MLCC versus Mica capacitor for high power RF transmitter

Log in or register to post comments
Go To Last Post
13 posts / 0 new
Author
Message
incal
incal's picture
Level: Hangaround
Joined: Fri. Aug 1, 2003
Posts: 416 View posts
#1
Posted by incal: Wed. Feb 5, 2014 - 12:16 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

For an LC series network of an high power (20KW) RF transmitter designed to resonate at 600Khz to 1.2MHZ, I compared some capacitor types.

The capacitor should be able to hold high voltages at high frequencies.

I canceled film capacitors , because they degrade the voltage blocking capacity severly wehn passing beyond 200kHZ. At 600Khz a pp film 2000V capacitor has only 60V voltage rating (datasheet wima)

The ripple current of the capacitor bank is 1200A @600KHZ

So I compared 2 types Mica and CG0/NP0 MLCC (ceramic class 1)

Which one fits best?

Mica: very expensive, but accurate, rare to find, but possible

MLCC: very economic 2000V MLCC 68pf 2000V is under 0.03euro . If I paralel 3000 of them I reach a capacitor level of 200nF @2000V with a ripple current rating of 1400A.
Downside: I am not sure how I design the creepage distance on the PCB with 1206 smd package. But if they produced it in this package I believe it should statisfy 2000V creepage distance level.

As some more experienced freaks in forum recommended mica in this forum and has not recommended MLCCs I wanted to ask what can be the downside of the MLCCs?

________________________________ We dream of a world where current does not need the voltage to flow.

Tags:

Learning and Information, General Electronics
Top
david.prentice
david.prentice's picture
Level: 10k+ Postman
Joined: Sat. Feb 12, 2005
Posts: 28649 View posts
Location: Wormshill, England
Posted by david.prentice: Wed. Feb 5, 2014 - 12:29 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Surely this exactly the ideal place to use high voltage Mica capacitors.

You are hardly going to use SMD components in the 20kW output stages of a transmitter.

Of course, 20kW RF transmitters are not consumer items. So you just have to pay the market price for your components. It is not worth risking your expensive output transistors/valves for the sake of cheap capacitors.

David.

Top
incal
incal's picture
Level: Hangaround
Joined: Fri. Aug 1, 2003
Posts: 416 View posts
Posted by incal: Wed. Feb 5, 2014 - 12:50 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I constructed a matching transformer between transistor chopper and the resonant network. It is also very effective for power limiting. (the ferrite core con not pass more than 30KW). So the secondary can also be short circuited or open loop.

So any bad capacitor will not harm. worst case will distance the cicuit from resonance point which is also not too critical .

But before I try , What is the downside of MLCCs (ceramic class 1 np0). Risk of crack / subharmonic ossilations that can convert to mechanical vibrations etc.
Any ideas?

________________________________ We dream of a world where current does not need the voltage to flow.

Top
Chuck-Rowst
Chuck-Rowst's picture
Level: Resident
Joined: Tue. Jul 19, 2011
Posts: 710 View posts
Posted by Chuck-Rowst: Wed. Feb 5, 2014 - 06:37 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

So, 1200 Amps divided between 3000 caps works out to 400 milliamps per 68 pF cap. Can a 68 pF COG/NP0 MLCC handle that level of 600KHz AC current? If so, how much will it heat up while doing so?

Also, NP0 caps are not nearly as temperature stable as micas over wide temperature swings. NP0 were originally formulated to counteract the temperature change in air-core inductors wound with copper wire (the copper expands and increases the diameter of the coil, changing it's inductance). They were not designed for the lowest possible tempco. Mica caps already existed at the time and fullfilled the need for a temperature stable capacitor.

No doubt, mica caps are getting harder to find, and more expensive to boot. One of their benefits is they have a virtually infinite shelf life, especially the type coated with vitreous ceramic or porcelain. So, if you find a surplus batch of unused units, they will quite likely be as perfect as they were when they came off the assembly line in World War I !

Top
ka7ehk
ka7ehk's picture
Level: 10k+ Postman
Joined: Fri. Nov 22, 2002
Posts: 19508 View posts
Location: Tangent, OR, USA
Posted by ka7ehk: Wed. Feb 5, 2014 - 08:38 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Is it an application where you need a precise capacitor value?

As others have noted, it is not likely that you will use SMT parts at that power level. I suspect that one of the important factors will be ESR or Q (they are equivalent, though high Q means low ESR). This directly relates to internal heating. Heating results in decreased breakdown voltage and increased failure rate and value change.

There may not be a definitive ceramic vs mica answer. That is because of different construction styles that do not depend on the type of dielectric. That is, in some cases, the package may be more important than the dielectric.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Top
incal
incal's picture
Level: Hangaround
Joined: Fri. Aug 1, 2003
Posts: 416 View posts
Posted by incal: Wed. Feb 5, 2014 - 08:57 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I know that 10nF 50V MLCC has 20mOhm ESR. The biggest advantage of MLCC is that there is a minima between 500KHZ and 1.5MHZ . The ESR decreases to a deep level of 10mOhm.

But I am not sure How I translate this to a 68pF 2000V MLCC.
I dont know if how much has the manufactorer decreased the paralel plates to go from 10nF to 68pF. Or how much has the manufactorer increased the dielectric distance to go up to 68PF and at the same time reach 2000V. So there are many parameters.

The manufactorer gives no clue. Says only that a thermal measurement must be evaluated.

I also think as a possibility to build my own from 25micron PTFE films sandwiched between aluminium plates.

But any idea how much ESR should have a 68pF 2000V cap if the same technology 1206 cased, 10nF 50V has 20mOhm at 100KHZ and 10mOhm at 1MHZ?
Any ideas? a rough estimation?

Mica caps seems perfect but they cost x60 than an MLCC cap bank (x3000 68pf) 200nF 1200A ripple current 2000V rating costs 120 dollar.(including PCBs and placement costs).

But the mica variant as I looked at digikey is not less than 6000 USD :(

________________________________ We dream of a world where current does not need the voltage to flow.

Top
david.prentice
david.prentice's picture
Level: 10k+ Postman
Joined: Sat. Feb 12, 2005
Posts: 28649 View posts
Location: Wormshill, England
Posted by david.prentice: Wed. Feb 5, 2014 - 10:42 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think you should explain where these "capacitors" are going to go.
If it is in the LC tank of a Class C output stage, it will need to be stable, low ESR, high voltage, high current.

I am sure that the several licensed Amateurs can suggest suitable components, probably from a WW2 surplus market.
If you are that tight for cost, you could make your own Mica capacitors.

I can't believe that you would ever consider a capacitor 'bank'. Think of the losses involved in your voltage sharing resistors!

David.

Top
incal
incal's picture
Level: Hangaround
Joined: Fri. Aug 1, 2003
Posts: 416 View posts
Posted by incal: Thu. Feb 6, 2014 - 08:59 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It is an induction heater.

Because of the balacing resistors , I changed my way to build a self made one just like this one but a little bit bigger in size.

[url]
http://www.ebay.com/sch/i.html?_...
[/url]

I plan to use 1 mm thick copper foils wrapped with 300um thick PTFE adhesive band. 20 of them stacked in paralel. The foils are 100mmx200mm

My concern is how is the internal inductance if I stack the foils in paralel. the celem unit above is also big, the question is how have they achieved to minimize the inductance (they claim to operate up to 100MHZ) so that the internal L of the box should be in nH range but how?

When the geometry is bigger , L must be bigger?!?. How do they get very low L?

________________________________ We dream of a world where current does not need the voltage to flow.

Top
RickB
RickB's picture
Level: Posting Freak
Joined: Sun. Jan 30, 2005
Posts: 1338 View posts
Location: Junction City, OR USA
Posted by RickB: Thu. Feb 6, 2014 - 04:48 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The link make no sense. Nothing to do with the subject.

Top
david.prentice
david.prentice's picture
Level: 10k+ Postman
Joined: Sat. Feb 12, 2005
Posts: 28649 View posts
Location: Wormshill, England
Posted by david.prentice: Thu. Feb 6, 2014 - 05:00 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

An induction heater is very different to a CW or SSB transmitter !
I doubt that you will worry about frequency stability.

There must be proven designs out there in Google land.
I would start from there.

Yes, when I looked at your "link", I did wonder whether the whole subject is a wind-up.

David.

Top
incal
incal's picture
Level: Hangaround
Joined: Fri. Aug 1, 2003
Posts: 416 View posts
Posted by incal: Thu. Feb 6, 2014 - 05:07 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

When I click the link I see a celem capacitor rated at 500V 22nF ripple current rating 150A.

Do you see something else?. I think i copied the link of the sorted celem result. it may look different on your computer. I give the link from the direct page.hope it works.

[url]
http://www.ebay.com/itm/1x-Celem...
[/url]

________________________________ We dream of a world where current does not need the voltage to flow.

Top
david.prentice
david.prentice's picture
Level: 10k+ Postman
Joined: Sat. Feb 12, 2005
Posts: 28649 View posts
Location: Wormshill, England
Posted by david.prentice: Thu. Feb 6, 2014 - 05:43 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes, your last link works fine.
Your original link is complete pants.

I am still confused by your 'values'.

After all a 600kHz LC tuned circuit is going to be nF and mH rather than pF and nH.

I suspect that induction heating is a specialist subject. OTOH, there must be plenty of app notes out there. Or you could simply take an angle grinder to a commercial design.

20kW is a fairly large power. It is equivalent to several cows. It depends on the volume of the target.

David.

Top
incal
incal's picture
Level: Hangaround
Joined: Fri. Aug 1, 2003
Posts: 416 View posts
Posted by incal: Thu. Feb 6, 2014 - 06:38 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My working conditions are

L=1uH C=14nF I=800A ac f=1MHZ
L=1uH c=24nF I=1200A ac f=600KHZ

________________________________ We dream of a world where current does not need the voltage to flow.

Top