15
\$\begingroup\$

I need to design a device that have to operate in an high pressure environment (Nitrogen gas). The operational pressure may vary from 1bar (atmospheric) up to 20..30bar gauge pressure. The regular working pressure will be around 10bar.

So, the device contains a switching voltage regulator with LM2674-5 that needs input and output capacitors with relatively high value - something like 100uF.

It is pretty obvious that the usual electrolytic capacitors with liquid electrolyte will probably be crushed by such pressures.

But what capacitors to use? Are the tantalum capacitors more pressure resistant?

\$\endgroup\$
  • 6
    \$\begingroup\$ generally, 'pretty obvious' things aren't. You do realise that liquid is incompressible? What you do have to worry about is gas or vacuum voids. If a component is not specified for some extreme environmental condition, then you shouldn't use it there. Read the data sheets. You may have to ask the manufacturers directly, they often have more data than they publish. \$\endgroup\$ – Neil_UK Jan 17 at 10:28
  • 1
    \$\begingroup\$ The liquid filled capacitors definitely can't be filled on 100%. There always should be some gas in order to allow the thermal expansion of the liquid. That is why I think that using liquid caps is not a good idea. \$\endgroup\$ – johnfound Jan 17 at 10:31
  • 2
    \$\begingroup\$ I've successfully used solid tantalum caps in under-sea equipment operating in the 100m-150m range (so 10-15 bar) for weeks at a time with no observed ill effects. \$\endgroup\$ – brhans Jan 17 at 13:09
  • 1
    \$\begingroup\$ Beware of pressure, and pressure-changes, distorting the silicon inside the epoxy packages. This distortion will (likely) cause imbalances in the threshold voltages of FETs (MOSFETs) and thus your precision analog circuits will have surprising OFFSET voltages. Ask the manufacturer about this. \$\endgroup\$ – analogsystemsrf Jan 17 at 14:17
  • 3
    \$\begingroup\$ @Neil_UK Of course liquids are compressible. So are solids. Look up what 'bulk modulus' means, and try to find any material that doesn't have one. More importantly, liquids compress between 10-100 times more than solids for the same pressure, which is enough to introduce meaningful mechanical strain under higher pressures. Water loses 0.14% of its volume at 30 bar, while steel loses 1/74th that volume. This means the pressure will exert much more strain on a container filled with liquid than one that is solid. Does it matter? Depends. But it shouldn't be ignored. \$\endgroup\$ – metacollin Jan 21 at 18:56
11
+100
\$\begingroup\$

This is said with significant caveats, but the only electrolytic capacitor options for a pressurized environment are ones with a solid electrolyte, so solid tantalum, tantalum polymer, or aluminum polymer capacitors.

Cornell Dublier, for example, specifically states that all of its aluminum electrolytic capacitors have an operational range of 1.5 atmospheres to 10,000 feet (source - page 9).

Aluminum electrolytic capacitors are not perfectly free of voids and their normal operation and initial anodizing ensure that there is a small amount of hydrogen gas already inside, straight from the factory. At modest pressures, any contaminants will be forced into the capacitor past its seals, potentially causing a short or altering the capacitance, and at higher pressures, they will simply get crushed inwards and guarantee a short-circuit failure mode.

Simply put, normal aluminum electrolytics are off the table entirely.

Now, this is where it gets tricky: when designing pressure tolerant electronics, for the most part, you are kind of on your own. What I mean by that is you are not going to find answers to questions like 'maximum operational pressure' of most components, even if you email the company. This is because such a niche is incredibly small and it is simply not worth the time and effort to test or qualify products under such unusual environmental circumstances.

There are a few (very few) companies that make a limited selection of high pressure-rated components like capacitors, some as high as 10,000 psi. These capacitors will be very expensive - I couldn't even find a price, you have to request a quote. If you have high enough volume, I would still expect them to cost well over $500-$1000 per capacitor. They're also huge, 50,000µF of tantalum capacitors, true 10,000 psi monsters. So actually finding pre-qualified parts that are practical is also, I would think, not a realistic option for you.

What this means is it is up to you to qualify components yourself. You have to use an educated decision and select a COTS capacitor, but no one can tell you for sure if it will work or how its properties or longevity will be effected in such an environment as yours. You have to test all of this yourself.

This is how most pressure-tolerant electronics have to be designed. You qualify the parts individually through your own testing, and then you further qualify the entire assembly together under testing, and then you either spend a lot of time and money required to get even a slight idea of the reliability or longevity of your set up, you you just hope for the best (and learn from what happens to the devices in the field - trial by fire if you will).

So you should also be keenly aware of what is at stake, and what the consequences would be if your board were to fail, and make sure that allowances are made so that, for example, no one's safety would be put at risk.

That said, for bulk electrolytic capacitance, solid tantalum capacitors would be your best bet for tolerating the pressure with minimal changes in performance.

Another option is to make sure you really need electrolytic capacitors at all. Ceramic capacitors rated for 10V and 100µF are readily available and not horribly expensive. This Murata capacitor is an option, for example. Just beware of the DC bias graph - most of the high capacity ceramic capacitors use dielectrics that exhibit the ferroelectric effect. Similar to ferromagnetic materials in the presence of a magnetic field, ferroelectric materials are analogous but for electric fields (and energy stored as an electric field is ultimately what the capacitor is ultimately storing). This means ceramic capacitors' effective capacitance drops under DC bias. So you would need to derate their capacitance and use more than one in parallel.

The gold standard in pressure-tolerant electronics has always been the polypropylene metal-film capacitor, but obviously these are much much too low value and simply not suited for any bulk-capacitance application. I thought I would note them here for completeness though.

In closing, aside from some fairly exotic high pressure, deap sea capacitors that are likely not practical for your application, the short answer to your question is that tantalum capacitors as well as most capacitors simply do not have a maximum operational pressure rating. Rating is emphasized on purpose here - do not mistake this to mean that they can operate at any pressure. They certainly have a maximum pressure they can be expected to operate at, but the rating itself simply will not exist.

Don't let all this discourage you, however. The pressures experienced by things like deep sea pressure tolerant electronics are much higher than 30 bar, and quality tantalum capacitors are the first choice here, and all of the purpose-made deep sea 10,000 PSI capacitors are likewise tantalum capacitors.

Just understand that the manufacturer is not at fault if or when the capacitors fail, and you still have to qualify them yourself. This doesn't just mean checking for failure, but making sure their various properties that are of importance to your circuit stay within acceptable levels.

Get some solid tantalum capacitors and test them yourself. You'll probably get it on the first try, but be prepared to try a few different brands or construction types.

Final notes: Other components can exhibit unexpected behavior in high pressure environments. Make sure you don't have anything that has a 'metal can' construction. One easy to overlook is quartz crystals - through hole or SMD, they have empty space inside the can and mechanical stress on the crystal will through the frequency way off, if it isn't simply destroyed.

Also, be wary of wet tantalum capacitors. You should avoid these. There is a common misconception that fluids are not compressible. This is simply not true - they're much harder to compress than gas but it is still compressible, as are solids. That's what bulk modulus is - the compressibility of a substance. Importantly, the difference in compressibility for liquids vs. solids is between 10-100, or 1 to 2 orders of magnitude. This means liquid will compress much more than solids, which would allow for potentially significant mechanical strain.

For water, it will compress by about 46.4ppm per atmosphere. So given volume of water will lose around 0.14% of its total volume if exposed to 30 bars of pressure. This won't make anything implode like a tin can, but for components with very brittle materials inside (like tantalum pentoxide), this could allow enough flex/strain to be worrisome. Solid electrolyte is what you want.

\$\endgroup\$
  • \$\begingroup\$ Well, interesting answer. Thanks for your effort. It didn't answered directly the question, but I didn't expected it anyway. :) Just one additional question. I am assuming that all tantalum smd chip capacitors (prismatic form factor) are solid electrolyte type. Is this right, or I should check the datasheets for every individual model/manufacturer? \$\endgroup\$ – johnfound Jan 22 at 0:59
  • \$\begingroup\$ Your assumption that a solid electrolyte is better than a wet one is plain wrong. Even the expensive capacitors you pointed are using wet electrolyte. Please see the specs: evanscap.com/pdf/TDD_REV_I.pdf paragraph 2.1 . \$\endgroup\$ – Dorian Jan 24 at 13:41
  • \$\begingroup\$ @Dorian: Of course it is possible a wet capacitor to be designed in a way to withstand huge pressures. But I have asked about a cheap, mass production elements. It is fairly stupid to use special, very expensive components, while it is possible to have the same device properly functioning with serial, cheap components. Isn't it? \$\endgroup\$ – johnfound Jan 24 at 13:49
  • \$\begingroup\$ I fully agree with everything else in your answer. But it's more likely for a solid capacitor to fail due a gas bubble than a liquid electrolyte one. The liquid will disperse the local tension to all capacitor case surface while a solid not. \$\endgroup\$ – Dorian Jan 24 at 14:07
5
\$\begingroup\$

Your problem may be solved by choosing a better design that operates >> 1MHz thus using a film cap capable of choosing one for your harsh environment.

Here is a reference from NASA for cryogenic testing on caps.

For example, while polypropylene, polycarbonate, and mica capacitors showed excellent stability when tested at liquid nitrogen, the solid tantalum capacitor exhibited an increase in its dielectric loss at that temperature. Most of the EDL capacitors experienced no change with ageing but seemed not to function at the extreme temperature.

Here is my suggested list of possible caps

You can find your own design from 1.5 to 3MHz to meet your requirements with a good battery source and film caps.

enter image description here

\$\endgroup\$
  • \$\begingroup\$ I wonder why NASA didn't report on electrolytic caps. I suggest the dielectric constant of wet dielectrics are NG at cryogenic temps. and Solid tantalum is more lossy which creates heat \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 25 at 7:02
  • \$\begingroup\$ I don't see low temperatures specified in the question. \$\endgroup\$ – Dorian Jan 25 at 11:05
1
\$\begingroup\$

As Neil_UK with his great wisdom and experience pointed out in his comment, this is not such a big issue as it seems.

How can a capacitor subject to a high pressure fail? You might think it would implode but this is simply not true if the capacitor is fully solid or filled with liquid because only gases are highly compressible. Liquids are much less compressible.

Also, the enclosure of an electrolytic capacitor is constructed to tolerate volume variations due to temperature changes. From 0\$^\circ\$ to 80\$^\circ\$ Celsius the volume growth is around 4%. I'm sure that a large margin is taken above this when designing a capacitor enclosure.

A 4% residual gas in a liquid filled capacitor (which is pretty much) will give a 3.8 % variation in total volume at 20 bar. Of course the thermal dilatation adds up but you see that they're in the same order of magnitude.

This is not true for a solid where the bubble cannot shrink to bring the bubble pressure to environment pressure because it's surrounded by solid, all pressure is concentrated on the bubblle's wall small surface.

enter image description here

The answer depends on your constraints, budget and reliability.

The expensive deep sea capacitors also require a long lifetime and low failure rate because the cost of replacing them is huge.

This might not be your case and metacollin's solution of using usual capacitors and testing them yourself might be good and cheap. Of course, not solid for the above reason and search capacitors with wider temperature range just because they have a much greater tolerance to electrolyte volume variations.

Also an old study that I found shows that the only usual grade components that really failed in high pressure environment (up to 70 bar) were components with air inside and weak cases like metal encased diodes.

\$\endgroup\$
  • \$\begingroup\$ Did you both to read the NASA report I linked? liquid nitrogen, the solid tantalum capacitor exhibited an increase in its dielectric loss at that temperature \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Jan 25 at 6:54
  • 1
    \$\begingroup\$ @SunnyskyguyEE75 There is nothing in the question about low temperatures. \$\endgroup\$ – Dorian Jan 25 at 10:25
  • \$\begingroup\$ If the gas bubble decreases its volume, either the liquid should increase its volume, or the outer case should decrease its volume. Because Vg + Vliq = Vcase always. \$\endgroup\$ – johnfound Jan 25 at 12:15
  • \$\begingroup\$ @johnfound The capacitor case is flexible to accomodate thermal dilatation or shrinking. Vcase is not constant. \$\endgroup\$ – Dorian Jan 25 at 13:59

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.