Hi!

I'm looking for some expertise in circuit modifications for ESD protection. I am trying to protect an analog switch (part number DG412) from ESD damage. The switch is connected to a banana jack input. We know that the analog switch survives below 4kV, but anything higher and the switch fails. We first connected a TVS diode to the banana input and tied that to ground. It seems to survive at an 8kV ESD blast, but there is an issue with the high capacitance that the TVS diode creates. Since I am using a cap sense chip, the TVS diode to ground will falsely trigger my cap sense.

I did some digging online and found a steering diode configuration using two fast response diodes and the same TVS diode. Connecting it in the configuration shown below solves the capacitance issue, however, at 8kV ESD test the configuration fails. We have tested this numerous times using different ground points for the TVS anode side, but it seems to continue to fail. I have included the information sheet about the steering diode configuration below for a reference.

Please let me know if you spot anything wrong with my setup! I would appreciate any and all help on this (hair pulling out) project! Thank you in advance! Please let me know if you need any additional information about my setup.



Here is a list of attachments I have included:

- "Steering Diode specs" include the link to the datasheet and main specs for the TVS and fast recovery diodes.
- Datasheet for the analog switch (DG412).
- Diagram of the steering configuration attached to the chip. I have also included a picture of this configuration as it appears on my board. The cathode of the TVS is connected to pin 13 (32V) on my switch

Steering diode config taken from "Understanding Steering Diode" (attached)

 

You will need to add a series resistor or PTC to the input.
https://circuitcellar.com/ee-tips/circuit-protection-ee-tip-116/
Some way to limit the input current is needed to protect the diodes. Just the diodes can work for low current sources, but adding the series resistor is a good idea anyway.

 

You will need to add a series resistor or PTC to the input.
https://circuitcellar.com/ee-tips/circuit-protection-ee-tip-116/
Some way to limit the input current is needed to protect the diodes. Just the diodes can work for low current sources, but adding the series resistor is a good idea anyway.

@kendal Sorenson
I assume by "fail" you mean that the circuit misbehaves or is damaged, but that the TVS diode and the "steering diodes" survive? If so, a resistor between the diodes and the Protected Device, with limited voltage appearing across the resistor, may be useful. A ferrite bead could be used in addition to or replacing the resistor. A resistor from the banana jack to the diodes would likely fail as most small resistors survive only a few hundred volts before there is arc-over. Unless you know that the Protected Device input is being damaged, you should also consider that your static discharge is likely disturbing circuit ground considerably, possibly causing damage elsewhere than at the input. Also, lightning bolts--both large and very small--are prolific generators of RF energy that goes pretty much wherever it feels like going.

 

You have asked a very valid question. Check this article, it should point you in the right direction.

 

See

 

What are the expected input voltage range under normal operation? What is the circuit impedance beyond the switch? Is the problem only when the input is not connected, or even when a source is connected?
An arrangement that can clamp the input at 1 volt can provide a good bit of protection, but if your application demands a range of +/- 10 volts ten a different arrangement is needed. So please provide a bit more information, otherwise all you get "is best guesses", some good and some not so good.

 

@kendal Sorenson
I assume by "fail" you mean that the circuit misbehaves or is damaged, but that the TVS diode and the "steering diodes" survive? If so, a resistor between the diodes and the Protected Device, with limited voltage appearing across the resistor, may be useful. A ferrite bead could be used in addition to or replacing the resistor. A resistor from the banana jack to the diodes would likely fail as most small resistors survive only a few hundred volts before there is arc-over. Unless you know that the Protected Device input is being damaged, you should also consider that your static discharge is likely disturbing circuit ground considerably, possibly causing damage elsewhere than at the input. Also, lightning bolts--both large and very small--are prolific generators of RF energy that goes pretty much wherever it feels like going.

Hi Teekay,

You nailed it. The steering diodes are always intact after the ESD blast test. I have tested this twice. On the first test, the circuit malfunctioned. I took off the diodes, the circuit returned back to normal, then put the same diodes back on and the circuit appeared to be okay. I'm not sure what happened there. Maybe a grounding issue like you mentioned, but the circuit was hard grounded during the test. On the second test, the chip failed, so returning the circuit back to normal required a chip replacement. The diodes are still good. It looks like I got some good feedback from you guys (thank you!). For the next test, I plan on adding a series 1k resistor to the input (schematic below). Regarding your comment about the ground disturbance affecting the circuit, would this be an issue even if my circuit is connected to a hard ground?

Some other additional info: the board is two layers (no ground or power plane). The circuit is measuring resistance. It starts as a voltage divider. So 32V divides between some going to an analog circuit (where the resistance measurement occurs) and the rest outputting to the banana jack. So the more resistance on the input of the banana jack, the more current goes through the analog circuit. Adding the 1k resistor did throw off the resistance measurements by 1k, but we can likely adjust that in the firmware if the resistor helps. Thank you for the help! If you have any other suggestions in light of this additional info I provided please let me know!

 

Hi Teekay,

You nailed it. The steering diodes are always intact after the ESD blast test. I have tested this twice. On the first test, the circuit malfunctioned. I took off the diodes, the circuit returned back to normal, then put the same diodes back on and the circuit appeared to be okay. I'm not sure what happened there. Maybe a grounding issue like you mentioned, but the circuit was hard grounded during the test. On the second test, the chip failed, so returning the circuit back to normal required a chip replacement. The diodes are still good. It looks like I got some good feedback from you guys (thank you!). For the next test, I plan on adding a series 1k resistor to the input (schematic below). Regarding your comment about the ground disturbance affecting the circuit, would this be an issue even if my circuit is connected to a hard ground?

Some other additional info: the board is two layers (no ground or power plane). The circuit is measuring resistance. It starts as a voltage divider. So 32V divides between some going to an analog circuit (where the resistance measurement occurs) and the rest outputting to the banana jack. So the more resistance on the input of the banana jack, the more current goes through the analog circuit. Adding the 1k resistor did throw off the resistance measurements by 1k, but we can likely adjust that in the firmware if the resistor helps. Thank you for the help! If you have any other suggestions in light of this additional info I provided please let me know!

@Kyndal Sorenson
Rather than the 1K resistor, you could try one (or several in series) ferrite bead; these increase the inductance of the wire passing through them and thus limit high frequency (but not DC) currents. For your application it would likely be best to choose beads that claim to be non-conducting. There is a huge selection of beads available; I cannot help you much in selecting an appropriate bead. Perhaps another AAC commenter can help.
As for a hard ground, such a ground exists only in theory. In reality every conductor has inductance and resistance. If a tiny 8KV lightning bolt strikes a 1/4" diameter copper rod at its end, the voltage seen even 1" away from that end will be different than is seen at the end. This is partly due to the inductance of the rod and the extremely high frequencies generated by an electric arc. A PCB trace of 0.1" width that runs from point A to "ground" two inches away has significant inductance and can easily affect results of ESD testing. Proper PCB layout and construction to minimize ESD problems is an art and a science. There is a lot of info available at AAC (search "ESD", "RFI") and online (use Google search or equiv.). Depending on the ESD regulation you intend to meet, an actual (tiny, short, weak) arc may or may not be present; no arc generally means less RF noise is generated.

A related note: When using the "steering diodes" scheme, it is imperative that a low ESR cap be connected from the +PS to ground at the same point as the diodes...and as you are aware, that ground must be a low resistance, low inductance ground...at least until it reaches a ground point sufficient to dissipate whatever ESD energy might be present. Every inch of ground path matters.

 

@Kyndal Sorenson
Rather than the 1K resistor, you could try one (or several in series) ferrite bead; these increase the inductance of the wire passing through them and thus limit high frequency (but not DC) currents. For your application it would likely be best to choose beads that claim to be non-conducting. There is a huge selection of beads available; I cannot help you much in selecting an appropriate bead. Perhaps another AAC commenter can help.
As for a hard ground, such a ground exists only in theory. In reality every conductor has inductance and resistance. If a tiny 8KV lightning bolt strikes a 1/4" diameter copper rod at its end, the voltage seen even 1" away from that end will be different than is seen at the end. This is partly due to the inductance of the rod and the extremely high frequencies generated by an electric arc. A PCB trace of 0.1" width that runs from point A to "ground" two inches away has significant inductance and can easily affect results of ESD testing. Proper PCB layout and construction to minimize ESD problems is an art and a science. There is a lot of info available at AAC (search "ESD", "RFI") and online (use Google search or equiv.). Depending on the ESD regulation you intend to meet, an actual (tiny, short, weak) arc may or may not be present; no arc generally means less RF noise is generated.

A related note: When using the "steering diodes" scheme, it is imperative that a low ESR cap be connected from the +PS to ground at the same point as the diodes...and as you are aware, that ground must be a low resistance, low inductance ground...at least until it reaches a ground point sufficient to dissipate whatever ESD energy might be present. Every inch of ground path matters.

Hi Teekay,

Thank you again for your help! I drew a diagram illustrating what I think you mean regarding placement of the Low ESR cap and the ferrite bead. Does that diagram look correct based on what you are saying?

 

Hi Teekay,

Thank you again for your help! I drew a diagram illustrating what I think you mean regarding placement of the Low ESR cap and the ferrite bead. Does that diagram look correct based on what you are saying?

@Kyndal Sorenson
Your sketch is very close. If you intend to use the "steering diodes", then the path ought to be from the jack to the steering diodes to the bead(s) to the IC input. Remember that ground is part of the path; wider traces and shorter traces yield lower inductance and better grounding. Ground is not just a symbol; it is a major element of the path that the ESD current will take. Ideally the jack, the ground for the steering diodes and for the bypass cap of the steering diodes, and the ground of the IC will share a large copper area tying them all closely together via a low inductance and low resistance path. The goal is that the steering diodes will intercept and at least somewhat reduce the amplitude of the incoming spike before it is applied to the bead(s), resistor(s) and IC. There is nothing wrong with retaining a resistor in series with the bead on the assumption that the steering diodes limit the amplitude enough that the resistor does actually resist rather than being bypassed by an arc. In that sense, the resistor should have as high a voltage rating as possible (generally meaning it's physically large). 0603, 0805 and similar SMDs have in general very low voltage ratings (it is possible and reasonable to use several R's in series to achieve a higher rating). If your ESD test requirement does require creating arcs wherever possible (i.e. all user-accessible conductive surfaces, including gaps in enclosure coverage), then minimizing the accessiblity of such surfaces is an important part of ESD immunity. Arcs are wicked!