High frequency capacitor selection

[ZeroResistance]

I need some capacitors for high frequency resonant oscillators, these capacitors also need to handle high frequency resonant currents and should be of low loss. The frequency range I am looking at is between 1 Mhz to 15 MHz.

I was looking into polypropylene capacitors for these and while looking into data sheets I found something strange.
The AC Voltage rating of the capacitor drops with frequency. So If I select a 250VAC / 1000VDC, 1nF capacitor. At a frequency of 1Mhz the AC voltage rating drops to 90VAC rms.

I want a capacitor at least for 13.56 MHz frequency with AC voltage of 250V rms. What would be a suitable capacitor for this frequency?

There are a few graphs in the datasheet that I have attached.
Also a equation that doesn't make sense to me  "AC voltage: f <= 1000 Hz; 1.4 x Urms + UDC <= Ur"

[sourcecharge]

V-caps are teflon and copper.

They are expensive though, and the lowest cap is 10nF.

Never tried them but I'm guessing they would do the trick.

https://www.v-cap.com/cutf-capacitors.php

[ZeroResistance]

V-caps are teflon and copper.

They are expensive though, and the lowest cap is 10nF.

Never tried them but I'm guessing they would do the trick.

https://www.v-cap.com/cutf-capacitors.php

That's interesting never heard of these before, and I thought that polypropylene capacitors had the lowest loss particularly for high frequency applications. Aren't there any polyproylene capacitors that work at this frequency range with a suitable volatge range of 250Vrms. Or is connecting capacitors in series to get the desired voltage rating the only way out?

[Twoflower]

I think the cap selection of choice for high frequency requirements are Silicon Caps: https://www.murata.com/products/capacitor/siliconcapacitors

[helius]

Also a equation that doesn't make sense to me  "AC voltage: f <= 1000 Hz; 1.4 x Urms + UDC <= Ur"

It is talking about peak instantaneous voltage. 1.4 x Vrms + Vdc is the (positive) peak of a sinusoid.

[ZeroResistance]

Also a equation that doesn't make sense to me  "AC voltage: f <= 1000 Hz; 1.4 x Urms + UDC <= Ur"

It is talking about peak instantaneous voltage. 1.4 x Vrms + Vdc is the (positive) peak of a sinusoid.

OK, so if Urms = 250VAC and if I'm correct UDC is the DC level on which this Sine wave is superimposed on? correct?
Then that equation is 1.4 * 250 + UDC (lets say UDC is 0)  so that comes to 350V.
So is that equation trying to say that 350 should be <= Ur for a frequency of less that 1000Hz? What about frequencies above 1000Hz?
And what is Ur? is it the max rated voltage of the capacitor?

[ZeroResistance]

I think the cap selection of choice for high frequency requirements are Silicon Caps: https://www.murata.com/products/capacitor/siliconcapacitors

Yes they seem to be for high frequencies in the GHz range.. but can they handle high sinusoidal currents running thru them? Something in the order of 10 to 20Amps peak.

[Alex Nikitin]

http://www.atceramics.com/

Cheers

Alex

[ZeroResistance]

OK, so good alternatives available  from Silicon, Teflon and ceramic capacitiors. However I would like to know why does the capacitor rated AC volatge decrease as freqency increases? Anyone has any info regarding this?

[helius]

So is that equation trying to say that 350 should be <= Ur for a frequency of less that 1000Hz?

Affirmative.

What about frequencies above 1000Hz?

The permissible peak voltage typically decreases with higher frequencies, as the capacitor's impedance decreases until the resonant frequency is reached. The specific phenomenon that weakens the voltage withstood is called "dielectric loss".

DF [dissipation factor] is frequency-dependent and displays a small increase with increasing frequency.

And what is Ur? is it the max rated voltage of the capacitor?

Yes. It will be spelled out in the datasheet: capacitors can have a Surge Rating higher than the Working Voltage rating (only the latter is usually printed on it), which would be the limit in this case.

[ZeroResistance]

The permissible peak voltage typically decreases with higher frequencies, as the capacitor's impedance decreases until the resonant frequency is reached. The specific phenomenon that weakens the voltage withstood is called "dielectric loss".

DF [dissipation factor] is frequency-dependent and displays a small increase with increasing frequency.

Ok, so all capacitors will face this roadblock right? At higher frequencies eventually their Xc will be a limiting factor right? For eg. for a 1nF capacitor the Xc comes to 160 ohm for a frequency of 1Mhz. So how come the ceramic capacitors are used for RF power at frequencies in the 10's of Mhz to 10's of Ghz?

[iMo]

At higher frequencies eventually their Xc will be a limiting factor right? For eg. for a 1nF capacitor the Xc comes to 160 ohm for a frequency of 1Mhz. So how come the ceramic capacitors are used for RF power at frequencies in the 10's of Mhz to 10's of Ghz?

All types of capacitors have a quality Q. The quality depends on "parasitic" serial/parallel R, L, and frequency. The same with ceramics for 0.01-100GHz.
The 1nF polypropylen bulk Wimo could have 500nH L, 2000ohm R (because of its mechanical construction).
The ceramics for xxxxMHz are of different construction - those are leadless pieces of ceramics of special shapes. The 1nF may have 0.xxnH L and 0.xxohm R for example.

[Alex Nikitin]

The permissible peak voltage typically decreases with higher frequencies, as the capacitor's impedance decreases until the resonant frequency is reached. The specific phenomenon that weakens the voltage withstood is called "dielectric loss".

DF [dissipation factor] is frequency-dependent and displays a small increase with increasing frequency.

Ok, so all capacitors will face this roadblock right? At higher frequencies eventually their Xc will be a limiting factor right? For eg. for a 1nF capacitor the Xc comes to 160 ohm for a frequency of 1Mhz. So how come the ceramic capacitors are used for RF power at frequencies in the 10's of Mhz to 10's of Ghz?

http://www.atceramics.com/technical-notes.aspx

Cheers

Alex

[iMo]

@ZeroResistance: better you define the required capacity (in nF/uF) at specific frequencies, max currents (Ip-p), voltages(Vp-p), and other params you are aware of, and the experts here may identify the proper caps for you easier.

[dmills]

The reason the applied voltage limit drops with frequency, is that you are really hitting a current limit!

Xc drops with increasing frequency, so the voltage you can apply without exceeding the maximum rated current (Yes, there always is one) drops with increasing frequency.

What you probably want are 'doorknob' capacitors used in high power transmitters, these are specified for RF current handling and voltage.

ATC make excellent parts but mainly for higher then the wiggly DC that is 13.56MHz.

Regards, Dan.

[ZeroResistance]

@ZeroResistance: better you define the required capacity (in nF/uF) at specific frequencies, max currents (Ip-p), voltages(Vp-p), and other params you are aware of, and the experts here may identify the proper caps for you easier.

Problem being I looking for various values from 1000pf to 100pf, i might connect these in series or parallel to get the desired tuning.

[Twoflower]

I was suggesting that you should get in contact with Murata to get a complete data sheet. Unfortunately I don't have the max peak current or other values than shown on the web. They seem to be very stable over a wide range of values. But your capacitance ranges might strike them from the list.

[T3sl4co1l]

C0G ceramic or silvered mica.

If you're doing that much voltage into any significant amount of capacitance (let alone 1nF), you're probably looking at doorknob or pot type capacitors, possibly with water cooling.

Maybe just salvage a diathermy machine (that sounds like what you're doing, anyway!).

Tim

[bson]

A 90V 15MHz resonator?  That sounds like volcano-dwelling mad scientist stuff to me...

[ZeroResistance]

The reason the applied voltage limit drops with frequency, is that you are really hitting a current limit!

Xc drops with increasing frequency, so the voltage you can apply without exceeding the maximum rated current (Yes, there always is one) drops with increasing frequency.

What you probably want are 'doorknob' capacitors used in high power transmitters, these are specified for RF current handling and voltage.

ATC make excellent parts but mainly for higher then the wiggly DC that is 13.56MHz.

Regards, Dan.

At 1Mhz the Xc of a 1nF capacitor is 160 ohms.
So If I take the actual rated voltage of 250V rms the current is 250/160 = 1.5A rms.
and If I take the max allowed volatge of 90V rms @ 1Mhz the current is 90/160 = 0.56A rms.

I still don't see why 1.5A * 1.414 = 2A peak should cause damage to a capacitor, especialy a pulse rated capacitor. What am I missing here?

[JS]

At 1Mhz the Xc of a 1nF capacitor is 160 ohms.
So If I take the actual rated voltage of 250V rms the current is 250/160 = 1.5A rms.
and If I take the max allowed volatge of 90V rms @ 1Mhz the current is 90/160 = 0.56A rms.

I still don't see why 1.5A * 1.414 = 2A peak should cause damage to a capacitor, especialy a pulse rated capacitor. What am I missing here?

Pulses are short duration and every once in a while? As opposed to constant RMS rating... Most devices have a very different spec for pulse than for nominal value, you will find, for some semiconductors, the SOA in the datasheet, which shows the different ratings at different exposure times.

JS

[iMo]

The pulse current through a capacitor is mainly given by the parasitic resistance/inductance of the capacitor, impedance of the pulse source, rising/falling times of the pulse's edges and the amplitude of the pulse. With a "perfect capacitor" and a "perfect pulse source" the current is unlimited high.

[ZeroResistance]

The pulse current through a capacitor is mainly given by the parasitic resistance/inductance of the capacitor, impedance of the pulse source, rising/falling times of the pulse's edges and the amplitude of the pulse. With a "perfect capacitor" and a "perfect pulse source" the current is unlimited high.

Ok, So if I connect the said 1nF capacitor to a 1Mhz AC signal source at supply 230Vac rms wavefrom to it what kind of current waveforms should I expect? would'nt it be just a plain old sinusoidal current waveform.

[iMo]

This is a 1nF capacitor with R=10mOhm, L=10nH, source is 0/100V square wave (100Vp-p) of 10MHz with 1ns r/f edges and internal resistance of 10mOhm.
The current is ~1200Ap-p.

[T3sl4co1l]

Ballpark speaking, the RMS current rating is usually 1/5 to 1/100 the pulse rating.  If a pulse or dV/dt rating is provided at all.

RMS current limit is thermal, obviously.  It can be worse at higher frequencies, due to skin effect and such; CDE 940C caps for example are quite prominent in this way (rated for lots of peak amps, but not very much RMS at 100kHz, and even less at 400kHz -- I know, I've melted them before).

Mind that ceramic chip caps aren't very conductive (they aren't made of Al2O3, or much metal!), so they can't handle the same power level that a chip resistor (typically Al2O3 or metal body) of the same size would.

I have plans to build an induction heater with C0Gs; it'll take about qty 300, wired in parallel.  It's a pretty good bit of reactive power handling, though: 100 VA each (30kVA total) isn't bad at all for parts so small!

Tim