Diode Clamp Protection for CD4053B

[JDW]

After accidentally burning down a couple CD4053B chips, I realized my need for at least a resistor on all the I/O pins.  That's because there is the possibility the I/O could be connected to 12V when the CD4053B isn't powered, thereby passing current through the on-chip clamping diodes.  Since this will be used an a 12V automotive environment, I decided to add some external clamping diodes and two resistors to each I/O as shown in the schematic below (only 1 of the 3 switches will be used, hence everything else is tied to GND). 

Breadboard testing works fine with 1N4148 diodes shown in the schematic despite the fact the Vf voltage drop is larger than Vf of the on-chip diodes, and 1N4148 diodes are a tad cheaper in high quantities (4000pcs+) than a high voltage Schottky diode pair like the BAS70-06Q-7-F.  A voltage spike of 60V isn't unthinkable in a car, hence the need for something like a 1N4148 or a 70V BAS70 instead of a 30V BAT54S.  That particular BAS70 can only handle 15mA continuous current, but even if the car battery voltage reached 16V (which is a voltage level you won't see in a car even when charging the battery), the worse case current flow even with the resistors at the low end of their 5% tolerance would be 14.6mA.  However, it probably would be best to have a diode pair with a higher forward current rating, hence my current use of the 1N4148, which again can handle up to 150mA and up to 100V peak reverse voltage.

A digital data signal will be passed through CD4053B switch, and the switch will be controlled by an MCU to make or break that signal path.  I do not know if the same signal speed or signal voltage will be maintained from car to car, so keeping the resistance low in my protection circuit is desirable.  Since I have 1.22k on both sides of the switch, that's 2.44k total resistance right now, plus the 120Ω or so resistance of the switch itself.

I wanted to gather some thoughts on whether my 1N4148 choice (4ns recovery time, 100V rated, Forward current of 150mA) is a reasonable good idea, or if a high voltage (yet low cost) Schottky diode pair should be used instead.

Thank you.

[Ian.M]

If a 60V or even a 30V spike ever gets through to V2, the CD4053 will be toast anyway so a low voltage series Schottky pair is entirely adequate, and will offer much better protection due to its lower Vf and much faster operation than any silicon PN junction diode.

Take another look at ZD1,D1.  During a negative going transiernt, ZD1 will clamp at  approx -50V, with approx 15V on C1, putting a reverse 65V across the 60V rated D1.  You probably should use a unidirectional TVS diode for ZD1.   Also, if you want it to have a reasonable chance to survive a load dump, use min 1W wirewound resistors for R1, R13 (to  handle the expected >5W peak overload) and much beefier Zeners (e.g. 3W) after them with more margin below the abs. max. rating of their loads.

[JDW]

If a 60V or even a 30V spike ever gets through to V2, the CD4053 will be toast anyway so a low voltage series Schottky pair is entirely adequate, and will offer much better protection due to its lower Vf and much faster operation than any silicon PN junction diode.

Thank you for sharing your analysis.

R13, ZD3, C10 and C11 were chosen to ensure that a 60V transient entering at +Vin will NOT make it's way to +V2 (i.e., will not raise +V2 above 20V).  That part of the circuit was built to keep the voltage at +V2 to 20V or less.  So with that in mind, I am of course designing this with the understanding that the CD4053 would not be toast for a short-duration 60V transient at +Vin (where there is also C1 to attenuate the said 60V transient).  And that is why I am not so quick or eager to choose a 30V or even a 40V Schottky clamping diode pair for the I/O pins of the CD4053.  That is also why I am curious as to the best "higher voltage Schottky" diode pair for the I/O pins of the CD4053.  After doing a bit more thinking on it today, I am considering use of the low-cost 60V 200mA Rhom RB521SM-60FHT2R for D6, D7, D8 and D9.

Thank you for pointing out the possibility of a negative going spike possibly putting 65V across D1.  An ease and low-cost fix for that would simply be to use a 400V 1N4004 or similar instead of a 60V Schottky. The larger Vf voltage drop across a 1N4004 wouldn't be a problem for this circuit, especially with external clamping diodes on all I/O pins of the CD4053.

Use of a Unidirectional TVS at ZD1 would not be the most prudent choice insofar as the current placement of D1 is to the right of ZD1.  If D1 (which acts as Reverse Polarity Protection) was placed at left of ZD1 AND if D1 is also a high voltage 1N4004 or similar, then a unidirectional TVS would be perfectly safe.  But I made the current design in hopes of using a Schottky at D1, mainly for the benefit of a lower Vf.  Even so, like I said, using a 1N4004 for D1 in the current design would work fine.  Alternatively, I could just use a 100V 1A Schottky (for reverse polarity protection) like a V1FM10-M3/I or an SL110PL-TP, which probably would be a more prudent choice than the 60V 2A DFLS260-7 I use right now in terms of reverse voltage rating and cost.

As to your recommendation about using 1W-rated 100Ω resistors at R1 & R13, I actually use a 1W carbon resistor and an 1SMA4750 1W Zener in a current design that has been mass produced for the last 3 years -- a design that hasn't had any electrical issues.  But that 1W resistor eats too much PCB space, and a 0.5W Zener is a tad cheaper, and for the redesign saving space and cost wherever possible is important.  But even a 1W resistor is going to see more than 1W for a short duration.  And that duration is key.  So even if I use a 1/4W 100Ω resistor, if the duration of the spike is short (dozens of milliseconds), the resistor would not heat enough to burn from the current we are talking about in a Load Dump.  And seeing Load Dumps are rare, it would not be a case of the said 100Ω resistors being hit often with such high current.

I would appreciate hearing any further thoughts you might have in light of this.

[Ian.M]

As you say your analysis proves V2 cant exceed +20V, with respect to ground, and the clamping diodes in question clamp the signal line between V2 and ground, the peak reverse voltage across any signal clamping diode cannot exceed 20V+Vf where Vf is the forward voltage across the other clamping diode in the pair, when its passing its peak current while clamping a transient.  Therefore *IF* your analysis of the max. possible V2 voltage is correct, 30V rated diode will be sufficient for the signal clamping.

On the subject of the resistors, it all depends on their pulse overload rating and the duration of the transient.  However, as you note, load dumps are rare.   Its a trade-off you should probably be discussing with your marketing department re: the possible negative publicity impact, and with your insurance provider.

I agree a higher voltage diode for D1 would be an easy fix.  If you fit a regular silicon PN diode, you've got enough downstream capacitance to handle the reverse recovery without excessive disturbance at V1, though if you choose not to use a Schottky for D1, a soft recovery diode may be advisable to avoid blasting your device with internally generated broadband pulse EMI due to snap recovery during large negative going transients.