3
\$\begingroup\$

I'm studying switching converters and want a bench setup that will allow me to reliably observe the various voltage waveforms (output, inductor, switch) on my oscilloscope.

I discovered pretty quickly that just hooking up my 10x probe with witches hat and ground pigtail wasn't going to work very well, picking up large inductive spikes and hiding the rest of the action in noise:

This image is from the Analog Devices app note here on measuring switching regulator output ripple which is pretty much exactly the sort of thing I'm trying to do. The results I get are quite similar.

enter image description here

That app note advises a setup like this, which I'd like to build up:

enter image description here

The basic idea is a length of \$50\Omega\$ coax soldered directly to the output capacitor (SMD in this case), fed through a DC Blocking Capacitor (the tubular aluminum coupling on scope input 2), and using the built-in \$50\Omega\$ input impedance of the scope.

My scope (Rigol DS1054Z) doesn't have a \$50\Omega\$ option, but I have a feed-through terminator that accomplishes the same thing.

What I don't have is a DC Blocking Capacitor fitting.

So I was wondering. How hard could it be to put together at least a fairly usable one?

So my main question is how to size the capacitor and what type/dielectric would be suitable? I'm happy to order the right type and wait, but would also like to slap in the closest thing I have on hand that would work just to get a feel. The effective bandwidth would optimally be something like 10kHz to 100MHz.

I'm kind of supposing I would use an axial capacitor, solder it inline with the coax center conductor, shrink wrap that, then use some copper foil to wrap it and solder the coax braid to that, then shrink wrap the whole thing, maybe after wrapping some 2 liter Coke bottle plastic around it for mechanical strength.

If SMD caps were going to do the trick I could use a small bit of veroboard in roughly the same sort of way.

Anybody got any pointers on the capacitor selection or construction approach that might help me out?

\$\endgroup\$
  • 2
    \$\begingroup\$ No AC coupling on the scope? \$\endgroup\$ – uint128_t Jan 31 '16 at 2:16
  • \$\begingroup\$ @uint128_t -- The problem is the \$50\Omega\$ termination, which is basically a \$50\Omega\$ resistor from the coax center conductor to ground. This happens prior to the internal blocking capacitor and would pull a lot of current at the say 5-48V output voltages I'll be measuring. So a DC blocking cap needs to be added prior to the \$50\Omega\$ termination. \$\endgroup\$ – scanny Jan 31 '16 at 2:19
  • \$\begingroup\$ Ah, forgot about that. Makes sense. \$\endgroup\$ – uint128_t Jan 31 '16 at 2:20
1
\$\begingroup\$

Something really small (maybe that could be assembled between two close-spaced side launch connectors) and that has low enough impedance at your low end frequency not to affect the 50 ohm load too much.

So if Xc = 1/(2*pi*1E4*C) = 1 ohm, Cmin ~= 15uF.

You can get 1210 10uF X7R caps. One or two of those in parallel- check the impedance curves at your highest frequency and parallel with lower values if required. If you need much more than 50V-63V it will become more difficult.

\$\endgroup\$
  • 2
    \$\begingroup\$ Then double or triple that value because you'll have the whole regulator output as a DC bias across these, massively affecting the X7R effective capacitance. muRata and other manufacturers have great graphs for this. \$\endgroup\$ – pipe Jan 31 '16 at 10:15
  • \$\begingroup\$ This is just the ticket, thanks Spehro! :) I think I'll rig up a BNC feed-through out of a 25 x 25 x 50 mm extruded aluminum enclosure I've ordered and give that a try. I hacked together an ugly prototype with a 1uF tubular film capacitor I had on hand, placed in a LDPE tube and wrapped with aluminum foil and tape for shielding and that definitely proves the concept; the low-level noise is reduced to the scope's noise floor and I can see something useful now. I'm not sure what to make of the waveform I'm seeing now, but that's a new question :) \$\endgroup\$ – scanny Feb 1 '16 at 19:24
  • \$\begingroup\$ @pipe Good point, but I was pretty conservative with the 1\$\Omega\$ so decided not to get into that. \$\endgroup\$ – Spehro Pefhany Feb 1 '16 at 19:39
0
\$\begingroup\$

A few uF of capacitance should be plenty, but 20-30nF may even be adequate (bigger cap will magnify lower freq. waves more, relative to higher freq. waves).

Tantalum caps are what I hear the most ppl swear by, but any decent quality non-polarised, solid state cap should work for "playing with."

Try soldering a small strip of pcb at the end of your plug, with a break across the cladding to solder caps across; then attach one end to your scope plug & the other end to coax ctr conductor. Wrap the board & your coax jacket with removable conductive tape (over insulation btwn tape & pcb, of course); then experiment with different value caps until you find one that you like enough to order a tantalum & build a more permanent connector around it.

\$\endgroup\$
  • \$\begingroup\$ Tantalums are a form of electrolytic and are polarized en.wikipedia.org/wiki/Tantalum_capacitor. How would that work when the polarity of the DC may be in either direction? What did you mean specifically by "solid state" cap? just not a liquid electrolyte? \$\endgroup\$ – scanny Jan 31 '16 at 4:11
  • \$\begingroup\$ My error on the tantalums maybe (I blame faulty addressing in neural data storage). End result: you want a non-polarized cap that's not an aluminum, electrolytic, or EDL/supercap/ultracap (or anything else with a liquid electrolyte that can leak/evaporate/etc. & lower your effective capacitance) \$\endgroup\$ – Robherc KV5ROB Jan 31 '16 at 4:42

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.