# 3.3V overvoltage limiter circuit topology

I need to design, using discrete components only, a shunt voltage limiter that drains little or no current (less than 2mA, the lower is better) at 3.3V, and ensure that, when driven at more than 4V, it sinks at least 200mA. No integrated circuit can be used for such purpose (TL431 would solve this immediately). The circuit should remain functional with a delta (positive and negative) of 50°C from the ambient temperature of 25°C and within component tolerances. In addition, the circuit should be kept as simple as possible.

Input parameters:

• Guaranteed non triggering voltage < 3.3V
• Guaranteed triggering voltage > 4.2V
• Clamping current > 200mA

I tried to explore many different options with zeners, resistors, mosfet, bipolar transistors, and compensated transistor pairs, but I was unable to work out a proper solution.

ALD suggests a very simple solution that may fit within requirements by using ALD111933 (http://www.aldinc.com/pdf/VoltageClampCircuits.pdf) however it’s a rather exotic component with a noticeable cost per part.

Can you suggest circuit topologies that could help me to solve the problem?

EDIT: Some background might be needed to understand the question:

1. The requirement of not using integrated circuits is related to EN60079 (ATEX/IECEx) requisites. While there might be discordancies about this point, most certification agencies should reject designs where integrated circuits are used as components rated against section 7, even if replicated their fault cannot be countable. This should be the position of most notified bodies, officially endorsed by Intertek (http://www.intertek.com/uploadedFiles/Intertek/Divisions/Commercial_and_Electrical/Media/PDF/Hazardous_Locations/Preventing-Disasters-Through-Intrinsic-Safety.pdf, integrated circuits are assumed to fail in any condition and those faults are non countable).

2. Why this solution should be technically archiveable is because a similar circuit should be the basis of RS485-IS protection circuit described here: https://www.profibus.com/download/profibus-rs485-is-user-and-installation-guideline/ in order to archieve a Uo of 4,2V with a 3,3V transceiver, there must be a similar voltage clamp. Note that if all 32 transceivers on the network would sink 20mA per diode, the power supply should be capable of sourcing nearly 2A just for the protection diodes. Or, each of the circuits should include a DC/DC to avoid overloading the bus).

• Build a discrete TL431? Just a zener? Is this homework? – winny Jul 8 '18 at 14:01
• "No integrated circuit can be used for such purpose" - why not? Engineering requirements need to make sense... – brhans Jul 8 '18 at 14:02
• I would have expected your comments... thank you for pointing it out. The circuit needs to be verified formally against failures of single components and component ratings against EN60079 regulations (see §7.6d of IEC 60079-11 Ed 6). Countable faults and infallible assembles are explicitly requiring that the circuit is well known and each source of fault is described in detal. ICs. are explicitly banned for the purpose. Please tell me if I should elaborate with more details... – Fab Jul 8 '18 at 14:19
• @winny it's a low voltage protection circuit designed for explosion proof environments. I know products that already solved the issue but unfortunately I don't have one on hands to disassemble. I hope there are suggestions that may point me in a possible direction. I am not trying to get the homework done... – Fab Jul 8 '18 at 14:28
• Series-parallel zener-resistor arrangement should solve your needs and be tolerant to first order faults. – winny Jul 8 '18 at 14:30

The circuit needs to be verified formally against failures of single components and component ratings against EN60079 regulations (see §7.6d of IEC 60079-11 Ed 6). Countable faults and infallible assembles are explicitly requiring that the circuit is well known and each source of fault is described in detal. ICs. are explicitly banned for the purpose.

It sounds like you are trying to build a zener barrier and zener barriers use two zeners for category i(b) and three zeners for i(a) intrinsic safetly category. Zeners are not regarded as infallible and barrier designers use more than one and each one is individually rated to perform without the others.

Zeners of course fail (they are not regarded as infallible) and the same is true for ICs so, I would avoid the problem you appear to be creating for yourself and work along the lines of using an IC and duplicating or triplicating the circuit in order to reach the level of intrinsic safety required for the device.

ICs are NOT explicitly banned BTW.

• Thank you for your reply. What you write about zener and infallible assemblies of diodes is well known. Using ICs as countable faults looks not feasible according to requirement of EN60079-11. I could be wrong (and I hope so). May I ask you if you have proof that ICs can be treated as a countable fault against EN60079-11? – Fab Jul 8 '18 at 15:01
• I've used duplicate transistor and op-amp circuits for an i(b) application that was approved by BASEEFA back in the late 80s. I don't have copies of the I.S. requirements any more so I'm stuck as to being able to read chapter and verse on the subject. – Andy aka Jul 8 '18 at 15:13
• You may look at Intertek's interpretation here: intertek.com/uploadedFiles/Intertek/Divisions/… (page 6 § 3). I've seen implementation with ICs but IC approach should not have been accepted according to this strict interpretation. This was also the position of a member of a member of TC 31 we've discussed with – Fab Jul 8 '18 at 15:19
• I don't see anything in the link that worries me - they talk about zeners but that doesn't mean a circuit of adequate integrity couldn't perform a zener function. You should speak with the approval authority for your area (as I did back then). – Andy aka Jul 8 '18 at 15:31
• I was meaning this extract: "Any component or construction not complying with the standards is assumed to fail to open or short conditions. As far as the standard is concerned, a semiconductor is a diode, zener diode, FET, Transistor, or simple PN junction device. It does not include integrated circuits because what is inside the integrated circuit is an unknown. Integrated circuits are assumed to fault to any state between their pins and this is not considered as a countable fault." – Fab Jul 8 '18 at 15:33

There is a crowbar circuit in my copy of "The Art of Electronics" that may fulfill your requirements:

simulate this circuit – Schematic created using CircuitLab

I did a quick simulation with D1 being a 3.3V zener diode and it seemed to be working quite well.
Because you have 0.9V difference between guaranteed triggering and non-triggering you should not have too much problems with the "knee" of the zener diode I/V-curve.