Hi quite a few people asked for a full up to my lead flipper in video LinkedIn down below. In the end, if you haven't seen it and you should where I tore down one of these a dodgy cheap-ass the bottom of the barrel, constant current panel drivers here and how it was causing bad flickering on my LED panel lights and basically it was like a hundred percent ripple on the output and then I compared it with a ripple free or flicker free one. Granted, this one is twice the power 48 what's versus 24 watts but still, the cause was of course the lack of any output constant current regulation here. constant current regulation is done from the primary side and not only is there a lack of output filter in here compared to over here, but there's also a lack of input filter in which might cause an issue as well.

So anyway, let's do a follow up and see. quite a few people wanted to know how we could improve something like this or how we could hack it to fix it now. I Wouldn't recommend hacking these to fix it because there's a lot of design considerations, thermal considerations with your diodes, for example, thermal considerations with your transformer, and other things that you really don't want to mess with because these things are cheap. I Just recommend tossing something like this in the bin and just getting a non flicker free or a ripple free version of it.

But anyway, let's just do some experiments with this and see what's what now. Quite a few people commented that the input filter cap in quote marks was only 4.7 micro farad's and it was only a 50 vote job'. Well, if it was a 50 vote job and it was directly across a full wave bridge rectified mains, it would be toast. Is it explode straight away? But it's not actually across the mains, It's not doing the filter in its well, it's filtering, but only for the controller chip.

So here's the capacitor here. Here's the full wave bridge rectifier diode after it goes through the common mode choke Here, you can see that goes through these two series resistors here. so there's nothing in the chip data sheet that says that has like an internal Zener - like clamp the voltage or anything like that across there. So it's actually I getting its power from the feedback coil here which goes through this diode and then to the chip.

So it's using that after it boots up to regulate the voltage across there which will be the chip. maximum is 40 volts in the datasheet. so it's it's something under 40 volts. So a 50 volt rated cap is fine.

So that capacitor there has absolutely nothing to do with the input filtering. The only input capacitors we have are these two here and they're basically being used as a as part of the common mode choke. Yeah, there's not much doing those playing along at home. Both of those are a hundred and fifty nano farad.

so two of those in parallel? Well, yeah, 300 nano farad's not much. So a lot of people said. How about we just whack a big of course high voltage raided filter cap on the input? Because yeah, sure enough, for the lows that we've got, then this amount of input filtering is going to do Bug rossol. Your ripple is going to happen on your high side here.

Well, okay, well let's experiment with that. First thing we'll do is actually measure this bridge the output of the bridge rectifier here and see what our ripple is. And because we're measuring the ungrounded primary side of this instead of the isolated transformer secondary, You'll blow up your oscilloscope if you try and probe this side. So you need a proper high voltage probe available in the Eevblog store, of course.

Anyway, so we use this baby and we can safely probe anything on the primary side here. so we'll probe directly across the output of the common mode filter there. I've got this set to my divided by a hundred range. Of course, you set your scope to divide by a hundred as well for your probe.

and there we go and we can see that we have a maximum peak there of three hundred and forty five volts. There's our ground point there and bingo Look at that. We've got ourselves a full wave rectified job'. of course it's a hundred Hertz It's full wave rectified.

So there you go. This is at the full 20 watt output load or whatever it is. So yep, that's three hundred nano Farad's here. They do and much is it.

But the good thing about having a small amount of input capacitance is that you're going to have a good power factor. and this thing about this chip anyway is designed or advertised as having good power factor and that would be probably a requirement to get that. New South Wales government contract you couldn't have a poor fout power factor converter most likely I'm just guessing I haven't looked into the you know the requirements, legislation or less it a crap. just over 24.

What's there here you go? 245 volts here here in the lab. It is on the near to the as high side as you can get anyway 140 milliamps. So let's go PF Power Factor Naught point Nine five. that's not too shabby at all.

Any bureaucrats going to be happy with that. So that low power factor is going to be a combination of the low input capacitance here plus power factor features of the chip there which are you know switches things at the zero point and stuff like that. So if we put our output toroidal so we can measure the current, we see that our current waveform there at 200 milliamps per division. In fact, I've got that AC coupled so we'll change that to DC and you can see that there you go.

There's the ground point so it's a hundred percent Ripple Pretty horrible. So let's try a few experiments. Add some input and output caps in various combinations and see if we can improve that without killing our power factor. Okay, let's try some extra output cap.

Not a huge increase, but we had a 330 mic 50 volt before. It will add a 470 mic 50 volt. and for salaries from another bit of gear of course I hope I've got the polarity right. If not, well, good game.

Well there we go look at that. We have improved it bit it's not going a hundred percent to zero. Now you see, it's improved it up a huge amount. but you know and that really hasn't changed our power factor at all.

And as I mentioned, thermals on things like your diode here are going to be a potential issue. So I have removed the output filter cap I've had it been, had it running for a bit. Let's get the thermal camera on there. What do you know? The hot spot is that diode and we're talking Yep, 72 74 and rise in maybe I haven't left it on there long enough, but it's yeah, it's getting up there.

that's called 74. Oh Let's go overboard. 2200 Mike 63 volts thank you very much. And Bingo there is our current waveform.

sorry I've taken off the voltage probes. You can see it's a much further away from ground now. So yeah, it's not. What is it? 50% Ripple Now instead of a hundred percent.

and by the way it, we're still drawing 24 what's there? And let's have a look at our diode. A Little Diode II There, she's about the same. So there you go. there's no more stress on that dirty transformer is not I Trust me, it's really not hugely hotter.

So a huge amount of our increased output capacitance there practically an order of magnitude increase and pretty much overkill. Like you know that is asthma. You wouldn't make it that big on a 24 one panel like this and it's still got that's still like 50 odd percent. That's that's terrible.

Muriel And no surprises for guessing that hasn't changed our power factor or our input voltage waveform either. It's so obviously there's only a limited amount you can do by fixing the output capacitance like this because we've got no secondary side conversion over here. Nothing. It's all done via the primary side and and we've got bugger all input capacitance.

So let's add some income in capacitance over here and see what's what. Once again, we're going an order of magnitude. You've got to love orders of magnitude in engineering. Nippon Chemi-con for the Wind Once again salvaged from an old board and should always have like scrap boards lying around so that you can salvage parts like this because not everyone's going to keep like a component been full of various size and rating caps like this, especially on the mains input side unless you're working on that sort of stuff daily.

But okay. I Found a hole to put that into. Unfortunately, the pitch wasn't correct and there's not enough clearance around their support of jumper over. Definitely got that correct.

That's a negative on the bridge rectifier and I've confirmed that with the Voltmeter as well. Let's go into the negative: the cap. Don't to screw the pooch on that one. Anyway, let's pair that up and give her a ball.

Oh by the way, I've disconnected the output cap again. so we're in the original configuration. Just add in 33 mic input capacitance and tada. look at that.

That fixed it like a winner. Winner chicken dinner. No worries whatsoever. and our output ripple is pretty decent.

And look at our input voltage. look at that, there's not. You know there's a bit of ripple on that, but no worries. You know how it was going almost down to ground before because there was only 330? None of Farad's there.

So the input capacitance is the well. It fixes the problem and our output ripple is really quite nice. even with the original was a 330 micro Farad output cap. so it's still drawn out 24 watts, but our power factor won't want wah wah.

Thanks for playing it. Struck down 2.5 8. That is a horrible, horrible power factor. So really, from the utility, you're drawing almost double the amount of our current through the network copper, even if you're not being charged for it.

I've done a whole video on that, linked in at the end and down below. if I remember to do it, that's really bad and then if the government know about that, well, they probably wouldn't approve. So you might not be able to sell that with a power factor of nought. Point 5 8 or you know, under point 6.

That's pretty terrible compared to what 0.95 or whatever we had before. So yeah, that fixes our Flickr problem essentially. and I turned off my studio lights and as you can see, can't get any pick up any real flicker on this anymore. As if I look at the lights up there, they're the existing ones.

I'll switch my main lights on. There you go, you can see those flicker. now. this one ain't because we're getting.

you know, not much ripple on that. It is still. if you had decent measurement I gear you know you couldn't measure the flicker on that probably. but you know it's it's basically fixed.

It's low ripple once again. The output bridge rectifier there 75. the main switching chippy down in there even though it's on the back side of the board bridge. rectifier there neither.

We're talking about the center spot there, it's neither here nor there and our under the bottom. it's doing all right. So really, the only effect of increasing the input filter cap is not only you make it physically larger, of course, more expensive increases your bond cost, but it kills your power factor. and that could kill your lucrative government contract.

And if you think that the flicker free version solves the problem by input capacitance, nope, it doesn't. There you go, it's a little piddly, you barely see it, but 4.7 micro farad, 400 volt. They're solving it because it actually does proper secondary side current regulation over here. Quite a significant topological differences between these this one and this cheap-ass job'.

And if you want to know what the power factor of this particular one is, it's even better than the other one. Point, Nine, Seven, four. thank you very much. Now, of course we could try and correct this power factor correction problem by but passive means using some inductors, for example.

but it's kind of like polishing a turd pretty much because we don't know the or what impact this these mods have had to the efficiency of something like this. We saw that potentially the main switching debate I see is potentially getting hotter. There's internal losses in the switching transistor and stuff. Don't know what's going on there.

You'd have to measure the whole performance, You'd have to characterize the whole thing. And it's just know you are polishing a turd and you could do it as an academic exercise of course, but as a practical solution. No, it's just easier to go out and buy a proper flicker-free version now. I Actually found this Ti part which is a TPS 92 Three One four for those playing along at home and it does look to be a near identical part to the on Bright one.

In fact, they look at the typical application schematic it's practically identical and it's got all the same stuff offline primary side sensing controller, but this one says with PFC or power factor correction now with inherent in it When when they have to use the word inherent PFC it means it's it. Kind of does PFC like how exist in design Here, it actually had a good power factor correction because it does inherently do that, but at the expense. but because there's a single-stage converter, it's essentially going to have that trade-off limitation between your output ripple current and your power factor correction, so you know it's not that great. So if at this part, the good thing about this is that there's a few extra things in here.

See, it's got some additional clamping across the primary side transformer tap here. It's got a Zener diode clamped, but it's essentially the same. We've got our auxilary winding here and just a single half wave rectifier output and it does exactly the same thing. It's got all the same stuff, it's got quasi resonant switching, and you know all the sort of stuff that the on Bright one says.

And if we have a look and it does actually have a complete schematic down here, you know a complete application. Example schematic we've gotten as some additional filtering in here, but apart from that that looks pretty much the same. Here's the compensation capacitor down here and that value There does the or attempts to do the power factor correction so maybe we could tweak something like that on our or on brighter design perhaps but we don't have the relevant data. she doesn't have the relevant information on that and it's given us typical output capacitance figures as well.

Once again, input capacitance is only 47 nano Farad's and the good. The other good thing is that it lets us calculate the output capacitance and ripple. So it gives you the example here of for 30% ripple current. which is a heck of a lot right? And you still need four hundred and eighty, you know for seventy micro Farad's on the output and that still gives you 30% ripple current.

So yeah, you get your nice power factor correction, but your ripple, it sucks. So really, if you want to do this properly, you need a two stage converter instead of a single stage converter and you might get something. This is maybe a bit overkill. you can get low-cost ones and this, but this is a the LM 3454 example is they let drive with active power factor correction.

None of this inherent power factor correction rubbish. And and you can do what phase do you mean So you can do like phase control dimming of the thing as well and you can implement it as a single stage or a two stage design here. And here's the basically this is what we're doing here. Here's our secondary LED driver and that is essentially what I said of having the constant current regulation on the secondary side.

You're doing it as a second stage and this does active power factor correction. You'll notice. here's the main switching FET down here and they've got another switching FET here that is doing some active power factor correction. So this is the puppy you want if you want to do it properly.

but of course it's much higher cost than the simple single stage solution designs. Internal block diagram A Pretty comprehensive and complicated. We won't go into the details, this is not the place to do it, but you can actually implement it as I said as a just a single. You know this typical single stage design, typical flyback application like that.

and we won't get into power factor correction because I've done that in another video. But suffice to say, this one uses active power factor correction. and here you go. They specifically tell you single stage design low-cost download so you want the high end down light, you go for the two-stage design which is more expensive.

And here's the two-stage lead driver. Look at this. it's even got soft start. Fantastic! We've got EMI filter up here, very nice.

Then we've got all the active power factor correction and drive, we've got control feedback and then we've got the second stage lead driver here. they're using and of course they're pushing their own parts in home rather than 34 own. IB Can use any constant current lead driver you like and that just inherently gets rid of all the ripple and it solves a problem. And here's your opt.

A couple of feedback here, but that's expensive as it said in the on bright one. I think you need the opto coupler you need LM V4 3-1 typically and then you need to buy the constant current lead driver and well there there's some extra stuff of for dimming if you're doing that, but all of that adds significant cost as well as the active var. power factor correction stuff adds a very significant cost. You could easily triple your bomb costs or something like that by adding in these parts.

and by the way, this schematic might look a bit weird. It might look like all this stuff is physically on the primary side, but it's not. you will notice. V out here is connected to V Out here, which is on the isolated side of the transformer.

so they just didn't have a landscape format schematic so that's just a good well. A poor example of laying out a schematic well. you kind of. I I Would have liked physically put some dashes down here to sort of like you know, show that it's primary side secondary side I would added that those notes to the schematic there.

But anyway, yeah. two-stage output, current regulation. So as for our on bright turd here, well you know it's probably equivalent to that TI part. You're not going to get any better.

Once again, there is that compensation capping there so you could try and experiment with that and they've got some detail on that. The duration of the turn on period t on is generated by comparing the internal fix or T of wave with the voltage on the comp in. during steady state operation, voltage on the company can't be slowly varying due to a large external capacitor connected the copy and therefore the turn on time T on is constant in a flyback topology which is what we've got. constant turn on time and quasi risen.

The operation provide higher power factor and low THD Yeah, okay, how do we calculate that right? They don't tell ya no, there's just there's there's nothing in here. There's nothing in here to tell you how to calculate that compensation capacitor unlike the TI documents. So you know what are you supposed to do? like? Yeah, you can just suck it and see experimentation. Who cares.

I'm not gonna polish this turd if it's done. and just like get a proper flicker-free one. Yeah, you can probably shoehorn this thing to work, but no. I don't recommend it.

As I said, you'd probably have to do full characterization again. I Wouldn't want to put these things up, but you know a dozen of these things up in my roof that are hacked and just oil measures. okay in the bench when you put it up on the roof and the overheats and does what it? No, no thanks, no no, just don't do it so. I Hope you like this relatively quick second video.

Looking at this again and if you did, please give it a big thumbs up. As always, you can discuss it down below in the comments or over on the EEV blog forum. Catch you next time.