new polymer caps

Further information here:

Am 04.09.25 um 19:26 schrieb john larkin:

Be careful with "DC" current through polymer caps.

I once built a low noise amplifier that had to do without feedback for stability reasons, There was a negative real part of the input resistance, even in the circuit from AOE3.

CS stage, cascode, opamp, feedback to the sources so that may look like CS but it is no longer CS.

To get at least a stable o/p I fed the sources from a fixed current mirror and bypassed the sources with large electrolytics with fb removed. That was a *bad* idea. The circuit seems to show every electron that defects through the electrolytics. Oscon Polymer was worst by far: a 1/(f**2) corner at a few KHz and so much noise that the 89441A went into clipping. That's not 1/f!!!

Nippon Chemical ALU was way better but the only thing that worked in a usable way was a AVX wet slug tantalum for €100, Sprague wanted twice that. 1/f**2 corner at 10 Hz.

stability problem with feedback: <

The green s11 plot leaves the unit circle ---> not stable at these frequencies.

Same thing in LTspice: <

------ flickr gets worse by the week. The now limit pic size to just a bit more than stamp size for free accounts. I'll have to soup up my web site. :-(

Cheers, Gerhard.

That is the first time I have seen a data sheet that tells me not to use the product! John

Am 04.09.25 um 20:13 schrieb Gerhard Hoffmann:

Ooohps: <

(Cascode omitted)

G.

"The storage period before opening the moisture barrier bag is two years after manufacture,and after opening the bag is seven days. If these conditions are exceeded, then the outer package may be damaged due to thermal stress during mounting caused by moisture absorption by the package."

The warnings are indeed educational!

John :-#(#

I did something similar, though *with* the cascode and without an opamp in the loop.

The real part of the impedance went negative around 70MHz, which was fine, because it was outside my bandwidth target. I muffled that by inserting a constant impedance bridged-T. That made it stable despite the purely reactive signal source.

Jeroen Belleman

Presumably those from packages that had been opened long ago could be 'rehabilitated' by leaving them in a dessicator for a week or two or by heating them to their maximum operational temperature for a few days before using them.

The cracking due to moisture can be subtle, not at all like popcorn, and so easily missed, except that the part misbehaves. I've had this happen.

The manufacturer provides the time-temperature schedule to prepare their parts for soldering. It's far hotter than a dessicator.

Joe

A vacuum dessicator gets the water out quite a bit faster than a regular dessicator. If you throw in some heat, it works even faster. The problem is physically adsorbed water, and the higher the temperature the higher the vapour pressure, but if there aren't any permanent gases around to reduce the mean free path of the water molecules they move pretty fast.

Sounds like Panasonic got new cap design and a new set of lawyers.

Try Dropbox.

Did you find that the polymers created noise?

I like polymers

Am 05.09.25 um 17:12 schrieb john larkin:

They did in my circuit. One comparison may not be statistically significant for the general case, but enough to think about if there should be problems.

It seems the circuit is a magnifying glass.

There was no 1/f corner but 1/f**2 and that gets out of hand quite fast. 40 dB/decade. Maybe a less ideal CCS would have helped. Anyway, the effect wandered clearly with capacitor type over 3 or 4 decades of noise corner frequency. Just keep it in the back of your mind.

Evading the feedback somehow felt like a personal defeat so I took a bit more gain from the cascode and replaced the op amp by a BFS19S follower to drive the low impedance sources. Burns some precious battery power. But it seems to be fast enough and avoids the 1/f noise of the 3 GHz CFA.

BFS19 is EOL as Digikey wrote me last week. :-(

cheers, Gerhard

The old CS13 type wet slugs had sulfuric acid inside and a silver case. Eventually, the acid would eat the case and it would leak.

A vacuum desiccator can be a lot more damaging than desiccation at atmospheric pressure. Any object which is pressure-sensitive could suffer permanent damage when subject to even a partial vacuum. I wonder how many data sheets refer to the pressure sensitivity of any components or equipment they make? Are there any vacuum-sensitive? If there were, and it wasn't mentioned, you'd probably never know that the device failure was due to vacuum damage without a lot of testing. It's a lot safer to spend more time desiccating at NTP than pushing it with a vacuum and heating.

Bubble wrap is the only example I can think of. As a graduate student I ran a vacuum line for years, and nothing I worked with ever got damaged by being pumped down.

None that I know of. Anything that is convection cooled can't be operated at anything like full power under vacuum - that could be problem for electron beam testers, and you could need a thermal bridge to get heat out of a working integrated circuit while you were using an electron beam to look at the voltages on it surface. It didn't stop Motorola from using one to inspect prototype 68000 processer chips with a Lintech electron beam tester. It apparently knocked three months off the development time, or so Lintech liked to claim.

Since most devices wouldn't fail, there mostly wouldn't be anything to know.

It's even safer not to do anything at all. If you don't understand much, and don't want learn, that can be the preferred solution.

Sulphuric acid doesn't eat silver. Electrolysis can move silver around, and that can create the occasional hole.

Concentrated sulphuric acid is an oxidising agent and that can dissolve silver, but you wouldn't put that into a tantalum electrolytic capacitor.

A room temperature desiccator would take ages. You need heat to accelerate the kinetics of the process.

Dry air is essentially equivalent to vacuum in this sort of situation, because the amount of water is trivially small. It isn’t like making instant coffee.

Cheers

Phil Hobbs

"Trying is the first step towards failure" ~ Homer J. Simpson

The mean free path of water molecules is lot longer in vacuum desicator than it is a NTP - their progress towards the desicant is still a drunkards walk, but with much bigger steps, so they get there faster.

You don't have to get the pressure all that low to create a situation where most of the residual gas molecules are water, and then you get an appreciable pressure gradient, and even faster transfer.