Input protection stage - suggetions/improvements

[ricko_uk]

Hi,
I designed the attached input protection stage. I want to make sure it is as robust as possible as if it was designed for safety critical application. So please feel free to highlight any issues with it and suggest any improvement.

Power comes from an external brick-type PSU providing 24V at 10A. 8A are used by another part of the circuit and 2A are used by the regulator shown.
The external brick PSU obviously has its protections but the reason for having this input protection circuit is that - as some of you correctly pointed out in another one of my posts - a user could plug in any other external PSU with higher voltage and/or opposite polarity. This would make sure the system is protected. Again, if even better protection can be achieved please let me know.

These are my consideration for the circuit and for choosing those specific chosen components:

Surge Protection
1)TVS D4 has typical failure mode as short under overcurrent or overvoltage conditions. The wording "typical" on the datasheet does not sound "guaranteed" or "certified" to perhaps there are better parts?
2) If the TVS fails short or it causes too high current due to overvoltage the fuse blows. Should the fuse be fast acting or slow-blow?

Reverse Polarity
3) I think that is pretty straight forward. Not sure if it can be improved?

Overvoltage Protection
4) I am wondering if the values are correct and also whether the Q4 should be a MOSFET or not. Mainly because it could operate in a linear/transition region which could cause the power mosfet Q5 to also operate in the transition region. That could result Q5 to dissipate a lot of power. Is that correct? How can I ensure that Q5 is always either fully on or fully off? Perhaps using a darlington instead of Q4?

Any feedback and suggestions for improvements to make it more robust?

Thank you as always!!

[moffy]

I think I would try to combine the Reverse Polarity & the Overvoltage into one pass element, i.e. having Q4 or some equivalent acting on Q2. It is possible that your overvoltage scheme could turn into some sort of unstable/stable linear regulator, depending on gains etc. What might be better is have it switch off for a period of time when an overvoltage event occurs, then try to switch on again later.

[bdunham7]

I think the fuse would be fast blow unless there is some inrush due to normal operation (not accidents or surges) and the PSU can handle it.

Your reverse polarity section alone will dissipate ~13W and drop ~1V @ 10A--you'd be better off with a large Schottky.

Your overvoltage section is going to start kicking in well before 24 volts as I see it.  At 24V, Q4 V_GS is 4 volts, which is well above its threshold so R_DS will be much less than 10K and.....

I think a smaller, lower current V_ref, a comparator and a voltage divider ought to work out better and should be able to drive Q3 directly if you choose wisely.  You also have to decide how much overvoltage you intend to protect against.

Edit:  You also might want to add an output capacitor to your buck converter!    And pay close attention to the impedance of the traces from the ground side of the capacitor to the diode and the coil. 

[ricko_uk]

Thank you bdunham7 and moffy

[jonpaul]

Impossible to critique without knowing the Environment and transients conducted and radiated

eg home lab, machine CAM shop, military service, vehicles?

Can respond further

Jon

[mc172]

If you're going to add a comparator you could get over and undervoltage protection, should that matter to you, for the price of a few resistors if you configure it as a window comparator.

There are surface mount FETs available with much lower on resistance such as the IRF9310.
A Schottky is still going to have to dissipate around 1.5-3W at 10A which is heatsink territory.

[ricko_uk]

Thank you jonpaul and mc172,

@jonpaul,
the application is lab equipment that is connected to gas monitory system. And because it has an external brick-type PSU, I want to ensure that no matter what the end user does the system can handle it safely and at worst shut down in a safe manner AND ideally inform the user that the PSU is not the original one and/or there is something wrong with it. That include it being handling safely someone plugging in some other random PSU with inverted polarity and/or much higher voltage or even a low voltage but not rectified PSU.

@mc172
As you pointed out even using a schottky would dissipate quite a bit of power. Requiring possibly a heatsink which I would definitely prefer not to use.
The IRF you suggest has really low Rdson which is nice but only handles up to 30V. But I have found several other ones that have higher voltage.

Yes the window comparator for undervoltage is a good suggestion. But would have to think how to wire it to be safe when a reverse polarity condition occurs. I thought of a bridge rectifier providing power to the comparator but perhaps there are better ways?

Thank you both!

[bdunham7]

But would have to think how to wire it to be safe when a reverse polarity condition occurs. I thought of a bridge rectifier providing power to the comparator but perhaps there are better ways?

I tried to just answer your questions directly rather than make broad design critiques, but one apparent goal of yours I sort of don't like is the idea that the device should tolerate wrong PSUs and such.  This can lead to a lot of complication and unintended consequences and make a product more expensive, less reliable and less efficient.  I think just using one large regular silicon rectifier (across the inputs) along with a TVS and a fast blow fuse is the appropriate protection for accidental reverse polarity, AC and gross overvoltage inputs.  If a user screws up that badly, your responsibility IMO is to prevent injury or fire where possible.  In this case, blowing the fuse puts the idiot user on notice that they have made a serious error.  Rescuing them from their own stupidity comes at a cost.

[jonpaul]

Serious lab equipment needs to be fail safe and not single point fail. Especially in a lab with gasses and potential safety issues.

Use of an external power brick (mass produced Chinese junk) is not recommended.

For dedicated lab equipment I  suggest a professional modular PSU designed for medical or lab use rather than putting babd aids on a cheap consumer power brick.

Jon

[ricko_uk]

Thank you both,

I agree with you both. But the PSU must be external one (there is no space inside the enclosure for a modular one - that would have been ideal) and is already a medical type by XP.

The issue is if someone plugs in a different one, damage the equipment and then claim under warranty. That is something the client asked specifically to do all I can on the design side to prevent from happening.

Given the requirements of having an external PSU, what strategies/architecture would you suggest I implement to make it fail safe? I think that by definition having a single external PSU would always introduce a SPOF point.
If you have any other suggestions/guidelines they are always welcomed!

Thank you

[bdunham7]

The issue is if someone plugs in a different one, damage the equipment and then claim under warranty. That is something the client asked specifically to do all I can on the design side to prevent from happening.

Well, if that is the design criteria, you're stuck with it.  I would probably just go medieval and use a bit of circuitry to confirm 'good' power and then have it operate a relay.  Cheap(ish), easy, reliable, robust and efficient.

[mc172]

The reverse polarity protection and OV FETs aren't able to be combined because they have to be wired backwards relative to each other. You want the body diode to conduct for the former and not for the latter. In the reverse polarity condition the opposite is true, the body diode in the OV protection FET (which would be your only FET if you combined them) would conduct and not do any protecting. Hence your comparator goes after the rev. polarity protection and isn't ever subjected to any reverse polarity. The issue is that it will potentially be exposed to overvoltage so needs its own OV protection to suit.

And yes you're right, the 9310 is only 30V, it was just one I rattled off from the top of my head, I should have checked it, sorry.

[jonpaul]

"The issue is if someone plugs in a different one, damage the equipment"

1/ Warramty and instructions must sepcif the PSU and connector. VOID if wrong.

2/ Just a fuse is OK, a simple rev pol rectifier for RP if desired.

3. Choose PSU with special connector to avoid the millions of PSU woth 3.5 mm etc concentric, we see many NAS with DIN PSU connectors.

Jon

[jonpaul]

MR2525L, MR2535L
 Motorola auto alternators load  dump diode

jon

[ricko_uk]

Thank you both again!

[t1d]

Include a proprietary power plug shape that simply will not accept any other non-OEM power supply?

[PartialDischarge]

Some suggestions,
Use 2 10V zeners in series instead of the 20 V part, since you are already using it.
A window comparator can be made with low powerish tl431
I’m missing Hf filtering at the input, LC should be a must