User-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.12; rv:52.0) Gecko/20100101 Thunderbird/52.6.0
On 2/28/18 12:50 PM, Myfirstteslacoil via Tesla wrote:
Hi all,
I made a 'post' a while back about my Ricci transformer that wouldn't work.
After trying long it turned out to be working fine. It is a 9-50 trafo.
Resonant cap value is 0.017684 uF. (I live in EU)LTR is then: 0.026526 uF.
My first question is concerning the LTR value. Is it safer to go higher than this value? Or does it have no impact on safety (and performance)?
I did some calculations using a 0.15 uF 2kV cap for the MMC. If I use 2 strings in parallel with 12 caps in series I get a capacitance of 0.0249 uF. Using 11 in one string gives a capacitance of 0.0272 uF.
These options are both very realistic to me but which is the better setup? 2 caps more doesn't really disadvantage me in terms of costs. However a 3 string of 17 is really goes to the boundary. (It would give 0.0264 uF)
For the MMC I want to use the most popular caps but they seem to be out of stock in my country. I decided to look for similar ones. I found one from Kemet.
This is the link: https://nl.mouser.com/ProductDetail/Evox-Rifa-KEMET/PHE450SR6150JR06L2?qs=sGAEpiMZZMv1cc3ydrPrF85GIl804cbNdYhnhUR0dRE%3d
Data sheet looks like they're metallized film on polypropylene, which is
good, but the real question is what the internal construction looks like.
Do they have the "self healing" internal design (I couldn't find it in
the data sheet, but, then, I don't think that's on the 942 data sheet
either - it's somewhere else)
I'd compare the internal dissipation against something known to work
like the 942 series. if the "tan d" at highish frequencies (100kHz) is
comparable (or lower!) then that's good.
The derating curves in the Kemet part look like they're taking into
account heating from internal dissipation. If there's similar curves
for the 942, you could compare.
Life time is missing from this page while it is present on the CD 940C series caps. https://nl.mouser.com/ProductDetail/Cornell-Dubilier-CDE/940C20P15K-F/?qs=11kF6y5Z3OhbthK%252bpYczHA%3D%3D
It does have a higher VAC of 700. Is this beneficial? And most importantly has anyone ever used these? If so can you recommend them? They are significantly cheaper.
In the TCML archives there was already one asking the same. As a reply someone answered: "I don't think those capacitor will work for Tesla coils. If you look at the curves of applied voltage vs frequency on the web site, these capacitors roll off at much lower frequencies than we work with. In other words we couldn't apply a high voltage to them at primary/secondary frequencies."Links that were provided by him and other repliers do not work anymore. (http://www.kemet.com/kemet/web/homepage/kechome.nsf/file/PHE450%20Series/$file/F3294_PHE450.pdf).
However Bert Hickman said the following about the caps: "The series is rated for high frequency applications with high current stress. Sounds like a potential TC MMC app to me... :^)."
I did find a page containing the information however I do not know if this is the same that was on the site once.
After some digging I found it in someone's PhD Thesis. In the datasheets is the page it's about the PHE450 series. To find it quickly for you it's on page 8 of the Datasheets. The Datasheets begin on page 259. ; )
This is the link of the paper: https://dspace.lib.cranfield.ac.uk/bitstream/1826/8574/.../Aldhaher_S_Thesis_2014.pdf
This person used one of the PHE450 series: https://www.youtube.com/watch?v=Uh3sLMV2jeY (The caps don't look similar to the picture on Mouser)
The pictures on Mouser (or Digikey, or Newark, or Farnell) aren't always
accurate - I'd believe the datasheet drawing.
If someone successfully used the PHE450, that's a good sign - maybe you
could email him and ask.
Be aware that there are a wide variety of caps that will *work*, but not
*work well* or might have a short life. A lot of people have used beer
bottles in a bucket of salt water for a TC capacitor.
You're digging up the right kind of information, but at some point,
tesla coiling isn't what you would call a mainstream application of
these parts. You might just have to try them and see if they work.