TVS Diode Usage and Diode breakdown voltage

[injineerka]

Hi everyone,

I am designing the circuit shown below. Here is an overview of what I am trying to achieve:

The maximum input voltage is 22.5 V, and the circuit should work with reverse polarity, hence the use of the bridge diode.

The TVS diode is an SMF4L24A from Littelfuse, and it is intended to handle transients. The TVS diode's clamp voltage is 38.9 V. I understand that the clamp voltage is not constant and can vary depending on different conditions.

I believe the TVS diode should be bidirectional, considering that the input voltage is supposed to be polarity-agnostic. Please correct me if I am wrong.

Since the clamp voltage of the TVS diode is 38.9 V, while the VRWM of the diode is 30 V, is there a possibility of damage to the diode during a transient?

A colleague suggested that the transient is very fast and therefore might not damage the diode, but neither of us has a clear understanding of whether this is correct.

We could use a higher-rated diode, but we are working with a space-constrained board and would prefer to keep the small package of the current diodes.

We also considered using a bridge-diode IC, but decided against it due to supply-chain concerns.

I would appreciate it if someone could review my assumptions and clarify these questions.

Thank you!

[iMo]

..
I believe the TVS diode should be bidirectional, considering that the input voltage is supposed to be polarity-agnostic. Please correct me if I am wrong.
..
Thank you!

Why not download and read the datasheet? The DS says it is unidirectional..

[HwAoRrDk]

If you are going to use a TVS diode to clamp over-voltage on a power supply input, ensure you have some kind of current-limiting or circuit-breaking element upstream of the TVS diode. Otherwise, in the case of a sustained over-voltage (e.g. a PSU with output higher than 22V is accidentally connected), your TVS diode will short-circuit the power supply, and either release the magic smoke itself, or overload the PSU (or both!). I suggest you add a fuse (either single-use or PTC), and size your TVS diode such that it will not be overloaded before the fuse acts.

You say your bridge diodes have 30V V_RWM. I assume this means you're using schottky diodes, as they typically have a 30V reverse rating. Any particular reason you need schottky? Regular silicon diodes typically have reverse ratings of 100V+.

Why not download and read the datasheet? The DS says it is unidirectional..

Yes, you need SMF4L24CA for bidirectional.

[Jeroen3]

The TVS diode converts the transient into heat. If the transient carries too much energy, the diode fails.
For example by having too much current (low source impedance) or have a duration exceeding the maximum energy dissipation of the diode.

I think the main purpose of the diode where you've drawn it is to protect the bridge rectifier diodes.

Since TVS diodes fail-short, you need to add protections upstream or accept a burned pcb.

[injineerka]

I know that it is unidirectional. What i meant was that it should be bidirectional TVS instead of unidirectional.

[injineerka]

Thanks for the reply. It is current limited upstream. Yes it is indeed schottky diode, and the main reason was the small size and low Vf. But I might need to reconsider it again and look for smaller package of standard diode.
Thanks again, it was helpful insight.

[Terry Bites]

I bumped into this just a minute ago.
www.ti.com/tool/download/TVS-RECOMMENDATION-CALC/1.0.0
Set target device to "new"

[tszaboo]

A colleague suggested that the transient is very fast and therefore might not damage the diode, but neither of us has a clear understanding of whether this is correct.

We could use a higher-rated diode, but we are working with a space-constrained board and would prefer to keep the small package of the current diodes.

A diode breakdown normally alone doesn't damage a rectifier diode. It's just going to conduct when it's not supposed to.
You need to have a definition of what the source of this transient voltage though. An ESD event from a human is different than a IEC surge pulse train. It's also different from an electric motors back EMF, or having the wrong voltage connected to the input of a device constantly.
So to answer correctly, you should first ask the correct question. What are we protecting against?

[PGPG]

You have to know against what pulse you are trying to protect.
Imagine you got 100V pulse at your V1 + pin but not versus V1 - pin but versus GND.
Nothing will prevent this pulse from going to your load.

Do you assume your V1 source can generate high voltage pulses or do you assume nearby storm strike will generate pulse at both V1 pins versus your GND?

In TVS datasheet there should be information (or chart) how long pulse of specified current the diode should survive.
Typical relation I expect would be that 10 times higher current pulse will have to be 100 times shorter.

[injineerka]

A colleague suggested that the transient is very fast and therefore might not damage the diode, but neither of us has a clear understanding of whether this is correct.

We could use a higher-rated diode, but we are working with a space-constrained board and would prefer to keep the small package of the current diodes.

A diode breakdown normally alone doesn't damage a rectifier diode. It's just going to conduct when it's not supposed to.
You need to have a definition of what the source of this transient voltage though.

An ESD event from a human is different than a IEC surge pulse train. It's also different from an electric motors back EMF, or having the wrong voltage connected to the input of a device constantly.
So to answer correctly, you should first ask the correct question. What are we protecting against?

Thanks for the reply. basically we are considering these types of transients such as IEC ESD compliance test, hotplugging the PSU, lightning surges etc. and we are not sure if this will damage it or not or how will the diode react to it. in your opinion what do you think the issues will be given these type of ESDs? but we are leaning more towards going with higher rated voltage diode probably with Vr more than 60V.

[injineerka]

V

You have to know against what pulse you are trying to protect.
Imagine you got 100V pulse at your V1 + pin but not versus V1 - pin but versus GND.
Nothing will prevent this pulse from going to your load.

Do you assume your V1 source can generate high voltage pulses or do you assume nearby storm strike will generate pulse at both V1 pins versus your GND?

In TVS datasheet there should be information (or chart) how long pulse of specified current the diode should survive.
Typical relation I expect would be that 10 times higher current pulse will have to be 100 times shorter.

So we are mainly using the TVS for ESD protection such as lightning, IEC compliance. and assuming that the TVS diode can handle the transient, then it will clamp the voltage at it Vc which is 38.9V, but this voltage is higher than the reverse voltage of the bridge diodes, so my question is, will this transient voltage damage the diode? and what kind of damage will it cause?

[PGPG]

V

Imagine you got 100V pulse at your V1 + pin but not versus V1 - pin but versus GND.
Nothing will prevent this pulse from going to your load.

So we are mainly using the TVS for ESD protection such as lightning, IEC compliance. and assuming that the TVS diode can handle the transient,

Just think about my 100V example. Your TVS will not notice it at all and this 100 will go to your load (I am ignoring at this moment a problem what diodes are - assume they are 1000V).
Don't mix all the time ESD with lighting as we don't know against what you are trying to protect your circuit.
ESD is from your finger - is very fast (nanoseconds), can be even 25kV (3kV ESD from your finger you even not notice, but device can be damaged).
If ESD from your finger be put into your V1 pin it will be searching a way to GND and not to second V1 pin so it will ignore your TVS. Your TVS protects V1 against being forced to higher voltage between its + and - pins but not protects your load.
When we speak about lighting effects we speak about surge pulse. It is much slower (microseconds) and longer (so much higher energy). As you are probably not building energy supply for building you need not assume direct lighting but only indirect effects mainly by lighting currents traveling through GND. In such case it strongly depends on where are located your two devices (V1 source and load). The situation is much different if they are located in separate buildings then if they are both in the same building (with common grounding system).

ESD protecting diodes must be very fast and with low ESL (equivalent series inductance) because very high dI/dt of pulse can produce high voltage at ESL. I remember reading (20+ years ago) of someone who measured 200V pulses generated by ESD at low voltage Surge diode (it was possible because its high ESL). But ESD protecting diodes need to withstand smaller energy that surge protecting diodes. Surge protecting diodes are bigger but you don't worry of having because of it higher ESL as these pulses have much smaller dI/dt.
Except ESD and Surge you should also consider Burst pulses, but as they are 'between' these two if you protect against ESD and Surge you in most cases are also protected against burst.

TVS diodes do not protect devices alone. They 'cooperate' with serial impedance (it can be cable impedance or intentionally inserted serial element).

EMC standards regarding your device specify against what pulses levels you should protect its pins.

[injineerka]

Thanks PGPG. That gave me a lot to think about.