I had a very similar problem back in 2010. In analog ion chambers dose rate is measured via transimpedance amplifier as chamber current (~10^-13A....10^-10A) and absolute received dose is measured as a tottal charge passed through chamber. That is done via precize integrator, and one of the problems was to find (and prove via measurement) ultralow leakage capacitors, also radiation hardened ones since ionising radiation disharges capacitors via production of charge inside capacitor insulator. Absolute hell. The instrumentation was so sensitive to everything. I swear to God it was almost sensitive to look at. 😂 At the end we did it by using kind of a sample hold circuit with electrometer amp to measure the disharge rate at known current. We had a guy from NIST working with my team. One of the most talented engineers I ever worked with. :) P.S. Also if you have the time and will, try finding capacitor leakage tester from youtuber and a fellow engineer Mr.Carlson. He did a great job designing very very sensitive indicator.
Yes, Mr Carlsons tester is definitely a cut above the rest, I built it and it's fantastic, has a forecast feature also. you have to be a patreon to get access to his design but I think he has a video demonstrating it somewhere.
A good thing about the resistor across an LED is that the forward voltage of the LED and the resistor set a point where there is no LED current. The proportional change in LED forward voltage with temperature is lower than you get with the E-B drop of a transistor.
If you have just an ordinary digital multi-meter that has an input impedance for voltage measurement of 10MOhm you can connect it via the capacitor to a dc source. The voltage you see in the display stands for the current of 100nA per Volt. (1/10Mohm) This is also 100pA/mV or 100fA/µV. For faster testing it is good to have a smaller resistor like 10kOhm to shunt the input of the multi-meter at the start. This method is not so common (I do not know why, maybe as leakage measurement is older than having digital multi-meters).
A long since obsolete Fluke 8060 A had a 2'000 nS conductivity scale. Conductivity - what is that, you may ask. In fact, instead of the IEC naming "siemens", the Americans used to call it "mho", for 'inverted' ohm, or 1/ohm. So, the 2'000 nS full scale has a resolution of 1 nanosiemens. The meter uses 9 V battery, so it cannot measure high voltage leakages, but the 1 nS already corresponds to 1 gigaohm. I used that to select some very low leakage diodes from a batch of 1N4148 parts. Well, nowadays I would just buy FDH 300 family parts (assuming they are still available). And, on the other hand, I have a Keithley electromeer, as well as an HP 4329A that has both high voltages and 10^15 ohms, or so, capability. Yet, I am finding the blue LED designs quite nice and interesting for DIY activities.
In his original video from 3 years ago, tomtestkit said, he adjusted that shunt resistor in that circuit, with that particular LED, to light under a leakage resistance of approximately 250 megaohms or lower, not for a particular capacitor value. @ 12:25 It would be impossible to "set up" a shunt resistor on a circuit like that to accurately cover indicating leakage in random value electrolytic capacitors that's why device manufactures like Sencore and Sprague ( along with a few others) created tables with "accumulated" capacitor manufacturer leakage rating info for components of different voltage rating and capacitance values.
I have a University Degree in Physics and Chemistry and I saw an Electrometer ONCE during a high school Physics class. The Warning Label stated: "Maximum Input Voltage 1 V". The readout device was a 0 to 100 microammeter.
Measuring leakage at the rated voltage is how I discovered fake rated chinese electrolytics. They were marked 16V but the leak was terrible, so I lowered the test voltage to 10V and the leak current dropped to a few micro-amps, just the normal value for an ok electrolytic. Just to prove they were re-marked as higher voltage, I put them at 16V as the sleeve said, and the ones I tested started to overheat and finally exploded!.
@@kensmith5694 Actually, they were correct in capacitance and ESR, so they were not terribly bad. I will just use them for small unimportant pojects...
I have a Keithley 160 Digital Multimeter with Nixie tube display. I wondered why it went down to nA and you have answered why at the end of the video. The DM was an estate sale find that was in the garage and not for sale with the rest because it didn't have a plug. My husband saw it, just saw digital multimeter on it, picked it up for $10 and i found an old monitor plug for it at home. We were surprised to fire it up and it had a tube display. He brought me home other great finds from that sale too. I love your channel! I watch often but comment little because I'm a newbie ee. I especially love the analog stuff. P.S. I also love chip of the day!
According to both Nichicon and Chemi-Con, electrolytics should be tested for leakage current at their rated voltage, and allowed to run under that voltage for five minutes. Chemi-Con ratings for KNG series, 105 Deg C: 6.3-100 VDC: I=0.03CV or 4μA, whichever is greater. 160-450 VDC: CV≦1,000 I=0.1CV+40 max. CV＞1,000 I=0.04CV+100 max. So, within the above voltage ranges and capacitances, it can range from 4 uA to 100+ uA max. A setup to supply 600 VDC at 1 mA, and using both a 50 uA and 200 uA meter, switchable between the two, along with a 0-600 VDC voltmeter, would be sufficient to cover the entire range. The 50 uA meter would be used for some not electrolytic as well. For metalized polypropylene, according to Illinois Capacitor: 30000 MΩ x uF “Not to exceed 3 GΩ” They do not give a maximum current, only a resistance. However, knowing that the maximum resistance is 3 gigaohms, and using a rating of 600 VDC the current would be: 2.0 × 10-7 amperes or 0.0000002: 200 nanoamperes.
I just use a multimeter and a power supply or more in series to achieve the desired voltage, i agree with your opinion about the test voltage, i always try to measure on the "high side voltages", i usually use the uA range but sometimes i also resort to the intrinsic 10Meg resistance in the voltage range, but very rarely. As you said practically all electrolytic has some leakage, i tend to estimate the allowable leakage according to the capacitance, usually i accept some couple of uA to tens of uA. It has been proven to be effective for my applications over the time.
My fluke 27 multimeter has a leakage tester that measures in the Giga ohms range. Of course, this is not testing rated voltage of the capacitor. Much like Mr Carlson's tester. I agree with testing at rated voltages cuz the behavior of insulation breakdown at a higher voltages like arching may be an undiagnosed issue.
To me this is a 'nice to have', but if a capacitor is suspect or old I just change it. If its old it could fail any time after testing. For the price of pocket change its not worth the risk. However this is an interesting topic and still educational so thank you.
On the other hand, new capacitors can have unusually high leakage as well. It’s just a luck of the draw. Typical capacitor specs are very conservative but production parts are much better. Then your app depends on that and when an occasional worse but in-spec part comes along there’s trouble.
The challenge is that for very low currents the input impedance is hard to control. A better approach if you want to improve accuracy would be to use a modern cmos opamp that has a very high input impedance.
The opamps for pH measurements have the highest impedance among easy obtainable parts. 100Meg impedance of measurement electrode is typical value, the impedance of opamp is much higher. If one needs to measure something in that range - use them. PH and reference electrodes are usually connected to 2 amps, connected like a buffer and all the other signal processing is done from their outputs.
Oh well, but it was typical for electrolytics to have mA of leakage back in time even when they were new. Caps would vary but in the range of uA. Modern devices are much much better so you need to actually build at least two test ranges into your tester. I agree many of the old caps and electrolytics will break down at rated voltage, often after some days or weeks even if they tested good initially. If it is a high voltage or current circuit I always change old devices.
Are you sure it was blue and not green LEDs that work best for this? I tested this with various color LEDs from an old Christmas string. To get a noticeable glow from the blue, it took 25uA. The green was brightly lit at under 1uA. I also tried UV LEDs and, although slightly better than blue, green is still the clear winner with the selection I have on hand. Thanks for sharing this trick!!!!!
@@IMSAIGuy OK, thanks for the confirmation. The old green ones I tried (according to my DMM) are noticeably lit down to around 300nA. Nothing else I have in the junk box comes close. What current do your blue ones light up visibly at? Maybe the age of these old LEDs has something to do with it because I tried a newer green one and the blue worked way better in that case. I guess I had better save these few old green ones I have left!
I asked on a electronics forum and someone made a good point. Some LEDs are also good at producing current in the presence of light, and these are terrible at that. This may explain why they worked so good with low current, no photo-induced current to counteract.
Thank you. I came across a chroma 11200, the 650V one. It says that one can test other components like diodes and transistors with it. What is your opinion on this model.
A picoammeter is a highly specialized instrument designed specifically for measuring extremely low currents, typically in the picoampere range or lower. It excels in applications requiring high precision, such as semiconductor testing. On the other hand, an electrometer is a more versatile instrument capable of measuring both low and high currents, as well as high-impedance voltage measurements. Electrometers find applications in a wide range of fields, including physics experiments, medical physics, and environmental monitoring. The key distinction lies in the specialized sensitivity of a picoammeter for precise low-current measurements, while an electrometer offers broader measurement capabilities across various current and voltage ranges.
@@andymouse Hi: AFAIK we called an instrument, or meter, that measured current, an ammeter... after Ampere... and the unit of current the amp. If it measured milliamps, we called it a milliammeter... if it measured picoamps, we called it a picoammeter. If the instrument could measure charge in coulombs (and maybe current, voltage, and resistance... but at least charge ... we called it an electrometer. If it could only measure static charge we called it a coulometer. AFAIK...an electrometer has an integrator (or something that acts like an integrator) in it... but an ammeter doesn't. And an electrometer measures charge on the scale of picoamperes.
Very interesting video. I would like to see this put up against Mr Carlson's model that shows capacitors to be bad at 30 volts that other high voltage testers test as good. Without giving away what's in "the secret sauce" his detector uses about 80 M ohms of input resistance to check at it's highest sensitivity. I would like to see that Keithly put up against it. Is Carlson's secret just that if you have enough sensitivity, you can pick up ANY leakage, even that which is "acceptable"? His is based on stacking up transistors in a similar method like you've done in your schematics, YET it doesn't condemn new components. It would be cool if you built his and compared it to the Keithly meter. I would love to see that. I am also curious if you had an op amp or jfet at the input of a version of his tester instead of a 3904 if it would be too sensitive to be practical? On the electrolytic capacitors, I have a variable high voltage power supply and a 100 micro amp analog meter that I put in series to test them and the meter dances all over the place while charging the capacitor up to it's rated voltage. I've noticed that it takes about ten minutes to really settle down, even with a good cap, but it always fluctuates even when its down to only a few micro amps of current. This video makes an interesting discussion, so feel free anyone who wants to comment on this. Thanks.
I had a very similar problem back in 2010. In analog ion chambers dose rate is measured via transimpedance amplifier as chamber current (~10^-13A....10^-10A) and absolute received dose is measured as a tottal charge passed through chamber. That is done via precize integrator, and one of the problems was to find (and prove via measurement) ultralow leakage capacitors, also radiation hardened ones since ionising radiation disharges capacitors via production of charge inside capacitor insulator. Absolute hell. The instrumentation was so sensitive to everything. I swear to God it was almost sensitive to look at. 😂
At the end we did it by using kind of a sample hold circuit with electrometer amp to measure the disharge rate at known current.
We had a guy from NIST working with my team. One of the most talented engineers I ever worked with. :)
P.S. Also if you have the time and will, try finding capacitor leakage tester from youtuber and a fellow engineer Mr.Carlson. He did a great job designing very very sensitive indicator.
Yes, Mr Carlsons tester is definitely a cut above the rest, I built it and it's fantastic, has a forecast feature also. you have to be a patreon to get access to his design but I think he has a video demonstrating it somewhere.
A good thing about the resistor across an LED is that the forward voltage of the LED and the resistor set a point where there is no LED current. The proportional change in LED forward voltage with temperature is lower than you get with the E-B drop of a transistor.
If you have just an ordinary digital multi-meter that has an input impedance for voltage measurement of 10MOhm you can connect it via the capacitor to a dc source. The voltage you see in the display stands for the current of 100nA per Volt. (1/10Mohm) This is also 100pA/mV or 100fA/µV.
For faster testing it is good to have a smaller resistor like 10kOhm to shunt the input of the multi-meter at the start.
This method is not so common (I do not know why, maybe as leakage measurement is older than having digital multi-meters).
A long since obsolete Fluke 8060 A had a 2'000 nS conductivity scale. Conductivity - what is that, you may ask. In fact, instead of the IEC naming "siemens", the Americans used to call it "mho", for 'inverted' ohm, or 1/ohm. So, the 2'000 nS full scale has a resolution of 1 nanosiemens. The meter uses 9 V battery, so it cannot measure high voltage leakages, but the 1 nS already corresponds to 1 gigaohm. I used that to select some very low leakage diodes from a batch of 1N4148 parts. Well, nowadays I would just buy FDH 300 family parts (assuming they are still available). And, on the other hand, I have a Keithley electromeer, as well as an HP 4329A that has both high voltages and 10^15 ohms, or so, capability. Yet, I am finding the blue LED designs quite nice and interesting for DIY activities.
In his original video from 3 years ago, tomtestkit said, he adjusted that shunt resistor in that circuit, with that particular LED, to light under a leakage resistance of approximately 250 megaohms or lower, not for a particular capacitor value. @ 12:25 It would be impossible to "set up" a shunt resistor on a circuit like that to accurately cover indicating leakage in random value electrolytic capacitors that's why device manufactures like Sencore and Sprague ( along with a few others) created tables with "accumulated" capacitor manufacturer leakage rating info for components of different voltage rating and capacitance values.
true, but most people don't have a 250M resistor laying around
Really enjoyed this; having a bit of a project in a video adds zest.
I have a University Degree in Physics and Chemistry and I saw an Electrometer ONCE during a high school Physics class.
The Warning Label stated: "Maximum Input Voltage 1 V".
The readout device was a 0 to 100 microammeter.
Measuring leakage at the rated voltage is how I discovered fake rated chinese electrolytics. They were marked 16V but the leak was terrible, so I lowered the test voltage to 10V and the leak current dropped to a few micro-amps, just the normal value for an ok electrolytic.
Just to prove they were re-marked as higher voltage, I put them at 16V as the sleeve said, and the ones I tested started to overheat and finally exploded!.
Yes and you also can't trust the capacitance number on them.
@@kensmith5694 Actually, they were correct in capacitance and ESR, so they were not terribly bad.
I will just use them for small unimportant pojects...
I have a Keithley 160 Digital Multimeter with Nixie tube display. I wondered why it went down to nA and you have answered why at the end of the video.
The DM was an estate sale find that was in the garage and not for sale with the rest because it didn't have a plug. My husband saw it, just saw digital multimeter on it, picked it up for $10 and i found an old monitor plug for it at home. We were surprised to fire it up and it had a tube display. He brought me home other great finds from that sale too.
I love your channel! I watch often but comment little because I'm a newbie ee. I especially love the analog stuff.
P.S. I also love chip of the day!
ua-cam.com/video/_8sn2DJL2I0/v-deo.htmlsi=cQ99JgF_Tjr2sn_F
The darlington circuit shown is the basis of Mr Carlson Lab Capacitor Leakage Tester project (ua-cam.com/video/LhovRIM5xAo/v-deo.html)
According to both Nichicon and Chemi-Con, electrolytics should be tested for leakage current at their rated voltage, and allowed to run under that voltage for five minutes.
Chemi-Con ratings for KNG series, 105 Deg C:
6.3-100 VDC:
I=0.03CV or 4μA, whichever is greater.
160-450 VDC:
CV≦1,000 I=0.1CV+40 max.
CV＞1,000 I=0.04CV+100 max.
So, within the above voltage ranges and capacitances, it can range from 4 uA to 100+ uA max.
A setup to supply 600 VDC at 1 mA, and using both a 50 uA and 200 uA meter, switchable between the two, along with a 0-600 VDC voltmeter, would be sufficient to cover the entire range. The 50 uA meter would be used for some not electrolytic as well.
For metalized polypropylene, according to Illinois Capacitor:
30000 MΩ x uF “Not to exceed 3 GΩ”
They do not give a maximum current, only a resistance. However, knowing that the maximum resistance is 3 gigaohms, and using a rating of 600 VDC the current would be: 2.0 × 10-7 amperes or 0.0000002: 200 nanoamperes.
I just use a multimeter and a power supply or more in series to achieve the desired voltage, i agree with your opinion about the test voltage, i always try to measure on the "high side voltages", i usually use the uA range but sometimes i also resort to the intrinsic 10Meg resistance in the voltage range, but very rarely. As you said practically all electrolytic has some leakage, i tend to estimate the allowable leakage according to the capacitance, usually i accept some couple of uA to tens of uA. It has been proven to be effective for my applications over the time.
I test them the same way: ua-cam.com/video/H1qMV_hTGGw/v-deo.htmlsi=DV7tbJ5IWyLfmLfO
Dave Cad, the technique made famous by the host of EEVBlog? Love both of y'all's work.
It's not DaveCad if it has straight lines.
Great tutorial. I don't think I knew what an electrometer was.
ua-cam.com/video/h4ktoYdkjlI/v-deo.htmlsi=pGGE3L_ljRoxJiDt
My fluke 27 multimeter has a leakage tester that measures in the Giga ohms range. Of course, this is not testing rated voltage of the capacitor. Much like Mr Carlson's tester. I agree with testing at rated voltages cuz the behavior of insulation breakdown at a higher voltages like arching may be an undiagnosed issue.
To me this is a 'nice to have', but if a capacitor is suspect or old I just change it. If its old it could fail any time after testing. For the price of pocket change its not worth the risk.
However this is an interesting topic and still educational so thank you.
On the other hand, new capacitors can have unusually high leakage as well. It’s just a luck of the draw. Typical capacitor specs are very conservative but production parts are much better. Then your app depends on that and when an occasional worse but in-spec part comes along there’s trouble.
there are expensive caps that you won't be changing willy-nilly
While you can't 100% rely on the beta of the transistors, could you not use some form of feedback to set it to a specific amount?
The challenge is that for very low currents the input impedance is hard to control. A better approach if you want to improve accuracy would be to use a modern cmos opamp that has a very high input impedance.
The opamps for pH measurements have the highest impedance among easy obtainable parts. 100Meg impedance of measurement electrode is typical value, the impedance of opamp is much higher. If one needs to measure something in that range - use them. PH and reference electrodes are usually connected to 2 amps, connected like a buffer and all the other signal processing is done from their outputs.
Yes you can. It is called bias. You can look up transistor circuit biasing to see how.
It is the way to go.
Very interesting indeed!
What about the Blue ESR meter? I have a kit for one I never built.
Oh well, but it was typical for electrolytics to have mA of leakage back in time even when they were new. Caps would vary but in the range of uA. Modern devices are much much better so you need to actually build at least two test ranges into your tester.
I agree many of the old caps and electrolytics will break down at rated voltage, often after some days or weeks even if they tested good initially.
If it is a high voltage or current circuit I always change old devices.
great content.
Are you sure it was blue and not green LEDs that work best for this? I tested this with various color LEDs from an old Christmas string. To get a noticeable glow from the blue, it took 25uA. The green was brightly lit at under 1uA. I also tried UV LEDs and, although slightly better than blue, green is still the clear winner with the selection I have on hand. Thanks for sharing this trick!!!!!
blue! green leds are some of the most inefficient
@@IMSAIGuy OK, thanks for the confirmation. The old green ones I tried (according to my DMM) are noticeably lit down to around 300nA. Nothing else I have in the junk box comes close. What current do your blue ones light up visibly at? Maybe the age of these old LEDs has something to do with it because I tried a newer green one and the blue worked way better in that case. I guess I had better save these few old green ones I have left!
I asked on a electronics forum and someone made a good point. Some LEDs are also good at producing current in the presence of light, and these are terrible at that. This may explain why they worked so good with low current, no photo-induced current to counteract.
@@Channel-xb4xt I worked on LEDs for 10 years. I don't thing that explanation is correct.
ua-cam.com/video/ClVLgFO52cY/v-deo.htmlsi=ZqfVxiNHNl7rLvL6
remember this is an april 1 video. all is true but the antenna part
Transistors also leak... depending on temperature...
hell, I leak sometimes
Thank you. I came across a chroma 11200, the 650V one.
It says that one can test other components like diodes and transistors with it.
What is your opinion on this model.
never seen one
i get usiyan application voltage if you're picking a voltage, but are there capacirors that will leak at one voltage but not another? (within reason)
yes, it is non-linear
Can someone explain the difference between a picoammeter, and an electrometer. (I have a feeling there is a significant difference.)
A picoammeter is a highly specialized instrument designed specifically for measuring extremely low currents, typically in the picoampere range or lower. It excels in applications requiring high precision, such as semiconductor testing. On the other hand, an electrometer is a more versatile instrument capable of measuring both low and high currents, as well as high-impedance voltage measurements. Electrometers find applications in a wide range of fields, including physics experiments, medical physics, and environmental monitoring. The key distinction lies in the specialized sensitivity of a picoammeter for precise low-current measurements, while an electrometer offers broader measurement capabilities across various current and voltage ranges.
Technically an electrometer measures charge. The term is often misused for picoammeters
@@kensmith5694 Ya... charge and current so easy to get confused, in a timeless world. But, who's timing it?
@@andymouse Hi: AFAIK we called an instrument, or meter, that measured current, an ammeter... after Ampere... and the unit of current the amp. If it measured milliamps, we called it a milliammeter... if it measured picoamps, we called it a picoammeter. If the instrument could measure charge in coulombs (and maybe current, voltage, and resistance... but at least charge ... we called it an electrometer. If it could only measure static charge we called it a coulometer. AFAIK...an electrometer has an integrator (or something that acts like an integrator) in it... but an ammeter doesn't. And an electrometer measures charge on the scale of picoamperes.
read: assets.testequity.com/te1/Documents/pdf/keithley/KeithleyLowLevelHandbook_7Ed.pdf
why can't I use my 0-100VDC PS & just put my Fluke 189 uA setting inline (in series) and measure 100V & below caps current flow (leakage) that way?
you can ua-cam.com/video/H1qMV_hTGGw/v-deo.htmlsi=VyRdOg7rTlWBPIsU
Which part am I going to actually am I going to learn being an automotive person Slow the down sorry for being slow.
I’m surprised you did not mention Mr Carlson’s capacitor tester
I think that's the point ;)
@@AshleyDawson-k4v Yup, roger that.
White LEDs seem to be even more sensitive
a white led is just a blue led with phosphor on top. so should be the same
Good
👌👌👌👍
Someone needs to take the camera zoom controls away from this.
I don't mind
Very interesting video. I would like to see this put up against Mr Carlson's model that shows capacitors to be bad at 30 volts that other high voltage testers test as good. Without giving away what's in "the secret sauce" his detector uses about 80 M ohms of input resistance to check at it's highest sensitivity. I would like to see that Keithly put up against it. Is Carlson's secret just that if you have enough sensitivity, you can pick up ANY leakage, even that which is "acceptable"? His is based on stacking up transistors in a similar method like you've done in your schematics, YET it doesn't condemn new components. It would be cool if you built his and compared it to the Keithly meter. I would love to see that. I am also curious if you had an op amp or jfet at the input of a version of his tester instead of a 3904 if it would be too sensitive to be practical?
On the electrolytic capacitors, I have a variable high voltage power supply and a 100 micro amp analog meter that I put in series to test them and the meter dances all over the place while charging the capacitor up to it's rated voltage. I've noticed that it takes about ten minutes to really settle down, even with a good cap, but it always fluctuates even when its down to only a few micro amps of current.
This video makes an interesting discussion, so feel free anyone who wants to comment on this. Thanks.
In tomorrow's video, I show opamp that would be used in a more modern version (instead of darlington). I should design a board.
Great! I can't wait. Take care.@@IMSAIGuy
It's hard to compare if he never would like the schematic public
Poors man tester 😂😂
When I see the word "electrometer" I think of a thing called a field mill. Kinda want to build one. en.wikipedia.org/wiki/Field_mill