LED Driver High Side Sense resistor

fenix-store
L

LEDboy

Newly Enlightened
Joined
May 28, 2004
Messages
18
Location
UK
Hello,

I have trying to develop a LED driver circuit for 3 Luxeons in series with one side grounded. This is for a car, so the driver needs to work within a 12V lead acid battery range.

I have been experimenting with a high side sense resistor 0R22, a PNP all in series with the LEDs. Then using a differential amp to read the sense voltage. Then comparing a reference voltage with the diff' voltage to drive the PNP.

I have not had much luck.
Has anybody else tried this? or know of a better circuit?

Cheers
James
 
Umm...
Andrew Wynn and Mr. Al have a low dropout linear current regulator. I think I remember Andrew saying it could be used at one amp.
 
Hi there,

Here's one...
You'll have to use a higher power transistor for your
application and put the three LED's in series, but the rest
is the same.
leslieschematicv3-01.gif


Take care,
Al
 
Al,

If you want a little faster response, and have the ability to measure the light output rise time and fall time, I believe a 220pf capacitor across R4 to be the right value to speed up your constant current source a tad, without causing overshoot. The circuit also works great with a TLC0272.
 
Thanks for all the replies,

The circuit I am after is similar to Mr Al's but the LEDs are grounded, and the sense resistor is on the high side along with the transistor. This is to allow for use for LED fixtures which have a common ground.

Thanks for the link chimo (MAX4073), I'm in the UK and my suppliers don't stock that part but they do a MAX472 which is similar, which looks like it could do the job, but it's expensive at $6.68 each. I will order some anyway and give it ago.

Another design I tried is using the LM334 with the current boost PNP/FET, but instead of using a floating load, I put the load from the ground to the LM334 V-/sense resistor. This gives me a drop out of about 2V, better than a LM317 3V drop out, but still way too high. I am sure that the 2V drop out could be improved, but not by enough.
Another idea I had was to use a small switching circuit to create a supply rail a few volts above the supply and use this to power the LM334 V+. Then by taking a signal from this it could then drive a FET.
This would be good if it worked as it could give a very low drop out and would be cheap if you want to create a PCB with multiple drivers, as the switching supply could be used for each driver.
However I have not tried this and have no circuit ideas, has anybody else?

James
 
Is there any reason that you're stuck with the LEDs being directly tied to ground? Mr. Al's circuit will work fine if you can just put a tiny bit of resistance between the LED- and ground. So will the LM334 current regulator we've discussed before. In both cases, you can get 0.1V of dropout. Of course, you'll want to put a 3A 100V or higher diode in series with everything to protect it against negative surges that can happen in a car's electrical system.
 
Yes unfortunally so,
I have a bank of LEDs which are wired in series strings of 3, and all common to ground and have 12 LEDs in total for each fitting. The aluminium case is connected to ground. I have around 20 of these fittings, so do not wish to change them.
The main issue is the case shorting with another ground, this won't be a problem with a true 0V connection.
What I am really trying to create is a portable power pack with at least 4 driver circuits and built in charger and batteries. The pack will be able to be re-charged via the car lighter socket, or the car can run the lights direct. Because I can't guarantee where the fitting will be placed, I must ensure that its ground referenced. I also only have a 5 pin connecter to each fitting, so it rules out floating loads.

James
 
I think I have a solution. It uses Mr. Al's circuit, and reverses everything, so instead of being referenced to ground, you're referenced to V+.

Here's how:

1) Move R1 so that it's between U2/R2-R3 and ground. This makes the 1.25V drop by the U2 reference diode be from V+ to R1. Also, swap R2 and R3 - now the reference point for the non-inverting input (+) for the op-amp is referenced 0.1V below V+.

2) Swap R5 and the Luxeon(load) - now the load is tied directly to ground, and R5 causes an 0.1V drop from V+ at desired current.

3) Move the tap for the non-inverting input (-) of the op-amp to the low-side of the current sense resistor R5.

4) Swap the transistor for a PNP or a P-channel MOSFET.

I think that should do the trick - I sketched it out here at workr real quick - Mr. Al should be able to verify that I haven't messed something up.

[edit]
Put a blocking doide between V+ and everything to protect the circuit from the car's electrical system.
 
Looks like you captured everything I said correctly. Good luck with the testing.

One hint - use 4-5 1N4001 diodes in series to simulate a Luxeon instead of an actual Luxeon, just to be on the safe side (esp with an "untested" regulator design). I've blown a few testing out various things - and I now use either some diodes, or about 4 ohms of resistance to simulate a luxeon load.
 
One more thing, and hopefully Mr. Al could comment on this,

I would also experiment with placing a 100K ohm resistor in parallel with some capacitance (1uF?) and placing those between the gate and source of your MOSFET, then replacing R4 with 4.7K ohms. I think this will help reduce inrush current when power is first applied - the time constant will be a few milliseconds.
 
Hello again,

evan:
Yes, that's a very good circuit transformation which should
work out very well as redrawn, keeping a few things in mind:

1. The selected op amp works up to a max of 14v, so it
wont work out too well in a car where the voltage goes
higher, and possibly contains some even higher spikes.
A new op amp will have to be chosen or something else.
I dont know if it would be enough to insert a diode
in series with the supply and some filter cap to keep
spikes low. Remember the op amp alone cant be regulated or
else V+ wont reach up to +Vcc from the battery, unless
perhaps you were to change the Vref (making it lower)
and use a resistive voltage divider from the bottom of
the sense (0.1 ohms) resistor, but in that case you can
use the standard LM358 ic also.

2. The choice of using a rail-to-rail input op amp was a good
one, as one of the input voltages goes right up to +Vcc
for some time. If you change the design by adding four
resistors you can probably use a more standard op amp
however, but the cost is the four more resistors, or add
another reference diode and resistor to subtract a set voltage
from the bottom of the sense resistor. There's probably more
solutions to this also.

3. Resistor R4 can be eliminated completely if using a
MOSFET transistor. This will keep turn on surge low or
eliminate it completely. Shouldnt be any need for a
comp network of any kind. If you like you can check with
a scope but simulations look very very good.

4. Power diss in the pass transistor shouldnt be a problem
but you may wish to calculate it anyway just to make sure,
especially if you increase the current to the LED's.
Remember also that the max power in the transistor occurs
at high input line (14.5v or so).


Take care,
Al
 
I've got one of these in an automotive application:
http://focus.ti.com/docs/prod/folders/print/tle2142.html

I put a diode on the input power to protect against reverse transients, and a "high temp" electrolytic:
http://www.nichicon.com/english/seihin/alm_mini/daia_f.htm
used the 63V version and large value with a little resistance in series to the input to help filter any spikes on the way in. Watch out if you utilize a switch to turn on and off and also to apply power and remove it.

Watch out when using tantalums with Lead Acid cells, and any other high output low internal impedance batteries, many will ignite. There are some surge rated tantalums, but personally I don't trust them much anyhow with low impedance lead acid cells.

Also note some references are "slow" to come up, if you need the speed.
 
Hi there Newbie,

[ QUOTE ]
NewBie said:
I've got one of these in an automotive application:
http://focus.ti.com/docs/prod/folders/print/tle2142.html


[/ QUOTE ]

That's a very nice op amp chip, with pretty impressive
spec's and i wouldnt mind getting a couple, but for
this exact application i wouldnt recommend using it.
If you read my previous post i was mentioning that the
chip that *was* originally chosen had rail to rail inputs,
so it could easily handle a 0.1 ohm resistor (with only
maybe 35mv drop) connected from Vcc+ directly to one
of the inputs. With a more standard chip like that one
more resistors would be required to allow it to operate
normally in a high side sensing circuit scheme like this
one.
Perhaps you can locate a chip similar to that one only
with rail to rail inputs?

The speed is pretty impressive too on that chip, and they
even spec a time value for coming out of saturation...
something i'd like to see on all op amp data sheets!!!


Take care,
Al
 
Mr Al,

Thanks - I was unsure about how things would power up (if the op-amp would be pulling down initially, turning the transistor on full) - good to know that shouldn't be a problem - makes the circuit simpler
 
Thanks for all your input evan & Mr Al,
I have built up the circuit as previously posted but changed the following R1 = 0R22, R2 = 3k3, R5 = 100k, R4 = removed, LM385 is a 1.2V part & C1 removed.
This gives me a measured 356mA and a drop out of ~100mV. However bear in mind this circuit has been tested in a solderless breadboard, and with such a low sense resistor it is bound to change when built on veroboard.
When I first powered up this circuit it did not work, this I found out was due to having my power supply set to 20V for a short time, and this blew my op-amp, but its failure mode saw its output voltage rise with supply until a point where it would start falling again.
Anyhow a new op-amp sorted that out, and it seems to be functioning fine at 15V, even if the spec is 14V max. I don't know if I just happen to be using a good one!
As for C1, I have scoped about and cannot see any ringing, and the start up overshoot is around 10uS, which I'm quite happy with.
The only real problem is the op-amp supply voltage, I had difficulty finding one with a higher supply voltage at a sensible price. It also remains to be seen what happens when I load the other 3 amps in the package at 15V.

I am very pleased with this circuit, low drop out, grounded load, low part count & cheap, especially with a 4 channel version.

So I will build up a 4 channel Vero board version and give it proper testing.

Regards
James
 
[ QUOTE ]
MrAl said:
Hi there Newbie,

[ QUOTE ]
NewBie said:
I've got one of these in an automotive application:
http://focus.ti.com/docs/prod/folders/print/tle2142.html


[/ QUOTE ]

That's a very nice op amp chip, with pretty impressive
spec's and i wouldnt mind getting a couple, but for
this exact application i wouldnt recommend using it.
If you read my previous post i was mentioning that the
chip that *was* originally chosen had rail to rail inputs,
so it could easily handle a 0.1 ohm resistor (with only
maybe 35mv drop) connected from Vcc+ directly to one
of the inputs.
Take care,
Al

[/ QUOTE ]

I was looking at your low side sense resistor schematic with the resistor hooked from ground...


http://hometown.aol.com/xaxo/images/leslieschematicv3-01.gif
.
.
.
.
How about:
http://focus.ti.com/docs/prod/folders/print/tlv2371-q1.html
(for auto stuff you'll need to protect it better)

Or, http://www.analog.com/en/prod/0,,759_786_OP284%2C00.html

These are interesting for other circuits:
http://focus.ti.com/docs/pr/pressrelease.jhtml?prelId=sc05086&prSection=overview
 
Thanks for the links,
The TLV2374IN has a maximum supply of 16.5V, and is less than half the cost of the TS924, is fully rail to rail and it's widely available.

Cheers
James
 
Hello again,

Newbie:
Ok, no prob /ubbthreads/images/graemlins/smile.gif The newer link looks especially good for
this particular app using the high side sense scheme
around 12v or so.

James:
Sounds like you had some good results with this thing.
Let us know how it works out once you get the new chips
if you can...


Take care,
Al
 
You must log in or register to reply here.

Similar threads

K
Circuit to distribute LED driver current
Replies
9
Views
1K
Dave_H
D
BigusLightus
LED driver input voltage?
Replies
8
Views
999
alpg88
alpg88
G
Convoys Buck-Boost driver burns after bad contact
Replies
1
Views
519
Monocrom
Monocrom
Poppy
high speed charger fail
2
Replies
25
Views
2K
hsa
H
A
Any new "high end" chargers in the near future?
Replies
6
Views
2K
knucklegary
knucklegary

Latest posts

Top