Hotwire LDO driver! 13mv drop with 1185

Update: found the chip to make this driver.. read on!

==

Hey folks.. working on turning the 'nano' driver into a voltage regulating FET controlling constant voltage driver for use in hotwire lights.

After Modamag did his testing on the Maglite switch and i did follow-up testing on a couple switches here.. and measured 0.15ohm to 0.16ohm...that equates to about a 1/2Volt loss on the switch in a Mag85.. and that equates to about a 200 lumen drop in output! craziness.

So... enough of that i'm goin' FET found a most excellent FET with RDSon of 0.006 ohm with Vgate=10 and 0.009 ohm with Vgate=4.5. (They are $7-8).. so i looked some more and found a much more common FET that is 0.026ohm at vGate of 5V... $2.30

So.. going to build an FET-based driver for my hotwires.. was going to just use it as a pure switch but with like $1 more in parts i can have it regulate the voltage at least just a little bit (i.e. hold the voltage to 11V or less with the mag85 or the 1331 bulb to keep it from popping when the Vbat is 12.7).

In any event, my idea was to take a 100kohm pot and trim the Vbulb and put the wiper into the input of the 35mV LDO driver MrAl designed that i use in the nano... using the 100k pot i use to trim the amplifier bias i should be able to find an input wiper voltage that is a little higher than the 4/100ths of volt i typically put into that circuit.. the amp bias can decrease the gain and i can put in some higher input voltage.. maybe 1/10th to 1/5th of volt.. that would be a 1.2:100 ratio on the trimpot.. maybe i'll use a 200k pot, or better use like a 2K resistor in series with 100k pot or something to set the feedback voltage.

In any event using this method i should be able to have a ludicrously low drop out voltage.. even at 3.3A of an 1185... 0.026ohm means 9/100ths of a volt drop... for the 1185 i'm considering using the $7 FET which would be a 2/100th of a volt drop on the FET...

I will be rewiring the switch of course and having some thick wire like 18 or even 14ga wire hooking things up.. and hope to wire it up such that the switch on the body actually just supplies the power to run the amp and w/o power applied the FET i'll put a resistor to pull down the gate and shut off the FET which will actually also be used as the power control.

The FET i found is N channel i think.. needs positive gate to turn on... goes on the 'neg' side of the bulb.. i don't have a problem for my designs with that.. i think i'd be interested in a P channel solution for 'the public' so they can't short out the battery if the switch is off and some how short the bulb pin that is always live.

I think i have a basic ckt design at home i'll try to post that to get thoughts.

Oh.. i have another design idea using the National voltage reg. part, but the dropout voltage is way to high for most designs, but ONE.. i want to drop 14.4V to 12.6 to run an 1166 bulb, and a 'stock' part will do wonders for that design...

I was trying to find a switcher or anything that will just take a feedback from a pair of resistors to set voltage but can run an EXTERNAL FET.. found a couple from National with 'low and high' FETs i didn't understand how those work..

I would love a single chip with 2 resistors to set voltage and external FET solution... or build one based on the nano ckt.. it's absolutely wonderful and should do the trick for what i want to do.

Thanks in advance.

-awr

Re: Hotwire LDO driver needed

Hi there Andrew,

So you're looking to build a higher current LDO then?

Take care,
Al

Re: Hotwire LDO driver needed

Andrew,

Here's a part I've been eyeballing for quite a while:

http://www.st.com/stonline/products/literature/ds/7335/tsm108.htm

It's for powering automotive accessories. Traditional step-down switchmode topography. It uses an external P-channel MOSFET. I can't imagine there being any power/current limitations, since that is ultimately decided by the inductor, switch, and diode.

I've also been eyeing it for LED applications, since it has the ability to limit current based on a sense resistor, and the current sense voltage is only 0.2V. So it might not be too ideal for a single luxeon, but for multiple in series, or a luxeon V, the 0.2V drop isn't too bad of an efficiency hit.

It also has programmable under and over-voltage lockout, so you can protect your rechargables.

And the price isn't too bad, mouser has them for $0.72, and digikey for $1.35

Re: Hotwire LDO driver needed

MrAl said:

Hi there Andrew,

So you're looking to build a higher current LDO then?

Take care,
Al

Click to expand...

Higher current.. you could say.. if 6 or 7 amps counts. not current regulated this time.. voltage regulated.. no series resistance that way.. i want to take your design and convert it to voltage regulation and have it adjustable at least 'set and forget'.. i.e. with small changes would like to make a 6.6V version a 7.2V version.. a 10.8V version a 12.6V version and a 14.4V version.. the same exact ckt should work with any of them just by changing the bias to the amplifier and the feedback voltage bridge.

Evan.. waaay too complicated and inefficient...

here's what i have in mind:




Here's the description:

with SW1 open.. C1 will drain through R3 pulling gate to ground = FET off.. this will put E2 (amp ground) high that'll make the amp 'off' ready for switch to be turned on.

Close SW1:
with FET off.. amp will 'see' high voltage at E2 so also high voltage at B2.. the amp will be 'high impedance' and current will flow into C1 through the 'low R' side of R1.. at some point Q3 will 'turn on' and allow current to pass through the bulb and the Drain V of Q3 will drop pulling down E2 (amp ground)... R2 will generate a feedback voltage... current will flow out of E1 through R2 to the 'amp ground' (Drain of Q3)... and since E2 dropped to turn on the amp.. feedback will develop based on the input at E1 and when current gets high enough the voltage at E1 will become high enough that B2 will rise turning on Q1 which will pull down C2 voltage (of course the gate of the FET).

With any luck R3 and C1 can be picked to get the 'turn on' to take 1/4 sec or so, and not cause any ugly oscillations.

Once 'cooking'.. as Vbat drops.. the R2 fraction will decrease, pulling down E1.. which will pull down B2 (amplified), which will decrease current in Q1 pushing up the voltage at C2... this will increase the current at Q3.. eventually Q3 should see virtually full Vbat as long as Q1/Q2 can develop that much voltage and the bias can be set just right.. at that point the device will be 'direct drive'.. the FETs i've picked have as low as 0.006 ohm RDSon.. and even the cheap one is 0.026.

the 'weird part' i think other than having positive on the bulb all the time... is that the 'ground' of the amplifier doesn't go to ground.. but through the FET.. it would seem the amplifier ground and input have to share the same reference for 0V... and i think that should work ok..

I'm not lookin' for perfection if there was a 10% swing in output over the life of the batteries i'm not going to cry...

The KEY is that i want these two things:

1) over voltage protection.. i.e. i want to clamp the Vbulb to 10.8V for example even when freshly charged batteries are 12.7.

2) absolutely the lowest resistance possible... battery-> bulb ->FET -> battery.. and i'll be using some hefty wires.. i would like the total circuit resistance from battery to battery to be on the order of 0.05 to 0.07 ohms max... right now the typically mag switch is like 0.15 ohms... it usually sucks between 100 and 200 lumen off most hotwire solutions.

So.. does the FET source have to be the same as the amplifier ground for this to work... maybe i have to flip the design over to being positive fed vs negative pull..

Like i said... a single IC 'voltage regulator' that is designed to just 'sense' a feedback voltage but control an external FET would be ideal.. just hook up a potentiometer from Vbulb to ground.. the wiper to the input and an FET (P channel i'm guessing) above the lamp.. the regulator monitors the voltage and turns up the gate if it's too low, simple as that... badda bing.

The result i'm looking for is basically an FET 'soft start' switch... but with the add'l bonus of holding back any peak battery voltage...

with the right deign a simple switch could be added to change the voltage setting.. i.e. using the same light host with 10.8V or 7.2V and just flipping a switch when you change the bulb to set the output.

-awr

Re: Hotwire LDO driver needed

Hi Andrew,

Since you have plenty of voltage to work with, perhaps you would like
to use an op amp for this design? Makes things very easy, very predictable,
and very temperature stable.


Take care,
Al

Re: Hotwire LDO driver needed

andrewwynn said:

Higher current.. you could say.. if 6 or 7 amps counts. not current regulated this time.. voltage regulated.. no series resistance that way.. i want to take your design and convert it to voltage regulation and have it adjustable at least 'set and forget'.. i.e. with small changes would like to make a 6.6V version a 7.2V version.. a 10.8V version a 12.6V version and a 14.4V version.. the same exact ckt should work with any of them just by changing the bias to the amplifier and the feedback voltage bridge.

Evan.. waaay too complicated and inefficient...

here's what i have in mind:

Here's the description:

with SW1 open.. C1 will drain through R3 pulling gate to ground = FET off.. this will put E2 (amp ground) high that'll make the amp 'off' ready for switch to be turned on.

Close SW1:
with FET off.. amp will 'see' high voltage at E2 so also high voltage at B2.. the amp will be 'high impedance' and current will flow into C1 through the 'low R' side of R1.. at some point Q3 will 'turn on' and allow current to pass through the bulb and the Drain V of Q3 will drop pulling down E2 (amp ground)... R2 will generate a feedback voltage... current will flow out of E1 through R2 to the 'amp ground' (Drain of Q3)... and since E2 dropped to turn on the amp.. feedback will develop based on the input at E1 and when current gets high enough the voltage at E1 will become high enough that B2 will rise turning on Q1 which will pull down C2 voltage (of course the gate of the FET).

With any luck R3 and C1 can be picked to get the 'turn on' to take 1/4 sec or so, and not cause any ugly oscillations.

Once 'cooking'.. as Vbat drops.. the R2 fraction will decrease, pulling down E1.. which will pull down B2 (amplified), which will decrease current in Q1 pushing up the voltage at C2... this will increase the current at Q3.. eventually Q3 should see virtually full Vbat as long as Q1/Q2 can develop that much voltage and the bias can be set just right.. at that point the device will be 'direct drive'.. the FETs i've picked have as low as 0.006 ohm RDSon.. and even the cheap one is 0.026.

the 'weird part' i think other than having positive on the bulb all the time... is that the 'ground' of the amplifier doesn't go to ground.. but through the FET.. it would seem the amplifier ground and input have to share the same reference for 0V... and i think that should work ok..

I'm not lookin' for perfection if there was a 10% swing in output over the life of the batteries i'm not going to cry...

The KEY is that i want these two things:

1) over voltage protection.. i.e. i want to clamp the Vbulb to 10.8V for example even when freshly charged batteries are 12.7.

2) absolutely the lowest resistance possible... battery-> bulb ->FET -> battery.. and i'll be using some hefty wires.. i would like the total circuit resistance from battery to battery to be on the order of 0.05 to 0.07 ohms max... right now the typically mag switch is like 0.15 ohms... it usually sucks between 100 and 200 lumen off most hotwire solutions.

So.. does the FET source have to be the same as the amplifier ground for this to work... maybe i have to flip the design over to being positive fed vs negative pull..

Like i said... a single IC 'voltage regulator' that is designed to just 'sense' a feedback voltage but control an external FET would be ideal.. just hook up a potentiometer from Vbulb to ground.. the wiper to the input and an FET (P channel i'm guessing) above the lamp.. the regulator monitors the voltage and turns up the gate if it's too low, simple as that... badda bing.

The result i'm looking for is basically an FET 'soft start' switch... but with the add'l bonus of holding back any peak battery voltage...

with the right deign a simple switch could be added to change the voltage setting.. i.e. using the same light host with 10.8V or 7.2V and just flipping a switch when you change the bulb to set the output.

-awr

Click to expand...

If you are trying to "hold back" the battery, in your words, the element that would be doing the holding back is the FET. It will be burning up the "power" you are trying to hold back, until the cells get down to a voltage close to your output voltage.

Thats probably why Evan9162 was recommending a switcher.

What is your battery input voltage on fresh charged cells, and what is your bulb voltage? What is the current to the bulb at this time?

BTW, there are plenty of MOSFETs out there with 0.007-0.001 ohms on resistance...

Re: Hotwire LDO driver needed

One of the projects is a 7.2V nominal running a 6.6V bulb at 3.75A.

The battery will start at 8.4V and drop to 7.0V.. meaning.. I need a voltage Drop on the FET of 1.8V to start and 0.4V at the end.. at 3.75A that means 6.75W of power dissipated on the FET to start (but quickly dropping to 2.25W for most of the battery life)... at the end of life, it will be 1.50W..

The system will put 25W to the bulb.. so.. 25/(25+6.75) = 78% efficiency but really that will last for seconds.. 25/(25+2.25) is the average.. 90% efficiency.

That doesn't factor in some inevitable wiring losses.. the FET will probably go to 100% saturation before the batteries die which is what i want to happen.

Most of the circuits would be more like this:

10.8V nominal battery -> 10.65V nominal bulb..

Basically i only want a ckt to hold back the 12.7V battery voltage of a fresh charge to protect the bulb from those ludicrous startup currents and initial 'overdrive'..

Typically a 'mag85' solution would have as much as 0.15ohm resistance.. 10.8V nominal Vbat menas about 10.3V Vbulb... putting an FET in that can bring the switch resistance down to more like 0.05 means.. 10.64 Vbulb and the lumen output goes from 1047 up to 1169. The problem is.. with 0.05ohm ckt resistance and a starting Vbat of 12.7.. means a peak of 12.54Vbulb and that's 37% overdrive on the bulb vs 16%.. and though that 'maths out' to over 2000 lumen.. bulb life drops to 2.0 hrs.. which 'in reality' means most likely 2 seconds.. i don't trust bulb life rating lower than about 8 hrs.. that's when they start getting very finicky. (instaflash).

I'm looking for something 'dirt simple'.. like the nano driver.. 4 components! I'm working on converting the nano driver to hotwire driver.. using the same basic concept. I don't care if there is some variation, just want to hold back those ludicrisp startup voltages with fresh batteries so i can pare up bulbs that would normally not be possible in hosts thta will eat them for lunch w/o the ckt..

Example..

Typical 7.2V nominal soluition: 1111
Typical results: well.. people plug 7.2V into the WA calculator and say 'i have a 880 lumen light'... well.. here's what the 'realistic' calculator says...

A brief flash of about 1000-1100lumen when 8.4V is unleashed and about 7.66V gets to the bulb... that quickly drops to about 7.2Vbat = 6.66Vbulb and about 672 lumen (using stock PR bulb base).. with a bipin base using stock switch.. probably more like 737 lumen.. but with FET switch.. more like 806 lumen (average).

however.. if you just put a FET switch in and no LDO soft-start ckt... as much as 8.12V will get to the bulb initially.. and though the 1342 lumen does sound nice... 'til you see the 2.6hr estimated life.. because anything lower than about 7 really is a high-risk for instant flash.

LDO/FET combo with 7.2V and 1111 means.. dead-flat 806 lumen output for about 80% of the runtime.. dropping to about 730lumen over the last 20% of runtime... coincidentally just about what the non-FET solution has with bi-pin after about 5 minutes of use when the Vbat is about 7.2V.

(with stock switch, the 1111 should be good for about 600 lumen when the batteries are 90% depleted.. 82% the output).

Yup i've found some nice FETs.. TO-220 is the size i'm looking for and prices around $2.50 i'd like to find... the really low resistance FETs were more like $7. I'm hoping to make something that 'costs' less than $5 maybe can sell for $15-20 (depending on how much labor it takes) if i make them for people to drop in their own mag hotwires... but it will be totally 'open source' so people can make it themselves if they'd like.

-awr

Re: Hotwire LDO driver needed

andrewwynn said:

Like i said... a single IC 'voltage regulator' that is designed to just 'sense' a feedback voltage but control an external FET would be ideal.. just hook up a potentiometer from Vbulb to ground.. the wiper to the input and an FET (P channel i'm guessing) above the lamp.. the regulator monitors the voltage and turns up the gate if it's too low, simple as that... badda bing.

Click to expand...

I haven't dived deeply into your requirements but check out Micrel's MIC5156/5157/5158 series of Super LDO Regulator Controllers (http://www.micrel.com). They may be close to what you're looking for.

Re: Hotwire LDO driver needed

wow cool.. i found a similar part made by TI... but it has a gate output max of 9V... will be great for my lower voltage solutions but need something to run the 14V solutions.... looking closer at this model i think it's closer to a universal solution.. wow on closer examination looks like you saved the day so to speak (so i don't have to engineer two solutions).. the chip is bigger (14pin) than the other solution (8), but it's also cheaper and having ONE chip that will work at 6.6V, 7.2, 10.7 and 14.0V... is awesome.. because i have a light i would like to be able to change bulbs and just by flipping a switch i could convert from 6.6V to 7.2V.. too awesome.

I will be posting results of course.. the neat thing with this solution is that the FET past transistor is used as the on/off switch.. the 'mag' switch ends up just flipping the enable pin... the ckt draws like a micro amp when disabled.. stunning...

the Enable pin is 'high=on' which really makes wiring simple! just a resistor to pull it low when V not applied is all it takes and wire the mag switch to the EN pin.. the battery + terminal from the other side of the magswitch goes to both the Vcc on the chip and the Drain of the FET... than the lamp and back to ground.

Eureka! this is exactly what i was looking for.. i still would be interested in a 'simple' solution like a P-channel FET and a potentiometer to set the feedback.. no idea how much regulation that would have but if it can at least limit the maximum.. the other possibility i thought of is this:

using a zener to keep the voltage constant.. putting a zener in parallel with the bulb.. and somehow use that to control the gate of an FET... when the zener opens up at the set voltage it closes the FET gate .. just a thought.. really want a simple cheap solution like the likes of that.... maybe a zener for Vref.. a simple chip (no external devices).. that runs the gate of the FET..

anyhow thanks TONs that is just what i was looking for for my 'high end' solution. I hope to make some 'turnkey' switches like somebody can send me their magswitch and i'll upgrade it to fet-switch requlated to x, y voltage with a switch to select.

-awr

Re: Hotwire LDO driver needed

ps.. CPF ROCKS.. that took like 2 days.

Re: Hotwire LDO driver needed

Be sure to keep us up to date on what you come up with.

Hotwire LDO driver

oh how's this for up-to-date.

1) ordered the chips they are perfect and exactly what i was looking for..

2)..



that is the standard solution.. it would cost about $7 to build and give you absolutely dead-solid voltage output (1% or something).. using a pot to adjust the voltage means you could use different bulbs or 'back off' on the drive level if you want to make them last longer..

where to start?

C1, C2 are for the internal voltage boost ckt to push up the gate drive voltage higher than Vbat (the main problem with this kind of ckt is that to drive an FET the gate voltage has to be higher than the Source... but when the FET starts opening up suddenly Vsource is basically the same as Vbat.. this magic chip has a voltage tripler inside and will boost Vgate up to as much as 14V over Vbat.. the FET i picked wants at least 10V, i think it can handle 20, have to double check.

C3... it's where that gate voltage (output) is stored.. it takes a moment to charge that capacitor when you first turn on the ckt so it is the 'soft start' regulator.. if you want a slower start use a higher value (i'm going to put in a 10µF cap)... ps notice i can type µ now whoo hoo.

the EA is the feedback loop.. it has an internal Vref of 1.235V.. the voltage bridge of R2/R1 is set to make 1.235V when Vout is what you desire, the formula is on the schematic... the chip prefers an R2 of 10k, but i wanted to be able to set full range of 6.5V to 14.4 with just one pot, so 9.0 is substituted.

C4 is the output cap... probably completely unnecessary with this ckt.. the bulb will get 'pulses' but it won't care, the average will keep the voltage at whatever you set it to... i will probably try it with and w/o the cap.

'Flag' is very cool.. it monitors the output and if it drops below 92% of set voltage it will go TTL low... you can pipe that right back to the 'EN' enable pin which will shut the ckt down.. for this to work of course you need to have a resistor between Vbat and EN, that is R3.. R3 is not necessary without the switch.. EN is a very high impedance pin... the whole ckt pulls like 1 µA when disabled.

now.. 92% of full voltage that works like this.. I would maybe set my 1185 to be 10.7V (so that even with any connection losses at 'nominal' voltage of 10.8 on the battery i get full regulation.. now.. 10.7x92% = 9.844 which sounds awfully high but with NiMH they are 95+% exhausted at 1.1V/cell.. that is what i use for cutoff.. that's 9.9v.. so that's perfect... Lithium Ion.. they are 'stone cold dead' as far as useful output at 3.5V/cell.. and that is 10.5V..

It will be a bit weird to have the light 'just shut off' but i'm used to it most of my lights basically do that.. some voltage recovery will happen if you shut if off and you'll be able to turn it back on in a bit.

If you opt to not have that cut off ckt (which will prevent destruction of bat packs from over draining).. you just don't put in the wire from 'flag' or R3..

Here's where it gets interesting...

The chip ALSO includes over current protection.. you can for example.. pre-set a maximum current.. say for an 1185=3.35A.. it will automatically curb voltage if it's ever higher (i think it actually will work for a dead short.. duty cycle will just be like 0.01%).. in any event.. R4 is the optional 'sense resistor'.. and this is how it works..

The chip is constantly monitoring the 'drain' input to the FET and also the Vbat... if it 'sees' a difference greater than 35mV it will turn down the output 'til the output = 35mV. the formula is on the schematic for calculating the Rsense value.

I have the example for the mag85.... it wants 10.5miliohm resistance for that to work out.. I put some AWG mohm/ft values in-case it's possible to just use the lead wire.. or to make your own resistor since it's kind of hard to find a 10.5miliohm resistor... just take 4.884 inches of 24ga wire and bingo.. don't worry that it's pushing 3.35A.. it's supposed to drop voltage... that particular solution will generate 117mW on the bit of wire.. i would probably pot it so i can heat-sink it to the wall.. 1/5th of a watt is not a lot but it'll get warm and there is no air inside the light to cool it off.

A more interesting solution is that by total random coincidence.. the SWITCH inside a maglite is just about exactly 10.5 miliohm.. I've measured 11mohm w/o progold and 10mohm with it.... you can actually use the switch as your Rsense.. you loose about 20 lumen (1.5%) but it's really sneaky and clever.. oh.. you'll lose that same 1.5% using the 'bit of wire' trick... resistance is resistance.

3) here is the magswitch Rsense trick schematic:



The only difference is that the Vcc is no longer connected to the 'Drain' and goes directly to Vbat (the chip is always powered.. this goes before the switch)... the mag switch now actually hard-disconnects the FET power, just coincidentally since it's not necessary.. just using it that way to limit the current to the lamp..

Note that it still has the 'flag' ckt wired.. R3 and the wire between pin 3 and 14 is still optional (but recommended).

==============

Well there ya go.. one kick-butt linear voltage regulator for Hotwire.. designed to go around or through the mag-switch your choice (mine i will go around the switch.. I have some solutions that need more current than 3.35A.. and also want to take advantage of the 4miliohm FET i found!).. but for the mag85 solution.. this is the bomb with 'version b'.. because when the bulb is dead-cold i'm betting that the current spikes higher than 3.35A which would not be the case with this 'hack' and it would make the bulb last longer when pushed to the limit.

For me.. many of my bulb choices work ok with Direct Drive... but they are 'super hot' when i first turn them on.. and when the Vbat gets into 'nominal' there is a 15-18% drop in output... I want to keep the voltage of fresh batteries 'at bay' to make the bulbs last longer and have an average brightness that is absolutely constant from full to dead batteries but have that level be maybe 5-10% dimmer than the 'holy cow i'm melting my filament' brightness you get from fresh batteries and an 1185!

Thanks for the tip.. i already ordered the chips i should have a working version within a week.

I have some really awesome combinations lined up.. 800+lumen from 1111 in a 7.2V host.. 900+lumen from a 1160 in a 7.2V host.. and 900+lumen from an 1166 in a 14.4V host.. (the last two are just plain impossible without a voltage regulator because the voltage difference is too great).. the neat part is that DEAD battery voltage is higher than the Vbulb in BOTH of those cases... the light will stay in full regulation the ENTIRE life of the batteries.. a 900 lumen hotwire light that doesn't dim in the least and has soft-start and constant voltage and current protection so the bulb is never hammered by startup surges! I can not wait!

-awr

ps.. I still am trying to solve the problem with a more simplified design... an FET and a dual NPN or PNP transistor for an amp.. or using a 'feedback' amp which has a Vref built in.. or zener diode.. I want the 'po' man' Vreg.. it could be built from $3 in parts i know it.. and a lot easier to wire than a 14 pin IC as well!

if this test goes well i think i'll start offering turnkey switches for about $20-25 if i can find the time to make them.. you'd just need to tell me the voltage you want and which bulbs you wanted to run.

-awr2

Re: Hotwire LDO driver

Excellent! I'm glad the chip looks good for your design.
IMHO, the solution you have is pretty darn simple already. You could spend a lot of time trying to reduce costs a little bit and lose some of the great features of the MIC5156/5157/5158 chip too.

Unless you're like me and the searching/designing is actually the fun part.

Re: Hotwire LDO driver

Cant wait awr2!!

if you end actually making a run..

I'd like one please

You'll have to make sure you heatsink the hell out of that FET. You'll be dissipating several watts (5-6) initially, and even more during the startup pulse - with a current limit of 3.35A, initial power dissipation in the FET could be 30-40W until the bulb filament warms up

I'd say imposing a current limit on a bulb doesn't make too much sense.

Andrew,

Good job dude!. Thanks for all of your hard work

Like cy sad, if you end up building these for sale, please do let us know

Will

I do intend to make a 'drop in' for magD switch just swap the whole switch, probably for PR and for bi-pin.. probably with a 'core exchange' deal.. hope to be able to price it about $20 to $25 (similar to LED drivers). I think this will be the cat's meow for hotwire mag mods.. 1% regulation and auto-shutdown and having the possbility of current limiting to boot (a bit trickier.. might just set them to 4A or something.. i don't know what an 1185 takes at startup but those lower voltage high watt bulbs pull more than 3.7A running!

The capabilities are really just dependent on the FET and i found some winners.. the one i have in the queue can handle instant power dissipation of like 150+W and constant 70 or something (of course only by being heat-sinked to death).. but plan to heat-sink it directly to the side of the body of the light with arctic alumina.. it will be interesting to see if you can feel a 'hot spot' there. The p-channel FETs with single-digit RDson i've only found itty bitty ones that can handle like 2.5W, fortunately i don't need P-ch for this design... i found a 4miliohm FET that actually is an SMT device but heat-sinked can handle the loads mentioned above.

I ordered both DIP and SOP for the chip so i can build up the ckt nice and easy on my bread board, and build some 'deadbug' drivers from the SOP that will be easier to fit inside the light.

We have something very sneaky in the works, but sworn to secrecy, just stay tuned.

-awr

evan9162 said:

You'll have to make sure you heatsink the hell out of that FET. You'll be dissipating several watts (5-6) initially, and even more during the startup pulse - with a current limit of 3.35A, initial power dissipation in the FET could be 30-40W until the bulb filament warms up

I'd say imposing a current limit on a bulb doesn't make too much sense.

Click to expand...

not just 'warming up'.. some of the designs.. like running 1160 from 7.2V.. means like 3.75Ax (8.4-6.5) = 7.15W starting ... and 2.6W continuouis once the bats are warmed up.. the FET i picked are like good for 75+W. It will be very interesting to measure the startup voltage on the FET when set to current limit. How much current normally spikes through the likes of the 1185 or 1160 when first turned on and stone cold?

Yeah this is a pretty simple design, i think it'll fit in about a penny size ckt... but uses a pretty pricy IC... the raw parts cost is about $7-8... it could be done for $2-3 with some simple solution and 'do the basics'.

The current limit just happens to be 'built in' to the chip... it's a 33mV drop on a mag85 that is otherwise unnecessary so i will probably leave it out, unless in testing i find there is like a 5A spike at startup and in that case i'll leave it in... for reasons mentioned before.. bulb life and battery surge (not kicking in protection ckts)... the FET can handle a lot more power so 'why not' i figure.

-awr

Initially, the bulb will probably have 0.25 ohms or less of resistance - so you could see startup surges over 20-30A for the first few ms. You'll have to check the safe operating region of the FET to see if it can handle 150-200W pulses for a few ms.

The FET can only handle 75W when attached to a huge heat sink. You need to look up the specs for the thermal resistance junction-ambient (no heat sink), and the maximum power dissipation with no heat sinking. For surface mount packages it's something like 0.5W with no additional copper besides the minimum mounting surface, and for TO-220 it's usually like 1W with no heat sink attached to the tab.

Are you willing to reveal what FET you're planning on using? (unless I missed it in the above posts)