DC Power input & filtering & TVS for ESD

[hcglitte]

Hi,

I would like to suppress noise which is picked up on the charging lead (wires) from entering into my product/circuits.
Also, I would like to protect the circuits from ESD events.
I have attached a schematic of how I plan to implement this. I want to use a ferrite bead on both positive and negative leads.
Ground noise generated on my PCB will be suppressed by the negative bead.
The TVS is placed directly after the connector. The problem is, I have not found any reference schematic where both TVS and dual ferrite beads have been used.
There is only a TVS followed by one bead (on positive). And there is reference schematics where dual beads are used - but no TVS.

Using both schemes would imply that the TVS must be placed before the two beads.
If the TVS was placed after the beads, a very high voltage would build over the beads possibly destroying them.

However, since I have not found any similar designs online, I am a bit worried that I may have missed something...
Any input is much welcome

Regards,

HC

[T3sl4co1l]

FYI:

(I'm going to start copy and pasting this, since it probably can't be repeated enough!)


Ferrite beads are rarely suitable for their claimed purposes.  The problem is they don't give much impedance (until very high frequencies, where they are mostly resistive anyway), and they saturate easily with DC bias.  You might add a ferrite bead to a power supply connection, and have it work perfectly well at light load, but then it does nothing at higher currents.  Sadly, ferrite beads are rated by thermal limit, not electrical limit, so it's impossible to tell how much current is actually reasonable.

In almost all cases, you should use an inductor instead.  The inductance is higher, the peak impedance is much higher, and it will remain normal up to rated current.  It's also a reminder that you are forming an LC resonant circuit, and need to dampen that circuit with losses: usually from capacitor ESR.

The best places to use ferrite beads are on signal lines, to provide modest attenuation (maybe < 10dB) of high frequency, interfering (RFI) signals.  This is commonly done on signal outputs for reduced emission, signal inputs for improved susceptibility, or some internal signals for filtering or parasitic purposes (occasionally, a ferrite bead is used instead of a gate resistor, for driving MOSFETs).


So, ferrite beads not so great, but, a common mode choke including both power leads is quite reasonable (the balanced current ensures the part does not saturate due to DC current flow).

The TVS should follow the inductance, because during transients or hot-plugging, voltage drop occurs across that inductance.  Following that inductance with capacitance means that, the voltage drop is probably going to swing around and become voltage overshoot, causing stress on your capacitor and attached load!  Placing the TVS at the capacitor will do a fine job clamping things.

There's also the issue of: what situation are you protecting against?  If automotive transients, then a TVS won't do enough (there's one elephant in the room: load dump).  If it's transients from long cable runs, a TVS might again be inappropriate (you might need big MOVs or GDTs, plus a ground path!).  If it's battery reversal, a unidirectional diode will do (as long as the input is fused), but a series diode might be simpler and more gentle (it doesn't blow a fuse or risk damage when reversed, it just doesn't do anything).

BTW, for hot plugging, be careful to avoid parts that are sensitive to current spikes.  The most important being tantalum capacitors.  A capacitor with modest ESR is probably preferable, to minimize surge current and dampen ringing: a common electrolytic does wonders.

Tim

[Jeroen3]

Put a fuse or PTC right at the connector. If you connect a 24Vdc bus on it, or reverse it (use a unidirectional tvs) the circuit the fuse will blow. Just make sure you have a big enough TVS to not explode before the fuse does.
Sometimes a bridge rectifier is used to create an input that has no polarity, and does not leak current back from the inputs caps when the power has a dip.

[T3sl4co1l]

FYI, PTC fuses aren't that good.  They take forever to react (near 100ms under fault), are fairly fragile (check the maximum open voltage and peak fault current limits), and are very difficult to blow using semiconductors (if you want to clear a transient fault with a TVS, or crowbar a supply).

A fuse of an appropriate type is definitely wise, though!

Tim

[hcglitte]

Hi Tim, thank you for your thorough reply.

I was not aware of the DC current bias effect of ferrite beads. I read some now. I have seen that some are targeted against "high current" applications and some are more aimed at digital signal conditioning.

I did not get in what order of sequence you meant that the DC-input jack, TVS, inductors and capacitor should be placed in.
I believe you meant this order:

DC-JACK -> Series Inductors (both negative & positive) -> TVS -> Capacitor ?

I read this document from TI: www.ti.com/lit/an/slva680/slva680.pdf

This document states that the inductance from the ESD source into the TVS should be minimized. And it says that the inductance after the TVS towards the "victim" should be increased. This would justify the order of sequence the components are placed in my original schematic, I would think...? Also, if the inductors were in front,  couldn't there be a possibility that they could become destroyed by an ESD event?

This link: http://www.analog.com/en/design-center/reference-designs/hardware-reference-design/circuits-from-the-lab/cn0328.html#rd-description
Figure 7, does have the TVS-like suppressors according to my original schematic also. The pdf they refer to also states the the additional "TVS's" may be left out. This means that the first one must be there, and then it becomes similar to my design with the exception that they use inductors instead of beads (your point exactly!) and they have included a PTC as well -  but that is off topic for this discussion.

Jeroen3: I know PTC's are slow. I'm not sure what you wanted to combat with the unidirectional TVS and PTC solution. I understand what you meant design wise though. If the solution proposed was to replace the PMOS, I guess it would work (since the TVS is fast), but it would still allow a small negative voltage drop over the rest of the circuitry before the PTC kicks in.

HC

[T3sl4co1l]

You're not usually concerned about ESD strikes to a power line: the filtering is beefy enough to take care of that already, and the ground path is directly into the entire circuit ground, carrying the high frequency currents harmlessly over the circuit like a Faraday cage.

Surges such as IEC 61000-4-5 are more the domain of power supplies and large filter caps and TVS diodes.  Or more applicable standards for respective purposes, automotive for example.

If you aren't concerned about any external surges at all, and only wish to protect against hot-plugging transients, then the location of maximum voltage occurs at the first filter cap, and that is where the TVS should be placed.

Tim