Hi Everybody!

I am currently interning at a company before I return to university for my masters and I am trying to help solve a problem of resistors (sometimes) blowing out on a board when the boards are connected to a power source (physically connected).

Here is what I have figured out so far:

The total power waveform for parallel resistors is shown below (ie. There are two parallel resistors so each experiences half the peak power).


The circuit is shown below as well as some assumption made to simply it are shown below (note, the actual resistors are now 680 Ohm not 220 Ohm):

What I think is happening:

So what seems to be happening is that the 110Vdc to 24Vdc converters input pin is taking time to charge up from 0V to 110V. This means that during startup the equivalent load behaves like a capacitor with negligible inductance and resistance. (see circuit and simplified circuit above). This capacitive load during startup also means there is a voltage pulse of 110V across the resistors which dies down asymptotically. This voltage pulse ofcourse cause a power pulse in the resistors which could potentially damage the resistors (shown first figure).

My questions are:

1) Is the circuit design flawed, leading to the blowout? (The simple low power 1206 resistors seems like a inelegant solution at best)
2) If yes to 1) how can the design be improved?
3) Since the PCB have already been printed, is it possible to replace 1206 package chip R25, R26 and the 2512 package fuze (F1) with other components that may solve the problem?

Some additional info:

110V input can vary between 70V and 150V.
Output power of load is about 1.6W (meaning about 15mA input on the 110V line)
The resistors are 1206 package 0.5W 680 Ohm resistors. So it makes sense that they cannot handle the 50W peak reliably.

 

More info: The problem only seems to occur during startup, ie. when the 100Vdc power is connected to the circuit.

 

Replace them with NTC resistors
Calculate rush-in current in (R in series with the 100uf) + chip consumption+ load
Then take ntc open resistance to calculate the rush-in power. use tnc able to handle 2 x that power.

 

More info: The problem only seems to occur during startup, ie. when the 100Vdc power is connected to the circuit.

Why is the Remote On/Off pin connected to a capacitor? Shouldn't that be pulled High or Low? I would imagine that leaving it floating is the wrong option.

 

Replace them with NTC resistors
Calculate rush-in current in (R in series with the 100uf) + chip consumption+ load
Then take ntc open resistance to calculate the rush-in power. use tnc able to handle 2 x that power.

That sounds that a great I idea, expecially if I can find 1206 package thermistors. What is your reasoning for the 100uF. Is it because that is the largest input capacitor? I assume the open resistance is when the thermistor is at cold and at the minimum resistance?

Why is the Remote On/Off pin connected to a capacitor? Shouldn't that be pulled High or Low? I would imagine that leaving it floating is the wrong option.

Thank you for pointing that out, I will have a look at the datasheet and let you know.

 

What you need to look at is the single pulse rating of the undisclosed resistor you are using. Resistors are rated for single pulse, and continuous pulses which are an overload in the steady state, but the net energy of a short duration is acceptable.

The pulse in your figure is HUGE. A Vishay resistor of same size and value is only rated for about 10-5 second duration pulse. Go check what you are actually using.

One trick to put in your pocket, since the footprint is defined by the existing boards, is to stack up multiple resistors one over the other until you get an acceptable debating on the inrush current.

 

To cope with the start up pulse of current you could choose to use a "Pulse Withstanding Resistor"...check this out...
http://www.digikey.co.uk/Web Export..._738/PDF/Stackpole_PitchPack.pdf?redirected=1

 

100uf will be loaded from null volt to V input max via a resistor this a non linear process.
power I^2 x resistance
but read http://hyperphysics.phy-astr.gsu.edu/hbase/electric/capchg.html
look at vishay to find the one you need

 

I am smelling homework. Above looks like a homework assignment.

 

What is the value of Z1? If that's the symbol for an NTC thermistor, it shouldn't be across the power source.

 

Looks like a varistor.
In my book wasting 50W on two 1206 resistors is a bad idea, even if it is just momentary. At least try changing them for anti-surge types.

 

The only difference between the symbols for a varistor and an NTC thermistor is the -t notation that might have been omitted by a careless schematic creator or a conscious decision was made to use a generic symbol.

It seems like whoever designed it was trying to be very thorough. A varistor in parallel with a TVS, with series fuse and current limiting resistors. Or they weren't very competent and just threw everything, including the kitchen sink, at the problem.

Surface mount NTC thermistors are more likely to be for temperature sensing and not surge limiting.

Are the parallel resistors before the fuse intended to protect the fuse?

 

What you need to look at is the single pulse rating of the undisclosed resistor you are using. Resistors are rated for single pulse, and continuous pulses which are an overload in the steady state, but the net energy of a short duration is acceptable.

The pulse in your figure is HUGE. A Vishay resistor of same size and value is only rated for about 10-5 second duration pulse. Go check what you are actually using.

One trick to put in your pocket, since the footprint is defined by the existing boards, is to stack up multiple resistors one over the other until you get an acceptable debating on the inrush current.

Thanks for your feedback, I will find out the exact resistor we are using on Monday. I only know they are 0.5W and that they are highly unlikely to be rated for such a large power pulse.

To cope with the start up pulse of current you could choose to use a "Pulse Withstanding Resistor"...check this out...
http://www.digikey.co.uk/Web Export/Supplier Content/Stackpole_738/PDF/Stackpole_PitchPack.pdf?redirected=1

I managed to find these pulse resistant devices after some googling and it does seem like exactly what we need. I have been looking at the CRCW-HP e3 (Pulse Proof, High Power Thick Film Chip Resistors) and the D/CRCW-IF e3 (Pulse Proof Thick Film Chip Resistors). Only the 2502 (6mm) packages seem to fit the requirement but wont fit the 1206 (3mm) solder points.

What I do not understand is why the designer would use such severely under-designed components or why he even use the series resistance solution in the first place. They are the largest contributer to efficiency loss in the system and for what? I am sure the purpose is to limit the maximum instantaneous or power pulse during startup (since Pmax = Vmax^2/R = 150^2/330 = 68W). Is this a standard solution to limit the inrush current to a DC to DC?

The only difference between the symbols for a varistor and an NTC thermistor is the -t notation that might have been omitted by a careless schematic creator or a conscious decision was made to use a generic symbol.

It seems like whoever designed it was trying to be very thorough. A varistor in parallel with a TVS, with series fuse and current limiting resistors. Or they weren't very competent and just threw everything, including the kitchen sink, at the problem.

Surface mount NTC thermistors are more likely to be for temperature sensing and not surge limiting.

Are the parallel resistors before the fuse intended to protect the fuse?

I read up on NTC thermistors and a few hours ago, and it does not seem like the proper application in the schematic. I can only guess that it must be a varister or the designer made a large blunder.

I do not think the designer intended to protect the fuse, though it is rather strange the resistors are blowing and not the fuse. I think it is a slow-blow fuse. I believe the purpose of the resistance is to limit the inrush current / maximum instantaneous power / Pmax of the power pulse.

 

I do not think the designer intended to protect the fuse, though it is rather strange the resistors are blowing and not the fuse. I think it is a slow-blow fuse. I believe the purpose of the resistance is to limit the inrush current / maximum instantaneous power / Pmax of the power pulse.

Using resistors is a poor design choice. An inrush surge limiter would dissipate less power. Using a fixed resistance would be okay if that resistance was switched out after the surge event; and the resistors were sized to survive.

Seems like the designer was trying to be thorough, but missed the obvious efficiency issue and didn't implement surge limiting correctly.

 

Replace them with NTC resistors
Calculate rush-in current in (R in series with the 100uf) + chip consumption+ load
Then take ntc open resistance to calculate the rush-in power. use tnc able to handle 2 x that power.

The circuit posted by the TS shows Z1 MOV protection - its seems a little puzzling that anyone could be aware of the need for that, yet oblivious to the need for NTC surge suppression.

 

The circuit posted by the TS shows Z1 MOV protection - its seems a little puzzling that anyone could be aware of the need for that, yet oblivious to the need for NTC surge suppression.

The most puzzling part is that the only conceivable reason for the resistor would be current/power surge suppression, yet the resistors selected were low power resistors with virtually no pulse capability.

As Dennis suggested earlier, it does not seems surface mount NTCs for surge suppression exist. So it seems we will need to stack 1206 pulse resistant resistors on top of each other until I meet the pulse specification. And if that doesn't work, the PCBs must be reprinted to allow for surge suppression NTCs.

 

I am smelling homework. Above looks like a homework assignment.

Everything an intern does looks like a homework assignment. What is your point? The information and the analysis was more than sufficient.

 

The most puzzling part is that the only conceivable reason for the resistor would be current/power surge suppression, yet the resistors selected were low power resistors with virtually no pulse capability.

As Dennis suggested earlier, it does not seems surface mount NTCs for surge suppression exist. So it seems we will need to stack 1206 pulse resistant resistors on top of each other until I meet the pulse specification. And if that doesn't work, the PCBs must be reprinted to allow for surge suppression NTCs.

The NTC can be put anywhere in the AC path. Not sure if inline NTC parts exist, but a through hole sub panel would certainly solve the problem.

 

Everything an intern does looks like a homework assignment. What is your point? The information and the analysis was more than sufficient.

Thanks! Yes I I assure you all this is not homework. The company normally uses a E&E consultant to design their electronics so nobody at the company knows exactly what is going on in the circuit or how to best solve this problem. I have only been here 3 weeks and will be be here for 3 more, I am normally tasked with building simple test equipment, but I decided to take some initiative and try solve this problem over the weekend, since it seems to be causing the engineers alot of stress.

 

The NTC can be put anywhere in the AC path. Not sure if inline NTC parts exist, but a through hole sub panel would certainly solve the problem.

What do you mean by AC, the input is 110Vdc and the output is 24Vdc. Also what is a through hole sub panel. My apologizes, I did try googling that first.