Voltage coefficient of HV ceramic caps / High Voltage / Forums

Forums

4hv.org :: Forums :: High Voltage

« Previous topic | Next topic »

Voltage coefficient of HV ceramic caps

1 2 next

Move Thread

LAN_403

woodchuck

Tue Jan 05 2016, 05:49PM

Registered Member #39190 Joined: Sat Oct 26 2013, 09:15AM
Location: Boise National Forest
Posts: 65

For many years I've been vexed by the fact that all but C0G/NP0 ceramic capacitors 'lose' capacitance as the DC voltage across the plates rises. In a worst case situation a 4.7uF capacitor might have only a 0.33uF capacitance once the rated voltage has been reached.

Further adding to the frustration is the fact that the three character code (eg X7R) says nothing about this voltage coefficient. The code only refers to things like precision and temperature stability - not voltage stability.

So, has anyone tested cheap Chinese HV ceramic capacitors (such as these http://www.ebay.com/itm/20KV-1000PF-High-Voltage-Ceramic-Disc-Capacitors-HV-5pcs-/130664159798?hash=item1e6c30da36:g:2x0AAMXQhpdR3PdZ) for voltage stability? I would expect a 20% loss by the time 20 kV was reached, and might even tolerate a 40% loss. But an 80% loss would be right out!

Sulaiman

Tue Jan 05 2016, 08:59PM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

From the few experiments and a fair bit of reading that I've done,
I think that ceramic capacitors with high C.V/volume have poor voltage characteristics,
so I would expect the caps that you linked to will almost certainly have a large loss in capacitance at rated voltage.

woodchuck

Tue Jan 05 2016, 10:36PM

Registered Member #39190 Joined: Sat Oct 26 2013, 09:15AM
Location: Boise National Forest
Posts: 65

This makes perfect sense. People wouldn't be making big, fat, heavy HV ceramic caps if small, thin, light caps would work just as well. Call it wishful thinking... Most scams occur when gullible (or greedy) people think they're getting the deal of a lifetime.

So I just ordered some caps with N4700/ZM dielectric from a major player who publishes a graph of DC bias characteristics for them. They weren't outrageously expensive and they lose only a little over 20%.

Thanks Sulaiman.

BTW: Who needs a varactor diode anymore? Just use a couple Y5V ceramic caps.

woodchuck

Fri Jan 08 2016, 12:58AM

Registered Member #39190 Joined: Sat Oct 26 2013, 09:15AM
Location: Boise National Forest
Posts: 65

It was bugging me not knowing more about mystery capacitors from China so I arranged a test. I connected two 15kV 1000pF disc ceramic caps (that I ordered from China in 2014) in series to make a 500pF cap. This was attached to a capacitance checker (making sure to ground its chassis). Sure enough, 500pF was measured.

Then, I slowly increased the voltage of the node between the two capacitors using an external supply. The voltage was slowly cranked up all the way to 10kV. I'd expected to see a notable decrease in capacitance. Instead I saw a 20% increase.

This is very puzzling. I'll have to repeat the test with a different capacitance meter - perhaps a 555 timer connected to the series capacitors.

Along with that test, I intend to leave the capacitors charged to 10 kV for many hours to see if dielectric depolarization has a time element.

Russ Edmonds

Fri Jan 08 2016, 05:04PM

Registered Member #46264 Joined: Sun May 11 2014, 05:27PM
Location: Tucson, Arizona USA
Posts: 61

I would ground the outer ends of the two HV caps as you apply voltage to their inner node then remove the ground connection at the ends and connect the meter. If you don't do this you might be inducing voltage in the capacitance measuring circuit of the meter and effecting the capacitance reading. Be careful, this sounds a little dangerous for your meter!

woodchuck

Fri Jan 08 2016, 06:09PM

Registered Member #39190 Joined: Sat Oct 26 2013, 09:15AM
Location: Boise National Forest
Posts: 65

I was concerned about the meter, too. So much so that before performing the test mentioned above, I did a little dress rehearsal...attaching a 220k resistor from one outer capacitor lead to ground. Then I put a voltmeter across that resistor while slowly increasing HV to the center node of the cap combo. What I observed on the voltmeter reassured me that it was probably safe to do the real test.

Nevertheless, it was a gamble. That's why I used a relatively inexpensive but fairly trustworthy Tenma meter. :)

Still, I think the 555 timer idea is a better idea for future tests. The plan is to power it by battery and attach a speaker (via electrolytic cap and resistor) to the 555's output. Then I'll use an app on my cellphone to measure the frequency of the tone coming out of the speaker. Nothing valuable will have to get near the HV.

I've also ordered small quantities of a few other HV ceramic caps. Since I'm on the trail, I may as well complete the picture. (Yeah, too many metaphors.)

johnf

Fri Jan 08 2016, 06:39PM

Registered Member #230 Joined: Tue Feb 21 2006, 08:01PM
Location: Gracefield lower Hutt
Posts: 284

I'm not surprised that the capacitance goes up as the voltage increases.
Most ceramic dielectrics exhibit some piezo effect. I learnt the hard way many years ago when I decoupled a tune line into a vco with a ceramic cap. I now had a vco that was extremely sensitive to vibration that modulated the output.
also if you drive a CW stack at audio frequencies you will hear the capacitors sing at the drive frequency meaning that they are changing shape Constricting)with the applied voltage.
However caps loose capacitance with rising frequency due to internal losses with include dielectric absorption and resistive losses in the metalised layers

Russ Edmonds

Fri Jan 08 2016, 10:44PM

Registered Member #46264 Joined: Sun May 11 2014, 05:27PM
Location: Tucson, Arizona USA
Posts: 61

The 555 timer idea got me interested in doing a voltage effect on capacitance experiment myself. I used a .33 and 5 nF ceramic caps rated at 1Kv in series in a 555 circuit. I used a transformer, diode and capacitor from a microwave oven for the HV power supply. I put 3 megaohms of resistance on the output of HV power supply to limit current output. With the power supply not connected to the middle node of the ceramic caps the output frequency of the 555 circuit was 2.027 KHz. When I put the lead from the power supply on to the middle node the caps the frequency dropped to 1.6 KHz because of the capacitance loading of the power supply. To overcome this I had to first set the voltage on the power supply and then briefly connect and disconnect the power supply to the middle node and take a frequency measurement. After converting frequency to capacitance here are the results

Voltage C (nF)
0 .320
95 .322
200 .321
302 .317
410 .313
536 .306
613 .302
739 .295
830 .288
917 .283
1008 .274

I get a 14% decrease in capacitance at 1 KV
Russ

woodchuck

Sat Jan 09 2016, 06:29PM

Registered Member #39190 Joined: Sat Oct 26 2013, 09:15AM
Location: Boise National Forest
Posts: 65

This is a fascinating result! So the capacitance loading of the power supply was causing the erroneous results. I figured I had a high enough resistance between the HVPS and the center node. Not so. I guess there isn't "a high enough resistance." Anything with a resistance low enough to charge the capacitors being tested will have a low enough resistance to influence the outcome.

So a quick disconnect has to be provided... Russ, thank you very much for solving this mystery! And thanks for results that don't lead to insanity. :)

johnf

Mon Jan 11 2016, 07:47PM

Registered Member #230 Joined: Tue Feb 21 2006, 08:01PM
Location: Gracefield lower Hutt
Posts: 284

I found this to add to the discussion

]2006_01_arrowasiatimes_- _voltage_coefficients.pdf[/file]

This effect seems to be more apparent with the newer high dielectric constant dielectrics

1 2 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob