OK, chris, can you do this?Originally posted by philba:
If some one has a claim about a circuit - show the schematic and describe its function.
"Perfect" LED current limiting challenge
Re: "Perfect" LED current limiting challenge
Re: "Perfect" LED current limiting challenge
Hi Chris.
I'd suggest you take a look at the OSRAM datasheet yourself.
http://catalog.osram-os.com/media/_en/G ... 9735_0.pdf
Page 4, Radiant Intensity Ie in Axial Direction
Radiant Intensity at 100mA, time 20mS = 75 mW/sr
Radiant Intensity at 1A, time 100us = 400 mW/sr
Notice the difference in time and current.
In the 100mA case, 100mA * 20mS = 0.002 mA/S
In the 1A case (1A=1000mA), 1000mA * 100uS= 0.1mA/S
So we have a difference of 50 times the current going in per second, 0.1/.002=50.
So, if we have 50 times the current, we should be getting 50 times the mW in light output.
In the case at 100mA, we had 75mW/Sr. So, 75mW/Sr * 50 = 3750mW/Sr
You don't get anything even close to the 3750mW/SR at the 1A pulse level, but only 400mW/Sr.
(400mW/Sr)/(3750mw/Sr)= 0.1067 or 10.67% of the expected result, if the output intensity was linear.
Furthermore, we didn't account for the Vf of the LED going up considerably at the higher forward voltage. This would show an even a greater loss.
Where you might be getting quite confused here, is if you have an item like a camera. Some cameras will take a 1mS snapshot every 20mS. Here you would get to take all your power and concentrate it into the time the camera "sees" the image. I think this is where alot of folks get confused about all this stuff. The human eye doesn't take short snapshots, but integrates light over time, which is an entirely different ball of wax, all together.
Page 5 also shows a Forward voltage vs. current, at a 20uS pulse width. Notice how the Vf rises to 3V at 100A pulses. W=V*A, 3V*100=300W With 0.1A pulses, the Vf is 1V. 1*0.1=0.1W
If we were to re-do the calculations above, using the power consumed, you'd find there are even greater losses
Radiant Intensity at 100mA, time 20mS = 75 mW/sr
Radiant Intensity at 1A, time 100us = 400 mW/sr
At 100mA, the Vf is 1V. 100mA*1V=0.1W We are doing this for 20mS, we have 0.002.
At 1000mA (1A), the Vf is 1.2V. 1000mA*1.2V=1200mW but we do this for 100uS , we have 0.12.
0.12/0.002 = 60 -so we should have 60x the output if the part was linear.
75mW/Sr * 60 = 4500 mW/Sr -but we only have 400mW/Sr, or 8.9% of the expected output.
The sitation gets much worse at 100A currents, as the Vf goes even higher.
Again, folks often get quite confused about the camera thing, and LEDs, were you can pulse the power just during the shutter time, which increases the output, as you only produce the light during the time the camera "sees". An example
http://homepages.inf.ed.ac.uk/rbf/CVonl ... ghting.pdf
I'd suggest you take a look at the OSRAM datasheet yourself.
http://catalog.osram-os.com/media/_en/G ... 9735_0.pdf
Page 4, Radiant Intensity Ie in Axial Direction
Radiant Intensity at 100mA, time 20mS = 75 mW/sr
Radiant Intensity at 1A, time 100us = 400 mW/sr
Notice the difference in time and current.
In the 100mA case, 100mA * 20mS = 0.002 mA/S
In the 1A case (1A=1000mA), 1000mA * 100uS= 0.1mA/S
So we have a difference of 50 times the current going in per second, 0.1/.002=50.
So, if we have 50 times the current, we should be getting 50 times the mW in light output.
In the case at 100mA, we had 75mW/Sr. So, 75mW/Sr * 50 = 3750mW/Sr
You don't get anything even close to the 3750mW/SR at the 1A pulse level, but only 400mW/Sr.
(400mW/Sr)/(3750mw/Sr)= 0.1067 or 10.67% of the expected result, if the output intensity was linear.
Furthermore, we didn't account for the Vf of the LED going up considerably at the higher forward voltage. This would show an even a greater loss.
Where you might be getting quite confused here, is if you have an item like a camera. Some cameras will take a 1mS snapshot every 20mS. Here you would get to take all your power and concentrate it into the time the camera "sees" the image. I think this is where alot of folks get confused about all this stuff. The human eye doesn't take short snapshots, but integrates light over time, which is an entirely different ball of wax, all together.
Page 5 also shows a Forward voltage vs. current, at a 20uS pulse width. Notice how the Vf rises to 3V at 100A pulses. W=V*A, 3V*100=300W With 0.1A pulses, the Vf is 1V. 1*0.1=0.1W
If we were to re-do the calculations above, using the power consumed, you'd find there are even greater losses
Radiant Intensity at 100mA, time 20mS = 75 mW/sr
Radiant Intensity at 1A, time 100us = 400 mW/sr
At 100mA, the Vf is 1V. 100mA*1V=0.1W We are doing this for 20mS, we have 0.002.
At 1000mA (1A), the Vf is 1.2V. 1000mA*1.2V=1200mW but we do this for 100uS , we have 0.12.
0.12/0.002 = 60 -so we should have 60x the output if the part was linear.
75mW/Sr * 60 = 4500 mW/Sr -but we only have 400mW/Sr, or 8.9% of the expected output.
The sitation gets much worse at 100A currents, as the Vf goes even higher.
Again, folks often get quite confused about the camera thing, and LEDs, were you can pulse the power just during the shutter time, which increases the output, as you only produce the light during the time the camera "sees". An example
http://homepages.inf.ed.ac.uk/rbf/CVonl ... ghting.pdf
-
Chris Smith
- Posts: 4325
- Joined: Tue Dec 04, 2001 1:01 am
- Location: Bieber Ca.
Re: "Perfect" LED current limiting challenge
Philba, your eyes closed again?
Already posted the circuit several posts back?
******************
The Osram post had nothing to do with Leds in a plasma state or the original perfect Led post.
The osram runs in the slow lane where as the 10 amp pulser is in the Nano second light speed lane. Dont confuse the two.
People are confusing TWO [actually three] issues here, One the original request for a "perfect" led, and my after post[s] of Cranking out high amperages into a led, as they do exist and have always existed.
Don’t confuse the two, I made no comments as to the latter other than "It is not absurd" that a led can handle one amp, or 100 amps and that they have been doing this since the 60s.
That’s the only correlation between these two posts.
Some here want chicken little to explain to you that it cant be done, has never been done, or its absurd to think anything past the “manufactures data sheet” is what a LED can handle. The Osram post shows that is a crock, [1 amp vs the standard DC current] and so does my experimentation from the 80s along with a driver that can do the job.
<small>[ January 01, 2006, 03:55 PM: Message edited by: Chris Smith ]</small>
Already posted the circuit several posts back?
******************
The Osram post had nothing to do with Leds in a plasma state or the original perfect Led post.
The osram runs in the slow lane where as the 10 amp pulser is in the Nano second light speed lane. Dont confuse the two.
People are confusing TWO [actually three] issues here, One the original request for a "perfect" led, and my after post[s] of Cranking out high amperages into a led, as they do exist and have always existed.
Don’t confuse the two, I made no comments as to the latter other than "It is not absurd" that a led can handle one amp, or 100 amps and that they have been doing this since the 60s.
That’s the only correlation between these two posts.
Some here want chicken little to explain to you that it cant be done, has never been done, or its absurd to think anything past the “manufactures data sheet” is what a LED can handle. The Osram post shows that is a crock, [1 amp vs the standard DC current] and so does my experimentation from the 80s along with a driver that can do the job.
<small>[ January 01, 2006, 03:55 PM: Message edited by: Chris Smith ]</small>
Re: "Perfect" LED current limiting challenge
Ian, you might try looking at this buck switcher, with integrated MOSFETs, and this buck converter can go all the way to 100% duty cycle:
http://focus.ti.com/lit/ds/symlink/tps62050.pdf
Figure 4 sounds the closest to your application (I assume you'd use the adjustable version-this one is fixed, but will give us decent numbers to start with), which at the 20mA output with 3.5-6Vin you were talking about, should put you in the 90-95% efficiency range.
You could monitor the LED current with a INA193 ( http://focus.ti.com/lit/ds/symlink/ina193.pdf ) series part and disconnect the current path, or rig up a constant current source with a low on resistance MOSFET, creating a LDO that has less than 0.1V dropout (and you can make an CC LDO with much lower dropout than 0.1V). Setting your LDO CC source for 30mA, would mean that it is fully on, unless the DC-DC hiccups, and then the CC circuit would limit the current.
Another approach would be to set the DC to DC to the maximum Vf the LEDs will ever have, plus the less than 0.1V CC LDO, and set the LDO such that it only allows 20mA to flow.
You could also do a CC DC to DC switcher, but then you'd have no protection in the event of a failure. A hybrid would be a CC DC to DC, and putting the less than 0.1V drop CC LDO in the path. This would protect in the event of some failure.
As I understand it here, Ian has to run off a 6V source. As Chris mentioned, he'd be burning up 44% of his power if he were just to resistor or LDO drop the power. Yes it is simple, but sometimes battery life is very important. Also, he probably wants a regulated current, which a simple brain dead resistor setup would not provide.
One advantage, as Ian mentioned, for the DC-to-DC approach is that with a 6V input, and at the spec'd 90% efficiency off the datasheet, he'd be pulling about 13mA off the cells.
One of the drawbacks of the DC-to-DC Converter, is it's complexity. But with this converter, you can go clear to 100% duty cycle, which allows you to run the input power right on down to the LED Vf (as long as your battery pack will run down that low). Fortunately, the complexity becomes much simpler, once you integrate most the components within the chip.
This particular switcher chip will tolerate up to 10V on it's inputs.
You may want to also look at:
http://focus.ti.com/docs/prod/folders/p ... 62100.html
This series is available in a variety of switching frequencies, depending on whether space or efficiency is your primary desire.
TI, Linear, Maxim all have parts that can do simular functions, and you'll find some great application notes for using them for driving LEDs. Oh, and of course, so does National, Sipex, Semtech, On-Semi, Micrel, and others.
Micrel's site:
http://www.micrel.com/page.do?page=prod ... vers.shtml
Semtech (put LED in for search term):
http://www.semtech.com/search/searchResults.do
Sipex:
http://www.sipex.com/content.aspx?p=power_app_notes
<small>[ January 01, 2006, 08:23 PM: Message edited by: Jarhead ]</small>
http://focus.ti.com/lit/ds/symlink/tps62050.pdf
Figure 4 sounds the closest to your application (I assume you'd use the adjustable version-this one is fixed, but will give us decent numbers to start with), which at the 20mA output with 3.5-6Vin you were talking about, should put you in the 90-95% efficiency range.
You could monitor the LED current with a INA193 ( http://focus.ti.com/lit/ds/symlink/ina193.pdf ) series part and disconnect the current path, or rig up a constant current source with a low on resistance MOSFET, creating a LDO that has less than 0.1V dropout (and you can make an CC LDO with much lower dropout than 0.1V). Setting your LDO CC source for 30mA, would mean that it is fully on, unless the DC-DC hiccups, and then the CC circuit would limit the current.
Another approach would be to set the DC to DC to the maximum Vf the LEDs will ever have, plus the less than 0.1V CC LDO, and set the LDO such that it only allows 20mA to flow.
You could also do a CC DC to DC switcher, but then you'd have no protection in the event of a failure. A hybrid would be a CC DC to DC, and putting the less than 0.1V drop CC LDO in the path. This would protect in the event of some failure.
As I understand it here, Ian has to run off a 6V source. As Chris mentioned, he'd be burning up 44% of his power if he were just to resistor or LDO drop the power. Yes it is simple, but sometimes battery life is very important. Also, he probably wants a regulated current, which a simple brain dead resistor setup would not provide.
One advantage, as Ian mentioned, for the DC-to-DC approach is that with a 6V input, and at the spec'd 90% efficiency off the datasheet, he'd be pulling about 13mA off the cells.
One of the drawbacks of the DC-to-DC Converter, is it's complexity. But with this converter, you can go clear to 100% duty cycle, which allows you to run the input power right on down to the LED Vf (as long as your battery pack will run down that low). Fortunately, the complexity becomes much simpler, once you integrate most the components within the chip.
This particular switcher chip will tolerate up to 10V on it's inputs.
You may want to also look at:
http://focus.ti.com/docs/prod/folders/p ... 62100.html
This series is available in a variety of switching frequencies, depending on whether space or efficiency is your primary desire.
TI, Linear, Maxim all have parts that can do simular functions, and you'll find some great application notes for using them for driving LEDs. Oh, and of course, so does National, Sipex, Semtech, On-Semi, Micrel, and others.
Micrel's site:
http://www.micrel.com/page.do?page=prod ... vers.shtml
Semtech (put LED in for search term):
http://www.semtech.com/search/searchResults.do
Sipex:
http://www.sipex.com/content.aspx?p=power_app_notes
<small>[ January 01, 2006, 08:23 PM: Message edited by: Jarhead ]</small>
Re: "Perfect" LED current limiting challenge
Funny, that has a problem in conjunction with this discussion. on the mosfet and the diode, there are no datasheets to be googled for - just general specs. The little info I could find on the laser diode said it is stud mounted and thus is capable of dissipating significantly higher heat levels than a simple LED. You claim its the same but it's really hard believe that's an apples to apples comparison. A typical stud package like the to48 can dissipate something like 100 watts with a decent heat sink. A standard 5 mm LED is typically rated at 150 mW. A wee bit of difference, ne c'est pas? Also, I could find very little usefull info on the mosfet - no Rds, no switching times for example. To work in that circuit, the mosfet needs to have an Rds of < 900 mOhms and switching times of less than 10 nS. Pretty amazing specs for a really old mosfet. Especially since we are talking about some reasonably high frequencies. Its kind of odd that the circuit you chose uses parts with scant detail.Originally posted by Chris Smith:
Philba, your eyes closed again?
Already posted the circuit several posts back?
I'd like to see a circuit that you claim will directly drive a simple LED in a 5mm package to the at least the 10A level. The one you posted doesn't do it. You made claim of much higher current, I'd also like to see that.
Re: "Perfect" LED current limiting challenge
Both the VN64GA and the SG2002 are very obscure devices.
The VN64AG was made by Siliconix, which is now part of Vishay. It was one of a series of MOSFETs made by Siliconix where the channel was placed on the sides of a "V" shaped groove in the silicon. This geometry eventually lost out to the planar "HEXFET" structure developed by International Rectifier. It was probably harder to produce and possibly more expensive. It did have one interesting feature, however. The drain-to-gate capacitance was exceptionally low, which made it easier to drive at short rise times or high frequencies. The package was a TO-3, which is also somewhat unusual for this type of transistor. This device was dropped from the 1985 Siliconix data book, which probably indicates that it had been discontinued by that time.
With high gate drive (15 volts), the drain voltage was probably about 2 volts at 10 amps, indicating a channel resistance of about .2 ohms. The typical curves indicate a saturation voltage of 2.5 volts with 10 volt gate drive.
At 10 amps, the total circuit voltage drop would be about 10 volts, leaving 5 volts across the laser diode.
The SG2002 laser diode is also a rarity. RCA made a few laser diodes, but not very many. They were usually in a copper stud mount case. The ungrounded diode lead came through the middle of the stud. Power dissipation of this package may have been in the 10 watt range, far above that of most contemporary LED and laser diodes. The threshold current could easily have been in the 5 amp range or higher. The 5 volt diode drop would indicate that the internal series resistance was probably under .5 ohms. It also indicates that the input power was around 50 watts, which would preclude CW operation.
The Osram diode mentioned, the SFH 4301, would only get a pulse of about 5 amps in this circuit, since its series resistance is substantially higher, at about 1.8 ohws. The allowable pulse width would also be narrower, possibly down to the 10 microsecond range.
Chris misses the basic point. The objective that Ian suggested was to maximize battery life. This has to be done by maximizing the efficiency of the LED. Pulsed operation of present LEDs does not lead to increased efficiency. On the contrary, the power required for the same total light output increases substantially. Therefore, pulsed operation is not a viable solution to the question that Ian posed.
The VN64AG was made by Siliconix, which is now part of Vishay. It was one of a series of MOSFETs made by Siliconix where the channel was placed on the sides of a "V" shaped groove in the silicon. This geometry eventually lost out to the planar "HEXFET" structure developed by International Rectifier. It was probably harder to produce and possibly more expensive. It did have one interesting feature, however. The drain-to-gate capacitance was exceptionally low, which made it easier to drive at short rise times or high frequencies. The package was a TO-3, which is also somewhat unusual for this type of transistor. This device was dropped from the 1985 Siliconix data book, which probably indicates that it had been discontinued by that time.
With high gate drive (15 volts), the drain voltage was probably about 2 volts at 10 amps, indicating a channel resistance of about .2 ohms. The typical curves indicate a saturation voltage of 2.5 volts with 10 volt gate drive.
At 10 amps, the total circuit voltage drop would be about 10 volts, leaving 5 volts across the laser diode.
The SG2002 laser diode is also a rarity. RCA made a few laser diodes, but not very many. They were usually in a copper stud mount case. The ungrounded diode lead came through the middle of the stud. Power dissipation of this package may have been in the 10 watt range, far above that of most contemporary LED and laser diodes. The threshold current could easily have been in the 5 amp range or higher. The 5 volt diode drop would indicate that the internal series resistance was probably under .5 ohms. It also indicates that the input power was around 50 watts, which would preclude CW operation.
The Osram diode mentioned, the SFH 4301, would only get a pulse of about 5 amps in this circuit, since its series resistance is substantially higher, at about 1.8 ohws. The allowable pulse width would also be narrower, possibly down to the 10 microsecond range.
Chris misses the basic point. The objective that Ian suggested was to maximize battery life. This has to be done by maximizing the efficiency of the LED. Pulsed operation of present LEDs does not lead to increased efficiency. On the contrary, the power required for the same total light output increases substantially. Therefore, pulsed operation is not a viable solution to the question that Ian posed.
Re: "Perfect" LED current limiting challenge
I see the reference to 511 on the schematic in pencil, which would be the IRF511. The two transistors written in pencil on there are 123 (2N2222A sub) and 159(forget which one that is), both made by ECG and NTE. NTE123 and NTE159.
IRF511
One link:
http://www.uoguelph.ca/~antoon/gadgets/irf511.htm
ELECTRICAL CHARACTERISTICS (All types unless otherwise stated)
---------------------------------------------------------------------------
Igss Gate-source leakage 500nA (forward) -500nA (reverse)
Idss Drain current, Vg=0v 250uA
Id-on On state drain current 5.6A IFR510, IFR511
4.9A IFR512, IFR513
Rds-on Drain-source "on" Res. 0.4-0.54 ohms (device ON resistance)
Cis Input capacitance 135pF (at Vds=12v, Cis=180pF)
Cos Output capacitance 80pF (at Vds=12v, Cos=130pF)
Td-on Turn-on delay time 8-11nS ) These parameters define
Tr Rise time 25-36nS ) how fast the MOSFET turns
Td-off Turn-off delay time 15-21nS ) on and off when gate is
Tf Fall time 12-21nS ) driven with a square wave
Vsd Diode forward voltage 2.5v (dropped across the source-drain
due to the internal diode)
http://us.geocities.com/radio_qrp/index.html
A part very close to the IRF511 is the IRF510 (which is a tad faster):
http://www.dhmicro.com/Datasheet/irf510.pdf
Here is a full IRF511 datasheet:
http://www.molalla.net/~leeper/irf511.pdf
Same part from another vendor:
http://www.molalla.net/~leeper/irf511_f.pdf
<small>[ January 02, 2006, 12:19 AM: Message edited by: Jarhead ]</small>
IRF511
One link:
http://www.uoguelph.ca/~antoon/gadgets/irf511.htm
ELECTRICAL CHARACTERISTICS (All types unless otherwise stated)
---------------------------------------------------------------------------
Igss Gate-source leakage 500nA (forward) -500nA (reverse)
Idss Drain current, Vg=0v 250uA
Id-on On state drain current 5.6A IFR510, IFR511
4.9A IFR512, IFR513
Rds-on Drain-source "on" Res. 0.4-0.54 ohms (device ON resistance)
Cis Input capacitance 135pF (at Vds=12v, Cis=180pF)
Cos Output capacitance 80pF (at Vds=12v, Cos=130pF)
Td-on Turn-on delay time 8-11nS ) These parameters define
Tr Rise time 25-36nS ) how fast the MOSFET turns
Td-off Turn-off delay time 15-21nS ) on and off when gate is
Tf Fall time 12-21nS ) driven with a square wave
Vsd Diode forward voltage 2.5v (dropped across the source-drain
due to the internal diode)
http://us.geocities.com/radio_qrp/index.html
A part very close to the IRF511 is the IRF510 (which is a tad faster):
http://www.dhmicro.com/Datasheet/irf510.pdf
Here is a full IRF511 datasheet:
http://www.molalla.net/~leeper/irf511.pdf
Same part from another vendor:
http://www.molalla.net/~leeper/irf511_f.pdf
<small>[ January 02, 2006, 12:19 AM: Message edited by: Jarhead ]</small>
Re: "Perfect" LED current limiting challenge
After all the investigation I'm thinking this concept may not be possible, at least for the configuration I'm using.
Using charge pump controllers..........
These devices seem to be rated on total current rather than total power. I see "95%" efficiency rating based on driving a 20mA LED using only 22mA. Besides, they want to boost voltage rather than buck it.
Using buck converters.........
Buck converters shut down and drive at 100% duty cycle when the dropout voltage gets close. In addition buck converters have a dropout voltage where no current will flow. Using a common configuration of 4 D cells at 6.4 volts and a white LED with a possible Vfo of 3.4V seems to be an efficiency nightmare.......
If the converter shuts down to 100% duty cycle at 2V that means I only get buck operation between 6.4 and 5.4 volts. In addition, if the dopout voltage of the converter is 1V I lose ALL battery power below a total battery voltage of 4.4V. These numbers are a bit rough, and maybe I could squeeze out a few more 100mV here and there but that's where things seem to be......
Battery V. LED V. V Shutdown V. DropoutV
6.4--------3.4----3.0----Buck operation
5.4--------3.4----2.0---No buck
4.4--------3.4----1.0------------- No current
Since there's still a bit of power in the battery after 4.4 V I'm not sure there's any good gonfiguration using a buck converter.
Direct LED drive.........
I thought to switch the LED through an inductor at 30mA but I don't think there's enough in the inductor at this low a current to be useful.
So far it looks like there are ways to efficiently drive a LED with a battery, but not a white LED with 4 standard alkaline cells.
<small>[ January 02, 2006, 07:54 AM: Message edited by: ian ]</small>
Using charge pump controllers..........
These devices seem to be rated on total current rather than total power. I see "95%" efficiency rating based on driving a 20mA LED using only 22mA. Besides, they want to boost voltage rather than buck it.
Using buck converters.........
Buck converters shut down and drive at 100% duty cycle when the dropout voltage gets close. In addition buck converters have a dropout voltage where no current will flow. Using a common configuration of 4 D cells at 6.4 volts and a white LED with a possible Vfo of 3.4V seems to be an efficiency nightmare.......
If the converter shuts down to 100% duty cycle at 2V that means I only get buck operation between 6.4 and 5.4 volts. In addition, if the dopout voltage of the converter is 1V I lose ALL battery power below a total battery voltage of 4.4V. These numbers are a bit rough, and maybe I could squeeze out a few more 100mV here and there but that's where things seem to be......
Battery V. LED V. V Shutdown V. DropoutV
6.4--------3.4----3.0----Buck operation
5.4--------3.4----2.0---No buck
4.4--------3.4----1.0------------- No current
Since there's still a bit of power in the battery after 4.4 V I'm not sure there's any good gonfiguration using a buck converter.
Direct LED drive.........
I thought to switch the LED through an inductor at 30mA but I don't think there's enough in the inductor at this low a current to be useful.
So far it looks like there are ways to efficiently drive a LED with a battery, but not a white LED with 4 standard alkaline cells.
<small>[ January 02, 2006, 07:54 AM: Message edited by: ian ]</small>
Re: "Perfect" LED current limiting challenge
Hello again,
Thanks to Jarhead for all the great links! Very
interesting and informative.
Chris:
I'd have to agree a little in that i'd like to
see a higher current circuit too...
say a 50 amp pulse at least?
Ian:
I think you missed one of my posts which stated that you should be able to
drive a series inductor to 'make your own' buck. You would drive a low
Rs N-Mosfet with your controller ic which would switch the inductor. A catch
diode (Schottky) would also be required. The output of the circuit feeds
the LED and you monitor voltage across the LED with a simple filter circuit.
This feedback tells your controller what to do.
This effectively gives you a zero dropout buck regulator, so that when the
voltage goes down to 4 volts you should still be able to drive a 3.5v LED.
You didnt specify how accurate you needed this circuit however, in that if
the LED brightness changes by just a little bit over battery life does
that matter or not? What is the LED doing--just an indicator or some sort
of photometric application? It's important for you to tell us this so we
know how accurate you need this circuit.
Take care,
Al
Thanks to Jarhead for all the great links! Very
interesting and informative.
Chris:
I'd have to agree a little in that i'd like to
see a higher current circuit too...
say a 50 amp pulse at least?
Ian:
I think you missed one of my posts which stated that you should be able to
drive a series inductor to 'make your own' buck. You would drive a low
Rs N-Mosfet with your controller ic which would switch the inductor. A catch
diode (Schottky) would also be required. The output of the circuit feeds
the LED and you monitor voltage across the LED with a simple filter circuit.
This feedback tells your controller what to do.
This effectively gives you a zero dropout buck regulator, so that when the
voltage goes down to 4 volts you should still be able to drive a 3.5v LED.
You didnt specify how accurate you needed this circuit however, in that if
the LED brightness changes by just a little bit over battery life does
that matter or not? What is the LED doing--just an indicator or some sort
of photometric application? It's important for you to tell us this so we
know how accurate you need this circuit.
Take care,
Al
LEDs vs Bulbs, LEDs are winning.
Re: "Perfect" LED current limiting challenge
Ian,
Did you read the post where I mentioned this buck chip?
http://focus.ti.com/lit/ds/symlink/tps62050.pdf
Vin(min) = Vout(max) + Iout(max) * ((Rds(on)(max) + Rinductor))
Using that formula, the circuit will buck all the way down to 3.400328V, at which point it will enter 100% duty cycle. Assuming your LED Vf is 3.4V It can also take up to 10V in.
Using an inductor with as low of a resistance as possible, like the Sumida CDRH127/LD or lower will allow it to buck as close as possible to your Vf.
http://www.sumida.com/en/products/pdf/CDRH127.pdf
One bonus of this converter is that it has a battery voltage lockout, that you can set to whatever point you want, so that you can avoid over discharging the battery back.
There is also an enable input that will allow you to shutdown the converter if you want.
Your comment on Buck converters seem oriented towards a certain buck converter, not all buck converters are the same.
Many buck converters will not go to 100% duty cycle. There are plenty of them that will only go to 85% duty cycle. In this case it would only buck down to 4V, at which point it would go into linear mode. Linear Tech makes lots of parts that have this issue. Some of the parts won't even switch to linear mode, but just go down to 85%, and just stick with that, and your output voltage falls.
The TI part I mentioned is not like that.
TI offers free samples, so you could just go order one and try it out.
Did you read the post where I mentioned this buck chip?
http://focus.ti.com/lit/ds/symlink/tps62050.pdf
Vin(min) = Vout(max) + Iout(max) * ((Rds(on)(max) + Rinductor))
Using that formula, the circuit will buck all the way down to 3.400328V, at which point it will enter 100% duty cycle. Assuming your LED Vf is 3.4V It can also take up to 10V in.
Using an inductor with as low of a resistance as possible, like the Sumida CDRH127/LD or lower will allow it to buck as close as possible to your Vf.
http://www.sumida.com/en/products/pdf/CDRH127.pdf
One bonus of this converter is that it has a battery voltage lockout, that you can set to whatever point you want, so that you can avoid over discharging the battery back.
There is also an enable input that will allow you to shutdown the converter if you want.
Your comment on Buck converters seem oriented towards a certain buck converter, not all buck converters are the same.
Many buck converters will not go to 100% duty cycle. There are plenty of them that will only go to 85% duty cycle. In this case it would only buck down to 4V, at which point it would go into linear mode. Linear Tech makes lots of parts that have this issue. Some of the parts won't even switch to linear mode, but just go down to 85%, and just stick with that, and your output voltage falls.
The TI part I mentioned is not like that.
TI offers free samples, so you could just go order one and try it out.
Re: "Perfect" LED current limiting challenge
sorry. double post. sigh.
<small>[ January 02, 2006, 07:20 PM: Message edited by: philba ]</small>
<small>[ January 02, 2006, 07:20 PM: Message edited by: philba ]</small>
Re: "Perfect" LED current limiting challenge
The IRF511 won't do it. It has 20 nS rise and 20 nS fall times and also has 25 nS turn on and turn off delays. I think the best that circuit could do is on the order of 50 nS pulses with a really ugly pulse shape.Originally posted by Jarhead:
I see the reference to 511 on the schematic in pencil, which would be the IRF511. The two transistors written in pencil on there are 123 (2N2222A sub) and 159(forget which one that is), both made by ECG and NTE. NTE123 and NTE159.
IRF511
-
Chris Smith
- Posts: 4325
- Joined: Tue Dec 04, 2001 1:01 am
- Location: Bieber Ca.
Re: "Perfect" LED current limiting challenge
Well, philba, I guess you have to argue with the author?
Im sure all the books you have written on the subject, trumps Mr Rudolf F Graf?
Good luck knocking your head against that wall?
I suggest you spend less time trying to disprove old history, which leaves you with more time for your own education, ...Ehh?
What there's a thought,... learning,...instead of arguing with [old] history?
Considering My Little Green Led [of 20 ma] didn’t even have a Metal case to dissipate that terrible heat build up and it didn’t even burn out at 18 volts in, which means it ran at slightly more than 10 amps.
Gee, From what I hear,.... I think the metal case on the laser is for stabilization over time and temperature, what do you think philba?
I hear laser diodes drift, lose their signal, and the phone lines go down if they over heat or even get past warm?
Ive even heard that some run TEC coolers so that they dont drift but a few nano meters off wavelength as they operate? Something about multi mode transmissions? Can you confirm this, phil?
I hear they even frequency shift and shut down the fiber optics if they get warm, let alone hot?
I guess you would know, All that experience you have on the subject and all, Going back to the 80s or perhaps earlier?
Something about modulating at GHZ and temperature/ frequency drift of the laser? Only takes a few extra nanos to screw up your day.
Were you confusing this with absolute temperature rise or even mistaking that with a 3 cent LED and then worrying if the frequency is a little off color or what ever?
IS it still bright, is all you have to really ask.
Let us know how you fare? Im sure we all can learn from your personal experience on this subject?
Mr Al, I have already sent you some extrapolations [I think i sent you a copy?] that can send you to the next level.
Improve your FET, and improve your timing circuits. There are plenty of FETs out there that surpass the standard over the counter parts like the IRF 511, and they cost a lot more, but they aren’t for the faint of wallet.
The EEM data books are full of them.
You can use such things as tunnel diodes, microwave diodes etc to trigger the base of the final transistor, but you have to know how to couple part A with part B, and get it right with out too much ringing or impedance mismatch.
The basics are in that diagram, if you want more you can up the voltage as per diagram, or replace just the FET with a better spec, and then up the amps by using 100 or more volts.
BUT,....Every time you tweak one figure, You must consider all the others.
OR, you can simply buy a circuit for $10,000 and be done with it?
I can look up the part number for that?
This is a experimental board.
I can lead you to water, but I cant make you drink.
Just In case you want to purchase a Nice 500 amp pulser here is where you can empty your wallet.
One of many vendors......Just type in "high amp laser pulsers" and stand back.
http://www.avtechpulse.com/
The price list is here:
http://www.avtechpulse.com/prices/
<small>[ January 02, 2006, 10:17 PM: Message edited by: Chris Smith ]</small>
Im sure all the books you have written on the subject, trumps Mr Rudolf F Graf?
Good luck knocking your head against that wall?
I suggest you spend less time trying to disprove old history, which leaves you with more time for your own education, ...Ehh?
What there's a thought,... learning,...instead of arguing with [old] history?
Considering My Little Green Led [of 20 ma] didn’t even have a Metal case to dissipate that terrible heat build up and it didn’t even burn out at 18 volts in, which means it ran at slightly more than 10 amps.
Gee, From what I hear,.... I think the metal case on the laser is for stabilization over time and temperature, what do you think philba?
I hear laser diodes drift, lose their signal, and the phone lines go down if they over heat or even get past warm?
Ive even heard that some run TEC coolers so that they dont drift but a few nano meters off wavelength as they operate? Something about multi mode transmissions? Can you confirm this, phil?
I hear they even frequency shift and shut down the fiber optics if they get warm, let alone hot?
I guess you would know, All that experience you have on the subject and all, Going back to the 80s or perhaps earlier?
Something about modulating at GHZ and temperature/ frequency drift of the laser? Only takes a few extra nanos to screw up your day.
Were you confusing this with absolute temperature rise or even mistaking that with a 3 cent LED and then worrying if the frequency is a little off color or what ever?
IS it still bright, is all you have to really ask.
Let us know how you fare? Im sure we all can learn from your personal experience on this subject?
Mr Al, I have already sent you some extrapolations [I think i sent you a copy?] that can send you to the next level.
Improve your FET, and improve your timing circuits. There are plenty of FETs out there that surpass the standard over the counter parts like the IRF 511, and they cost a lot more, but they aren’t for the faint of wallet.
The EEM data books are full of them.
You can use such things as tunnel diodes, microwave diodes etc to trigger the base of the final transistor, but you have to know how to couple part A with part B, and get it right with out too much ringing or impedance mismatch.
The basics are in that diagram, if you want more you can up the voltage as per diagram, or replace just the FET with a better spec, and then up the amps by using 100 or more volts.
BUT,....Every time you tweak one figure, You must consider all the others.
OR, you can simply buy a circuit for $10,000 and be done with it?
I can look up the part number for that?
This is a experimental board.
I can lead you to water, but I cant make you drink.
Just In case you want to purchase a Nice 500 amp pulser here is where you can empty your wallet.
One of many vendors......Just type in "high amp laser pulsers" and stand back.
http://www.avtechpulse.com/
The price list is here:
http://www.avtechpulse.com/prices/
<small>[ January 02, 2006, 10:17 PM: Message edited by: Chris Smith ]</small>
Re: "Perfect" LED current limiting challenge
The circuit that Rudolpf Graf reprinted was from a 1983 Siliconix data book. He quite properly cited his sources in a separate table in the back of the book. The Siliconix book, in turn, cited a 1979 Siliconix ap note by Dave Hoffman, which includes several other drive circuits of increasing performance and complexity.
The VN64AG was out of production by the time Graf published this circuit in 1985. The IRF511 is not a particularly close equivalent. It has 1/4 the power dissipation capability and 2 times the channel resistance of the original part.
The green 5 mm LEDs sold by Hewlett Packard in the early 1980's (such as the HLMP-3519) show a series resistance of about 16 ohms. With this driver circuit, the pulse current would be limited to less than 1 amp. The maximum rated DC current was 30 milliamps, with 90 milliamps allowed under pulsed conditions. This appears to be a conservative rating, since the data sheet allows a pulse of 90 milliamps for up to 2 milliseconds.
I note with amusement that the 500 amp, $10,000 pulse generator mentioned by Chris would only drive this diode to about 3 amps without a matching network (1 amp with a matched, 50 ohm transmission line).
As a historical note, the radar transmitters at the end of World War II were pulsed magnetrons capable of peak powers of several hundred kilowatts. The duty cycle was typically less than .1 percent, making the average power a few hundred watts. By the late 1950's, CW magnetrons at about the same power level were being used in microwave ovens. The average microwave oven today puts out as much energy as the average radar transmitter in World War II at a small fraction of the cost, and several times the lifetime.
The VN64AG was out of production by the time Graf published this circuit in 1985. The IRF511 is not a particularly close equivalent. It has 1/4 the power dissipation capability and 2 times the channel resistance of the original part.
The green 5 mm LEDs sold by Hewlett Packard in the early 1980's (such as the HLMP-3519) show a series resistance of about 16 ohms. With this driver circuit, the pulse current would be limited to less than 1 amp. The maximum rated DC current was 30 milliamps, with 90 milliamps allowed under pulsed conditions. This appears to be a conservative rating, since the data sheet allows a pulse of 90 milliamps for up to 2 milliseconds.
I note with amusement that the 500 amp, $10,000 pulse generator mentioned by Chris would only drive this diode to about 3 amps without a matching network (1 amp with a matched, 50 ohm transmission line).
As a historical note, the radar transmitters at the end of World War II were pulsed magnetrons capable of peak powers of several hundred kilowatts. The duty cycle was typically less than .1 percent, making the average power a few hundred watts. By the late 1950's, CW magnetrons at about the same power level were being used in microwave ovens. The average microwave oven today puts out as much energy as the average radar transmitter in World War II at a small fraction of the cost, and several times the lifetime.
Re: "Perfect" LED current limiting challenge
Funny, I thought this thread had morphed back to a discussion of how to get hundreds of amps through a little, garden variety LED for cheap. Sure, one can build or buy stuff that can deliver that level of pulsed power but we are talking about RF power engineering - neither easy nor inexpensive. And to what end? To get a relatively small increase in luminous output when just adding an extra led or two and using conventional techniques will do the same at miniscule factions of the cost.
Who is online
Users browsing this forum: No registered users and 48 guests