What went wrong? Home Lamp mod.

[Steelwolf]

I had a halogen lamp, 240VAC going through a 240-12 transformer, rated 20VAC, to a 20W halogen lamp. Simplicity in itself. It's running approx. 1.7Amps, right?

Anyway, I tried to put in a dimmer switch. Since I couldn't find any pots to work for 2A, I used a IRF520 MOSFET. To get DC, I ran the output from the transformer through a bridge rectifier diode rated for 400V 8A and smoothed the output using a 2200uF 16V electrolytic capacitor. From this, through this setup, to the lamp and back.

During the trial run, the body of the lamp got really hot. Hot enough to get burnt if the finger were left on it too long. At first I thought it was because the bridge rectifier couldn't handle the current, but now I think it might be the transformer heating up. Whatever the case, I did a long test run and the lamp died. Preliminary testing suggests that all the added compoenents are alright, but the transformer may have been fried. DC was supplied at a point past the transformer and the dimmer and lamp works. The bridge rectifier was tested on a DMM and is displaying all the usual readings.

Any suggestions on what went wrong? Why does the transformer heat up so horribly?

 

[Ray_of_Light]

Yon simply cannot do that! By rectifyng AC and smoothing, you are feding a 12 V lamp with 16 V or more, depending on the transformer. So, the transformer gets overloaded, and the lamp blew up.

You need to use a TRIAC and a zero-crossing detector chip, on the low voltage side of the transformer, to correctly dim the lamp.

If you want to use your DC setup, which resemble a stabilized power supply, you must double the power of the transformer (40 VA even if the lamp is 20 W), put a big heatsink on the mosfet, and make sure that you do not get more than 12 V to the output.
BTW, it is an highly inefficient setup. In addition, DC will kill the lamp faster than AC.

Hope it helps

Anthony

 

[Brock]

I am confused. A halogen lamp shouldn't care if it is getting AC or DC as long as the voltage is ok. I used to have some 10w and 20w halogen lights set up to run off 12v AC from a transformer and then automatically switched over to 12v DC from a battery if the mains went out. I couldn't tell the difference on AC or DC.

Maybe I missed something?

 

[Steelwolf]

Oh crap! /ubbthreads/images/graemlins/ohgeez.gif I totally forgot about the RMS/sine wave calculation! ...Yeah, 16V, I needed higher VAC transformer. No wonder it was overheating.

Can you tell me more about the TRIAC and zero-crossing detector chip? Or show me a schematic?

And can you explain a little more about why DC kills the lamp faster? Oh, and about the "highly inefficient setup", why is that? When hooked up to my lab power supply, the setup drew no more than 1.6A at 12V at full brightness and was drawing no current when totally dimmed. This was what I calculated. Has it something to do with the primary side of the transformer coil?

As you can see, I have little knowledge about AC circuits, and what little I knew, I've forgotten. Please enlighten me? /ubbthreads/images/graemlins/confused.gif Thanks heaps.

 

[linearintigrator]

yes, keep in mind that the DC voltage out of a full wave bridge rectifier is 1.4 times what the RMS AC voltage in is. also for safety, the capacitor should have a voltage rating at least double what the highest DC voltage you expect it to handle. and lastly, you should install a fuse on the low voltage side of the transformer to protect it from "shorts" or overloads. dave

 

[INRETECH]

If you attemped to run the FET in Class A Operation (Not completely saturated), its going to get *REALLY* hot

 

[Ray_of_Light]

It is a known problem of the filament lamp that the electrons cloud creates a migration of the impurities contained in the filament.
This "migration" is non-existant if you use AC. Standard lamp are specified for AC use. Halogen are less prone to DC damages, and unless the purity of the tungsten alloy is analitically defined, no average life can be specified. Cheap vacuum bulbs are the most prone to the bad effects of DC current.

About dimming an halogen bulb. Remember that the halogen cycle (the redeposition of evaporated tungsten on the filament) takes place only if the quartz envelope is above 200 Celsius. It is possible that the life of an halogen bulb is actually DECREASED if it is dimmed.

About the schematic for dimming. I have no possibility to post images, but any 800 V 4 A Triac in TO220 case, driven with a zero-crossing detector will suit. Also, you can use a UJT transistor and pulse transformer, but it is not your case. I will try to find a suitable schematic on the network and post the link. Suggest a Google search for "AC dimmer".

My statement on the "Highly inefficient setup" refers to the fact that you, by rectifying AC and stabilizing it with a "series" type regulator, loose in the transformer, the rectifier and the stabilizer at least 50% of the power.

This is totally avoided by using a "phase partializazion" setup, like the one that uses the Triac. The triac is turned on after after a number of the millisenconds after the sine wave started to raise (or drop) from zero, and is always fully conductive.
The equivalent intensity of the current flowing in the lamp is then proportional to the "angle of conduction", which in turn is related to how many millisec the Triac is turned on after the sine wave has crossed the zero.

Hope it helps

Anthony

 

[LED-FX]

Dimming on the LV side of a 12V lamp is a major pain in the a.., er, neck.

You lose between .7 and 1.4 V over the switching element, unless its a fet, which means you usually need to boost the supply above 12V to get full brightness.

Get electronic Dimmable LV lighting transfomers, so much easier. Inductive , wound transformers need hard firing dimmers or it`ll work for so long.... Don`t ask how I know this....

Dimming circuits Tomi Holger Engdahl`s fantastic resource:

http://www.epanorama.net/links/lights.html

HTH
Adam

 

[LED_ASAP]

It just doesn't make sens to me. Usually a transformer is very tolerant to overloading. The transformer in a desk lamp is likely rated for continueous service 24/7, it should be able to handle at least twice the rated load in a shorter period. If you feel the transformer heating up severely, it means you've got some serious overloading somewhere. I would recommend checking the following:

1) the wiring---is there any sign of short circuit?
2) the capacitor---as The Prophet pointed out, you will need to have some over head space for the 1.4x voltage---likely higher at dimmer setting. I would not use anything less than 25V.
3) is the heat-sink of the MOSFET touching anything it shouldn't touch? You can form a bad shortcircuit with those "hot" heat-sinks.

 

[MrAl]

Hello there,

Hold on a minute!

The rms voltage out of a bridge is the same as the
ac input (minus the voltage drops of two diodes).
It's actually LESS then the input ac.

The problem is, the capacitor doesnt just 'smooth' the
output, it also stores energy that gets added to the
output during those times between cycle peaks.

What difference does this make?

Remove the capactor and all your problems are solved :-)
No more storage, no more extra energy delivered to the
output, no more transformer overheating, no more premature
bulb burnouts.

As a precaution, measure the output voltage with the
bulb connected to make sure it's around 12v dc.

Note1:
The halogens get pretty darn hot and usually need a
ceramic base.

Note2:
If you pulse width modulate the FET, you'll use
less power during dim periods, and the FET wont
get hot.

Take care and good luck with it,
Al

 

[Steelwolf]

Thanks guys. I've got to go through all that info again. I'm not sure I understood everything, especially in Anthony's post.

Anthony, I understand about the reduced lamp life due to tungsten evaporation. IIRC though, some of that effect can be mitigated by running the lamp at full temperature every so often. And anyway, I was planning to switch to a LED array at a later stage. What about the migration effect. Will that still happen when the lamp is run under rectified AC? I'm guessing that there will still be migration. And thanks for the suggestion. I'll do a search.

Adam, thanks for the link, I'm checking that up now.

Mike, I have used FETs in what you called a class A operation before, but I've never noticed any significant increase in heat. So that I can get a feel of the conditions that would cause significant heating, what sort of voltages and currents were you running at?

LEDSASAP, As far as I can tell, nothing is short-circuited. The capacitor is fine as well.

MR Al, what you're suggesting is that I should go back to running the lamp on AC, albeit rectified, so that the RMS voltage would again be 12V and I am not overstressing the transformer?


Just a little more background: The lamp was originally halogen, so the bulb is already in a suitable base. I wanted to keep the mod simple, and there is limited space for extra circuitry, so I did not go with PWM dimming. The halogen was to have been, at some later stage, replaced with an array of LEDs to maintain spectral composition even when dimmed. I also wanted to reduce flicker, which, I was told, can strain the eyes, and this flicker is because of the low 50Hz AC that is used. (There were a bunch of lighting products and TVs that used inverters to increase the frequency even higher, up to 100kHz I think, to reduce flicker and eye strain.) Hence my desire to go for full-wave rectification with smoothing. No flicker, no eye strain. /ubbthreads/images/graemlins/smile.gif

So now, I should be getting either a higher rated VAC transformer as a start, and looking in to those Triac circuits?

 

[Doug Owen]

[ QUOTE ]
Steelwolf said:


To get DC, I ran the output from the transformer through a bridge rectifier diode rated for 400V 8A and smoothed the output using a 2200uF 16V electrolytic capacitor.


Why does the transformer heat up so horribly?

[/ QUOTE ]

Coming in late (and no doubt missing something), is it possible that you used a diode (one junction, two leads) not a bridge rectifier (four junctions, four leads)? If so, the answer is I think here?

Prolly part of that way you use our common (stollen) language, here a diode is singular.

BTW, that's not much of a cap for two amps. Your ripple voltage must be several volts? I suspect yer flickering and don't know it because the light can't cool off fast enough......

Doug Owen

 

[MrAl]

Hello again Steelwolf,


I didnt really think you were going 'back' to anything, i thought
you had the circuit already built up and all you had to do
was remove a capacitor.

Since you are running on 50Hz you may wish to provide some
capacitor smoothing, but one thing to keep in mind is that
the filament inside the bulb has a mass that is capable of
storing heat energy. This means that between cycle peaks
the filament actually stays hot, to some degree. Whether
or not it stays hot enough to reduce all flicker would be
a test you would have to perform. With no capacitor
connected, plug the thing in and look at it. If you see any
flicker some smoothing would be a good idea. If you dont
notice any flicker, try moving your head from left to right
while keeping your eyes from moving inside their sockets.
If you notice any flicker, you may wish to provide some
smoothing.
If you find you need some smoothing, then you should first
try your original cap value. Although the calculations
for voltage droop show that that value (2200uf) doesnt do
much, it may be just enough when combined with the heat
storage capacity of the bulb filament as noted above.
If that doesnt do it, double the value (another 2200uf)
and try again. 4400uf should be enough to reduce most
flicker.
The other problem is, once you connect the cap, you have
to also reduce the setting on the pot, or find some other
way to reduce the output. You cant run it full steam with
the cap connected because as previously noted it will
burn out the bulb quick. The best way is to use a lower
voltage transformer, or take some turns off the secondary.
If you cant do this, then you will have to turn the pot
to a lower value for testing. For a permanent fix,
you would connect a resistor in series with the top of
the pot so that it prevents the user from turning it
higher then a certain point. An easyer way to do this would
be to connect a second pot in series with the first, at the
top of the pot, then turn pot 1 all the way up, and adjust
pot 2 to full resistance. Then, turn on the circuit and
adjust pot 2 to allow only 12 volts output with the bulb
connected. Pot 2 could be a trim pot, while pot 1 would
be a normal pot with a knob accessable to the user.


Now a word on FET heating:

You'll find that for some pot settings the FET wont
get hot at all while for other settings
the FET gets EXTREMELY hot. This means that you will
need a rather large heat sink for the FET, so i cant see
this solution working in a very very small space, although
i dont know how small a space your app requires. Ten
square inches of surface area would be nice for the heat
sink, but the more the better. If that fits in your space,
that's all there is too it. If not, you'll have to go with
pulse width modulation either with the FET or with a Triac.

Just in case you go with the triac, it's good to know that
you wont want to connect it to the primary of the transformer,
because triacs dont work well into transformer primaries for
various reasons. If you connect it to the output, you'll
need a special circuit in order to be able to drive the
bulb from zero brighness to max, which will be almost
the same circuit as a pulse width modulator except you
wont be able to smooth the output very easily.

Another approach if you can afford it, is to buy a variac.
A variac will adjust the brighness from zero to max without
any additional circuitry, although they are a little on the
high priced side...probably about $60 for what you need.

If you are going to convert to LED's in the future, then
you'll want to provide a dc output anyway.

Take care, and good luck with it,
Al

 

[Steelwolf]

Doug O: No, it is definitely a bridge rectifier. Shouldn't have included the word "diode". It has 4 junctions, 4 leads. And you're right. For 2 amps, I think the cap should have 10 times more capacity?

Mr Al: Thanks for the suggestions.

10 sq. inches is more space than I have available. But I figured that since I would be switching to LEDs, and the LED array shouldn't really take that much current, I thought I would be safe with a simple setup like what I described.

Anyway, it wasn't the FET that died. It was the transformer. And yes, the final output I desire is DC, so that I can attach LEDs. So apart from the brightness control circuit (which I think, maybe I can squeeze in a PWM using a NE555 IC), the question might be better put as "How do I achieve true DC from a 240V 50Hz input".

Perhaps the best solution would be to remove the transformer altogether, run the 240VAC through a bridge rectifier, buffer that with a few mains capacitors, run that through a PWM circuit which will limit the voltage to 12V and allow dimming, and finally buffer that output again with a large capacitor. Is that what those little wall warts do? Hehehe, I might just buy some cheap wall wart and plug that in. /ubbthreads/images/graemlins/smile.gif

BTW, it is only after fitting LEDs that I'm worried about lamp flicker. Anything that uses filaments for light would have to keep the filament heated to a certain temperature to achieve a certain light output. So even if the power delivered were pulsed, and the pulse-width were not too wide, the mass of the filament would even it out. But LEDs have almost instantaneous response, so a pulsed supply would show flickering except at very high frequencies.

BTW, we talk alot about PWM, is there such a thing as amplitude modulation for power supplies?

 

[UK Owl]

At 50 Hz you do not need to worry too much about flicker in a lamp as you will hardly notice it. Just think of all those fluorescent tubes you see. Each one will flicker at 50 Hz unless it has a fancy electronic ballast that re-generates a supply at a higher frequency.

 

[eluminator]

Are you sure fluorescents flicker at 50 Hz. I'm betting they use both halves of the AC cycle like incandescents. That would put the flicker at 100 Hz. 50 Hz flicker would be noticable to the eye.

 

[MrAl]

Hello again,

Yes, they flicker at twice the line frequency unless
they use half wave rectification without filtering.

Take care for now,
Al

 

[JSWrightOC]

Wiring the LEDs to the AC mains without isolation is NOT recommended!! /ubbthreads/images/graemlins/icon15.gif

The setup you described does not offer any electrical isolation between the mains input and the LED supply. So, if you swapped Line and Neutral your "negative" LED supply would be at Line potential! This is the reason why most ungrounded appliances have polarized electrical plugs (at least, this is common in the United States) but you'll also notice there is little or no exposed metal, and it is isolated with respect to the mains connection anyway.

A wall wart typically uses a small transformer and optionally a bridge rectifier and capacitor inside. Newer DC-output warts use a miniature *isolated output* swithcing power supply. If space is an issue, you might want to try some kind of a brick-style switcher inside, or power from a wall wart. Either way, PWM is probably your best idea for driving the halogen or the LEDs--use a frequency of at least 100Hz and you shouldn't see any flicker in the LEDs at all. If you happen to notice flicker, just change the timing capacitor in your oscillator circit so it runs even higher. /ubbthreads/images/graemlins/grin.gif

 

[Steelwolf]

UKOwl: That was approximately the point I was trying to make. There were all these claims about how even though we can't actually conciously see the flicker, it is still there, and our eyes will suffer for it. So they were promoting these fancy schmancy lights that use ballasts or inverters that raised the frequency very high. I figured with what I was planning to do, getting full DC or at least a frequency high enough to be almost DC would be almost part and parcel of the design. So why not just include it? /ubbthreads/images/graemlins/smile.gif

JSWrightOC: No worries mate! /ubbthreads/images/graemlins/grin.gif To begin with, Aussie plugs are polarised. The plugs use blades instead of pins and each one has a different angle in relation to the other, so there is absolutely no way to put in the plug differently.

I'm not sure what you are worried about with the "line and neutral" bit. There will be a 4 post isolation on/off switch. I think I forgot to mention this. Is that what you're worried about? This switch I found, should ensure that when the lamp is switched off, all the rest of the circuitry will definitely be cut off from mains. That would be safe enough, wouldn't it?

I would like to find out exactly how those wall warts are wired up. They are so light that I don't think they could have a transformer inside. On the other hand, they don't seem to be rated for high (1A, 2A) currents. 1A seems to be max, 800mA, 500mA is more typical. Those that handle 2A and up are very heavy and probably contain a transformer with a high VAC rating.

About PWM. I think all the circuits I have are symmetrical pulse-width. So increasing frequency, or decreasing the pulse-width, would increase the power out. Does anyone know about asymmetrical PWM? So frequency can be high, but the off to on time ratio can be adjusted to give the exact power I require? Utilizing the 555 timer IC would be a plus.

BTW, I may be misunderstanding the concept, but PWM controls only the pulse width, not the amplitude, right? So if I ran a PWM direct from the mains, I have to be prepared for shots of 240V? Which means my capacitors have to be rated for 240V?

And also, even if I used a transformer to lower the voltage, the AC pulse is still present. And even if I rectified it, there would still be oscillation, so that at any given point in the cycle, the voltage could be any where between 0V and 16V. (With me so far?) So how will this affect a PWM circuit? Would it be possible that, if I ran the PWM at an exact frequency to coincide with the 0V point of the cycle, I may not see any power coming? Or more realistically, that the power delivered could vary a fair bit? Which means that it is preferable, and maybe even essential, to get as close to DC as possible before going through a PWM circuit?

 

[JSWrightOC]

If you plan on driving an LED array, your best bet is to use a step-down transformer, rectify the output, then filter it with a capacitor. Even though the DC will fluctuate somewhat with the pulses feeding the capacitor, it should not be enough to cause any concern. This will certainly not interfere with the PWM, especially if you use a high frequency (a few hundred cycles per second).

As for PWM circuits, there have been ones built around 555 timers as well as CMOS hex inverter chips. Both work well. The 555 variants are a bit trickier to build and yet maintain the same frequency, but I know they exist. The whole idea is to change the duty cycle of the output, while maintaining frequency and amplitude.

Driving an incandecent lamp with a PWM circuit will result in a heap of energy savings at lower settings, but doesn't offer you a whole lot of benefit when driving LED arrays. Short of a variable resistor, it is perhaps the easiest method however.

No matter what you do, I still think it would be wise to isolate your circuit from the mains supply, for sake of troubleshooting and experimenting.