Testing compact electronic ballasts

[GBDamo]

Currently in the process of changing lamps, all 32w GX24p-3.

Every one has had to have the ballast changed.

New lamps in, not working.

New ballast in working.

Ive phoned the ballast supplier and they report there are no lamp incompatibility issues.

The customer is pushing back on cost, as theve only ever had to change the lamps, why are they all suddenly broken?

The fittings are about five years old and used 12 hours a day, 7 days a week.

They are fitted above a suspended ceiling in a heated area so i cant see any environmental issues, no damp.

Anyone heard of poor quality lamps causing ballasts to fail?

Any idea on how to go about testing an electronic ballast?

Typically, on electronic ballast fittings,i work on the 'if its not the lamp its the ballast' method and replace but the number out here is becomming a worry.

Any ideas much appreciated.

 

[James]

some ballasts,
if you change the tube with the power still on
(not that any of us would do such a thing) wont start back up until the power has been removed and turned back on. some type of failed lamp detection that locks out further attempts to restart the lamp.

I have also known some of the very early electronic ballasts to fail completely if the tubes are changed with power on.

 

[telectrix]

rewire the internals and fit LED tubes.

 

[DT1991]

I think if you do the maths Damo that's not bad for a ballast to be finally failing.

This coupled with possible use of cheap lamps in the past could be a contributing factor to the situation you have now.

Edit I miss read the time they was on.
Ignore my poor maths

 

[telectrix]

heat damage causes ballasts to cook after a few years, esp. in cheap fittings. i was called out last week to a tripped 10A MCB feeding 6 florries. 2 of them had cooked ballasts. approx 6 years old.

 

[GBDamo]

rewire the internals and fit LED tubes.

Compact flourescents, havent seen a LED replacement for these, altgough not looked overly hard.

 

[James]

can you not lay in some flat panel led 600 x 600 to fit the grid?

 

[GBDamo]

Well, I currently have an old one and a new one on the bench and would like to test them.

My worry is due to the nature of the electronics they will not output until a load is detected.

Anyone any experience of testing them?

 

[GBDamo]

This is proving quite an ask, I've been googling for a good hour with no joy.

 

[Charlie_]

I’ve not tested one before but I think you would need a high voltage/high frequency scope to test the output with a good hf tube fitted

 

[GBDamo]

I’ve not tested one before but I think you would need a high voltage/high frequency scope to test the output with a good hf tube fitted

Yes, it's looking more likely that I'm not going to be something I can do in my shed.

 

[James]

am I being dumb here? power up both the new and old ballasts,
try a new tube in each in turn. note result
try a suspect tube in each. note result.
there are no serviceable parts inside the ballast so you are really only trying to find out if they work?

 

[i=p/u]

Know by hot or cold if goosed

 

[GBDamo]

am I being dumb here? power up both the new and old ballasts,
try a new tube in each in turn. note result
try a suspect tube in each. note result.
there are no serviceable parts inside the ballast so you are really only trying to find out if they work?

These are 4pin compact fluorescent lamps, 2x32w off of one Electronic Ballast.

I have done all the obvious on site but was wondering if they can be bench tested and outputs measured.

It would appear that is a little hopeful without some serious test kit and a mocked up lamp holder.

My multimeter and volt stick just wont cut the mustard here I fear.

 

[GBDamo]

Know by hot or cold if goosed

Hot = goosed

or

Cold = goosed

what about tepid?

 

[i=p/u]

Put ur multimeter and volt sick in series and if lights up ,Beer

 

[i=p/u]

Come back later maybe if Luke warm

 

[i=p/u]

Come back later maybe if Luke warm

 

[Lucien Nunes]

By definition, if the old ballast won't drive new good tubes, it's faulty. That much you will already know before it reaches the bench. What you really want to find out is why they failed, which can normally only be done by internal examination and possibly detective work. External tests probably won't reveal anything.

 

[GBDamo]

By definition, if the old ballast won't drive new good tubes, it's faulty. That much you will already know before it reaches the bench. What you really want to find out is why they failed, which can normally only be done by internal examination and possibly detective work. External tests probably won't reveal anything.

Cheers Lucien, you're correct. I'd like to say they have failed because of 'X' but that appears beyond my capabilities.

It's just odd that of the nine lights out the three I got to all had failed ballasts, there are another six to do which could all be lamps but it just seems weird.

I'm getting "the last company just changed the lamps"..."Why are they all broken now?"

My only response so far is possibly age or duff lamps used previously.

Just worried I'm missing something simple.

 

[Lucien Nunes]

Do you know that the old lamps are dead too? Have you tried them with the new ballasts? If you can prove that the old lamps still work, then it must be clear to the customer that the fittings are only dying because of the ballasts. If the lamps are dead too, then it's harder to identify which is the chicken and which the egg.

One common factor that would be implicated with a sudden mass failure of ballasts would be an overvoltage. I dimly recall a lot of fried electronic transformers for LV halogens that we traced to an intermittent neutral connection in a 3-phase DB, that had caused occasional overvoltages on one phase that fed mainly the lighting.

If you can get one of the old ballasts open and fancy posting some pics, we might be able to do forensics....

 

[GBDamo]

Do you know that the old lamps are dead too? Have you tried them with the new ballasts? If you can prove that the old lamps still work, then it must be clear to the customer that the fittings are only dying because of the ballasts. If the lamps are dead too, then it's harder to identify which is the chicken and which the egg.

One common factor that would be implicated with a sudden mass failure of ballasts would be an overvoltage. I dimly recall a lot of fried electronic transformers for LV halogens that we traced to an intermittent neutral connection in a 3-phase DB, that had caused occasional overvoltages on one phase that fed mainly the lighting.

If you can get one of the old ballasts open and fancy posting some pics, we might be able to do forensics....

Already opened one up and can see no signs of failure.

 

[gazdkw82]

Usually when you change a ballast you need to turn the supply on and off. Sometimes even taking the tubes out and back in again, temperamental things those ballasts.

If they have been in use for the time you say then it's not uncommon for them to start failing. They were probably stored in a warehouse for years before being installed.

You could try testing the outputs with a meter and see what (if any) output voltage they are delivering.

 

[Lucien Nunes]

They were probably stored in a warehouse for years before being installed.

There seems to be a manufacture date of 03/13 on the casing.

So the first 3-second visual check is for tops blown off power MOSFETS, diodes, NTCs, fusible and low value resistors, and bulging at the top of the reservoir electrolytic (can't see this). Then, a 30 second poke around with a meter looking at those vulnerable parts for obvious failures (Q's & D's S/C, NTCs & R's O/C etc.) Also high value resistors that are subjected to continuous high voltage such as in any startup logic supply, which often fail but remain undetected with the unit working until the power is interrupted after which it will not restart.

To go further, we have to understand the method of operation of the controller IC, the L6585D made by ST, that operates both the driver half-bridge and the power input PFC.
Have a read of the following...
Datasheet:
https://www.st.com/content/ccc/reso...df/jcr:content/translations/en.CD00085137.pdf
App note:
https://www.st.com/content/ccc/reso...df/jcr:content/translations/en.CD00152938.pdf

We learn from this that it has active relamping detection, so should not have failed to restart due to lamps being changed hot, and also how its EOL detection works. But most importantly, how the startup and lamp presence detection works so we can look for the relevant voltages and waveforms.

How far do you want to go? Got a scope?

temperamental things those ballasts.

You could try testing the outputs with a meter and see what (if any) output voltage they are delivering.

Remember that a CFL ballast delivers a complex waveform that changes as the microprocessor inside analyses the behaviour of the lamp. If it thinks there is no lamp present, or an abnormal situation that endangers its output transistors, it will immediately react to protect itself. So before you can measure any output, you need a good lamp connected. Then, compare what you see with the datasheet or reasonable expected values, bearing in mind that the lamp drive frequency may be beyond the multimeter's ability to measure correctly, so a scope is often a better choice. If there is no output, and no evidence that the half-bridge is being driven, one has to look at the conditions surrounding the IC to see whether it is lacking one or more of its resources, if it is actively trying to protect itself against a perceived threat, or simply dud.

 

[GBDamo]

Thanks a million for this, I have no scope or to be honest the faintest idea what to do with one.

When I've sorted the frikkin fence I'll read more of the links you've posted.

Thanks again.