Design a short circuit protection for +15/-15 and +10/-10 DC power rails

[R1S8k]

Hi,

I'm working on repairing power electronics training boards. This training board has 3 power sources, 1-ph 13V AC, 3-ph 11V AC and +15/-15 and +10/-10 DC.

This is the main training board scheme -> on the left section of the board, the power supply blocks:
1. F1 and F2 are the AC fuses, so the AC blocks are protected.
2. The DC block as it's clear, not protected and the DC lines and direct connected to the parts of the board that are working with DC voltage.



The AC sections are protected with fuses and they are working OK, any mistake in the wiring, the fuse breaks and I just have to replace it and problem solved.

But the real problem is the DC section isn't protected against short circuit mistakes and that causes different problems; like, broken main voltage regulators, other ICs powered with the DC voltages, burnt resistors, transistors, etc.

Here are some pictures of burnt components because of wiring mistakes:










This is the DC schematic circuit:



My question now, is that what can I do to protect these DC lines ? I thought of using ploy fuses, where I can cut the DC voltage line and put a poly fuse in series with the +ve voltage line, would that work ?

 

[Delta Prime]

Hello there welcome to Maker Pro.
The answer is yes!
Now I'm going to run over to microcontrollers and answer your other question see you there. I'll race you!

 

[kellys_eye]

Diodes in series with the outputs will prevent 'backfeeding' although you might have to compensate for their volt-drop by adjusting the regulator outputs (i.e adjusting the regulator feedback resistors or putting another diode in the ground lead of the fixed DC regulators).

If the devices don't have built-in s/c protection, change them for devices that do else use extra devices configured as current regulators/limiters. The LM317/LM337 has such internal protection and can be set to whatever output voltage you require. Not sure why you chose a low-drop regulator for this application.

 

[crutschow]

What value of current-limit do you want?
A two-transistor limiter (example below) connected at the output of the bridge rectifier filter capacitors would work for that.
Note that M1 would need to be on a heat-sink for other than momentary shorts.

 

[R1S8k]

Thanks for the advices.

Diodes in series with the outputs will prevent 'backfeeding'

How 'backfeeding' is different than a short circuit ?

If the devices don't have built-in s/c protection, change them for devices that do else use extra devices configured as current regulators/limiters. The LM317/LM337 has such internal protection and can be set to whatever output voltage you require.

I can't change them, these boards aren't mine, it's for the college I work in. They are from a Dutch company named hera.

My work is to replace the broken components, and do a board check up. Changing components in a fixed PCB system would be a difficult task for me, even I do not have the permission to do that.

Not sure why you chose a low-drop regulator for this application.

This is the company design, I really don't know what the designer intention for components selection.

Do you mean the 15V LDO ?

 

[R1S8k]

connected at the output of the bridge rectifier filter capacitors

Do you mean between the bridge rectifier output and the voltage regulators ? If that's what you mean then I guess that would be a difficult one to do.

This the PCB:




On the red arrow area, it's difficult to insert your suggested circuit, but not impossible, I might check the board later.

On the yellow area, there are direct clear lines of the regulators output, this one I thought of to insert the protection circuit here.

But I don't know which path to take ? Should I only use poly fuses, or improve the circuit more to insert relays to cut off the power rails with a buzzer and some LEDs.

 

[kellys_eye]

The schematic you provide doesn't match the burned components you show. We really can't comment on why parts burn out unless you give more details.

I suspect (but can't actually prove) that some enterprising student connects an AC voltage to the DC output - is this possible(?) - I'd say yes. The board complexity also imples that some protection is part of its design so what's really going on?

And, what, exactly, IS that burned-out section? A component? A connector?

 

[R1S8k]

The schematic you provide doesn't match the burned components you show.

Yes, I agree with you, because the schematic I've put is for the power supply section.

The complete schematic is in the attachment.

I suspect (but can't actually prove) that some enterprising student connects an AC voltage to the DC output - is this possible(?)

I would say yes, but can't guess more because the experiments are divided each training session on a specific circuit; which would be an AC or DC circuit, but there won't be a circuit that include an AC and DC circuits together.

But the possibility I would assume is that a student might connect a point that has a potential voltage to the ground.

The board complexity also imples that some protection is part of its design so what's really going on?

There's a protection for the AC power supply only but there isn't one for the DC power supply and even the DC functional circuits.

And, what, exactly, IS that burned-out section? A component? A connector?

I think you mean the transistor.

 

[R1S8k]

I contacted the company which served us the training boards.

And the manager replied to me with this:

Dear Sir,
I am sorry for the trouble.
For R154, R155, R165 and R166 we are using meanwhile polyswitch fuses 1,35, RM5.08 Type: littlefuse for protecting the circuit.

You just use them instead of the wire resistors. They look like this example.

He mentioned that I should use them instead of the wire resistors, I also think using them in series with the resistors would be also ok.

I hope that solve the problem.

But also I'm thinking of adding an alert circuit with a red led and a buzzer for the students and the trainer to know that there's a mistake in the circuit.

 

[danadak]

In your schematic all those regulators have short circuit protection built in, but
sevetral cases exceeding an amp. But you can have a case where reg load is
excessive, and cause excess Pdiss. I did not check each datasheet to see if they
have thermal shutdown as well.

If you take the advice to use LDOs be careful to read data sheet concerning
output caps, and minimum ESR needed in the cap to avoid instability. Otherwise
they may oscillate and be useful in cooking eggs or making toast.

https://www.renesas.com/us/en/document/mah/usage-three-terminal-regulators

https://www.smcelectronics.com/DOWNLOADS/1980-VOLTREG.PDF

Regards, Dana.

 

[R1S8k]

but
sevetral cases exceeding an amp

Yep, that's what I'm talking about, there are different cases for getting broken components.

have a case where reg load is
excessive, and cause excess Pdiss

Absolutely, the regulator can supply a lot of current obviously and when I get a broken regulator, it's not fried, it only got broken maybe because it didn't shutdown.

And the regulator which get broken are the LM7810 and LM7910. The LM1085 didn't get broken.

So my guess is that there could be a short circuit on the +15V and it supplied a lot of current which is also going through the LM7810, and the current is bigger than the LM7810 can take, and causes the LM7810 to be broken from excessive forward current, and that of course can fry the components on the same path.

If you take the advice to use LDOs be careful to read data sheet concerning
output caps, and minimum ESR needed in the cap to avoid instability. Otherwise
they may oscillate and be useful in cooking eggs or making toast.

I can't use LDOs, I don't have the privilege to do that. I might take the company's employee who advised me to use polyswitches.

But I also want to add more components for alert purposes; like, a red LED and a buzzer.

I have to design a simple circuit where if the polyswitch is triggered, then a signal triggers the alert circuit.

My dream to solve this laboratory's problem and have a piece of mind to do other stuff

 

[kellys_eye]

and it supplied a lot of current which is also going through the LM7810, and the current is bigger than the LM7810 can take, and causes the LM7810 to be broken from excessive forward current,

It doesn't work like that. Excessive current can only be DRAWN through a device - you can't 'force it in' to a device. Only by shorting the 7810 output can it be made to 'overload' - but it has built-in thermal overload and internal current limiting so that's out of the question.

I note R155 and R166 burned out in your image above but the schematic show them in positions of relative 'safety' where, I would guess, only the application of an external source supply that destroys the op-amps (shorts them to ground) would account for those resistors burning out. It's more indicative of either 'idiot' users or deliberate attempts to destroy the equipment. Maybe the class leadership needs to be advised to take more notice of what the students are doing?

Poly fuses would help but I suspect the op-amps would be destroyed before the fuses worked - the resistors would be saved but the unit would still be defective until the op-amp(s) were replaced.

 

[danadak]

Also potential causes are switched L loads, or large output Cload when supply is turned
off and Cload charge gets dumped back thru output of reg, and all the internal protection
circuits no longer in effect.

Here is design recommendations for the venerable LM317 adjustable, but principles apply
to all regs :




Regards, Dana.

 

[R1S8k]

you can't 'force it in' to a device. Only by shorting the 7810 output can it be made to 'overload'

OK, I got that, because the LM1085 has two paths for its output, one to the application circuit and one to the LM7810.

So, if the LM1085 is shorted out, them that would be through the path it's shorted from.

And Luckily, I didn't get any broken LM1085, broken regulators are either the LM7810 and LM7910.

but it has built-in thermal overload and internal current limiting so that's out of the question.

As I got several LM7810 and LM7910 broken, so I guess the thermal protection didn't work.

I note R155 and R166 burned out in your image above but the schematic show them in positions of relative 'safety' where

Here where it gets complicated, we have 20 boards in the lab, as I opened several ones for repair purposes, I noticed that not each PCB is the exact one as the other ones, there are some differences, but I think not in the circuit lines, but in the resistors and capacitors category. The rest is the same.

Another important issue which is that the schematic manual I got from the company isn't the exact one on the board PCB I work on.

I was able to point many circuits from the schematic to the PCB circuits' lines, but some circuits aren't the same especially the ones with the op-amps.

I don't care eventually if I can fix the problems and close those boards for good, but before that I want to install some protection for the DC circuits so I don't get repair requests in the coming semesters.