I'm trying to design a simple switcher constant current circuit (using the LM2574-ADJ). It's easy enough if you put the sense resistor between the load and ground. However the snag is that the current sense resistor has to be on the high side of the load (there are 3 current supplies driving 3 devices with a commoned ground).
Does anyone know of a clever way of doing this without getting into differential amplifiers?
-- Tim Mitchell
Check out these switchers, might give you some ideas:
(using
between
resistor
Doesn't sound like you need much accuracy, so how about a pnp transistor and about 3 resistors to form a crude current mirror?
In article , budgie writes
No (sorry should have given more details), it's a supply for luxeon LEDs, but there can be a variable number connected in series. There's a
48V supply on one end and from 1 to 12 LEDs in series on the output. So a switcher is the only way to go otherwise too much power has to be dissipated if low numbers of LEDs are connected.-- Tim Mitchell
There is no need for a differential amplifier if you have access to an OA with RRIO that can work with your output voltage levels- those are the levels too unimportant for you to mention. The OA is configured as a simple voltage controlled current source, with input voltage IxRs, the output current of which is used to develop the feedback voltage. R2=1K would be in line with the 2574 fixed version value, so that IxRs/R1 should be calculated to be Vref/R2=1.23mA.
View in a fixed-width font such as Courier.
+-------------------+ | I | ==== | ----> | sw>-----UUUU--+----+---[Rs]-+ | | | | | [R1] | | | | | | | +----------+ | | | | | | | e | (-)| | | \| /|---+ | | pnp|--< | | | /| \|------+ | c | (+) | | Rs x R2 | | L | Vfdbk=I x -------
Mind telling why the LEDs *have* to be grounded?
This is a variation of an output voltage bootstrap regulation scheme posted by WH a while ago. Since your load is relatively light, the current can be regulated by an LM317 linear and the switcher can be used to bootstrap the total output so as to maintain enough voltage across the 317 for constant current source operation. The LED current is set at
1.25/42=30mA in this case.View in a fixed-width font such as Courier.
. .. . .. . .. Vin >--+ .. | .. +------------+ Vleds+3.3V .. | Vin | .. | sw |--------L1--+---------------+----------+ |---+ --- | .. | | D1 | | | .. | | | GND +-----+ .. | | | |LM317|-+ .. | | | +-----+ | .. | | | | | .. | UC2574-ADJ | | 42 | .. | etc | 2.7K | | .. | | | +----+ .. | | | D2 | .. | | +--------+---|
woops- adjust that L1-D1 junction:
View in a fixed-width font such as Courier.
. .. Vin >--+ .. | .. +------------+ .. | Vin | Vleds+3.3V .. | sw |-----------+-- L1----+------+----------+|--+ | --- | .. | | D1 | | | .. | | | GND +-----+ .. | | | |LM317|-+ .. | | | +-----+ | .. | | | | | .. | UC2574-ADJ | | 42 | .. | etc | 2.7K | | .. | | | +----+ .. | | | D2 | .. | | +---|
Fred Bloggs wrote: [..snip...]
Seems like this should be impervious to switching LEDs in and out while the circuit is powered up- the main thing is to make it impossible to blow the LEDs with excessive current- the 2.7K will have to be a 1W to withstand continuos short-circuited output- the LM317 protects itself:
View in a fixed-width font such as Courier.
. .. . .. .. Vin >--+ .. | .. +------------+ .. | Vin | Vleds+3.3V .. | sw |-----------+-- L1----+------+----------+|--+ | --- |in .. | | D1 | | +-----+ adj .. | | | GND |LM317|-+ .. | | 2.7K +-----+ | .. | | | |out | .. | | | D2 | | .. | UC2574-ADJ | +---|>|-----------+ | .. | etc | | | | .. | | | 42 | .. | | | | | .. | | | +-------+----+ .. | | | | | .. | | e\| |/e | .. | | |--+--| - .. | | /| | |\ V LEDs .. | | 1.23V | | | - .. | fb |------+--------------' '----+ | .. | gnd | | pnp pair | | .. +------------+ 1K bcm857bs 10k | .. | | bc857bs | | .. GND------+-------------+------------------------+-------+ . ..
In article , Fred Bloggs writes
They don't, but it's an rgb triad with the cathodes commoned, so I can't measure current on the cathode side.
These are luxeon LEDs which have a drive current of 350mA, not yer normal 20mA. The blues have a Vf of just below 4 volts, pretty much the whole 48V input is needed to drive 12 of them.
Thanks to everyone for the ideas.
-- Tim Mitchell
Are you also using the switcher to buck or boost the input voltage substantially? If not, I wonder why you are approaching a constant current task that way.
they make clever high-side current sense thingies that level-shift the signal down to the low side. digikey even sell them.
And if they do, others will too. Bear in mind that a differential amplifier is really a resistor mismatch detector.
Cheers Terry
And you also cannot stack these triads in series because you have to connect each anode to the common cathode of the triad above it which places all LEDs in the triad in parallel.
You seem to be having trouble with basic circuits.
Without an LED load the voltage soars to near Vin, which could be as high as 45V. What happens the instant the LEDs are attached? First a short high-current spike, followed by excessive LM317 voltage (if only one LED is used), etc.? Hmm, I wonder how the LM317 responds to a high voltage across its Vout and ADJ terminals for a short time? I see there's a 6V zener in series with 50 ohms inside the chip, but that resistor isn't a power resistor... Time for some experiments.
--
Thanks,
- Win
In article , Fred Bloggs writes
No, I'm just having trouble getting all the necessary information across...
The
-- Tim Mitchell
Winfield Hill wrote:
Too much current error with the dual PNP- make it:
View in a fixed-width font such as Courier.
. .. . .. .. Vin >--+ .. | .. +------------+ .. | Vin | Vleds+3.3V .. | sw |-----------+---L1----+------+----+-----+ |--+ | --- | |in .. | | D1 | | | +-----+ adj .. | | | GND | |LM317|-+ .. | | 2.7K | +-----+ | .. | | | | |out | .. | | | D2 | | | .. | LM2574-ADJ | +---|>|-----------+ | .. | etc | | 1N4148 | | | .. | | | | 42 | .. | | | \| | | .. | | | BC846 |---+----+ .. | | | e/| | .. | | e\| | | .. | | BC856 |---------+ - .. | | /| | V LEDs .. | | 1.23V | | - .. | fb |------+--------------' | | .. | gnd | | | | .. +------------+ 1K 10k | .. | | | | .. GND------+-------------+--------------------------+-----+ . ..
I am not getting that- ADJ terminal drops instantly to Vled, and the
42/2.7K voltage divider, through D2, makes for a small 0.016 x (48-Vled) or 0.75V worst case drop across the 42 ohm or 0.75/42=18mA- so you have the wideband emitter follower response in the '317. I'm pretty sure most '317's can tolerate 48V for a few msecs to discharge Cf.
Oh yeah - go simulate one, and see how crap it really is. I looked into it a while ago for a 48:9V buck converter. The transistor switching isnt too good, and gets a lot worse at high voltage. Without the current limit transistor the whole thing fries on an overload; but the Ilimit BJT basically operates in the linear region - when current limiting it pulls the base of the main switching transistor down to whatever voltage is required to get the requisite current - needless to say this does NOT help the switching BJT power dissipation. Half a days SPICE was enough to convince me to abandon the thing and do it properly.
Its OK if you're just a hobbyist, and dont really care about reliability. OTOH if you plan on selling it, be very afraid. It turns out there are VERY good reasons why we make buck regulators as complex as they are. Its a bit like driving really - any idiot can drive a car on a deserted, flat, straight piece of road. Going sideways around a corner in the pissing rain, into the path of an oncoming vehicle, is another matter entirely. Dont say I didnt warn you :)
Cheers Terry
Sideways thought.....
If a flyback converter is arranged to dump a fixed Joules/cycle, then making the Frequency proportional to Vout should result in a constant Iout.
-- Tony Williams.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.