The voltage rating of the capacitor matters here also. For this reason it is useful to use capacitors with much higher voltage rating than they will experience in a circuit - so the capacitance under bias is closer to the rated capacitance.
@magnuswootton6181 yeah that's definitely possible. Just checked a good brand x7r ceramic capacitor and at its rated voltage the datasheet shows -50% in capacitance. For this same part, if the voltage is kept below 20% of the rated voltage there is no change in capacitance.
Actually, you could use the worst type of dielectric type capacitors defects as "features" such as: - Temperature sensors - Piezoelectric microphones - High capacitance voltage-controlled capacitors. Even, the unstable capacitance-vs-voltage could be used as a modulation effect in filters, for music synths, as the signal grows in amplitude, the filtering effect changes because of the dielectric variance.
The lower cost versions have lower upper frequency capabilities. The 2 mhz version is close to 20 grand with minimal options. Note that your HP 4275 also has the capability to add DC bias.
Happy Thanksgiving to two of my favorite TH-cam engineers! I appreciate all of the knowledge and experience you both share with the world, it’s certainly enhanced my understanding and appreciation for the science and craft. ☮️
I wonder about the Chinese, not just with capacitors but there appears to be a general acceptance to take junk out of spec components and sell them as genuine. Also defective chips falsely marked with fake brand logos etc. that do not work properly. It is almost as if they think they might sell a bunch of components and most hobbyists don't check them and even if some are returned they might still make a profit over all.
Your first, stable 1 nF capacitor looked to my eyes very much like what I have normally seen in Mica type. They are famous exactly for their excellent stability. The old, very expensive GR capacitor standards, to my understanding, were built with mica capacitors.
The observed behavior is expected and normal for this type of capacitors and it is documented in part’s datasheet - The capacitance decreases as DC voltage applied to the capacitor increases. The capacitance variance is not linear and it gets worse as you approach the nominal voltage. Considering the capacitors you tested, probably the first one has a much higher nominal voltage, so the capacitance drop is smaller for the same applied DC bias.
Hi @IMSAIGuy, I have a self-interested idea/request for a video if you still have the marvellous 6? digit LCR around. We all know of the poor man's way of making a more precise resistor: Put 10 of them in series or in parallel to make a more precise 10x or 1/10th value, at least after any systematic bias in the OEM's own ideal measurement. I'm wondering if in practice this can work to a good or bad degree with capacitors; since we can't get 0.01% caps. So some experiments I've thought of: 1. If you accurately measure four different stable caps, then solder two of them in parallel and two in series, how 'spot-on' do their series and parallel empirical values match up to the formulas? 2. Hopefully if you have some spare nice bunches 1% or even 5% MLCC caps around, does soldering 10 of them in series to a board produce any noticeable improvement in accuracy to a 10th of their value? And similar for parallel 3. If Q2 is a positive answer, does it work well down to small values? Specifically, can ten 100pF be soldered in series to produce a well behaved 10pF? Or do parasitics tend to ruin the attempt Since you have the very nice meter it might also be interesting to know how the other measured properties change, or don't change, when trialling this procedure. If so interested, you could give a cursory check of how well inductors line up to the same idea just for academic interest (say, just following Q1), but somehow I think most are more interested in capacitors. Many thanks if you consider it.
Could you examine the issue of change in measured capacitance with changing frequency? Is it an effect of parasitics? Also, we can't afford this machine but perhaps you can use it to assess how accurate our nanovna can be at measuring capacitance and inductance, albeit at higher frequencies?
Please reply anybody ..We have that LCR meter but not have that fixture probe( where we put the the capacitor or inductor). So can we measure the inductance or capacitance without it? Kindly answer me 🙏
Do we need to add a DC bias when testing polarized aluminum electrolytic capacitors with the LCR meter to get the most accurate reading for capacitance? For example test with a DC bias = 1 VDC with a 0.5VAC (RMS) signal on top, which results in the test voltage across the cap DC+AC never going below ground, so the capacitor is never reversed biased.
I though an AC sinewave output centered on 0 volts, goes negative during one half of the waveform or does the LCR meter automatically shift up the DC bias?@@IMSAIGuy
I don't fully understand, when you switch on the DC bias, the light comes up but in the screen it clearly states: VDC OFF. Seen your results on the capacitor it is clearly doing something. Why is the external source only 10VDC, if you can do internally already more than that? Or is that just about the current? I have only experience with the old 4284A and the newer E4990A, but not this model. Never used external bias sources. If you want to be horrified about capacitor non-linearities, you can take a look at the SMD MLCC (X5R type for example) capacitors. They will gladly derate 50% off on the rated voltage. Also it will change with AC level. Component providers like Murata will just tell you that up front.
Ever see the PEAK test instruments. I like the design, but I don’t like the fact that every little test is a different instrument. Couldn’t you design an instrument with basic volt/amp/resistance + plus a pulse generator that would perform most of the testing multiple instruments perform?
how many watt will it take for china vs good cap to go from 0v(nothing) to 10v 90+% of the 100nf, looks like the china one used mA for the slow change up... so how much watt is it using to get to the same 90+% 10v of the 100n value
This is not any specially bad capacitors from china, this is standard behavior for high dielectric constant ceramic capacitors. These type of capacitors are used for power supply decoupling in high quality electronics due to their small size. They were just unsuitable for this application unless you want an VCO. Check the data sheet for some name brand X5R capacitors, then you can see that they behave the same. You could even buy some name brand X5R capacitors and make a video about it.
Sometimes they might be, but this certainly isn't the lesson to be learned. However, the price matters, more stable dielectrics will usually cost you more (if available at all) for the same capacity/voltage rating etc. And "cheap" is relative, there might be even cheaper. Btw.: I wouldn't consider X7R "cheap" for values above 10nF or so. In contrast, C0G would be expensive, imho ;)
This is specific to ceramic type caps. Easiest would be not to buy unknown parts, but something like Murata ones (from good sources ofc) that are characterized over full working ranges including voltages, bias, temp.. in a form of very detailed interactive graphs. They know everything about their caps, it is really impressive and worth the cost.
You could AC-couple a capacitance meter through a high value capacitor, say 1mF electrolytic (at least 10x larger than the DUT so that it has negligible effect on the measurement), and apply a DC voltage through a high-value resistor, say 1M or 100k, (too high will make the large cap charge up too slow, too low will influence the measurement result)
@w2aew has a video about measuring capacitors with an oscilloscope "#90: Measure Capacitors and Inductors with an Oscilloscope and some basic parts" I bet with some tinkering you could make a jig that can provide a DC bias so that you can measure this on the Oscilloscope.
This is absolutely standard behavior for high dielectric constant ceramic capacitor materials - part of the reason you never use stuff like X5R or X7R caps in timing circuits. Why are you acting like this is some sort of big surprise?
Those little low-cost disc caps typically use BaTiO3 as their dielectric material, because it's very cheap. It also suffers from very high levels of capacity depression with applied voltage as well as significant dielectric absorption. For example, Murata (who are generally considered a quality parts supplier) tell you to expect up to 50% capacity depression at full rated voltage. If you must use ceramic caps in this sort of application then use NP0/C0G types since they not only have way better tempco but also much less capacity depression.@@IMSAIGuy
This behaviour is an order of magnitude worse than XR caps, because these cheapest 'high capacity' small ceramics are made of a different dialectic. So it's not absolutely surprising but still noteable how bad they are.
Your results might be perfectly normal for this type of cap. However, they might be made of cheese, and therefore the alkalinity may be too high, or they simply didn't add enough rice.🤣👍
The voltage rating of the capacitor matters here also. For this reason it is useful to use capacitors with much higher voltage rating than they will experience in a circuit - so the capacitance under bias is closer to the rated capacitance.
so is he just over volting them?
@magnuswootton6181 yeah that's definitely possible. Just checked a good brand x7r ceramic capacitor and at its rated voltage the datasheet shows -50% in capacitance. For this same part, if the voltage is kept below 20% of the rated voltage there is no change in capacitance.
@@magnuswootton6181 unlikely, most small ceramics should be 100, 50V or even 35V, which 10V isn't even half of
@@ivolol but he could be being silly and overvolting them off camera, then saying they are bad capacitors afterwards.
Actually, you could use the worst type of dielectric type capacitors defects as "features" such as:
- Temperature sensors
- Piezoelectric microphones
- High capacitance voltage-controlled capacitors.
Even, the unstable capacitance-vs-voltage could be used as a modulation effect in filters, for music synths, as the signal grows in amplitude, the filtering effect changes because of the dielectric variance.
How about a Voltage Controlled Oscillator? As long as you can tolerate the utter lack of PPMs!
The lower cost versions have lower upper frequency capabilities. The 2 mhz version is close to 20 grand with minimal options. Note that your HP 4275 also has the capability to add DC bias.
future video
Happy Thanksgiving to two of my favorite TH-cam engineers! I appreciate all of the knowledge and experience you both share with the world, it’s certainly enhanced my understanding and appreciation for the science and craft. ☮️
I wonder about the Chinese, not just with capacitors but there appears to be a general acceptance to take junk out of spec components and sell them as genuine. Also defective chips falsely marked with fake brand logos etc. that do not work properly. It is almost as if they think they might sell a bunch of components and most hobbyists don't check them and even if some are returned they might still make a profit over all.
Your first, stable 1 nF capacitor looked to my eyes very much like what I have normally seen in Mica type. They are famous exactly for their excellent stability. The old, very expensive GR capacitor standards, to my understanding, were built with mica capacitors.
Barium Titanate ceramic capacitors can also exhibit Piezo effect, meaning they can produce a voltage under dielectric stress.
I wish you had used different frequcies and also ESRs. The will tell a lot about the behavior of caps.
The observed behavior is expected and normal for this type of capacitors and it is documented in part’s datasheet - The capacitance decreases as DC voltage applied to the capacitor increases.
The capacitance variance is not linear and it gets worse as you approach the nominal voltage.
Considering the capacitors you tested, probably the first one has a much higher nominal voltage, so the capacitance drop is smaller for the same applied DC bias.
Hi @IMSAIGuy, I have a self-interested idea/request for a video if you still have the marvellous 6? digit LCR around.
We all know of the poor man's way of making a more precise resistor: Put 10 of them in series or in parallel to make a more precise 10x or 1/10th value, at least after any systematic bias in the OEM's own ideal measurement.
I'm wondering if in practice this can work to a good or bad degree with capacitors; since we can't get 0.01% caps. So some experiments I've thought of:
1. If you accurately measure four different stable caps, then solder two of them in parallel and two in series, how 'spot-on' do their series and parallel empirical values match up to the formulas?
2. Hopefully if you have some spare nice bunches 1% or even 5% MLCC caps around, does soldering 10 of them in series to a board produce any noticeable improvement in accuracy to a 10th of their value? And similar for parallel
3. If Q2 is a positive answer, does it work well down to small values? Specifically, can ten 100pF be soldered in series to produce a well behaved 10pF? Or do parasitics tend to ruin the attempt
Since you have the very nice meter it might also be interesting to know how the other measured properties change, or don't change, when trialling this procedure.
If so interested, you could give a cursory check of how well inductors line up to the same idea just for academic interest (say, just following Q1), but somehow I think most are more interested in capacitors.
Many thanks if you consider it.
Caps, schmaps. Did anyone notice the LCR meter! Wow. What a beauty. 😊
Is there a fix/calibration session coming for your box? Sure hope so! Happy Turkey Day Mark!
just bought the turkey yesterday!
Could you examine the issue of change in measured capacitance with changing frequency? Is it an effect of parasitics?
Also, we can't afford this machine but perhaps you can use it to assess how accurate our nanovna can be at measuring capacitance and inductance, albeit at higher frequencies?
Please reply anybody ..We have that LCR meter but not have that fixture probe( where we put the the capacitor or inductor). So can we measure the inductance or capacitance without it?
Kindly answer me 🙏
Do we need to add a DC bias when testing polarized aluminum electrolytic capacitors with the LCR meter to get the most accurate reading for capacitance? For example test with a DC bias = 1 VDC with a 0.5VAC (RMS) signal on top, which results in the test voltage across the cap DC+AC never going below ground, so the capacitor is never reversed biased.
the LCR meter only uses positive voltages, so no problem.
I though an AC sinewave output centered on 0 volts, goes negative during one half of the waveform or does the LCR meter automatically shift up the DC bias?@@IMSAIGuy
@@windward2818 maybe you right but the voltage is very small so I wouldn't worry.
It would be interesting to measure diode (and maybe photodiodes and led) junction capacitance changing the reverse bias.
future video
I don't fully understand, when you switch on the DC bias, the light comes up but in the screen it clearly states: VDC OFF. Seen your results on the capacitor it is clearly doing something.
Why is the external source only 10VDC, if you can do internally already more than that? Or is that just about the current? I have only experience with the old 4284A and the newer E4990A, but not this model. Never used external bias sources.
If you want to be horrified about capacitor non-linearities, you can take a look at the SMD MLCC (X5R type for example) capacitors. They will gladly derate 50% off on the rated voltage. Also it will change with AC level. Component providers like Murata will just tell you that up front.
Ever see the PEAK test instruments. I like the design, but I don’t like the fact that every little test is a different instrument. Couldn’t you design an instrument with basic volt/amp/resistance + plus a pulse generator that would perform most of the testing multiple instruments perform?
how many watt will it take for china vs good cap to go from 0v(nothing) to 10v 90+% of the 100nf, looks like the china one used mA for the slow change up... so how much watt is it using to get to the same 90+% 10v of the 100n value
Will you do DC sweeps using matlab, labview of smd caps. And compare it with e.g. simsurfing?
I was excited to go out and buy an LCR meter at about the mid point of the video and then that price tag appeared. PHEW! Maybe in another life lol
This is not any specially bad capacitors from china, this is standard behavior for high dielectric constant ceramic capacitors. These type of capacitors are used for power supply decoupling in high quality electronics due to their small size. They were just unsuitable for this application unless you want an VCO.
Check the data sheet for some name brand X5R capacitors, then you can see that they behave the same. You could even buy some name brand X5R capacitors and make a video about it.
the capacitor I compared it too was a cheap to be used as a decoupling only ceramic x7r. it seems to do much better
@@IMSAIGuy,Was the bias voltage the same percentage of the rated voltage in both cases?
no idea
These generally have an even worse material than X5/7R, by some order of magnitude
Is the lesson here that Chinese parts are not of high quality? (price of parts may correlate to part quality)
I'd say that the lesson is to read the datasheet for ceramic caps ;)
the capacitor I compared it too was a cheap to be used as a decoupling only ceramic x7r. it seems to do much better
Sometimes they might be, but this certainly isn't the lesson to be learned. However, the price matters, more stable dielectrics will usually cost you more (if available at all) for the same capacity/voltage rating etc. And "cheap" is relative, there might be even cheaper. Btw.: I wouldn't consider X7R "cheap" for values above 10nF or so. In contrast, C0G would be expensive, imho ;)
Is there a relatively simple way to test this variability with standard test gear?
This is specific to ceramic type caps. Easiest would be not to buy unknown parts, but something like Murata ones (from good sources ofc) that are characterized over full working ranges including voltages, bias, temp.. in a form of very detailed interactive graphs. They know everything about their caps, it is really impressive and worth the cost.
You could AC-couple a capacitance meter through a high value capacitor, say 1mF electrolytic (at least 10x larger than the DUT so that it has negligible effect on the measurement), and apply a DC voltage through a high-value resistor, say 1M or 100k, (too high will make the large cap charge up too slow, too low will influence the measurement result)
@w2aew has a video about measuring capacitors with an oscilloscope "#90: Measure Capacitors and Inductors with an Oscilloscope and some basic parts" I bet with some tinkering you could make a jig that can provide a DC bias so that you can measure this on the Oscilloscope.
This is absolutely standard behavior for high dielectric constant ceramic capacitor materials - part of the reason you never use stuff like X5R or X7R caps in timing circuits. Why are you acting like this is some sort of big surprise?
the capacitor I compared it too was a cheap to be used as a decoupling only ceramic x7r. it seems to do much better
Those little low-cost disc caps typically use BaTiO3 as their dielectric material, because it's very cheap. It also suffers from very high levels of capacity depression with applied voltage as well as significant dielectric absorption. For example, Murata (who are generally considered a quality parts supplier) tell you to expect up to 50% capacity depression at full rated voltage. If you must use ceramic caps in this sort of application then use NP0/C0G types since they not only have way better tempco but also much less capacity depression.@@IMSAIGuy
This behaviour is an order of magnitude worse than XR caps, because these cheapest 'high capacity' small ceramics are made of a different dialectic. So it's not absolutely surprising but still noteable how bad they are.
Hey! You've found a great source of voltage-controlled capacitance! Oh the PLLs you could make! I dare you infact! 🤣
Those are chocolate capacitors. Keep chocolate for cakes.
Your results might be perfectly normal for this type of cap.
However, they might be made of cheese, and therefore the alkalinity may be too high, or they simply didn't add enough rice.🤣👍
I will never use varicap diodes again, these Chinese capacitors are better 😂
silver mica
👍