How to find out faulty part

[Jane11]

I need to repair a device that uses a dual synchronous
step-down switching regulator controller LTC3850.
Datasheet is here
http://cds.linear.com/docs/en/datasheet/38501fc.pdf
Let's suppose it uses the typical application design according to the attached picture.
**broken link removed**
How would you start to find out the faulty part if there is no
Vout1, nor Vout2? The input voltage Vin is correct.
Thank you for you ideas.

 

[Les Jones]

As both regulators are not working I would look for thing common to both regulators. The only thing I can see that is common is the internal 5 volt regulator (From the block diagram on the data sheet.) so I would check that there is 5 volts oon the INTVcc pin. If it is not present (Or the wrong voltage) I would suspect the 4.7 uf capacitor between that pin and ground or the two diodes connected to that pin.

Les.

 

[spec]

Hello Jane,

Could I establish one basic point about your problem: I know you say that you wan't to 'repair' your power supply so that word tells me, but could I ask all the same: has this power supply been working and failed or is it a new design that has never worked and you would like to know why. I only ask because diagnosing the fault is different to a degree for the two situations.

 

[gary350]

I use to repair circuits boards 40 years ago. I would look up the LTC3850 to see what the input voltages are and the output voltages are. First I would use a meter to test each pin of the LTC3850 if the voltage is wrong work your way up stream with the meter. Something is bad that gives the IC the wrong voltage or the IC is dead already.

I noticed once a certain circuit board always fried the same part on all the boards. So I learned not to test the boards just replace the part that is always bad. I started replacing the fried 1/2 watt resistor with a 1 watt resistors then there was no more boards to repair. Sometimes I wonder if factories do that on purpose so the customer has to buy more replacement boards.

 

[spec]

I use to repair circuits boards 40 years ago. I would look up the LTC3850 to see what the input voltages are and the output voltages are. First I would use a meter to test each pin of the LTC3850 if the voltage is wrong work your way up stream with the meter. Something is bad that gives the IC the wrong voltage or the IC is dead already.

Hi gary,

That's the way to fault-find: be methodical always check the salient circuit voltages first, especially the power rails, and then on the actual pins of the components concerned.

I noticed once a certain circuit board always fried the same part on all the boards. So I learned not to test the boards just replace the part that is always bad. I started replacing the fried 1/2 watt resistor with a 1 watt resistors then there was no more boards to repair.

I have done the same thing, especially on the 1950 thru 1980 TVs: mains droppers and cathode resistors.

Sometimes I wonder if factories do that on purpose so the customer has to buy more replacement boards.

It does seem that way, but I think, in the main, it is unlikely. More likely to be an error, incompetence, circumstance, or maybe management interference. I worked in product design and all four are more common than you would think. The manufacturer normally hopes to market the best product he can so it and the company gets a good reputation and you buy again and new customers buy. There are products that have a short life by design but I wouldn't say significant electronic products.

An example of what you say is laptop power connectors. That type of connector is an abomination and always has been. Why is it used: historical I expect. It can't be on cost grounds, because the cost of a decent connector would be trivial compared to the cost of the microprocessor, memory, multi-layer PCB, etc.

 

[granddad]

Garry obviously speaking from experience,

So I learned not to test the boards just replace the part that is always bad

when I worked for a living company had a psu in an cheque encoder machine , a non-polorized cap always failed about 1 year in, board cost 100 £s , I got a bunch of caps better voltage ( wired type ) and gave every engineer some, saved loads.

 

[Jane11]

Thank you all for help.
It was working device that uses LTC3850 but then something wrong happened and it does not work any longer.
Yes, I also checked voltages, replaced that LTC3850 + few FETs yet the same problem. One FET is rather hot, so I expect it supplies high current( I replaced that FET, yet still hot). The output of the LTC3850 is an input to other voltage convertors and finally to CPU. So I suspect it could be CPU problem :-(

 

[spec]

Hi Jane,

Going by the datasheet it is possible that the chip has gone into the fault mode when it turns on MOSFET of a pair on and the other off. That would cause one to get hot and the other to be cool as you say.

This may not be easy but I would suggest that you isiolate the two supplies from their respective loads and place resistors across the the two outputs that would draw 50 mA or so. Then turn the supply on and see what happens.

Have you checked the inductors? They can go open or short circuit. Also check to see if the iductors are cracked or the core has become loose.

It's a pretty tricky circuit to fault find remotely- bad enough hands on!

Good luck and let us know how you get on.