TVS for Protecting a 38Vmax DC-DC that needs to operate at 24Vin??......

[Smokey]

I have a design that's using a LMR33630-Q1 DC-DC converter.
https://www.ti.com/lit/ds/symlink/lmr33630-q1.pdf

That chip has a 38V absMax.  It will be running from nominal 24Vin, which is within the 3.8-36V range. 

--> Problem is that the TVS diodes to protect the input which have a max clamping voltage under 38V also have a Reverse Standoff voltage at or below 24V.  So this means that these diodes will most likely be conducting normally with a 24Vin.  Spice confirm this.  If my Vin comes in high at 26V (which might happen) then there is like 1.something watts in that TVS diode all the time.  That sucks.
Example: https://www.diodes.com/assets/Datasheets/ds19002.pdf

What can I do here?  I was looing at a combination of the next bump up in TVS Reverse Standoff Voltage, so a 26V (or 28V), which has a Max Clamping voltage in the 41.6-45.5V range.  If I use one of those in combination with a normal non-tvs zener at 35V (or even lower), spice says the voltage gets clamped below 38V for a load dump.  Not sure if I should trust the spice here though. 

Thoughts?

[johansen]

The data sheet shows 5uA at the rated vrm?

The max clamping voltage if im not mistaken is the voltage at the rated peak current, for a 24v rated part, which starts to conduct as low as 26.7 volts,, at 15 amps it will be 38 volts.

If 15 amps isnt enough or 38 volts too high, parallel more of them.

And yes you may need a 26v tvs to handle the nominal variations in 24v power supplies

[Smokey]

That's interesting.  The two parallel TVS does drop the max clamping voltage (according to LTSpice at least).  Cool.
Unfortunately it's not enough to fix needing a 26V TVS and protect the DC-DC.  I don't have room for like 4 of them

Is the normal 35V zener in parallel with the 26V TVS that terrible of an idea?

[jonpaul]

supply is in vehicle? ALT load dump transient? What is the transient threat?

We use MR2525, MR2535 Mot alt load dump for avaition and automotive bus 12/24/28V

Jon

[T3sl4co1l]

To some extent you can rely on abs. max. being somewhat conservative, i.e. it's rated 38V max continuous, in the same way the diode is rated 26V max continuous; breakdown is always some margin above there.  You don't know by how much, of course, and they never specify breakdown of the regulator.

Clamping is also at Ipp.  What are your surge conditions?  If much more relaxed (e.g. automotive pulses other than ISO-7637-2 pulses 1 and 5), expect tight clamping, closer to Vbr(max).  Conversely, if you don't have much pulse to handle but it's very intense (e.g. IEC 61000-4-5 as might be applied to long DC lines), you might be closer to the 8/20us waveform (not defined for this SMBJ but others have it), where peak voltage can be several times higher (MOVs are competitive!), simply because the energy is relatively low at the short duration.

If you do need to withstand pulses like load dump, consider a voltage limiter or electronic switch/fuse approach.

Tim

[Smokey]

Ya, I should have been more clear.  It's automotive, and the concern is load dump.
The vast majority of the time this thing will be running from 12V and that's what I originally designed it for (hence the 38V DC-DC and 24V TVS), but I picked up this application that needs to run at 24V and that sort of kills my design margin.  I'm trying to see if I can salvage this, or if I need to redesign the DC-DC to use a 60Vmax Vin regulator (which I'm not looking forward to).

[langwadt]

That's interesting.  The two parallel TVS does drop the max clamping voltage (according to LTSpice at least).  Cool.
Unfortunately it's not enough to fix needing a 26V TVS and protect the DC-DC.  I don't have room for like 4 of them

Is the normal 35V zener in parallel with the 26V TVS that terrible of an idea?

why do you need a 26V TVS?  V_BR is >26.7V for the 24V TVS

[thm_w]

If its automotive, two series batteries, wouldn't the max continuous Vin be more like 13.5V * 2 = 27V?
Depending on the float voltage.

[Smokey]

That's interesting.  The two parallel TVS does drop the max clamping voltage (according to LTSpice at least).  Cool.
Unfortunately it's not enough to fix needing a 26V TVS and protect the DC-DC.  I don't have room for like 4 of them

Is the normal 35V zener in parallel with the 26V TVS that terrible of an idea?

why do you need a 26V TVS?  V_BR is >26.7V for the 24V TVS

TVS can start to conduct at the Reverse Standoff Voltage. 

[T3sl4co1l]

What level? 24V load dump can go up over 150V. You need way the hell more than an SMB TVS to handle that!

Tim

[Smokey]

If its automotive, two series batteries, wouldn't the max continuous Vin be more like 13.5V * 2 = 27V?
Depending on the float voltage.

Ya, this may be the case.  The alternator is probably charging up to 28V.  Either way I need some actual margin so it will probably end up being a 30V TVS.

I think I'm going to just redesign the power supply to use a 60V max part.  You guys had all day to come up with an idea and be a savior, but you let me down!.... I kid.. I appreciate the ideas.

The SMB part was from the design when the requirements were 12V.  There actually are two in parallel on that design though.

[PCB.Wiz]

If its automotive, two series batteries, wouldn't the max continuous Vin be more like 13.5V * 2 = 27V?
Depending on the float voltage.

Ya, this may be the case.  The alternator is probably charging up to 28V.  Either way I need some actual margin so it will probably end up being a 30V TVS.

I think I'm going to just redesign the power supply to use a 60V max part.  You guys had all day to come up with an idea and be a savior, but you let me down!.... I kid.. I appreciate the ideas.

The SMB part was from the design when the requirements were 12V.  There actually are two in parallel on that design though.

If you have room for two, a suitable resister between them will expand he clamping.

I see TI claim to have 'better' TVS

https://www.ti.com/lit/pdf/slyy127

[Smokey]

What is going on with this part? 
https://www.diodes.com/assets/Datasheets/D30V0S1UG3LP20.pdf

The datasheet claims to have a Reverse Working Voltage of 30Vmax, AND an ESD Clamping Voltage at 70A of 30.5V?


The plot looks like the clamping voltage dips to what looks like 23ish voltage at low currents.


What gives?

[Marco]

Why don't they just make buck converters with back to back high voltage MOSFETs, which block both load dump and continuous/temporary reverse polarity, and some build in TVS to bridge the time for the MOSFETs to kick in? Lose a tiny bit of efficiency, lose a ton of external components at the same time. You'd think there's a market.

[PCB.Wiz]

What is going on with this part? 
What gives?

Get a part and measure it ?
The numbers seem to contradict for a simple clamping part, but there are break-back thyristor trigger parts that drop to lower voltage.
Imagine if a vendor combined (say) a 20V zener with a 10V breakback part, such a part could trigger > 30.5V and clamp at 25V 50A ?

Addit : yes, see fig 5 in their data, that does not quite go through their claimed 25V/50A but it does have a negative resistance action.

[temperance]

If more precision is required, a crowbar circuit made with a TL431 or some 30 V zener diode and PNP power transistor has a text book like knee voltage.

Or a two step approach. A TVS and after the TVS a low value resistor and shunt regulator with a zener and a PNP transistor.

[langwadt]

If more precision is required, a crowbar circuit made with a TL431 or some 30 V zener diode and PNP power transistor has a text book like knee voltage.

Or a two step approach. A TVS and after the TVS a low value resistor and shunt regulator with a zener and a PNP transistor.

depending on current why make it a crowbar? why not an (high voltage) NPN and a 30V zener, make a crude linear regulator

[Smokey]

I thought normal zeners aren't fast enough... 

[johansen]

I thought normal zeners aren't fast enough...

Nanoseconds not fast enough?

[Smokey]

Nanoseconds!  Talk to me again when you got some attoseconds!
https://www.nobelprize.org/prizes/physics/2023/press-release/

Either way, I think this actually solves my problem:
https://assets.nexperia.com/documents/data-sheet/PTVS30VZ1UPA.pdf

30V Reverse Standoff, 37Vmax Clamping Voltage.  Tiny package.

[temperance]

I thought normal zeners aren't fast enough...

The difference between a zener and a TVS is the thermal capacity which has been enhanced by a copper slug for a TVS diode.

depending on current why make it a crowbar? why not an (high voltage) NPN and a 30V zener, make a crude linear regulator

That's of course also possible. But considering the high output current (not stated but the buck reg in question is 3 A capable) the drawback is the power dissipation in the N type pass device. You might argue for A P type type voltage regulator. But those suffer from very bad power supply rejection because the base current or Vgs is referenced to the input and a transient will modulate the current / voltage while for a NPN or N-channel MOSFET the base current or Vgs is referenced to the output.

[temperance]

Either way, I think this actually solves my problem:

Nice find. Noted.

[bdunham7]

Ya, this may be the case.  The alternator is probably charging up to 28V.  Either way I need some actual margin so it will probably end up being a 30V TVS.

You are a lot more likely to see a nominal 24/28V system exceed 30V continuously (meaning for at least a few minutes) than you are to see a load dump actually hit 150V.  And in the case of a regulator malfunction, you could see 32 volts or more for quite a while.  Unless you have resistance in front of this device sufficient to allow your TVS to clamp continuously without melting, you're trading one vulnerability for another, and a more common one at that.  Clamping a load dump requires either limitation of the input current or some really big parts.  A 60V regulator would withstand almost every load dump short of a battery disconnect in a vehicle with a generator that lacks internal overvoltage protection. 

[langwadt]

I thought normal zeners aren't fast enough...

The difference between a zener and a TVS is the thermal capacity which has been enhanced by a copper slug for a TVS diode.

depending on current why make it a crowbar? why not an (high voltage) NPN and a 30V zener, make a crude linear regulator

That's of course also possible. But considering the high output current (not stated but the buck reg in question is 3 A capable) the drawback is the power dissipation in the N type pass device. You might argue for A P type type voltage regulator. But those suffer from very bad power supply rejection because the base current or Vgs is referenced to the input and a transient will modulate the current / voltage while for a NPN or N-channel MOSFET the base current or Vgs is referenced to the output.

guessing that the buck output voltage is much lower than the 24V input so the input current is much less than 3A

or with and extra transistor and a TL431 the pass element could be a P-fet for practically no loss in normal operation

[T3sl4co1l]

Either way, I think this actually solves my problem:
https://assets.nexperia.com/documents/data-sheet/PTVS30VZ1UPA.pdf

30V Reverse Standoff, 37Vmax Clamping Voltage.  Tiny package.

...But your problem was load dump?  How is that dinky thing going to absorb 10+ joules?

And yeah, they don't make any mention about what kind it is, but I would guess either a thyristor or snapback type.  Snapback can actually have negative resistance comparable to internal resistance and therefore a pretty flat V(I) curve, with a fairly modest step as it latches on.  Thyristor types I think only latch on hard (to a couple volts).

Tim