Circuit Protection & Input Filtering Tips

[saturdays_law]

Hi all,

First of all, this is my first post here, so thank you all who are willing to spare some of your time to read this post, I'll try to make up for it and help others in the future!

I have some questions/request for comments about a personal project which includes a DC/DC converter - specifically a **broken link removed** - to convert 48 V DC to 12 V DC.

I'm taking this part of the project as a chance to acquire/consolidate some basic knowledge about circuit protection and input filtering (input source, impedance, EMI filters, etc.), so my questions are related to the design of the input circuitry of the DC/DC converter.

The way I look at it, I believe I can divide my doubts/questions into 2 separate areas, Circuit Protection and Input Filtering.

I'll present each of the 2 cases in the following way:

1. Background of the problem
2. My proposal to solve the problem
3. My doubts/questions

1) Circuit protection

1.1) Requirements

I have a 48 V DC input voltage and my objective is to convert it to 12 V DC, with the capability of supplying 1.6 A (20 W of power).

My self-imposed requirements are to protect the circuit against the following:

1. Overcurrent protection - 1.5 A max. (as recommended by the **broken link removed**; according to the datasheet, at full load, the converter draws a max. of 490 mA.)
2. Overvoltage protection - 100 V max. (1 sec max.)
3. Reverse voltage protection

I've decided to use the max. absolute ratings of the THN 20-4812WI DC/DC converter - available on its datasheet & **broken link removed** - as a reference to the absolute values above.

1.2) My proposal

Figure 1 shows the design I've came up with. For now I'll only talk about the components located from the far left to TVS1.

Figure 1 - 48 V DC to 12 V DC Power Supply

I'll try to explain my reasoning behind the circuit shown above.

a) I've decided to use a PPTC (a.k.a. resettable fuse) due to its 'resettable' nature, which avoids the necessity of replacing a fuse every time it blows due to an overcurrent event.

The model is the TE Connectivity / Raychem RXEF075, holding the following values, which seem to fit my requirements:

1. Tripping current: 1.5 A
2. Holding current: 750 mA

b) The diode bridge protects the circuit against reverse voltage connections on pins X9-1 and X9-2. The model is the AM152.

c) For overvoltage protection, I've chosen a TVS diode, specifically the model SMAT70A-13-F, with the following values, which also fit my requirements:

1. Reverse stand-off voltage: 70 V
2. Breakdown voltage range: 77.8 V to 89.5 V
3. Clamping voltage: 100 V

The idea is to have the TVS diode in parallel with the DC DC converter and make it a short circuit upon a voltage surge near the value of 100 V.

Hopefully, any excess current that may result from this will also be felt by the PPTC, which will temporarily block the flow of current to the circuit.

1.3) My doubts/questions

Does the circuit shown in 1.2 - including all the reasoning behind my choices and considering the requirements shown in 1.1 - makes sense?

I confess I've never used nor have I ever evaluated the real-word performance of PPTCs over time. Do you find these adequate to answer the requirements shown in 1.1? Are the parameters of the chosen PPTC model adequate?

As explained in 1.2, the diode bridge protects the circuit against reverse voltage situations. I'm aware that this solution (1) reduces the voltage supplied to the DC/DC converter by 1.4 V and (2) draws a certain amount of power which, at full DC/DC converter load, translates to (0.490 A) * (1.4 V) = 0.686 W (max. load current given the THN 20-4812WI datasheet). Do the parameters of the diode bridge seem correct? What other solutions are there to provide reverse voltage protection, without the drawbacks of the diode bridge?

2) Input filtering

This is where I'm truly clueless... Anyway, and based on what I've been reading, I've divided my input filtering doubts/question into 2 categories: input source impedance and EMI filtering.

2.1) Input source impedance

I must confess that I haven't completely understood this topic, so the lines below may seem like a messy mixture of disconnected concepts to those who actually understand it

According to the **broken link removed** and some other documents that I've "read" (**broken link removed**,2) it is kind of a common rule that input source impedance should be small compared to the converter's input impedance.

2 even shows how large input voltage transients can be produced by using ceramic capacitors.

Regarding the input source impedance problem, the THN 20-4812WI application note recommends the use of a C-L-C filter (L = 12 uH & C = 10 uF/100 V) to minimize the effect of input source impedance.

I've simulated this filter on CircuitLab (see Figure 2), with a voltage step of 48 V DC as the input signal, in order to analyze it for voltage transients.

Figure 2 - Input C-L-C Filter

Results for L = 12 uH and L = 120 uH are shown in the attachments ("_01" for 12 uH and "_02" for 120 uH). Voltage spikes of > 90 V occur for both, although the circuit with L = 12 uH seems to stabilize faster to 48 V DC.

2.2) EMI filtering

I don't have strict requirements regarding this other than getting the thing to 'work' . Nevertheless, after reading the DC/DC converter's application note, I became kind of clueless (probably due to my lack of knowledge about input filtering, EMI filtering, etc.).

So, what should I be worrying about regarding EMI filtering?

2.3) My doubts/questions

a) I think I didn't really understand the input source impedance section of the THN 20-4812WI application note. The application note suggests the introduction of a C-L-C filter. My intuition tells me that the introduction of additional inductance - such as the 12 uH inductor of the C-L-C filter - should be avoided. According to the results of my CircuitLab experiment on 2.1, adding additional inductance only makes things worse.

So sorry if this question is basic but what are the actual advantages brought about by the C-L-C filter?

b) I believe a TVS diode, similar to the one mentioned in 1.2, would be useful at the output of the C-L-C filter. Would it be useful to simply move the TVS diode from the position shown in 1.2 to the output of the C-L-C filter? Should I keep the TVS diode at 1.2 and add a new one at the output of the filter?

c) Based on everything described and discussed so far, what is the adequate kind of input filter for my application?

Best regards,
saturdays_law

 

[mtwieg]

As for the input filter, it's much more complicated than you think... switching regulators have a negative input impedance within their bandwidth of regulation, which means that it has the potential to go unstable if the cutoff frequency of the input filter is too low (and its Q is too high). Actually modelling this accurately isn't feasible for you, since the dynamics of the DC-DC converter's feedback simply aren't known.

Keep in mind that the input filter's purpose is not really to filter out large transients on the input, but rather for filtering the high frequency EMI generated by the DC-DC converter. If that's all you want to do, then I would stick with the recommendations of the application note. A well designed DC-DC converter should have input feedforward, allowing it to reject large, low frequency disturbances in the input voltage. And for very large transients that could damage it, it's up to you to clamp those with your own protection circuitry.

I would move that TVS to before the diode bridge, and make it bidirectional. Putting an additional TVS after the CLC filter might help with undamped transients, if you're worried about that.

As for the PTC fuse, I feel it's not really the right choice. The DC-DC you selected has foldback limiting, so it should be able to take care of any fault condition on its output side. So your fuse should really only be concerned with faults on the input circuitry, or with the DC-DC converter itself. And when such a fault occurs it will most likely cause a dead short to ground. If that's the case, your PTC better at least be rated to withstand the full input voltage continuously.

In general, for protecting against rare but extreme faults (like a failure of one of the input components) I would go with a normal non resettable fuse. They're just more reliable, and require much less consideration than PTC fuses.