olled. At the bottom of the list I found that part which gave me the idea to roll my own. Using the same 2N2907 I'm using in the current limiter for the supercap (which you gave me the idea for with your soft start in Win's thread on the USB power limit) it works as good if not better than the On part.
sleep last night. Need to catch a few winks shortly.
Or is working on right now! What? You think he isn't still hanging out he re?
-- Rick C. -+- Get 1,000 miles of free Supercharging -+- Tesla referral code - https://ts.la/richard11209
Not sure what topology you are referring to. Is it the current boost shown in the TI data sheet, Fig 20 on page 11?
The total voltage available ranges from 5 down to 2 which is the point where other parts of the circuit stop performing in spec.
-- Rick C. -++ Get 1,000 miles of free Supercharging -++ Tesla referral code - https://ts.la/richard11209
Yes, it also appears in older NSC datasheets.
Not according to the datasheet, which gives a typical minimum operating voltgae (the V+ terminal) of 0.8 to 0.9 volts for under 100uA or under 1mA through the V+ terminal. Add 0.65V for the PNP and get 1.55 volts (the PNP base and a Vbe resistor, if present, don't need more than 1mA). If we keep the PNP's Vce above 0.2V, to keep it out of saturation, that allows for operation down to 1.86-volts with a 1.6-volt red LED, located between the collector and the Reference pin.
--
Thanks,
- Win
Sorry, still not following your topology. In Fig 20 you still need to allow for the total required by the LM334 (~0.9-1.0) plus the Vbe which is most of the 2.0 volts available.
If you are using another topology, I'm not picturing it. The diode might go between the Rs and the collector. Then the minimum drop out would be 64 mV plus the PNP saturation voltage.
-- Rick C. +-- Get 1,000 miles of free Supercharging +-- Tesla referral code - https://ts.la/richard11209
Where's the LM3909 when we need it?
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC / Hobbs ElectroOptics Optics, Electro-optics, Photonics, Analog Electronics Briarcliff Manor NY 10510 http://electrooptical.net http://hobbs-eo.com
Most but not all. 0.9V + 0.65V = 1.55V, well below 2V Base current under 1mA, so 0.9V, not 1.0V, but anyway... Choose a low-saturation Zetex = Diodes, Inc., PNP part.
Yes, of course that's where it goes. Wasted otherwise.
Right, with a good PNP, 150 to 200mV, very low. With only two active parts. How low do you want it to be?
--
Thanks,
- Win
Ok, you never said that. The figure in the data sheet simply shows an inli ne current source with a minimum drop out of 1.4 volts or so.
That's fine. But I'm good with the two PNPs and two resistors. Oh, how wo uld I turn the LM334 on and off? I don't see an obvious way of doing that unless the V- on the LM334 be broken from ground and tied to a CMOS gate ou tput. I don't want to just connect the entire string to the output since 1 mA will create additional offset in the CMOS output. I assume the current through the V- pin would be rather small by a factor of the beta of the PN P, so this the leg to lift.
Bottom line is unless the simulations are very wrong, the pair of PNPs do a very good job down to about 2.2 volts and at 2 volts the current is still above 0.7 volts. A bit of fade in the final minutes of the alarm will help to make it clear what is happening. Pulsing the sounder the way I am doin g helps to minimize the current drain from that. The LED current will be m ore than the sounder! I'd make that blink too at a 50/50 duty cycle, but t hat would require more parts. The logic that drives the sounder stops when muted. I could let the oscillator run. 1.6 Mohm at 128 Hz shouldn't brea k the bank on current. The enable current on the two PNP circuit is around 50 uA at 5V.
Changed my circuit to let the oscillator run in alarm condition even when t he sounder is muted. That frees up a NOR gate so it can invert a counter o utput and enable the LED when the counter is not held in reset. So the 1 m A is cut in half on average giving about 1.3 mA average current during alar m and 0.5 mA when in alarm and muted.
The 2.5 Farad supercap can probably be cut from 2.5 F to 1 F, but we'll see . It won't hurt to built up some of this before patting myself on the back . Not sure how I would even detect excess 100 uA currents on the bench. T he simulations should be pretty good though. The low voltage stuff concern s me since there is no reason to believe the simulation is accurate enough at minimum Vcc.
-- Rick C. +-+ Get 1,000 miles of free Supercharging +-+ Tesla referral code - https://ts.la/richard11209
olled. At the bottom of the list I found that part which gave me the idea to roll my own. Using the same 2N2907 I'm using in the current limiter for the supercap (which you gave me the idea for with your soft start in Win's thread on the USB power limit) it works as good if not better than the On part.
sleep last night. Need to catch a few winks shortly.
Thanks for the circuit and explanation.
George h.
Seems like an rrio opamp could control the LED current pretty well with very little lost voltage.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
Yes but it would need to work down to 2V and have a voltage reference, the LM10 might fit! I know Microchip to lowpower comparators with 1.2V references maybe they do an op-amp+ref combo too. But I think Win's LM334 suggestion is the best so far.
piglet
The low voltage is OK, and an opamp could sense, say, 10 mV in the current shunt. It would need a 10 mV voltage reference from somewhere, like divide down a bandgap or a diode drop or something.
There might be a way to use the drop in the LED as the voltage reference. In super-precision reference zener circuits, the current source for the zener could be based on the zener voltage.
Yeah, that might work.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
of
st
than expected. Some of the graphs in the data sheet talk about Ic/Ib being only 10. I guess that is not expressing the gain, it's pointing out the c onditions of the test. I already have a PNP on the board, so rather than a dd another line item to the BOM I configured two 2N2907 in the same arrange ment and it works the charm. With 1 mA of current at 5 volts the current o nly drops off toward the bottom, still 940 uA at 2.25 volts and 700 uA at 2 volts. The simulated diode drop for the LED is 1.56 volts at 1 mA current . We'll see how realistic this is.
's a huge improvement over a resistor and only costs two transistors, two r esistors and the transistors are already in the BOM.
An enable input is also required. I guess the reference could be powered f rom the enable input. But the BoM count is climbing. The two transistors is still the best in my opinion, especially since they are already on the B oM. Less than 10% variation down to 2.2 volts and about 0.7 mA at 2.0 volt s. It's not worth adding the LM334 to the BoM. One of the goals of this p roject is for the design to be easy to make using easy to obtain components . Not sure how that's going to work out really. But we'll see. Still, it puts an emphasis on not adding any components to the BoM as new line items if it is not required.
I'm not trying to get a highly accurate LED current. I'm trying to deal wi th a 5:1 variation in LED current. Heck, I expect this would work adequate ly with one transistor but it likely would require a new value of resistor so a new line item in the BoM. The two transistor circuit required one res istor to be added to the BoM. win-win
-- Rick C. ++- Get 1,000 miles of free Supercharging ++- Tesla referral code - https://ts.la/richard11209
A wheatstone-like bridge? Yeah, but it's necessary for the LED to be non-resistor-like, and LEDs aren't ideal diodes. Might have to use forward-biased diode instead (a Schottky would be best for regulation from the low voltage).
74HC14 works on 2V, just capacitor-couple the LED with a Schottky clamp, feeding anode and cathode from inverted signals, and the LED will stay lit until well below 2V (it's only spec'ed to work at 2V, but probably some function down to 1.5 or so).
No, at a constant current, the voltage drop across the LED can be used as the reference for the current source that makes the constant current into the LED. Easy.
An lit-up LED is a forward-biased diode.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
Actually, that doesn't need an opamp. The self-reference thing can be done with 8 cents worth of discretes and effectively zero headroom.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
Huh, OK care to share? I came back to look at piglet's circuit again.
Are you suggesting using the led as the reference?
George H.
Yes. Saves parts.
-- John Larkin Highland Technology, Inc Science teaches us to doubt. Claude Bernard
t
be
forward-biased
voltage).
dMore like it saves "part". Using two bipolars instead of this unique LND15
0 device while the two generic bipolars are cheaper. You can even buy them in one package. Since the bipolars have a lower drop out and so work bett er, there's no reason to bother with the LND150 device.-- Rick C. +++ Get 1,000 miles of free Supercharging +++ Tesla referral code - https://ts.la/richard11209
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.