Lowest Impedance with Capacitors

So don't use large ceramics? :-)

MKPs typically resonate at a few MHz with Rs ~ 50mohms. Not low enough?

Hah... I wonder... has anyone tried constructing a four layer PCB with the middle filler made of barium titanate? Who needs bypass capacitors when your +V and GND planes ARE the bypass!

Tim

Not from barium titanate, but 1mil (or so) dielectrics meant to be used in place of (many/most) bypass caps are not uncommon in high-end high-power digital powers (i.e., lots of large swtiching currents). I think the price is such that it isn't likely to show up in, e.g., PC motherboards any time soon, though. See:

---Joel

I've made designs with small Maxwell pulse capacitors that delivered 5kA without hesitation. They had pretty low esr. What's your size budget? How about price? eBay can be a low-cost source for getting just one or two pulse caps, if you search / wait long enough.

Is someone building a coin shrinker??

0.05ohms is probably good enough for my app. Bummer...fr around a few Mhz :( I suppose put a ceramic in parallel with the MKP cap for lower Z nearing 20Mhz.

One possible bug might be the barium titanate doesn't flex like a FR4 PCB...

D from BC myrealaddress(at)comic(dot)com BC, Canada Posted to usenet sci.electronics.design

Height no higher than 9mm. Perhaps around $2.00/cap max. will be budget ok..

I guess I could just compare dissipation factors on Digikey and pick the lowest DP. Z plots are not always on the datasheets.

D from BC myrealaddress(at)comic(dot)com BC, Canada Posted to usenet sci.electronics.design

Whoa, it takes much more than that to make a coin shrinker. But, yes, a few years ago I spent several thousand eBay dollars in a set of Maxwell caps sufficient to make a quarter shrinker. I'm only missing everything else, mainly the table-setup and the 1000kA switch. Oh, and also the 1M frame/sec camera!

As always, the question; what are you actually trying to do.

The $2 budget limit and actual app, duty cycle etc should have been in your first posting. The voltage rating is only a hint.

There are ways of reducing ripple and impedance artificially, but not at 20MHz. At that frequency, 0.05 ohm might be considered pretty good performance, or even too good, considering damping effects and the reactance of a 100nF part body.

If impedance vs frequency aren't on the data sheet, don't use the part where it matters, unless you know better.

RL

Gotta start somewhere.

Build one!

Try this one

Ignitron(s)? ;-)

Have Fun! Rich

I'm choosing capacitors for the input and output EMI filters for a smps design. I'm trying to figure out what my capacitor models should be like in LTspice. (Rac, Rdc, esl) I need good capacitor models to make noise attenuation simulations.

Yup. I'd rather look at Z plots than do bench testing.

D from BC myrealaddress(at)comic(dot)com BC, Canada Posted to usenet sci.electronics.design

Then I don't think electrolytics or other bulk energy storage in the power train will be a major consideration.

At higher frequencies, impedance can be manipulated in other ways than straight capacitive bypass. Often they have to be, due to practical or safety limitations.

As far as modelling goes, the problem is more often identifying the noise sources and coupling paths of the real physical environment. You've been using a sniffer probe. Is your display a spectrum analyser or are you limited to a scope?

RL

True.. For example: iirc.. Y capacitors have a 4700pf limit.

I'm just about to discover that :) It's the electronics of the electronics.

Limited to 2Gs/s DSO 2Mpoint scope. :(

Hopefully I can trust my makeshift sniffer probe. It's just a 0.5cm diameter single loop stuck on the end of a 75ohm tv coax cable terminated at the scope with BNC T and a 75ohm terminator.

smps design is the only electronics that makes me want to tie the power supply to my rear bumper and then go driving on the freeway..

D from BC myrealaddress(at)comic(dot)com BC, Canada Posted to usenet sci.electronics.design

That is an impressive camera to be sure. And a really nice price. Just about 1000 times too slow.

This can be an advantage in troubleshooting, as the shape of the interference waveform can be correlated to the source more easily, but it won't tell you where you are, as far as meeting real requirements goes. You could find yourself on a wild goose chase.

As long as the DSO will trigger on the interference, you can at least work on the conductive bands below 30MHz. I'd personally prefer analog in this kind of snipe shoot.

Make sure that you look at the interference in the actual test measurement paths regularly and keep careful notes of iterations and their effect. Get used to converting voltages and currents to dB, regularly. Then you'll know when you're in the ballpark, and what to ignore.

Have you got a crude isometric sketch of the box and intended internal subsection relative locations?

RL

Wow! 1200 fps, HD movies, and under a grand! Didn't Casio used to make cheap plastic keyboards for kids?

I'll compare on a 100Mhz analog scope.

Perhaps I'll start making DSO screen capture files to compare waveforms to verify emi improvements. I've yet to try out the network connectivity on my scope.

I guess I could make a sketch. I'm not sure why..

Is this to see what parts have capacitance to the metal box? Example: Mosfet bolted to box. iirc it's ~10pf for an isopak TO220 bolted to a box. I believe that's one potential source of conducted emi.

D from BC myrealaddress(at)comic(dot)com BC, Canada Posted to usenet sci.electronics.design

It's important to examine and comprehend as much as is possible from a screen shot. Iteration settings, date/time and your comments should be associated with the shot, in your log or other notes.

More thinking and fewer screen shots........

When considering network connectivity, also consider any implications of recording instrument connection to the test site. You may have to disconnect to measure, reconnect to transmit. Recording in scope memory on a battery-operated laptop rather than to an extended network might remove these issues.

For your records, for your own consideration and calculation,, on record and for ease of explanation/consultation with others.

It should be done before building, as it is cheaper than rebuilding later; will influence your thinking and design work; reduce your GUI keystroke count and slash the number of physical iterations, in the end.

RL