So I'm cleaning up an old Daybrite turret to put it into use, and while doing so, I decided to test the ballast in it (a GE 3xF40T12 ballast, Cat No. 89G777), and found that it tested somewhat out of spec, below are some of the numbers:
Current
Spec - Tested
1.2A - 2.3A
Watts
-NA- - 170w
Power Factor
>.9 - .62
Its also evident that the ballast ran pretty hot at one point, as it had an almost burnt smell to it, but despite all that, it still seems to work fine, I also don't have a 3xF40T12 ballast to replace it with, so I'd have to either use a 1xF40 ballast plus a 2xF40 ballast as a replacement, try to find one online, or buy one off someone here.
Thoughts?

This has all the appearance of failing transformer windings. It's probably even close to a catastrophic failure. The fact that its power factor dropped to 0.62 combined with excessive heat dissipation is the giveaway. The windings are likely shorted internally.

Kind of sucks, being an odd 3-lamp ballast. Some faults like dried wires, or even a bad capacitor can be fixed, if you feel confident enough to disassemble the ballast. But shorted transformer coils? That's a tough one... 

This has all the appearance of failing transformer windings. It's probably even close to a catastrophic failure. The fact that its power factor dropped to 0.62 combined with excessive heat dissipation is the giveaway. The windings are likely shorted internally.

Kind of sucks, being an odd 3-lamp ballast. Some faults like dried wires, or even a bad capacitor can be fixed, if you feel confident enough to disassemble the ballast. But shorted transformer coils? That's a tough one... 

With the low power factor I would rather guess for a capacitor short. Winding short would lead mainly to a real power increase, the PF won't change that much (or in case of a NPF ballast the PF may actually increase).
The paper capacitors are well known to degrade and fail over time. Mainly caused by moisture seeping in over time, the oil impregnation won't prevent that, mainly when non-PCB (as the PCB ones tend to have better sealing capacitor case, motivated by preventing PCB release, but it seals better both ways).

Question: How the ballast behaves (PF, current, power) when there are no tubes inserted? Are you able to compare that with a known good ballast?
If the current and mainly the real power are low or the same as a good ballast, the winding itself is very likely good and the culprit is the capacitor.
That is possible to replace, but it is a tedious and messy job. I would put the new capacitor externally (don't forget to update the wiring diagram on the label), as reliability wise the capacitor uses to be the weakest component. The burned winding is then usually just a consequence, if the ballast is left in place too long with the failed cap.

With the low power factor I would rather guess for a capacitor short. Winding short would lead mainly to a real power increase, the PF won't change that much (or in case of a NPF ballast the PF may actually increase).
The paper capacitors are well known to degrade and fail over time. Mainly caused by moisture seeping in over time, the oil impregnation won't prevent that, mainly when non-PCB (as the PCB ones tend to have better sealing capacitor case, motivated by preventing PCB release, but it seals better both ways).

I did suspect the capacitor at first, however, unlike every other ballast I've seen with a failing capacitor, the tubes still look like they're being driven at or near full power with no visible flicker, so either the capacitor only does PFC in this ballast and nothing else, or there's another fault.

Question: How the ballast behaves (PF, current, power) when there are no tubes inserted? Are you able to compare that with a known good ballast?
If the current and mainly the real power are low or the same as a good ballast, the winding itself is very likely good and the culprit is the capacitor.

I don't have any other 3-lamp ballasts to compare it to, nor have I done any tests with no lamps. I can possibly do the no lamp tests, post the results, and someone else that has a known working 3xF40T12 ballast can do the same test and compare.

With the low power factor I would rather guess for a capacitor short. Winding short would lead mainly to a real power increase, the PF won't change that much (or in case of a NPF ballast the PF may actually increase).
The paper capacitors are well known to degrade and fail over time. Mainly caused by moisture seeping in over time, the oil impregnation won't prevent that, mainly when non-PCB (as the PCB ones tend to have better sealing capacitor case, motivated by preventing PCB release, but it seals better both ways).

Question: How the ballast behaves (PF, current, power) when there are no tubes inserted? Are you able to compare that with a known good ballast?
If the current and mainly the real power are low or the same as a good ballast, the winding itself is very likely good and the culprit is the capacitor.
That is possible to replace, but it is a tedious and messy job. I would put the new capacitor externally (don't forget to update the wiring diagram on the label), as reliability wise the capacitor uses to be the weakest component. The burned winding is then usually just a consequence, if the ballast is left in place too long with the failed cap.

After reviewing a document on fluorescent ballasts troubleshooting, it seems your theory could be correct. There's indeed a capacitor in series with the tubes.

I did suspect the capacitor at first, however, unlike every other ballast I've seen with a failing capacitor, the tubes still look like they're being driven at or near full power with no visible flicker, so either the capacitor only does PFC in this ballast and nothing else, or there's another fault.

Maybe there's a test to try that could allow you to distinguish a cap vs transformer failure?

If you have an AC capacitor laying around (no matter how many uFs, but the more uFs, the more likely it'll give visible results, and that cap must be least rated for the mains voltage), you can try connecting it in parallel to the ballast, and then see how the measurements change. If the power factors raises, then the ballast's capacitor is the culprit. But if it drops, it means the load got further to a purely resistive load, and was capacitive to begin with, pointing towards a transformer failure.

Would that make sense?

Too bad it's an odd 3-lamp ballast. On standard 2-lamp ballasts the aforementioned capacitor can actually be measured externally, between one of the red wires and the ballast's white wire.

So I did a bit more testing, and one thing I'd like to start off with is the ballasts current seems to have dropped a bit from my last test, from 2.2A 170w to 2A 150w (but still with a PF of .62), this could be due to the fact its way warmer outside where I was doing these tests now compared to yesterday, though I'd usually give older ballasts I was testing about 5 minutes to warm up (as that tends to be ample time for most in-spec older fluorescent ballasts I've tested, but may have not given this ballast enough time to warm up for all I know), but anyway. I also hooked a 1.5uf film capacitor across the input of the ballast, and saw a small current and PF change from 2A, .62PF to 1.9A, .65PF. So it does seem to have a failing capacitor, I just don't know why it doesn't impact the lamps like it does on 2-lamp F40T12 ballasts.

Also, I did not know you can test a ballast capacitor externally on a 2xF40T12 ballast, so that's good info to know for future reference, thank you for that info.

So I did a bit more testing, and one thing I'd like to start off with is the ballasts current seems to have dropped a bit from my last test, from 2.2A 170w to 2A 150w (but still with a PF of .62), this could be due to the fact its way warmer outside where I was doing these tests now compared to yesterday, though I'd usually give older ballasts I was testing about 5 minutes to warm up (as that tends to be ample time for most in-spec older fluorescent ballasts I've tested, but may have not given this ballast enough time to warm up for all I know), but anyway. I also hooked a 1.5uf film capacitor across the input of the ballast, and saw a small current and PF change from 2A, .62PF to 1.9A, .65PF. So it does seem to have a failing capacitor, I just don't know why it doesn't impact the lamps like it does on 2-lamp F40T12 ballasts.

Also, I did not know you can test a ballast capacitor externally on a 2xF40T12 ballast, so that's good info to know for future reference, thank you for that info.

Keep in mind though that it's pure theory, based on documentation. If you have a multimeter with a capacitor range, it wouldn't cost much to actually test the theory. I would actually test it myself if I could. Though it's not possible at the moment. When my semester's over it's likely I'll have some time to do it.

There can be one of the ballast's coils in series with the capacitor as well, so any kind of precise measurement would probably be impossible. But I assume a ballpark measurement can still be obtained. Needless to say that with no value to compare the measurement against, the only valuable information to get out of this is whether or not the capacitor is shorted or open. And again, being an unusual 3-lamp ballast, finding that capacitor from the external wires is a guess game LOL.

So either you:

- read no capacitance & very low ohms -> cap is shorted
- read some capacitance & ohms rising to infinite -> cap may or may not be good
- read no capacitance & infinite ohms -> cap is open

And that implies you can find a way to get to the cap's terminal through the ballast's external wire.

I made it really easy didn't I? 

The lamp intensity drops, when the capacitor fails in the direction of loosing capacitance. This is rather normal capacitor wear, not direct fault,

But what I'm suspecting here is a capacitor going towards a short circuit fault, which leads to elevated current, overdriven lamp, overloaded primary winding and mainly excessive losses (this is, what ultimately overheats and fries the transformer in the ballast).

I understood you don't have access to another piece of the same ballast type, right?

Still you may take the measurements without the lamps connected (open circuit) and publish the results, it may still say something...

Sorry about that, I meant to put those results in my last post, but forgot to even do the no lamp test, anyway, here are the test results:
Watts: 9
Amps: .27
PF: .27
Now I've seen the starting capacitors in some GE 2xF40 ballasts short out, making the ballast into a single lamp ballast, and I think those capacitors were also made by GE.

Sorry about that, I meant to put those results in my last post, but forgot to even do the no lamp test, anyway, here are the test results:
Watts: 9
Amps: .27
PF: .27
Now I've seen the starting capacitors in some GE 2xF40 ballasts short out, making the ballast into a single lamp ballast, and I think those capacitors were also made by GE.

That was the case with several GE ballasts I found in my years of saving stuff being disposed of. Unfortunately I didn't have the knowledge back then to isolate the fault and possibly fix those 1960s ballasts (and at the same time get rid of those nasty PCBs!). A few of them were physically pristine, they would've made perfect display items.

Yeah, I tried to fix a GE ballast that had both a shorted starting capacitor and a shorted current limiting capacitor (not sure what else to call it) and things went wrong as soon as I opened it, because there were a couple of bond wires that were stuck to the lid and were ripped up by me opening it up. I was never able to figure out where the two wires connected to, so i kind of gave up on that ballast, even though now I could possibly get it working as a single lamp ballast (since I was able to do that when figuring out where those bond wires went), but the OCV would probably be too high for 1xF40T12.

Back on topic, does the 3 lamp ballast seem safe to use?, since the losses now seem to be in a more acceptable range for some reason (may have not given it enough time to warm up the first time, IDK).

Sorry about that, I meant to put those results in my last post, but forgot to even do the no lamp test, anyway, here are the test results:
Watts: 9
Amps: .27
PF: .27
Now I've seen the starting capacitors in some GE 2xF40 ballasts short out, making the ballast into a single lamp ballast, and I think those capacitors were also made by GE.

That des not sound like that much. Is the ballast warming up or just remained cold?
Assume with the lamps it get scorching hot...

If remained cold, it really looks like capacitors, either shorted or maybe "just" leaky (that means they are turning into resistors, dissipate a lot of power and overheat).

To check the capacitor leakage, normal multimeter won't do any judgement, nor a capacitor meter.
The leakage has to be tested at the peak operating voltage in order to prove it is good.
And even "mr Carlsons capacitor leakage tester" (tests at about 20..30V, but "requires" the exhibited resistance to be in the 10's of MOhm to pass, to compensate for the low test voltage) is useless here, because in the ballasts, capacitors are wired with parallel bleeder resistors (for safety reasons).

However: By an ohm meter you may identify leads, which either directly or indirectly go to a capacitor: You will see there the bleeder resistor, in the 100k..1Meg range. To make sure it is the intentional bleeder and not some isolation breaking apart, the resistance must be the same hen tested with a regular ohm meter, as well as with some higher DC voltage (multiple 9V batteries in series,...).
Check both polarities, to exclude an eventual starting diode (some lead type ballasts use a series RD combination connected parallel to the discharge, so it allows the peak voltage to double by charging the capacitor for ignition; those ballasts then refuse to start when the capacitor becomes leaky).

The 9w seems pretty normal to me since a lot of my 2xF40T12 ballasts have similar no lamp currents. I didn't test it for long enough to see if it would get really hot either with or without lamps, as I was concerned about the current draw, but I can leave it hooked up for a few hours to see how the ballast behaves.

So in an interesting turn of events, I now have three more of these ballasts in three fixtures that are identical to the one this ballast was in.

From that, I think now I can say the wattage is more or less in spec despite the bad capacitor, because the other three all tests within the same wattage (I also did an overnight test and the ballast didn't catch fire, soooo yeah)

Thank you Medved and Foxtronix for helping me, even though I feel i kind of wasted both your time (sorry).

... I feel i kind of wasted both your time (sorry).

What makes you think so?
You have asked, I was glad to give you my thoughts, you did not feel comfortable to go such deep way to disassemble and fix the ballast so you did not, that I see as a good decission (better than exceeding your capabilities and consequently cause some severe mess), so for me you are perfectly welcome...