IEC 60079-11 intrinsic safety and DC bias of ceramic capacitors

[sadboi]

Hey,

We are getting a board certified for IEC 60079-11 and trying to determine the intrinsic safety parameters of our power supply output. It's running at 12 V with an output capacitance of 20 uF, but accounting for DC bias reduces it by 80% to 4 uF.

If anyone has any experience in the certification process, are we allowed to account for DC bias for the calculation of effective capacitance? The standard documentation has no mention of DC bias of capacitors.

Thanks.

[floobydust]

Welcome to the forums 

They use the maximum capacitance possible so 10uF+10% or whatever your cap tolerance is, including temperature.
You get no slack due to DC bias because it's not controlled and in production you could switch cap manufacturers giving more capacitance than you expect.
From the spark tables that looks like Group IIC at best but I have no idea about your project.
If you are using lead-acid batteries, it's not 12V they use 16V for the max. system voltage.

[sadboi]

Hey thanks for the reply. We are trying to get certified for ia, Group IIB gases.

The 12 V is actually the output of a boost converter module, with 3x Zener diodes on the output in case the regulator faults. The 12 V output is used to power external sensors; we want the output capacitance minimised to be compatible with a wide range of sensors, so I guess we'll have to bump up the package size of the capacitor with a lower DC bias.

I do have another query, which is regarding an RF output of a modem that connects to an external antenna. The modem is not IS certified, but the max RF output power is 250 mW which is well below the 3.5 W limit of Group IIB according to IEC 60079-0, and has no internal voltage enhancing circuitry.

My question is under a fault condition, do we have to assume that the input VCC to the modem will pass through to the RF output? So let's say it will pass through; the manufacturer told us the sum of capacitances is 18 uF, and sum of inductances is 22 uH on the modem. The input VCC is clamped at 5 V with 2 uH and 86 uF capacitance in the worst case, so we have V=5V, C=104uF, L=24uH. Because we have a combined inductance + capacitance circuit, we have to used the universal output characteristic curves from Annex G. But Figure G.3a) the curve only show limits up to 10 uF capacitance and only for an inductance of 0.15mH. How can we prove our worst case parameters won't ignite a spark? I haven't seen any way to interpolate the curves to different inductance/capacitance values from my research.

Sorry for the wall of text, but I wanted to be as descriptive as possible. Cheers for your help.

[floobydust]

What is the current behind the 5V that could make it to the antenna? You have to consider that as well.

I have used the double capacitor barrier per 8.5 to stop unsafe DC power from getting to a cell antenna. It's two ceramic capacitors (which pads meet spacings) so they are physically big but around 1-10nF and high voltage parts, either 500V or 1500VRMS, depends on your incoming power if mains is involved.

I did a design that was off the spark tables. The certifier just sent the parameters to the head office lab that had a spark tester and they did the special test. It cost a few grand.
The hassle was that spark testers are finicky and not so reliable so they are frequently needing repair and you have delays and a possible lineup of people needing tests. There were two types of spark testers, one for high current with thicker tungsten wires like for 10A kind of thing as I recall.

For the 12V sensor outputs you can add resistance after the boost converter to help buffer the high capacitance it needs, so it's not all running to the outside. Or is it 12V on a board in hazloc?
It's hard for me to give much accurate advice without knowing everything like the zone and layout what is located where etc. so all I can do really is offer ideas.

[Tomorokoshi]

It looks like you are using Edition 7.0.

1. DC Bias
Does the circuit operate reliably with the 20uF that reduces to 4uF under bias? If you're paying for 20uF, but you are only getting 4uF, then perhaps just put in a 4.7uF aluminum capacitor anyway. Parallel with a 0.1uF ceramic if you need to.

2. RF Module
2.1. Does your agency fully understand what you are doing with the RF module?
2.2. Does the manufacturer of the RF module understand what you are doing?
2.3. Does the manufacturer of the RF module provide part numbers or only a total?
2.4. What is the procedure to lock down their design so a revision change doesn't add capacitance or inductance?
2.5. What about documentation for the factory audit process?

3. 5V Supply
Collecting data:
Uo = 5V (Probably 5.88V?)
Io = ? (Probably less than 700mA)
Po = ?
Capacitance = 104uF (Declared)
Inductance = 24uH (Declared)
Resistive limiting = ?

Going back to clause 9.2.6 "Circuits with both inductance and capacitance":
3.1. Apply 9.2.6 a) and run the spark test. (like floobydust pointed out. You might have to do that anyway.)
3.2. If you are applying 9.2.6 c), are you actually going to produce a FISCO device?

[sadboi]

What is the current behind the 5V that could make it to the antenna? You have to consider that as well.

I have used the double capacitor barrier per 8.5 to stop unsafe DC power from getting to a cell antenna. It's two ceramic capacitors (which pads meet spacings) so they are physically big but around 1-10nF and high voltage parts, either 500V or 1500VRMS, depends on your incoming power if mains is involved.

I did a design that was off the spark tables. The certifier just sent the parameters to the head office lab that had a spark tester and they did the special test. It cost a few grand.
The hassle was that spark testers are finicky and not so reliable so they are frequently needing repair and you have delays and a possible lineup of people needing tests. There were two types of spark testers, one for high current with thicker tungsten wires like for 10A kind of thing as I recall.

For the 12V sensor outputs you can add resistance after the boost converter to help buffer the high capacitance it needs, so it's not all running to the outside. Or is it 12V on a board in hazloc?
It's hard for me to give much accurate advice without knowing everything like the zone and layout what is located where etc. so all I can do really is offer ideas.

Oh awesome, I didn't notice the clause about the AC coupling capacitors. That makes things much easier, thanks a lot. Our board is completely battery powered, would we still need to satisfy the 500V dielectric strength test? It doesn't seem like a problem though...looking at Digikey I can see some 39pF 0603 caps with 500V dielectric. Also our board is fully encapsulated in an epoxy compound; only the RF output and sensor interface (12 V power + signals) will be exposed externally, so the separation distance won't be an issue.

Regarding the 12 V sensor output, we'll probably add some small resistance (~20 ohms) to reduce the effective capacitance, not too high to keep power dissipation low since it's battery powered. The spark test sounds annoying so trying to avoid that if we can.

2. RF Module
2.1. Does your agency fully understand what you are doing with the RF module?
2.2. Does the manufacturer of the RF module understand what you are doing?
2.3. Does the manufacturer of the RF module provide part numbers or only a total?
2.4. What is the procedure to lock down their design so a revision change doesn't add capacitance or inductance?
2.5. What about documentation for the factory audit process?

2.1. We haven't talked to an agency yet, still in the process of redesigning the board.
2.2. It looks like we're the first customer of this RF module trying to get IS certification; they did not have any prepared documentation...just got basic info from email correspondence. But these are great questions to ask the manufacturer, I appreciate the help.

[rancor]

Oh awesome, I didn't notice the clause about the AC coupling capacitors. That makes things much easier, thanks a lot. Our board is completely battery powered, would we still need to satisfy the 500V dielectric strength test? It doesn't seem like a problem though...looking at Digikey I can see some 39pF 0603 caps with 500V dielectric. Also our board is fully encapsulated in an epoxy compound; only the RF output and sensor interface (12 V power + signals) will be exposed externally, so the separation distance won't be an issue.

Regarding the 12 V sensor output, we'll probably add some small resistance (~20 ohms) to reduce the effective capacitance, not too high to keep power dissipation low since it's battery powered. The spark test sounds annoying so trying to avoid that if we can.

Separation distance under encapsulation still matters for your protection components so keep that in mind. You've mentioned your 12V supply is Zener clamped. All calculations for ignition current and allowed capacitance need to be made with you Zener clamp voltage plus Zener tolerance. Also if you don't have infallible separation and infallible clamping or current limiting assume all power, voltage and capacitance under the encapsulation is available at every output of the encapsulation. It might be easy to make that assumption and then just make the energy output from your exposed connections safe. 

[floobydust]

If you are potting the RF module or the board, it was so many problems.
Potting compound got in the RF modem and detuned everything. With big ambient temperature swings, sometimes parts got sheared off the board despite the use of "electronic potting compound" it had a large expansion coeff. Had to add a silicone coating first. Potting compound also changed the tuning on the antenna bias-T stripline. You can't upgrade the F/W in something potted. It's a brick. This was a board located in Zone 0 with cell modem.

Could not get a zener IS barrier to work due to the high power dissipation in a fault, so I went SCR crowbar and IS fuse instead.

Yes the barrier caps are 500V. Used 1111 [2828] Metric and can't recall if custom bit smaller pads to meet spacings. Murata GQM22M5C2H330JB01L or ATC700B330JT500

[jkostb]

Hi,

I have designed several intrinsic safe circuits, which were aproved by notified bodies. I can tell you that accounting for DC bias is NOT allowed. You even need to take into account the worst case capacitance including tolerances. For example if your nominal capacitance is 10uF with 20% tolerance than you need to to assume 12uF for your calculations. Furthermore if you don't have barriers, which prevent discharging into field lines, than you must  include all capacitors of the whiole board into account, because these can discharge into the field lines if parts fail.  Blocking capacitors are allowed as barrier, must you must ensure that these parts meet the creepage and clearance. This will become a problem if you have higher voltages.

Furthermore for all my  projects I passed the ATEX/IECEx certification without spark ignition testing, but then you must be able to prove by design that your circuit meets the 60079-11.

[jkostb]

I forgot to address your question regarding antenna. It is not possible to answer you question directly, because there is not enough info. Is your whole design intrinsic safe? Is it part of a circuit powered from mains (and therefore located in a flame proof enclosure)? If it is powered from mains and inside Ex-d enclosure the simple answer is: No it is not allowed. You need to use additional protection.

You wrote that your circuit has a boost circuit. An intrinsic safe circuit with boost converter and no clamping circuits will never pass  certification by notified body. Because these circuits can charge capacitors to very high voltages, and if these capacitors discharge into field lines, it can cause ignition