Suitable resistor replacements?

[daimoku]

Hey folks, been reading on the forums for a couple months now. Thanks for all the helpful posts. Now I have a question...

I recapped an Insignia 667-32HA37-20 power supply. The power supply has a heat issue and I found 4 resistors out of spec.

How do I determine the correct type of replacement resistor to order? Should I order wirewound, metal film, ceramic, etc? Also, I don't know the current drop so how can I figure out the correct wattage? Do I just measure the resistor width?

The resistor locations and proper values are:
R545 0.27 Ohms +/- 5%
R508 91.4 Ohms +/- 0.25%
R514 42 Ohms +/- 5%
R501 22.5 Ohms +/- 5%

Thanks for your help.

[Jack Crow]

D,
The big question is, will the P/S work as it is?
Are the voltage (s) correct?
Current limits right?
If it is, then my advise is to leave it.

If not then some things need to be looked at.

The problem with most 'consumer' DVM's is zero ohms and test leads.

Give this a think.
Take you DVM and short the leads togther. 00.0 ohms if your luckey.
Now put a fully isolated resistor in the path.
You get a reading.
Now the tricky part.
Did you know there are errors in that reading?

Bi metalic junctions make small voltage errors.
Just like tiny thermocouples.
Put enough in series and you have an error reading.
From the probe tip it is most likely made from brass with a chrome plate.
The resistor wire is most likely copper with a solder (lead tin) flash.
No telling what the resistor is made out of. Or how the wire attaches.

Junction one is the red probe to the resistor wire.
Junction two is the black probe to the resistor wire.

Those super small junctions exhibit tiny amounts of voltage either in the direction or bucking the direction of current from the DVM's ohms section.

Now that is assuming the junction points are ideal. Most are not.
Oxydation takes it's toll on copper, tin, and Pb. You got it, more error.

So your .27 ohm unit may be just fine.

Learned this the hard way when calibrating a relay test set years ago.
This machine was used to test relays used in military test gear.
Run a test signal through enough relays and all the bi metalic junctions add up fast.
The tester ran a current through the closed junction and measured the voltage drop across. (A four wire test) I forget what the pass/fail test spec was but it was small.

Then there is the rest.

Just what is 5% of .27 ohm? If I did the math right it's .0135 ohm, bet your dvm can't read that.
I may not be able to do it with my exotic HP 3456A meter in four wire mode either.

Some resistors need to be tested the hard way.
Power supply, series resistance, voltage drop measurment, and calculations.
Pain in the butt.

That's why I say test the unit first and see if it works.
Much quicker and more sane.

As for the heat issue, just what is it doing.
Drifting in voltage as the temp build?
How wide is your spec?
The best bet here is to look for zener diodes and other voltage regulating parts.
Set up to measure your output on the bench and drip tiny amounts of freeze spray on the zeners and see what happens. Had this problem a few times in Kuwait with generator parts that drifted out of spec over time.

Hope this helps.
Mike aka Jack Crow
Herndon VA

[daimoku]

Thanks for the advice Mike. I'm still learning so I need all the help I can get right now. You're right my meter leads introduce ~0.1 Ohm error in my readings.

The issue:
The tv powers on and works fine for an hour or two then it shuts off. It has to cool down then you can repeat the process. Two of the resistors measure out of spec by 5 orders of magnitude (MegaOhms instead of Ohms)...

R508 - white/brown/yellow/gold/blue - 91.4 Ohms 0.25%
R501 - red/red/green/gold/blue - 22.5 Ohms 0.25%

The other two (0.27 and 42Ohm) might be okay, but at least the 42 Ohm resistor measured 45.5 Ohms - slightly above the 5% upper limit.

What's the best approach to determine a suitable replacement resistor (both type and power rating)?

I will hit it with freeze spray and measure the output voltage as it heats up. My hunch is that the voltage varies inversely with temperature - but I'll take some readings to verify.

[steve2]

Are R508 and R501 reading that low out of circuit? Just looked and those are correct values. http://samengstrom.com/15924820/en/r...or_Color_Codes This works great. Just start clicking the color bar.

[daimoku]

Yes, R508 and R501 are giving MegaOhm readings instead of Ohm with one leg of each resistor removed from the circuit.

Thanks for the calculator - I had not seen that one before.

[Jack Crow]

D,
Ok, open resistors are as we say 'a clue'.
Change them out.
Your not getting any value keeping them in there.

Just a thought here.
Parts in our field are fairly cheap.
It might be to your advantage to simply order what you need from a supply dealer and get a few of each.

As for the Zeners, most times you can read the numbers on the side with the help of a low power microscope.

Type in that number to Google and get a data sheet and or specs for the part.
For example, just noseing around...

http://www.fairchildsemi.com/pf/1N/1N5226B.html

It's all out there, fairly cheap.

Labor is expensive, if your fixing stuff for a boss or things for the home.

Much luck.
Mike aka Jack Crow
Herndon VA

[Toasty]

I think there's something wrong with your reading of the color codes.

Going by the old(er) codes:

white-brown-yellow-gold is 910kΩ ±5%
red-red-green-gold is 2.2MΩ ±5%

I know that they present as 5 band, but 4 band with those readings makes sense.

If those 2 -supposedly- low ohm resistors had gone that high, I don't think the supply would be working at all.

Toast

[daimoku]

Thanks Toast - you're exactly right.

I didn't suspect the power supply would be working if that was the case either that's part of the reason I was confused. The resistors are within spec using those values.

Guess I'm back at square one haha

[severach]

Reading 0.27 ohms is easy with an ESR meter.

[momaka]

Originally Posted by Toasty

white-brown-yellow-gold is 910kΩ ±5%
red-red-green-gold is 2.2MΩ ±5%

I know that they present as 5 band, but 4 band with those readings makes sense.

+1
Moreover, the value for R501 makes perfect sense, since R501 is in parallel with the two safety X caps (yellow square blocks). Many power supplies use a resistor in the MegaOhm range for that spot.

As for the 5-band color codes on the resistors of your power supply - it probably indicates reliability (or something like that).

[Agent24]

5-Band is usually used for 1% resistors as far as I know....

[daimoku]

severach - Yeah, I used my Blue ESR meter to check that one. Good thought.

Agent24 - The calculator I used for those resistors suggested the blue band means 0.25% tolerance, but I'm no expert. That could be wrong or right for all I know.


Alright, I spent 30-40 minutes last night tracking down the real source of the overheating problem. The overheating problem is with the GPU/CPU or whatever chip is located under the heatsink on the mainboard (picture attached).

I don't have an oscilloscope, but the crystal oscillator is making a scratchy noise. If I hit the oscillator with some freeze spray, it quiets down for a few seconds then starts right back up.

Since the heatsink is epoxied on, I don't have a good solution for repairing this problem.

Any advice other than sit there with a can of freeze spray? haha If there's not a good solution, I may just part the TV out.

[Agent24]

Replace the crystal?

Originally Posted by daimoku

Agent24 - The calculator I used for those resistors suggested the blue band means 0.25% tolerance, but I'm no expert. That could be wrong or right for all I know.

That's probably right, makes sense, I didn't read the color code you posted properly. I have not had much to do with anything other than 1% and 5% resistors.

[daimoku]

Would a crystal cause the CPU/GPU to overheat?

[Agent24]

If it's making a weird noise it may be broken or running at the wrong speed.

If it was running too high it may be overclocking the CPU and causing overheating?

I don't really know, but your post seemed to suggest the crystal was faulty.

[daimoku]

Good point. That makes sense it would overclock the processor and cause the overheating issues. The scratchy noise does seem to suggest it might be faulty.

[Agent24]

I may be and quite likely am horribly wrong, so treat what I said lightly.

I personally would try replace the crystal if I suspected it, they're fairly cheap at least.

[daimoku]

Found some bad caps on the main board too. Gonna order some replacement caps and an oscillator.

The oscillator says H25.000M. Does that mean 25Mhz and are there any other specs I need to keep in mind?

Also, is orientation important for crystals? I forgot to denote the orientation before I removed it. I know I probably should have replaced the caps first then tested the board, but I wasn't thinking clearly (cold medicine lol)

[Agent24]

Sounds like 25Mhz to me., especially with the M on the end.

Orientation is not important as far as I know.

You may have to worry about load capacitance, if you get this wrong it could throw the value off or prevent oscillation.

The usual load capacitance is 12pF I think...

Other thing may be if the circuit expects a very accurate crystal, like resistors some are more accurate than others.

Best thing is try to get a schematic so you can be sure..


And of course, if clearance is an issue, don't get a tall crystal where only a short one will fit, and done get SMD when you need through-hole or vice-versa.

[daimoku]

Thanks for the quick replies.

Having trouble finding a datasheet for this crystal - it has radial termination style/mounts through hole, appears to have a 3.3V supply based on the main board PCB. Seems to be made by Ecliptek, but all the results I found are for SMD.

Here's a similar PN datasheet, but it is a SMD.

Here's a copy/paste of the datasheet specs so you don't have to read it:

• Nominal Frequency 25.000MHz

• Frequency Tolerance/Stability ±50ppm over -40°C to +85°C (Inclusive of all conditions: Calibration Tolerance at 25°C, Frequency Stability over the Operating Temperature Range, Supply Voltage Change, Ouput Load Change, First Year Aging at 25°C, Shock, and Vibration)

• Supply Voltage 3.3Vdc ±10%

• Input Current 3mA Maximum

• Output Voltage Logic High (Voh) 90% of Vdd Minimum

• Input Current Logic High (Ioh) -1.6mA

• Output Voltage Logic Low (Vol) 10% of Vdd Maximum

• Input Current Logic Low (Iol) +1.6mA

• Rise/Fall Time 6nSec Maximum (Measured at 20% to 80% of waveform)

• Duty Cycle 50% ±5% (Measured at 50% of waveform)

• Load Drive Capability 15pF Maximum

• Output Logic Type CMOS

I ordered this one from Digi-Key:

• Frequency 25MHz

• Frequency Stability ±50ppm

• Frequency Tolerance ±30ppm

• Load Capacitance 18pF

• Operating Mode Fundamental

• Operating Temperature -10°C ~ 70°C

• Mounting Type Through Hole

• Size / Dimension 0.447" L x 0.183" W (11.35mm x 4.65mm)

Everything matches up fairly well except for the load capacitance (15 pF for the Ecliptek and 18 pF for the one I ordered) and operating temperatures (smaller range on new one). If I understand right, the higher load capacitance will likely change the frequency. Is that correct?

I checked Mouser, Digikey, eBay and Newark, but I cannot find a 15 pF oscillator (unless I order SMD) with these specs.

The 18 pF oscillator seems to be the most common through hole/radial load capacitance. I'd appreciate any advice you folks can offer. Try the 18 pF oscillator? Or is there some other method I can use to ensure the oscillator is correct (have been unable to locate the board schematic).