High Power RF Capacitors

Thread Starter

Doceave

Joined Nov 16, 2019
49
Hi there

I have been asked to transpose an old schematic into EDA software for PCB manufacture. The circuit is essentially a 1kW 13.56Mhz RF generator.

My challenge, in addition to knowing nothing about this type of circuit (I do not plan to test / power this thing up myself), is finding components that are in active production for use in the circuit. My task is modelling the PCB and no more than that.

The most troubling components are the capacitors 13-16 and 19-25. Newark/Mouser/LCSC/RS-components --- parts not available or out of stock / I have not been able to identify the correct component from their stock lists.

1582721402938.png

Whilst I am able to find components that match in terms of capacitance and voltage rating, I am certain that this circuit requires larger package size components. The capacitors indicated by arrows with the 3838 packages are likely working hard and cannot be replaced with much smaller general purpose caps...

1582722076644.png

My request is thus for recommendations for rugged caps that could be dropped in given the application AND a supplier with stock :) No small ask.

Please do bear in mind that this is way far out of my area of expertise and I appeal for friendly pointers.
 

BobTPH

Joined Jun 5, 2013
8,808
Have you noticed that capacitors don’t have a power rating? This is because ideal capacitors do not dissipate any power.

Real capacitors, however, do, and the parameters you need to look at are ESR and ripple current. If you have these ratings for the original ones, look for a replacement with lower ESR and higher ripple current and you should be OK.

Size does not matter as long as those are sufficient.

Bob
 

Papabravo

Joined Feb 24, 2006
21,159
When it comes to RF circuits your approach is, in most major respects, doomed to failure. They must be designed and fabricated with particular care, that is not required of circuits that operate at lower frequency. I suspect somebody is setting you up for failure on a massive scale.
 

Thread Starter

Doceave

Joined Nov 16, 2019
49
I managed to find a distributor for the components as listed in the bill of parts.... https://www.richardsonrfpd.com/

Thanks, however, for the insight provided.

I would appreciate it anyone who reads this would trawl their favorite suppliers site for availability of the following and let me know if they have stock on hand:
700C470xxxxxx
700C820xxxxxx
100E271xxxxxx
100E220xxxxxx
100E100xxxxxx
 

Janis59

Joined Aug 21, 2017
1,834
Doceave: I pull an any number You gave into Google, for example 100E100xxxxxx + datash and found NO ANY. It means sth is not correct in that refecences.
More, I walked to the Richardson site and find no ANY capacitors at all.
 

Janis59

Joined Aug 21, 2017
1,834
RE: Doceave
For 47 pF 100 pF and 220 pF there are rather cheap alternative. Should calculate: Xc=1/6.3*13.56E6 Hz* 47E-12 F=250 Ohm. Then 2500 V / 250 Ohm = 10 Amp. 10 A * 2500 V=25 kVAR. Then thermal flux if tan(delta)=0,015=0.015*25K=15*25=375 Watts (too much). Go to the tan(delta) 0.0003 and get 7.5 Watts. What it illustrates? The first case is if You take the FR-4 glass-textolyte, having about 30 kV at 1,5 mm thickness, but >>10 mm safety zone must be applied at outer perimeter where plasma may hit the air (1 mm=3 kV). FR4 have around 3 pF/cm2. Second case is for Rogers Duroid.

Hope I shall not be engraved if I shall quote the article few sentences: ""We tested here three of their teflon base pcb products - all 1.5mm thick, two-sided boards - and started with the cheapest one: the duroid-TC350™ with 2.0 pF/cm2 at the price of $115 for 18”x12” plate and a loss factor of tgδ=0.0015 (Q=667). To ensure better natural cooling, copper cladding is produced with matted surface, thus larger surface area was obtained and better cooling, but on the other hand Focault effect was worsened, thus the heat production increased. Next examined was the duroid-6035HTC (high temperature material, (see chart-1 data at [7]) with 2.2 pF/cm2 and tgδ=0.0013 (Q=770). The cladding was glossy, but still costly - $240 for the same size. Last was the duroid-5880® material with 1.7 pF/cm2 and tgδ=0.0004 (Q=2500). The cladding was also glossy but its cost reached $325."" PS: I am using those at about 20 kV what is far off the maximum of allowable voltage at 1,5 mm. There exists as well the 0.5; 0.7;1; and 2; 2.5 and 3 mm versions.

So, let dare to buy the material with Q-factor to price ratio passing Your wishes and capabilities, and use as the capacitor. In the frame of 10Amp to 50 Amp it may have to have an air-cooling enough for rather normal Temperature.
 
Last edited:

Janis59

Joined Aug 21, 2017
1,834
RE Bertus:1) Looked in, rather impressive! The quality factor 10 000 is something unheard, however however... The tan delta thus the Q as well is proportional to the square root of frequency. Thus for case of Q it is inverse proportional. 13,56 MHz/1 MHz of given reference =1/sqrt(13.56)=2715. So, the actual Q-factor at Your application will be only the 2700, just roughly about the same of Rogers Duroid will give. Only the rhebus one must solve, seems that professional cap is smaller, nicer, only unclear what about current capability and what about the price.
2) Tried to look into second datasheet - Allaboutcircuits stop me to access the "restricted area". You seems had been pasted the file in the wrong PLACE. :)
 
Top