LED boost driver question

[bitrex]

There are a number of different inexpensive boost ICs of this style:

<https://i.stack.imgur.com/aHqvK.png>

for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

 

[Rick C]

On Friday, November 1, 2019 at 1:43:57 PM UTC-4, bitrex wrote:

There are a number of different inexpensive boost ICs of this style:

https://i.stack.imgur.com/aHqvK.png

for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an N channel device on the low side between the bottom of the LED string and the sense resistor/sense input junction. This will provide some added protection since as the current increases the drive to the N channel device is reduced and the current will be limited.

--

Rick C.

- Get 1,000 miles of free Supercharging
- Tesla referral code - https://ts.la/richard11209

 

[whit3rd]

On Friday, November 1, 2019 at 10:43:57 AM UTC-7, bitrex wrote:

There are a number of different inexpensive boost ICs...
for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Probably it's done by PWM, so the 'sense' resistor showing zero current
(when an active string is switched OFF) is just the same as the
startup condition, it should handle that. But, will your high-side
switch, in OFF state, handle the inductive kick of the first pulse?

You'd want any high-side switching to sequence in make-before-break fashion,
including at first power-up.

Two driver chips, two inductors and two Schottky diodes, using logic into the
shutdown input, seems a cleaner solution.

 

[bitrex]

On 11/1/19 1:59 PM, Rick C wrote:

On Friday, November 1, 2019 at 1:43:57 PM UTC-4, bitrex wrote:
There are a number of different inexpensive boost ICs of this style:

https://i.stack.imgur.com/aHqvK.png

for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an N channel device on the low side between the bottom of the LED string and the sense resistor/sense input junction. This will provide some added protection since as the current increases the drive to the N channel device is reduced and the current will be limited.

Yeah I think low-side would be preferable

 

[Tim Williams]

Sure. A make-before-break switch would be preferable, and relatively slow
switching (~us) so as not to dump the capacitor into the lower-voltage
string. No need for P type and level shifting, the base current and shunt
voltage at the low side are both small.

Heh, actually made a medical device years ago that basically did that, but
at higher current (we used a 12V MeanWell supply and two independent boost
controllers, one for each color) and with an MCU to set and measure the
operating current. (A specific blue wavelength is used for treating infant
jaundice, and the current is measured as dosage; the white was just for
general illumination.)

Tim

--
Seven Transistor Labs, LLC
Electrical Engineering Consultation and Design
Website: https://www.seventransistorlabs.com/

"bitrex" <user@example.net> wrote in message
news:s9_uF.1937$fL4.225@fx41.iad...

There are a number of different inexpensive boost ICs of this style:

https://i.stack.imgur.com/aHqvK.png

for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

 

[Winfield Hill]

bitrex wrote...

On 11/1/19, Rick C wrote:
On November 1, 2019, bitrex wrote:

There are a number of different inexpensive boost ICs of this style:
https://i.stack.imgur.com/aHqvK.png
for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an
N channel device on the low side between the bottom of the LED string
and the sense resistor/sense input junction. This will provide some
added protection since as the current increases the drive to the
N channel device is reduced and the current will be limited.

Yeah I think low-side would be preferable

There are a few issues. Your LED strings will no doubt have quite
different forward voltages. So I'd add a series resistor to the FB
terminal. At the moment of switching, the voltage on cap C2 (Fig 1
in the CAT4238 datasheet) will either create excess LED current, or
keep the LED string off. Inductor L is naturally a current source,
and the CAT4238 has a nice internal current-mode control loop. These
try to keep a constant current flowing into whatever the LED-string
voltage is. But cap C2 is a voltage-source trouble maker, and you
will want to minimize its value. At 40mA LED current, its voltage
changes 0.04/0.22uF = 180 mV/us. You can drop C2 a lot by accepting
a higher swing each cycle, especially if you add a ballast resistor
dropping say 3 volts. You can also add parts to finesse your LED
strings to be closer in voltage. Or use two boost converters, haha,
only 75 cents for the CAT4238 and 47 cents for the Sumida inductor.
OK, I get it, you're worried about space. Aha, a dual FDC6561AN
MOSFET, etc., in a SOT-23-6 package, only 25 cents.


--
Thanks,
- Win

 

[Lasse Langwadt Christense]

lørdag den 2. november 2019 kl. 17.03.48 UTC+1 skrev bitrex:

On 11/2/19 8:56 AM, Winfield Hill wrote:
bitrex wrote...

On 11/1/19, Rick C wrote:
On November 1, 2019, bitrex wrote:

There are a number of different inexpensive boost ICs of this style:
https://i.stack.imgur.com/aHqvK.png
for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an
N channel device on the low side between the bottom of the LED string
and the sense resistor/sense input junction. This will provide some
added protection since as the current increases the drive to the
N channel device is reduced and the current will be limited.

Yeah I think low-side would be preferable

There are a few issues. Your LED strings will no doubt have quite
different forward voltages. So I'd add a series resistor to the FB
terminal. At the moment of switching, the voltage on cap C2 (Fig 1
in the CAT4238 datasheet) will either create excess LED current, or
keep the LED string off. Inductor L is naturally a current source,
and the CAT4238 has a nice internal current-mode control loop. These
try to keep a constant current flowing into whatever the LED-string
voltage is. But cap C2 is a voltage-source trouble maker, and you
will want to minimize its value. At 40mA LED current, its voltage
changes 0.04/0.22uF = 180 mV/us. You can drop C2 a lot by accepting
a higher swing each cycle, especially if you add a ballast resistor
dropping say 3 volts. You can also add parts to finesse your LED
strings to be closer in voltage. Or use two boost converters, haha,
only 75 cents for the CAT4238 and 47 cents for the Sumida inductor.
OK, I get it, you're worried about space. Aha, a dual FDC6561AN
MOSFET, etc., in a SOT-23-6 package, only 25 cents.



Two switchers would be ideal but yes cost is the primary concern on this
lil' jerb. The two LED strings are not long, only two or three standard
5mm LEDs on each.

I had thought about hacking something out of discretes or a charge pump
with a logic IC or two but some of these LED drivers are only about 40
cent in small quantity, will be more efficient, it really doesn't seem
worth the hassle vs an off-the-shelf solution to bring the 2.7 volt
supply up to where it needs to be.

https://lcsc.com/product-detail/LED-Drivers_MT9284AS6_C181780.html
~4 cents in volume

https://www.buydisplay.com/default/led-backlight-driver-ic-chip-pt4110-in-sot-89-5-package

~20 cents

 

[bitrex]

On 11/2/19 8:56 AM, Winfield Hill wrote:

bitrex wrote...

On 11/1/19, Rick C wrote:
On November 1, 2019, bitrex wrote:

There are a number of different inexpensive boost ICs of this style:
https://i.stack.imgur.com/aHqvK.png
for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an
N channel device on the low side between the bottom of the LED string
and the sense resistor/sense input junction. This will provide some
added protection since as the current increases the drive to the
N channel device is reduced and the current will be limited.

Yeah I think low-side would be preferable

There are a few issues. Your LED strings will no doubt have quite
different forward voltages. So I'd add a series resistor to the FB
terminal. At the moment of switching, the voltage on cap C2 (Fig 1
in the CAT4238 datasheet) will either create excess LED current, or
keep the LED string off. Inductor L is naturally a current source,
and the CAT4238 has a nice internal current-mode control loop. These
try to keep a constant current flowing into whatever the LED-string
voltage is. But cap C2 is a voltage-source trouble maker, and you
will want to minimize its value. At 40mA LED current, its voltage
changes 0.04/0.22uF = 180 mV/us. You can drop C2 a lot by accepting
a higher swing each cycle, especially if you add a ballast resistor
dropping say 3 volts. You can also add parts to finesse your LED
strings to be closer in voltage. Or use two boost converters, haha,
only 75 cents for the CAT4238 and 47 cents for the Sumida inductor.
OK, I get it, you're worried about space. Aha, a dual FDC6561AN
MOSFET, etc., in a SOT-23-6 package, only 25 cents.

Two switchers would be ideal but yes cost is the primary concern on this
lil' jerb. The two LED strings are not long, only two or three standard
5mm LEDs on each.

I had thought about hacking something out of discretes or a charge pump
with a logic IC or two but some of these LED drivers are only about 40
cent in small quantity, will be more efficient, it really doesn't seem
worth the hassle vs an off-the-shelf solution to bring the 2.7 volt
supply up to where it needs to be.

 

On Fri, 1 Nov 2019 13:43:52 -0400, bitrex <user@example.net> wrote:

There are a number of different inexpensive boost ICs of this style:

https://i.stack.imgur.com/aHqvK.png

for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

You could put the strings in series, and short one or the other with a
small SSR or something. That allows both to be on, too.



--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[bitrex]

On 11/2/19 12:31 PM, Lasse Langwadt Christensen wrote:

lørdag den 2. november 2019 kl. 17.03.48 UTC+1 skrev bitrex:
On 11/2/19 8:56 AM, Winfield Hill wrote:
bitrex wrote...

On 11/1/19, Rick C wrote:
On November 1, 2019, bitrex wrote:

There are a number of different inexpensive boost ICs of this style:
https://i.stack.imgur.com/aHqvK.png
for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an
N channel device on the low side between the bottom of the LED string
and the sense resistor/sense input junction. This will provide some
added protection since as the current increases the drive to the
N channel device is reduced and the current will be limited.

Yeah I think low-side would be preferable

There are a few issues. Your LED strings will no doubt have quite
different forward voltages. So I'd add a series resistor to the FB
terminal. At the moment of switching, the voltage on cap C2 (Fig 1
in the CAT4238 datasheet) will either create excess LED current, or
keep the LED string off. Inductor L is naturally a current source,
and the CAT4238 has a nice internal current-mode control loop. These
try to keep a constant current flowing into whatever the LED-string
voltage is. But cap C2 is a voltage-source trouble maker, and you
will want to minimize its value. At 40mA LED current, its voltage
changes 0.04/0.22uF = 180 mV/us. You can drop C2 a lot by accepting
a higher swing each cycle, especially if you add a ballast resistor
dropping say 3 volts. You can also add parts to finesse your LED
strings to be closer in voltage. Or use two boost converters, haha,
only 75 cents for the CAT4238 and 47 cents for the Sumida inductor.
OK, I get it, you're worried about space. Aha, a dual FDC6561AN
MOSFET, etc., in a SOT-23-6 package, only 25 cents.



Two switchers would be ideal but yes cost is the primary concern on this
lil' jerb. The two LED strings are not long, only two or three standard
5mm LEDs on each.

I had thought about hacking something out of discretes or a charge pump
with a logic IC or two but some of these LED drivers are only about 40
cent in small quantity, will be more efficient, it really doesn't seem
worth the hassle vs an off-the-shelf solution to bring the 2.7 volt
supply up to where it needs to be.

https://lcsc.com/product-detail/LED-Drivers_MT9284AS6_C181780.html
~4 cents in volume

Woooooooooooooow. "These prices are Kraaay-ZEEEE"

https://www.buydisplay.com/default/led-backlight-driver-ic-chip-pt4110-in-sot-89-5-package

~20 cents

 

[Winfield Hill]

Winfield Hill wrote...

... I've made purchases from LCSC, easy to work
with, very respectable, kind of a Chinese Digi-Key.

I purchased a roll of 3Peak high-speed op-amps, excellent
parts at only 6 cents each. They also offered slower 36V
4MHz 15V/us op-amps, with 3mV Vos, only 1.3 cents each.


--
Thanks,
- Win

 

[Winfield Hill]

bitrex wrote...

On 11/2/19 12:31 PM, Lasse Langwadt Christensen wrote:
lørdag den 2. november 2019 kl. 17.03.48 UTC+1 skrev bitrex:
On 11/2/19 8:56 AM, Winfield Hill wrote:
bitrex wrote...

On 11/1/19, Rick C wrote:
On November 1, 2019, bitrex wrote:

There are a number of different inexpensive boost ICs of this style:
https://i.stack.imgur.com/aHqvK.png
for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an
N channel device on the low side between the bottom of the LED string
and the sense resistor/sense input junction. This will provide some
added protection since as the current increases the drive to the
N channel device is reduced and the current will be limited.

Yeah I think low-side would be preferable

There are a few issues. Your LED strings will no doubt have quite
different forward voltages. So I'd add a series resistor to the FB
terminal. At the moment of switching, the voltage on cap C2 (Fig 1
in the CAT4238 datasheet) will either create excess LED current, or
keep the LED string off. Inductor L is naturally a current source,
and the CAT4238 has a nice internal current-mode control loop. These
try to keep a constant current flowing into whatever the LED-string
voltage is. But cap C2 is a voltage-source trouble maker, and you
will want to minimize its value. At 40mA LED current, its voltage
changes 0.04/0.22uF = 180 mV/us. You can drop C2 a lot by accepting
a higher swing each cycle, especially if you add a ballast resistor
dropping say 3 volts. You can also add parts to finesse your LED
strings to be closer in voltage. Or use two boost converters, haha,
only 75 cents for the CAT4238 and 47 cents for the Sumida inductor.
OK, I get it, you're worried about space. Aha, a dual FDC6561AN
MOSFET, etc., in a SOT-23-6 package, only 25 cents.

Two switchers would be ideal but yes cost is the primary concern on this
lil' jerb. The two LED strings are not long, only two or three standard
5mm LEDs on each.

I had thought about hacking something out of discretes or a charge pump
with a logic IC or two but some of these LED drivers are only about 40
cent in small quantity, will be more efficient, it really doesn't seem
worth the hassle vs an off-the-shelf solution to bring the 2.7 volt
supply up to where it needs to be.

https://lcsc.com/product-detail/LED-Drivers_MT9284AS6_C181780.html
~4 cents in volume

Woooooooooooooow. "These prices are Kraaay-ZEEEE"

That chip appears to be a fine part, with a decent datasheet.
The MT9284 has the same architecture as the CAT4238, with a
current-mode loop inside, but it's a bit faster, and has a
higher-current switch. I've made purchases from LCSC, easy
to work with, very respectable, kind of a Chinese Digi-Key.
Select an inductor from LCSC as well. And your dual MOSFET.
Or, at that price level, you can use two separate converters!


--
Thanks,
- Win

 

[bitrex]

On 11/2/19 2:17 PM, Winfield Hill wrote:

Winfield Hill wrote...

... I've made purchases from LCSC, easy to work
with, very respectable, kind of a Chinese Digi-Key.

I purchased a roll of 3Peak high-speed op-amps, excellent
parts at only 6 cents each. They also offered slower 36V
4MHz 15V/us op-amps, with 3mV Vos, only 1.3 cents each.

With those prices on the boosts nowadays with what looks like good
availability I don't see any reason not to use one for each string, a
switching arrangement is over-complicated TOAB at that point.

The other question is whether I can tap a second diode plus cap off the
inductor and run a couple low-power quad RRIO op amps off the LED
booster. The current draw will be low with respect to the LEDs. I bet it
won't notice.

Application is a custom "big blinky" so the PWM inputs are real handy

 

[Klaus Kragelund]

Be careful about parts from LCSC

Many are Chinese versions of known designs and the lifetime is often not guaranteed, so you design it in, get to full production and then suddenly it is discontinued

 

[Winfield Hill]

Klaus Kragelund wrote...

Be careful about parts from LCSC

Many are Chinese versions of known designs and the
lifetime is often not guaranteed, so you design it
in, get to full production and then suddenly it is
discontinued.

From what I can tell, there are some good, reliable
semi companies in the Pacific-Rim mix. No doubt, a
few, competing at the sub-10-cent level, simply can
not sustain the effort, and their parts may collapse.
But I developed confidence in 3Peak, and aerosemi.com
looks good with lots of products. They don't have
an English website that I could see, so far, but the
datasheet was in English, so they may have one. The
MT9284 pinout matches the CAT4238, except they have
an extra OVP pin to provide safety protection. In a
pinch you could revert back to the CAT4238. But the
parts are so cheap, you can make lifetime purchases.


--
Thanks,
- Win

 

On Saturday, 2 November 2019 12:57:21 UTC, Winfield Hill wrote:

bitrex wrote...

On 11/1/19, Rick C wrote:
On November 1, 2019, bitrex wrote:

There are a number of different inexpensive boost ICs of this style:
https://i.stack.imgur.com/aHqvK.png
for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an
N channel device on the low side between the bottom of the LED string
and the sense resistor/sense input junction. This will provide some
added protection since as the current increases the drive to the
N channel device is reduced and the current will be limited.

Yeah I think low-side would be preferable

There are a few issues. Your LED strings will no doubt have quite
different forward voltages.

No problem at all when make before breaking strings in parallel -if anything it's a plus


NT

 

[Lasse Langwadt Christense]

søndag den 3. november 2019 kl. 12.28.43 UTC+1 skrev tabb...@gmail.com:

On Saturday, 2 November 2019 12:57:21 UTC, Winfield Hill wrote:
bitrex wrote...

On 11/1/19, Rick C wrote:
On November 1, 2019, bitrex wrote:

There are a number of different inexpensive boost ICs of this style:
https://i.stack.imgur.com/aHqvK.png
for constant current driving LED strings from e.g. LiPo battery.

Any problem with chip of that type having say two strings of different
color, perhaps white and blue, and switching between them using PNP or
Pfet on the high side, with a common current-sense resistor?

Looks like the sense voltage is only 0.3 volts. So you could use an
N channel device on the low side between the bottom of the LED string
and the sense resistor/sense input junction. This will provide some
added protection since as the current increases the drive to the
N channel device is reduced and the current will be limited.

Yeah I think low-side would be preferable

There are a few issues. Your LED strings will no doubt have quite
different forward voltages.

No problem at all when make before breaking strings in parallel -if anything it's a plus

you could even do with one switch, just when you turn off the lowest voltage string the other turns on