Reverse power flow through step-down DC-DC

[Unixon]

Here's something that's bugging me.  Looking at the schematic it looks like the 12V input polarity is backwards.

Schematic symbol may be wrong, the connector itself is wired as center-positive.

[langwadt]

fcb, gnuarm, thanks, good idea. I know this solution, but it can't be nicely applied to a production board.
I will try to integrate this kind of protection into a new board revision.

To illustrate the problem a little further, I've attached relevant part of PCB layout.
As you can see all power traces are very thick and necessary pads are pretty far away.
Wire budges are prohibited for production boards, all mods like that are only for prototypes.

So, from all solutions suggested until now, there are two definitely applicable:
1) raise UVLO threshold - OK - a 0603 resistor could be nicely soldered between R5 and R6;
2) clamp VIN+ with a zener diode - OK - a SOD80C zener could be nicely soldered on top of C3,C4 or between them, this would allow for better heat transfer away from the zener;
Of course, an experienced person would immediately know this is a budge, but it still looks good enough.

These mods don't solve the problem entirely, but they can buy enough time to rethink one's power supply connections.

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

[Unixon]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

Mod wires and other functionally good but ugly looking fixes are OK for boards that go inside an enclosure and don't affect the appearance of a product,
but if a board is a product by itself, you just can't sell an obviously repaired item as brand new, probably can't even give it away.

[langwadt]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

Mod wires and other functionally good but ugly looking fixes are OK for boards that go inside an enclosure and don't affect the appearance of a product,
but if a board is a product by itself, you just can't sell an obviously repaired item as brand new, probably can't even give it away.

but you can sell an item that doesn't quite work?

[Unixon]

but you can sell an item that doesn't quite work?

This board shouldn't be used that way in the first place and it is written in big red letters all over the manual.
However, a user still can physically connect supplies in the unrecommended way and get an unexpected result.
What we have to do is to make sure nothing goes horribly wrong even in that case.
Intermittent high power consumption is bad but probably acceptable, auto-destructing converter is definitely not.
I still hope there's a better solution that could be implemented in this layout.

[Unixon]

I have a little bit alternative idea about UVLO circuit implementation.
What if UVLO threshold is set by a zener diode and a resistor rather than a purely resistive divider?
It's kind of the same but may push EN pin into ground better at low voltages.
Or maybe it is worth doing even crazier thing like differential zener-capacitor divider for better noise immunity / hysteresis...

I still don't know how it creeps from off-state into full on-state... if it is possible to stop this intermediate mode then the problem may be solved.

[gnuarm]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

You know what the rule is and you know why.  The rule is to not allow wire modified boards in the field.  Wire modifications are not up to snuff in many regards including reliability and EMI for starters. 

That's why I thought the leaded resistor was a good idea.  Technically it's not a wire, it's a component with a bit of insulation added to one lead.   /;)

[fcb]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

You know what the rule is and you know why.  The rule is to not allow wire modified boards in the field.  Wire modifications are not up to snuff in many regards including reliability and EMI for starters. 

That's why I thought the leaded resistor was a good idea.  Technically it's not a wire, it's a component with a bit of insulation added to one lead.   /;)

What tosh.  A wire is as much a component as a leaded resistor.  Well applied 'bodge' wire will have little effect on reliability - costs more in production and aesthetics for sure.

You'd have to take a view on whether it is likely to affect EMC, but you wouldn't normally have gone through final EMC without a fully tested product would you??

[gnuarm]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

You know what the rule is and you know why.  The rule is to not allow wire modified boards in the field.  Wire modifications are not up to snuff in many regards including reliability and EMI for starters. 

That's why I thought the leaded resistor was a good idea.  Technically it's not a wire, it's a component with a bit of insulation added to one lead.   /;)

What tosh.  A wire is as much a component as a leaded resistor.  Well applied 'bodge' wire will have little effect on reliability - costs more in production and aesthetics for sure.

You'd have to take a view on whether it is likely to affect EMC, but you wouldn't normally have gone through final EMC without a fully tested product would you??

Not sure how you can claim a wire is as reliable as any other component.  That's a big reason why we have printed circuits and not wired chassis.  A leaded resistor may not be any more reliable, but it might serve as a way past the "rules" was my point. 

As to EMC, bugs can be found at any time.  There is no such thing as a "fully tested product".  As we all know, testing can show the presence of errors, but not the absence. 

I would expect any seasoned engineer to understand that.  Tosh indeed!   

[fcb]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

You know what the rule is and you know why.  The rule is to not allow wire modified boards in the field.  Wire modifications are not up to snuff in many regards including reliability and EMI for starters. 

That's why I thought the leaded resistor was a good idea.  Technically it's not a wire, it's a component with a bit of insulation added to one lead.   /;)

What tosh.  A wire is as much a component as a leaded resistor.  Well applied 'bodge' wire will have little effect on reliability - costs more in production and aesthetics for sure.

You'd have to take a view on whether it is likely to affect EMC, but you wouldn't normally have gone through final EMC without a fully tested product would you??

Not sure how you can claim a wire is as reliable as any other component.  That's a big reason why we have printed circuits and not wired chassis.  A leaded resistor may not be any more reliable, but it might serve as a way past the "rules" was my point. 

As to EMC, bugs can be found at any time.  There is no such thing as a "fully tested product".  As we all know, testing can show the presence of errors, but not the absence. 

I would expect any seasoned engineer to understand that.  Tosh indeed!   

I think you might have gotten the wrong end of the stick - here's some slightly condescending tosh:

A bodge/mod wire should be thought of as a component. It will have resistance, inductance and may introduce undesirable effects (noise pick-up, radiated noise, etc), but a well applied wire is no more or less reliable than a leaded resistor.

We have printed circuit boards to allow greater complexity and speed up manufacturing. In certain products they still use hand-wired components (valve/tube amps, high-voltage systems, heck even the insides of fairly modern washing machines have relays, sensors that form part of the 'circuit' that are off-board).

Disagree with the "EMC bugs can be found anytime" argument.  A well tested product (stable casing/hardware/firmware/software) should test the same regardless of when it was made and where it was tested.  So if it's compliant in Lab A on Day B it should be compliant in Lab X on Day Y - there is no spooky action. As usual, the devil is in the detail, and the quality/thoroughness of the testing should ensure this. If the device fails at a later point it should be possible to do a root-cause-analysis and discover what as changed.

[langwadt]

https://workmanship.nasa.gov/lib/insp/2%20books/links/sections/302_jumper%20wires.html

[gnuarm]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

You know what the rule is and you know why.  The rule is to not allow wire modified boards in the field.  Wire modifications are not up to snuff in many regards including reliability and EMI for starters. 

That's why I thought the leaded resistor was a good idea.  Technically it's not a wire, it's a component with a bit of insulation added to one lead.   /;)

What tosh.  A wire is as much a component as a leaded resistor.  Well applied 'bodge' wire will have little effect on reliability - costs more in production and aesthetics for sure.

You'd have to take a view on whether it is likely to affect EMC, but you wouldn't normally have gone through final EMC without a fully tested product would you??

Not sure how you can claim a wire is as reliable as any other component.  That's a big reason why we have printed circuits and not wired chassis.  A leaded resistor may not be any more reliable, but it might serve as a way past the "rules" was my point. 

As to EMC, bugs can be found at any time.  There is no such thing as a "fully tested product".  As we all know, testing can show the presence of errors, but not the absence. 

I would expect any seasoned engineer to understand that.  Tosh indeed!   

I think you might have gotten the wrong end of the stick - here's some slightly condescending tosh:

A bodge/mod wire should be thought of as a component. It will have resistance, inductance and may introduce undesirable effects (noise pick-up, radiated noise, etc), but a well applied wire is no more or less reliable than a leaded resistor.

There's the problem.  You keep comparing a wire to a leaded resistor.  No one is saying a leaded resistor applied in the manner of a fixup wire is a reliable component.  Please read what I write.

We have printed circuit boards to allow greater complexity and speed up manufacturing. In certain products they still use hand-wired components (valve/tube amps, high-voltage systems, heck even the insides of fairly modern washing machines have relays, sensors that form part of the 'circuit' that are off-board).

PWBs allow much higher reliability in the end product.  When I was a kid Zenith advertised their TVs were highly desirable because they were "hand crafted" and "no printed circuits", but the reality was that "hand crafting" was less reliable and printed circuits dominated after just a few years of commercial production. 

In mil/space applications printed circuits very quickly became the rule, again because of the reliability. 

There is literally no comparison.  The lower production cost is just a bonus.

Disagree with the "EMC bugs can be found anytime" argument.

Not what I said.  i talked about bugs being discovered after EMC testing was complete.   

A well tested product (stable casing/hardware/firmware/software) should test the same regardless of when it was made and where it was tested.  So if it's compliant in Lab A on Day B it should be compliant in Lab X on Day Y - there is no spooky action. As usual, the devil is in the detail, and the quality/thoroughness of the testing should ensure this. If the device fails at a later point it should be possible to do a root-cause-analysis and discover what as changed.

EMC results can be changed by any mechanical or electrical changes.  I see EMC issues fixed this way, they can also be caused this way.  The point is a bug in the system can be found after EMC testing is complete and often is.  Adding a wire to fix a bug is much more likely to impact EMI than is a change to the PWB.  Traces on the PWB are in close proximity to the power/ground planes so as to minimize radiation and pickup.  A flying wire is much less so. 

I think we have beaten this issue to death.   

[fcb]

EMC results can be changed by any mechanical or electrical changes.  I see EMC issues fixed this way, they can also be caused this way.  The point is a bug in the system can be found after EMC testing is complete and often is.  Adding a wire to fix a bug is much more likely to impact EMI than is a change to the PWB.  Traces on the PWB are in close proximity to the power/ground planes so as to minimize radiation and pickup.  A flying wire is much less so. 

"The point is a bug in the system can be found after EMC testing is complete and often is" - then your EMC testing is now practically worthless if you think that in fixing a bug you may have caused an EMC issue. Or put another way if you change the design (in any way shape or form) AND you think it might have some bearing on EMC compliance THEN YOU SHOULD RETEST for EMC. Otherwise what is going out the door might not be compliant.

So regardless of what form a bodge/mod may take (wire/resistor/whatever), it's affect on a products EMC must be considered.

[penfold]

Whilst I'm not aware of any standard that explicitly prevents the selling or use of PCBs with a bodge wire as people seem to be alluding to, all standards that I have worked to, however, definitely do state that the modification should be fully documented... which includes the usual design change process, FMEA (meaning 'impact assessment' not MTBF calc before someone misinterprets me!) and with that, EMC and electrical safety should be considered, the mod wire should also be present in the latest schematic and PCB/maufacturing data pack which should include instructions on how to implement said modification...

[Zero999]

Whilst I'm not aware of any standard that explicitly prevents the selling or use of PCBs with a bodge wire as people seem to be alluding to, all standards that I have worked to, however, definitely do state that the modification should be fully documented... which includes the usual design change process, FMEA (meaning 'impact assessment' not MTBF calc before someone misinterprets me!) and with that, EMC and electrical safety should be considered, the mod wire should also be present in the latest schematic and PCB/maufacturing data pack which should include instructions on how to implement said modification...

Yes, I don't see the problem with selling a PCB with a few wire modifications on, but it depends on the expectations of the customer. It certainly wouldn't bother me, if I asked for custom made product and the first run of PCBs had a few bodge wires on. In fact I would expect it to be the case for the first run. I'd certainly prefer it over an inferiour/marginal design.

Has the customer explicitly stated they don't want PCBs with modifications? If not, then just do it, in the neatest, most reliable way possible. Otherwise, scrap the boards and get new ones made. If that means making a loss, then learn from it and remember to build it into the cost, the next time a customer says they won't accept modified PCBs.

[mikerj]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

You know what the rule is and you know why.  The rule is to not allow wire modified boards in the field.

Which rule would that be?  It meets IPC standards if done correctly.

[Unixon]

UVLO via EN pin is finally working as planned 
Last time it didn't work properly because I've targeted wrong threshold voltage (VIN UVLO instead of EN V_hi  ).
3k resistor from EN pin to ground (10k to VIN) nicely stops the converter from starting up at 6.8V typically occurring at the converter input when it's reverse powered,
15V zener diode at input caps works as 2nd line of defense just in case UVLO levels aren't precisely met for a particular board.

[gnuarm]

who came up with the silly rule that a mod that doesn't quite work is preferable to a wire budge that works? if that is really the rules the fix is scrap the boards

You know what the rule is and you know why.  The rule is to not allow wire modified boards in the field.

Which rule would that be?  It meets IPC standards if done correctly.

Customers overrule IPC.  You know we are talking about customers' or company's internal rules.  IPC simply sets standards for levels of work quality.  They don't tell anyone what to ship or not ship.  That's one reason why they have multiple levels to their standards.

[Unixon]

Try a zener diode and a transistor:
Zener from input node to base of transistor base, transistor emitter to GND and collector to feedback node of the regulator. This should draw the control loop into the correct direction to stop reverse energy flow.
Or connect to collector to the enable pin, to turn off the converter if the input voltage gets too high.

I would probably go with the following schematic for new revision: [see attached]
This will keep DC-DC off at all times unless it is powered from an input connector with correct minimum voltage.
Obviously, EN pin divider should be different than 10k/3k, I'll find optimal values later.
It takes a little bit more parts to build this, but all of parts are low current and come in tiny packages.

[gnuarm]

Excuse my ignorance.  I don't recall all the details of this design.  Why can't R2 be connected directly to Vin rather than going through the transistor?  If the threshold is too low the Zener can be applied, no?  I'm not getting a full understanding of the operation.

[Unixon]

Excuse my ignorance.  I don't recall all the details of this design.  Why can't R2 be connected directly to Vin rather than going through the transistor?  If the threshold is too low the Zener can be applied, no?  I'm not getting a full understanding of the operation.

VIN or VIN+?
It can't be connected to VIN+ because this node could be supplied with reverse current.
But you're right about a different thing - this transistor could be omitted and R2 connected directly to D3 anode.
Now I don't know why I wanted this BJT... I guess I don't need R1 and R2 either.

[fcb]

The EN pin will have an ESD structure on it, it'll look like a pair of diodes limiting the the voltage on the EN pin between VIN+0.6V and GND-0.6V.  So if you have apply -12V through a 10K to the EN pin, you'll only passing a smidge over 1mA.  This won't hurt the device.

Worst case and the buck converter VIN is still at 12V when you reverse the polarity (unlikely), your going to be perhaps 2.4mA.  Again this will likely not damage the part.

The EN pin operates between 2.4V (ON) and 0.8V (OFF) according to the datasheet, so putting a zener (and resistor) in series with it would be OK if you need a more accurate turn ON point, the EN pin impedance to GND is 120K to 400K.  So test your zener for leakage at high temperatures if you need a more accurate switching, or perhaps add a second 10-50K between EN and GND.

[Unixon]

Oh, I'm overcomplicating this again...
I don't need a zener diode either, let's go back to classic divider.
So, only 1 new part + 1 low side resistor that should have been there in the first place.

[langwadt]

Excuse my ignorance.  I don't recall all the details of this design.  Why can't R2 be connected directly to Vin rather than going through the transistor?  If the threshold is too low the Zener can be applied, no?  I'm not getting a full understanding of the operation.

VIN or VIN+?
It can't be connected to VIN+ because this node could be supplied with reverse current.
But you're right about a different thing - this transistor could be omitted and R2 connected directly to D3 anode.
Now I don't know why I wanted this BJT... I guess I don't need R1 and R2 either.

I see no need for a zener, a resistor will do to divide the input voltage down to the 1.6V EN treshhold. Might not need the diode either, the series resistor will limit any reverse current



 

[Unixon]

Might not need the diode either, the series resistor will limit any reverse current

I'm not sure it is a good thing to leave EN pin open to this challenge...