That's why I like the time-domain measurement. It gives an instant visualization of ESL, ESR, and capacitance, all separated out. Since I'm doing pulse generators, time domain is where I live.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
The SMPS output cap I was testing was supposed to be an aluminum polymer I spec'd, ~30mohm, but the part populated and under suspicion measured ~1.5ohms, AIR. It wasn't subtle. The China crew had squeezed a few of the wrong shekels out of it.
The real part measured about right--I think I used a 100ohm series resistor to measure that.
rent of
FETs are amazing. Don't run 'em linear at rating though--BANG!
which of course is the reason for adding a small non-inductive cap across the leads of the electrolytic or nearby.
Semco had a customer that used hybrid caps in some designs where mica caps were potted at the base of an electrolytic cap. The leads of both were joined with in the potting and 1 set of leads brought out for PCB mounting. I think the idea was a space saving thought but something tells me that it may had been a little more to it, like custom PN# because the markings were their house numbers and they didn't correlate with what was inside..
Yeah, in the last week I've spent hours squinting at SOAR graphs. There's only a weak correlation between other specs and SOAR. The best indicator is gate capacitance, but that's still weak. A lot of apparently brute fets have pitiful power-time capability near their upper voltage limits; they have the equivalent of bipolar second breakdown, which people used to claim mosfets don't have.
I don't really trust SOAR curves. I always blow up fets myself.
formatting link
I need to explode a bunch of DPAKs next.
--
John Larkin Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
You'd be very shocked to see what is really used at the root of all this.
We had calibration equipment that was dated back to the 30,40's as the defacto to make sure the modern equipment was correctly representing the data. I can tell you that manufactures are all over the map with their gear, even the big names.
The federal Gov. made sure that we had such gear because production gear of modern day was not to be relied on and since we did a lot of military work, we followed through. Even though we had to send some of these out to calibration sites to correlate with their specs. One of them tried very hard to purchase a calibration instrument, instead of returning it. For a while we thought maybe it got damaged and they were covering their tracks.
When the company announced a closing date, they soon had several inquires from chinese manufactures, requesting the purchase of these items. So they must of knew something.
It all depends, of course. Newer is better than old. I haven't measured SMTs yet, so I don't know how they perform. I know alpoly is basically ideal (given its specs).
I once built a little 1W flyback converter with a 10uF aluminum radial filtering the output, which was at least a few ohms ESR. The ESR wasn't the worst part, though: the transient was squiggly, due to complicated ESL and ESR internals. I replaced it with a dry slug tantalum (the common yellow dipped radial kind), which had less ESR (though still far from zero), a teeny bit of ESL (basically lead inductance), and very flat time-domain characteristics -- the ESR is basically an ideal resistance, which is very nice to look at. The only downside is, the ESR makes heat...
I once conducted an experiment with a xenon flash tube and assorted capacitors: the fastest discharge (about 10us FWHP) was a nice big film cap. That's around the xenon ionization time of 5us, so it's about as good as it can get. The slowest was a generic 100uF 350V axial that looked like some
20 ohms ESR. The discharge just dragged on and on. Most new "general purpose" radials did better.
Snap-in types are actually impressively good: 220uF 400V discharged in only
20us, hugely more peak current than the 20uF film cap!
Tim
--
Deep Friar: a very philosophical monk.
Website: http://www.seventransistorlabs.com/
Any decent LCR meter will look like this on the workbench:
formatting link
And must use DDS and DSP processing. As Tim Wescott has been seen to write, "analog circuits are only applied to analog functions by people who don't understand microprocessors."
A must have, yes? It's got a dsp and all the associated software and it ONLY takes maybe 35-40 sq inches of board space, plus only a _few_ connectors and cables.
The schematic is a mere 5 pages to elucidate, too:
It doesn't do hardware floating point, so far as I'm aware. So any measure would be a measure of the software that was written for it. It's kind of like the ADSP-21xx in that way, except a little less a DSP and a little more of a microcontroller.
Given the option of attending the East Bay Sunday Garden Tour, I discovered that I absolutely had to build a laser driver prototype before Monday morning.
I used Phil's suggestion of using superglue to tack down the various FR4 islands. That's convenient, but I did have one pop off. I think that the copper surfaces to be glued should be scuffed some first, sandpaper or Scotchbrite.
formatting link
formatting link
This makes a sharp little tic! sound every time you pulse it.
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?
Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com
I'd like to see John's take on why he's doing his measurements with pulsed current rather than an LCR meter (which could be done with just one board, possibly inside of a USB dongle for crying out loud).
I do know that capacitors often don't conform well to a simple lumped- parameter model when you really start pushing for detail, but AFAIK an LCR model is fairly accurate for fast stuff -- it's when you want to get 16 bits worth of precision out of a measurement system that depends on the cap that you run into difficulty.
(Microprocessors, of course, obey lumped-circuit models as well as you make them, if you know what you're doing -- thank you for your insight, Jon :).
--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?
Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com
I don't trust LCR meters. They generally do something simple that is easily confused by the tangle of C, ESR, ESL, and shunt dielectric losses. You're lucky if a simple LCR meter gets the C right on a big 'litic, much less milliohms of ESR. The leads alone will have milliohms of their own.
A frequency-domain plot of Z vs F will tease out ESR, provided it can measure milliohms accurately.
The pulse thing tells me what I need to know, instantly, unambiguously.
--
John Larkin Highland Technology, Inc
jlarkin at highlandtechnology dot com
http://www.highlandtechnology.com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom laser drivers and controllers
Photonics and fiberoptic TTL data links
VME thermocouple, LVDT, synchro acquisition and simulation
Hmm. I was kinda forgetting that LCR meters usually just read at one frequency. Sigh. You'd need to get that Z (or Y) vs. F plot, then do some figuring.
(When I need an LCR meter I use a signal generator, a resistor, the part under test, an oscilloscope, a calculator or spreadsheet, and my brain).
Getting ESR from pulses still requires you to distinguish between the effective series resistance and any series inductance, but I suppose that there's generally a large enough difference in time constants to make that reasonable, if not even easy.
--
My liberal friends think I'm a conservative kook.
My conservative friends think I'm a liberal kook.
Why am I not happy that they have found common ground?
Tim Wescott, Communications, Control, Circuits & Software
http://www.wescottdesign.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.