impedance in that part of a circuit. That's why if we work with ESR-meter we may do that 'in-circuit', without unsoldering (and this is much faster).
When talking about using an ESR meter, it is only an AC ohmmeter. You have to consider all the people who read these posts and never post themselves and take into account every situation.
You don't know what all the test frequencies are, and 70- 140 KHz will read high resistance for most power supply capacitors. I can measure down to 50 mV, but this doesn't matter one bit. You are still measuring a circuit instead of the part.
This is really simple.
You are measuring the parallel combination of unknown parts at an unknown frequency (by your own admission) for a resistance value that is not at the specified values on the datasheet for the part (assuming you can find the datasheet for an 80 year old capacitor, or Chinese part from who knows where). If you are looking up the data sheet you have just wasted a lot of time and since you are dispensing with careful measurement, I know you are not looking this information up. In your case, speed=huge assumptions.
I will repeat this point and pound it home for you. You are using an AC ohmmeter that tests at some approximate level and frequency assuming you can set it somehow. You are sticking across the terminals of a capacitor in a circuit. You are assuming you are measuring the part the terminals are touching, but the test current is going everywhere. You do not know if there are any resonant elements in there. In fact, you don't know anything beyond where the test leads are connected. Everything else is a guess.
Your main attraction to an ESR tester is that you think it makes it easy for you. Ignorance is bliss is all I can say.
Would you test one resistor in a circuit with parallel components attached? The only difference is that the ohmmeter is probably more accurate than the ESR tester, and you are measuring at some unknown frequency that (from what you said) that is inappropriate for filter capacitors.
Okay, so you have a magic box that shows meaningless numbers on a display. It may as well have a large analogue meter with a "Good - Bad" scale. Not one reputable test equipment manufacturer makes one of these with actual specifications traceable to any standard. Not the last time I looked. So you are going to pin your reputation and charge your customers based on a toy.
Okay, fine. I do understand exactly where you are coming from, and I strongly disagree with you and laid out the reasons why. Finally, everyone should know that an in-circuit test is not accurate. Even the simplest thought experiment shows why this is so, yet you argue the point using arguments that clearly show you do not understand how an AC ohmmeter operates. I would have fired you.
That's it, you cannot justify your position at all. Time to hit the books and review ohm's law and AC circuit theory. Basic stuff. Debating this is a complete waste of time.
-Chris
You don't know what all the test frequencies are, and 70- 140 KHz will read high resistance for most power supply capacitors. I can measure down to 50 mV, but this doesn't matter one bit. You are still measuring a circuit instead of the part.
This is really simple.
You are measuring the parallel combination of unknown parts at an unknown frequency (by your own admission) for a resistance value that is not at the specified values on the datasheet for the part (assuming you can find the datasheet for an 80 year old capacitor, or Chinese part from who knows where). If you are looking up the data sheet you have just wasted a lot of time and since you are dispensing with careful measurement, I know you are not looking this information up. In your case, speed=huge assumptions.
I will repeat this point and pound it home for you. You are using an AC ohmmeter that tests at some approximate level and frequency assuming you can set it somehow. You are sticking across the terminals of a capacitor in a circuit. You are assuming you are measuring the part the terminals are touching, but the test current is going everywhere. You do not know if there are any resonant elements in there. In fact, you don't know anything beyond where the test leads are connected. Everything else is a guess.
Your main attraction to an ESR tester is that you think it makes it easy for you. Ignorance is bliss is all I can say.
Would you test one resistor in a circuit with parallel components attached? The only difference is that the ohmmeter is probably more accurate than the ESR tester, and you are measuring at some unknown frequency that (from what you said) that is inappropriate for filter capacitors.
Okay, so you have a magic box that shows meaningless numbers on a display. It may as well have a large analogue meter with a "Good - Bad" scale. Not one reputable test equipment manufacturer makes one of these with actual specifications traceable to any standard. Not the last time I looked. So you are going to pin your reputation and charge your customers based on a toy.
Okay, fine. I do understand exactly where you are coming from, and I strongly disagree with you and laid out the reasons why. Finally, everyone should know that an in-circuit test is not accurate. Even the simplest thought experiment shows why this is so, yet you argue the point using arguments that clearly show you do not understand how an AC ohmmeter operates. I would have fired you.
That's it, you cannot justify your position at all. Time to hit the books and review ohm's law and AC circuit theory. Basic stuff. Debating this is a complete waste of time.
-Chris
the cap was rated for 630 volts so i set my megger to 600vdc, still nothing wrong with it....anyway i can not go back now to test it at 2kv, the cap is now gone...
my regret is that now i will not have a chance to see that cap blow up...
Last edited:
the ESR meter is not used for measuring voltages but it is used to measure resistance, inductance and capacitance, and can identify the terminals of any three legged semiconductors, tell you what pins 1 2 and 3 are for....even tells you if your mosfet is an enhancement or a depletion type....a very handy device really...
to test an inductor that had a shorted turn, it will display just the resistance, and a wrong one at that, how to know? by comparing with an identical good one......and no inductance shown so you know the choke is bad....
never too late to use one...
i do not mean to say everyone should use one, only if you think you can use one...
to test an inductor that had a shorted turn, it will display just the resistance, and a wrong one at that, how to know? by comparing with an identical good one......and no inductance shown so you know the choke is bad....
never too late to use one...
i do not mean to say everyone should use one, only if you think you can use one...
Last edited:
just be careful, you can just be as guilty of the things you accuse others of...
Hi Tony,
Very true, and I am trying to be careful of that. That is one reason I explain everything carefully so it isn't just a statement. I have tried to state the why along with what I have been taught.
I want people to learn and think as well.
For inductors, the measurement you want is "Q". That is 1/DA. A shorted turn will really show up. You could also set up a ringing test. A good LCR meter is a lot easier to use and gives you the information you need directly.
Too bad that cap is long gone. I would love to know what was wrong with it. Some capacitors will heal from the heat of soldering. I would never trust it though.
-Chris
Very true, and I am trying to be careful of that. That is one reason I explain everything carefully so it isn't just a statement. I have tried to state the why along with what I have been taught.
I want people to learn and think as well.
For inductors, the measurement you want is "Q". That is 1/DA. A shorted turn will really show up. You could also set up a ringing test. A good LCR meter is a lot easier to use and gives you the information you need directly.
Too bad that cap is long gone. I would love to know what was wrong with it. Some capacitors will heal from the heat of soldering. I would never trust it though.
-Chris
Yes, of cause.
70 or 140 kHz - is close enough to typical 100 kHz (stated at the datasheet).
Yes, of cause.
I haven't had 80-year old electrolyte caps, but I had 50 years old soviet electrolyte caps a lot and they have very similar properties to actual general use caps, but only larger in size. The difference between a good cap, a tired cap, and a bad cap usually is clear enough (there is mostly quite a large margin between those three states). Usual circuit design is quite forgiving to some variation of part parameters.
Yes, of cause. It is going everywhere, but mainly it goes by the path I need.
E-e? Don't know what to say. I can only say: I agree with you again that - yes, I know less than a designer of that circuit (usually, but not always).
Yes, repair or service is a guess (usually it goes without having exact circuitry or w/o circuit at all).
Yes, I do that too. 🙂 For a resistor I'll have with 50% probability exact value, and 50% probability - some another value. And my task is to understand why there is a difference. Usually, we have to unsolder one leg of a resistor to check when it doesn't show its resistance "in-circuit". But there is a large difference in the amount of work between unsoldering all resistors or diodes, and unsoldering only part of them.
I don't have customers. I just do a repair or a service job sometimes for my own devices, or for devices at my current place of work, or for my friends, and mostly it is successful. I take devices from 'customers' very rarely and only when I aproximally know what a problem is and when I am sure enough that I can repair it.
Yes, of cause. I said several times that we are talking about repair or service, but not about 'accurate measuring' or metrology.
Last edited:
Hi Vovk Z,
All I am going to say is that you aren't measuring things properly. That's all.
The way you are doing things, you haven't got a clue really, and I bet you miss things. If the circuit works you would never know. A device isn't "repaired" until the operational issues are resolved ... and it is calibrated or set up to run at or below it's published specs for distortion and noise.
Most filter capacitors are characterized at 100 or 120 Hz, often you can see a spec for 10 KHz, but no higher. You appear to be working with switch mode power supplies only. As I said before, when talking about these things you have to include all possible uses. Most people will use this at line frequencies. That's the time when it's reading will be valid.
You should try to measure some accurate resistors with your tester.
-Chris
All I am going to say is that you aren't measuring things properly. That's all.
Exactly, a repair which should be a good repair. If I were to measure a component completely and accurately it would be completely removed from the circuit. The goal is to be able to measure a part at least to the accuracy of the test equipment at that range and frequency.
The way you are doing things, you haven't got a clue really, and I bet you miss things. If the circuit works you would never know. A device isn't "repaired" until the operational issues are resolved ... and it is calibrated or set up to run at or below it's published specs for distortion and noise.
Most filter capacitors are characterized at 100 or 120 Hz, often you can see a spec for 10 KHz, but no higher. You appear to be working with switch mode power supplies only. As I said before, when talking about these things you have to include all possible uses. Most people will use this at line frequencies. That's the time when it's reading will be valid.
You should try to measure some accurate resistors with your tester.
-Chris
If it is good or bad, or how good or how bad it is - I think it is rather a philosophic question. I can give you an example of my last repair, and we can see together if it was good or bad.
We have two devices "Canal-2" ("Канал-2") at one of the laboratories at my work, they are analog (tonal) channel simulators. They simulate analog telephone lines so students can practice how to measure basic line parameters (frequency response, attenuation, etc). And both two were rather bad but worked somehow (worked good/bad enough to carry out students laboratory work), and they totally died in the last year.
I'll attach a photo I made (I have only one photo, w/o front panel), so it was easier to understand what a device is.
It has regulated power supplies, and 5 large PCBs: mainly analog circuitry (old opamps, 741 clones), and one or two PCBs mixed or digital (old TTL IC) with a lot of mechanic switches and ICs.
We don't have (or couldn't find) any documentation.
I checked the power supply first, looked at PCBs, and realized that I have to check electrolyte caps first. So I took my ESR-meter and almost directly found dead caps (in-circuit). Middle-sized electrolytes were ok, but about half of all small electrolytes (10-47 uF 16V) were dead or between life and death. There was absolutely no reason to unsolder and measure them with RLC-meter. I had some amount of surplus caps and replaced all dead ones in the first device. The device started to work immediately - it was showing a signal similar to what we wanted to see.
The next day I bought more caps and replaced all electrolytes on PCBs in the second device.
Yes, of cause, there are a lot of possibly faulty switches in these devices - but we checked em first (before all of that) and they were not the main reason (it was done without my help, but I had checked them roo, w/o unsoldering, just by pressing on/off).
I haven't checked the parameters of all new caps, I just measured the ESR/capacity of a couple of them.
So, now we have two working devices instead of two dead ones. If it was a good or bad repair? Hmm, from my point of view it was a good job.
Attachments
Last edited:
Hi Vovk Z,
Given the age of those units, replacing capacitors was a good bet. You didn't really need any test equipment for that.
An oscilloscope would have allowed you to quickly chase signals through the unit. I would have probably pulled a couple capacitors to confirm my suspicions and simply replaced them all. This is the same conclusion you came to as well. The big difference would have been that I would have known exactly what condition the capacitors were in at the frequency range they actually operate in. I wouldn't measure each one since they are all probably in a similar condition - or will be soon.
So, have you measured the test frequency of your meter? Have you checked the accuracy against a fixed resistor? Those are two very basic things you need to know before checking your first capacitor.
For example, say you are testing at 50 KHz. You do know this is above the normal operating range of power supply capacitors. Therefore the meter isn't suitable for testing those types of capacitors. A 1 uF capacitor would measure higher ESR than the datasheet, but at least you can get a reading that means something. It would be fine for most film capacitors, but those are not normally the problem anyway.
To be more effective as a tool, you really need to characterize it at least approximately.
This one repair didn't need an ESR meter. All I needed was experience and a soldering iron really, the LCR meter only confirmed my suspicions and gave be real, supportable answers. The best your instrument allows you to say is that "they looked bad on my meter". That differs from being able to say with extreme confidence that they measured so much capacitance, and the dielectric was bad and measured x.xx for DA where the average good one in your case may have been 0.16 to 0.22. Poly-Aluminium caps would be a factor of 10 better approximately.
It's up to you with those kind of repairs. The ones I do are high quality audio or test equipment. If you are going to measure something, may as well do it properly and have real answers. The dielectric absorption test is critical in many situations, it certainly is when deciding which capacitors to buy and use for example.
What can I say, it seems to work for you. But your repairs are non-critical and signal fidelity isn't a factor. This is not true of most equipment, and given that this is an audio web site, we'll assume that the advice is for working on audio equipment. I can't recommend your methods as a result.
-Chris
Given the age of those units, replacing capacitors was a good bet. You didn't really need any test equipment for that.
An oscilloscope would have allowed you to quickly chase signals through the unit. I would have probably pulled a couple capacitors to confirm my suspicions and simply replaced them all. This is the same conclusion you came to as well. The big difference would have been that I would have known exactly what condition the capacitors were in at the frequency range they actually operate in. I wouldn't measure each one since they are all probably in a similar condition - or will be soon.
So, have you measured the test frequency of your meter? Have you checked the accuracy against a fixed resistor? Those are two very basic things you need to know before checking your first capacitor.
For example, say you are testing at 50 KHz. You do know this is above the normal operating range of power supply capacitors. Therefore the meter isn't suitable for testing those types of capacitors. A 1 uF capacitor would measure higher ESR than the datasheet, but at least you can get a reading that means something. It would be fine for most film capacitors, but those are not normally the problem anyway.
To be more effective as a tool, you really need to characterize it at least approximately.
This one repair didn't need an ESR meter. All I needed was experience and a soldering iron really, the LCR meter only confirmed my suspicions and gave be real, supportable answers. The best your instrument allows you to say is that "they looked bad on my meter". That differs from being able to say with extreme confidence that they measured so much capacitance, and the dielectric was bad and measured x.xx for DA where the average good one in your case may have been 0.16 to 0.22. Poly-Aluminium caps would be a factor of 10 better approximately.
It's up to you with those kind of repairs. The ones I do are high quality audio or test equipment. If you are going to measure something, may as well do it properly and have real answers. The dielectric absorption test is critical in many situations, it certainly is when deciding which capacitors to buy and use for example.
What can I say, it seems to work for you. But your repairs are non-critical and signal fidelity isn't a factor. This is not true of most equipment, and given that this is an audio web site, we'll assume that the advice is for working on audio equipment. I can't recommend your methods as a result.
-Chris
I think most repairs in the audio world are gross fails and not degraded signal integrity due to components running out of specs. And these simplified methods easily detect failed parts without "knowing what to do" - i.e. understanding the hole circuitry. Leaking coupling caps in tube circuits can be checked without removal by measuring grid voltage of the current stage - or measuring grid current after removal of the tube.
When I was a student I repaired a HP3000 computer with the simple component tracer of a Hameg scope, comparing TTL-pins of working pcbs with the defective ones. Without removing any chip for measurement. No circuit diagram available. I found and replaced all defective chips. All it took was a full weekend of measurements...
When I was a student I repaired a HP3000 computer with the simple component tracer of a Hameg scope, comparing TTL-pins of working pcbs with the defective ones. Without removing any chip for measurement. No circuit diagram available. I found and replaced all defective chips. All it took was a full weekend of measurements...