CCFL test circuit?

[mathog]

As I understand it when CCFL's age they draw more current. It seems like
some of the inverter problems discussed in this forum might be the result of CCFL's that have aged enough that the current they draw eventually damages the attached inverter. In this situation, one can swap caps and transistors in the inverter to affect a "repair", but since the high current draw remains, the inverter will eventually fail again.

So how to diagnose this situation?

It is easy enough to tell when a CCFL is blown by hooking it to a known good inverter. However, doing that with a high current CCFL will most likely not reveal anything out of the ordinary: the bulb would light "normally" and the test would not last long enough to destroy the inverter. My digital multimeter cannot even measure AC current, and most likely even if it had that setting, wiring it in series with the CCFL would blow up the meter.

Seems like some sort of dedicated circuit is required, one that could plug in between a good inverter and the bulb being tested. Googled for this and found some good/bad type testers like this one:

http://www.lcdparts.net/LST01.aspx

which I don't think would diagnose a low resistance CCFL. Also one analytic CCFL tester, which almost certainly would do the trick, but I couldn't find a price for it, and it looks really expensive:

http://www.nfcorp.co.jp/english/pro/...fications.html

Surely we can come up with a relatively simple/cheap circuit one could place in series with the CCFL which would emit a DC voltage that could be easily measured with a multimeter. For instance, placing in one of the CCFL power leads a small resistor in parallel with a series (diode + capacitor). Set the capacitor value fairly large, and the resistance such that the voltage drop across it is small compared to across the CCFL, but more than 0.7V, and the peak voltage on the resistor should eventually show up as a DC voltage on the capacitor (-.7V), which could be measured with a multimeter. This is simple and cheap, maybe not too accurate, and definitely not very safe, since both sides of that capacitor are swinging up and down by hundreds of volts, so there is plenty of room for blowing oneself up by touching a lead or the test instrument. To be useful we would also need to know what sorts of voltages to expect on good and not so good bulbs.

Anybody have a better idea?

[PlainBill]

Re: CCFL test circuit?

As I understand it when CCFL's age they draw more current. It seems like
some of the inverter problems discussed in this forum might be the result of CCFL's that have aged enough that the current they draw eventually damages the attached inverter. In this situation, one can swap caps and transistors in the inverter to affect a "repair", but since the high current draw remains, the inverter will eventually fail again.

So how to diagnose this situation?

It is easy enough to tell when a CCFL is blown by hooking it to a known good inverter. However, doing that with a high current CCFL will most likely not reveal anything out of the ordinary: the bulb would light "normally" and the test would not last long enough to destroy the inverter. My digital multimeter cannot even measure AC current, and most likely even if it had that setting, wiring it in series with the CCFL would blow up the meter.

Seems like some sort of dedicated circuit is required, one that could plug in between a good inverter and the bulb being tested. Googled for this and found some good/bad type testers like this one:

http://www.lcdparts.net/LST01.aspx

which I don't think would diagnose a low resistance CCFL. Also one analytic CCFL tester, which almost certainly would do the trick, but I couldn't find a price for it, and it looks really expensive:

http://www.nfcorp.co.jp/english/pro/...fications.html

Surely we can come up with a relatively simple/cheap circuit one could place in series with the CCFL which would emit a DC voltage that could be easily measured with a multimeter. For instance, placing in one of the CCFL power leads a small resistor in parallel with a series (diode + capacitor). Set the capacitor value fairly large, and the resistance such that the voltage drop across it is small compared to across the CCFL, but more than 0.7V, and the peak voltage on the resistor should eventually show up as a DC voltage on the capacitor (-.7V), which could be measured with a multimeter. This is simple and cheap, maybe not too accurate, and definitely not very safe, since both sides of that capacitor are swinging up and down by hundreds of volts, so there is plenty of room for blowing oneself up by touching a lead or the test instrument. To be useful we would also need to know what sorts of voltages to expect on good and not so good bulbs.

Anybody have a better idea?

Great minds think alike - (or at least my mind thing like yours).

First, what you describe happens rarely. Most inverters monitor the voltage across and the current through each CCFL. The inverter will shut down if either goes out of limits. A more common cause of shorted transistors seems to be bad solder joints. Of course, a design that does not measure current could also be responsible.

As for building a tester, it already exists on many inverters. Take a look at the attached schematic. This is from the application note for an OZ964 inverter controller. The same or similar design can be found on most inverters. And there is the dilemma. Often the controller shuts the inverter down before measurements can be made. What's more, there is no standard parts arrangement, and certainly no standard test technician to make the measurements.

While it is possible to force the inverter controller to ignore the fault condition, I'm reluctant to do that publicly for several reasons. The most important is that the technique is already known for some inverter controllers, and it is considered a fix for the problem, not a troubleshooting aid. I would rather not be blamed for a monitor flame-out.

I have been hoping to come across someone with the problem who has a monitor whose schematic is available. At least then I could create a sample procedure for troubleshooting the problem. Peering at fuzzy pictures trying to follow traces as they go under components is not fun.

PlainBill

[Krankshaft]

Re: CCFL test circuit?

Can't you just use a clamp meter over a CCFL supply wire to measure the current?

The CCFLs are supplied directly from the high voltage transformer and that's technically square wave AC right?

Clamp meters are made for measuring AC. That way you are safely decoupled from the HV circuit and have no direct contact with live lines.

[PlainBill]

Re: CCFL test circuit?

Can't you just use a clamp meter over a CCFL supply wire to measure the current?

The CCFLs are supplied directly from the high voltage transformer and that's technically square wave AC right?

Clamp meters are made for measuring AC. That way you are safely decoupled from the HV circuit and have no direct contact with live lines.

Well, if you can find one that will sense 5-10 ma at 100 Khz, that would work. You don't have to do the measurement on the output to the CCFL, it's just as easy to measure on the return line.

PlainBill

[Dgtech]

Re: CCFL test circuit?

Well well well.... I knew I'd find a post on here for testing CCFL's . Of course it was after I found a bad lamp in a IBM Thinkvision LCD monitor. The only way I was able to find it was simply taking a "give up guess". I had checked the inverter board and found no problems. Monitor was shutting off after a second. Now after taking the LCD completely apart, meaning lamps separated from unit, I connected the lamps back to just the board. When the unit powered up for a second, I noticed 1 of the 2 lamps on the bottom lamps were out with burn marks on the ends just as flourescent lamps normally do. My question is this:

Instead creating a circuit to test the lamps, how about a circuit to simulate the lamps to test the inverter?? A circuit that would keep the inverter on. With the symptoms that I had, if I had something that would mimic the lamp's load and keep the inverter on, I would have possibly assumed earlier that there was a bad lamp.

[mathog]

Re: CCFL test circuit?

Instead creating a circuit to test the lamps, how about a circuit to simulate the lamps to test the inverter??

A small test circuit would be nice, but most folks just use known good CCFL tubes. Of course, the whole idea for a CCFL test circuit is precisely to be able to determine when a CCFL is "known good", as opposed to "works in that other display", which might not be the same thing if the other display is a different model and its inverter is more tolerant of bad tubes.

[PlainBill]

Re: CCFL test circuit?

Well well well.... I knew I'd find a post on here for testing CCFL's . Of course it was after I found a bad lamp in a IBM Thinkvision LCD monitor. The only way I was able to find it was simply taking a "give up guess". I had checked the inverter board and found no problems. Monitor was shutting off after a second. Now after taking the LCD completely apart, meaning lamps separated from unit, I connected the lamps back to just the board. When the unit powered up for a second, I noticed 1 of the 2 lamps on the bottom lamps were out with burn marks on the ends just as flourescent lamps normally do. My question is this:

Instead creating a circuit to test the lamps, how about a circuit to simulate the lamps to test the inverter?? A circuit that would keep the inverter on. With the symptoms that I had, if I had something that would mimic the lamp's load and keep the inverter on, I would have possibly assumed earlier that there was a bad lamp.

That's another approach. The only problem is there is no 'standard' voltage and current for CCFLs. Also, there is no 'standard' connector.

Ideally, the test kit should consist of several parts.

First, an inverter to drive CCFLs without removing them from the panel. This should include test jacks to allow measuring the voltage across and the current through the CCFL under test. This would allow comparing the CCFLs and if only one is bad, catching it.

Next, adapters to allow measuring the voltage and current out of the inverter. This gets a little trickier, because you would want at least two.

Last, a voltage source to override the protection circuit on the inverter controller so you can keep the inverter running long enough to get a meaningful reading.

Of course, you also need adapters so you can connect the test jigs to the various CCFL / inverter connectors.

PlainBill

[Dgtech]

Re: CCFL test circuit?

Since the bottom CCFL was where the bad lamp was, I'll keep the good top pair for test.

[Dgtech]

Re: CCFL test circuit?

By the way, there is a very good article to read here on this site.
http://www.designnews.com/article/34...ght_an_LCD.php

[mathog]

Re: CCFL test circuit?

On thinking about the sorts of high frequency AC ammeter circuits that could be put in series with a CCFL, it dawned on me that I didn't really know what the drive waveform was like. If it is just a sine wave, or at least the same shape every cycle, and the control circuitry changes the peak to peak voltage then a relatively simple sort of circuit could be used. However, if the drive wave is pulse width modulated the circuit would need to be a lot more complicated in order to integrate the wave form.

For the "constant shape, varying magnitude wave" I'm thinking one could diode feed two fairly slow RC circuits (maybe 10 sec time constant) to pull out the desired information. That kind of circuit comes up to max measurement in one cycle and then holds it long enough for a DMM to settle (the diode charges the capacitor in the RC quickly, but the resistor discharges it slowly - quick sample, slow read). The first circuit would run between the two drive leads and have a large and small resistor such that the voltage across the smaller one was 1/10th (or whatever) of the total voltage. In parallel with the smaller resistor would be a diode driven RC circuit. This side would measure the peak drive voltage (at 50-100 volts, - Vdiode). The second side would have a smallish resistor (5K?) in series with the CCFL, and in parallel to that resistor another diode driven RC Circuit like the first one. Turn the display on, the peak wave voltage loads onto the capacitor in the first RC circuit, and the peak current (as I*5K - Vdiode) loads onto the capacitor in the second circuit. Read the two circuits with two DMMs in DCV mode (at $5 from Harbor Freight, they are cheaper than building a more complex circuit, plus I already have a couple on hand.) The ammeter will work as long as I*5K is >> 1V, roughly, and with ~5mA going through the CCFL there would be around 24V on the ammeter DIMM. If that is too much series resistance even a 1K resistor would probably work.

If the lamp never gets up to voltage it would show on the peak wave voltage DIMM. If the current goes very high (triggering an immediate shutdown) that would show on the ammeter DMM. By waiting for the circuit to settle back from the initial firing conditions on a lit display, one would also obtain the steady state information, which I'm hoping should show when one tube out of N runs at a much higher voltage or current than the others. That is, it would indicate a tube that is working, but just barely. Plus we are only talking a handful of resistors, capacitors, and two diodes, so nothing fancy is involved in terms of parts.

Lastly, by plugging in a 200K resistor (or thereabouts) instead of a CCFL, this should function as a dummy load. This would most likely be the most expensive part, since it needs to be a 5 or 10W resistor.

The intention being that one would NEVER touch this circuit when it was hot,with high voltages at high frequency running around inside it. It would be wired up, the display turned on, the DMMs read, and the display turned off. Then wait for the DMMs to both drop down to zero before touching it again.

[jotto]

Re: CCFL test circuit?

Why reinvent the wheel?

You have a built in tester with any inverter. You have your Vcc and switched voltage. You can use one power supply or two, I prefer to use two, if you want to use one, just use a resistor to drop the voltage to the required amount.

I use a dummy load for lamps, I can run 8 lamps of any size. 50k ohm 10 watt 7 each per lamp in series, tap off what you need to keep a balanced load.

Inverters use a circuit called a royer oscillator to monitor the load, any imbalance load.

Lamps will show a tinge of red when they are on the way out.

[budm]

Re: CCFL test circuit?

Most of the new inverter circuits used in the new monitor are not Royer any more, a lot of early designs (5 years or older) use that a lot.
I use spare lamps from broken panel, or just use old CFL.
https://www.badcaps.net/forum/showthread.php?t=19987

[caphair]

Re: CCFL test circuit?

Can you use smaller sized ccfls from old monitors to test monitors that use bigger sizes and vice versa? Or would this cause problems?

[stj]

Re: CCFL test circuit?

if the difference is not much.
obviously using a 15" tube in a 32" tv is going to overdrive the tube real hard and could cause a fault to be detected.