Errr, yes if you say so, but you've lost me. I am but a novice!

I posted a schematic traced from a commercial outdoor night light about a year ago - if I can find the time, I'll have a look for it.

 

I posted a schematic traced from a commercial outdoor night light about a year ago - if I can find the time, I'll have a look for it.

 

Thank you ian field
Interesting but I was also considering trying to design a circuit based around a LM358 and a MOSFET (IRFxxx) instead of fiddling about with transistors.
I shall work on it tomorrow and post it below.
See what you and others think..... to be continued....

 

Thank you ian field
Interesting but I was also considering trying to design a circuit based around a LM358 and a MOSFET (IRFxxx) instead of fiddling about with transistors.
I shall work on it tomorrow and post it below.
See what you and others think..... to be continued....

Note that the circuit used a photo diode - LDRs have been getting harder to find since RoHS, they contain highly toxic cadmium.

The PD is very similar to a type that was used in TV front panels for the remote control - its quite likely to be an IR PD, but the circuit was traced from a commercial light fitting with daylight cut off which worked just fine before it was ripped out for renovations.

 

Note that the circuit used a photo diode - LDRs have been getting harder to find since RoHS, they contain highly toxic cadmium.

Yes, I see the diagram you posted is using a photo diode but I have loads of the GL5528 LDR's which I purchased through AliExpress. They seem to be OK and lots of sellers there are selling them.

 

Yes, I see the diagram you posted is using a photo diode but I have loads of the GL5528 LDR's which I purchased through AliExpress. They seem to be OK and lots of sellers there are selling them.

I didn't say you couldn't get any - I ordered a couple of bags from China a while back.

But cadmium is highly toxic and LDRs are certainly becoming a lot less popular.

 

I like the 72 minute time constant.

ak

 

I was also considering trying to design a circuit based around a LM358 and a MOSFET (IRFxxx) instead of fiddling about with transistors.
I shall work on it tomorrow and post it.

As I said yesterday, I have designed a circuit to switch on and off a 3528 SMD 12v LED striplight.
I have chucked the idea of using transistors and have come up with this:


This is using an LM358 and an IRF530 MOSFET.
I have tested the LDR against the light levels and have noticed that the LDR is about 60k ohms when I want the striplight to turn on so I have tried to design the circuit as close to this theshold as possible.
The R8 100k resistor controls the hysteresis so that the striplight does not flicker when the light levels are close to the switch on/off theshold. If the hysteresis is too close, I can change this value, but I will test it and replace the R8 later if necessary.
VR2 is an trimmer to adjust the theshold switch on/off levels/sensitivity of the theshold values.

I hope it's ok, but I will be anxiously waiting for your thoughts and comments.

 

Other than a decoupling cap across the U1 power pins, looks fine. For the other half of U1, connect the two inputs in parallel with the two inputs shown and leave the output floating.

If I were going to optimize it for production, I'd recalculate R9 for a trip point of 50% of Vcc, make R6 and R7 the same value, and recalculate R8 for the amount of hysteresis needed. In this way, there are only two different resistor values instead of four. This lowers inventory costs, and reduces either manual assembly errors or placement robot setup times.

ak

 

With the values of the resistors on the hysteresis setup, your Off Voltage trip will be 8V, when your Ldr is at 60K, the preset needs to be at 24K approx which is around the centre position, looking good to me.

 

With the values of the resistors on the hysteresis setup, your Off Voltage trip will be 8V, when your Ldr is at 60K, the preset needs to be at 24K approx which is around the centre position, looking good to me.

Yes, thank you. This was exactly my thinking. This is also why I chose R7 and R8 values to more or less 'balance' out the voltage at the non-inverting input to about 8v in comparison to the inverting input.
I also like AnalogKid suggestion to decouple the power pins, but what size capacitor do you suggest - 0.1uF , 100nF or some other value?
Also to tie the unused inputs. I have had a look on several LM358 datasheets but it appears that most/many opamp datasheets do not mention what to do with the unused inputs...or maybe I've just missed any mention.
If I tie the unused inputs to ground, would that be suitable?
This circuit is in no way a production design, it's just home designed and built for my purposes and will only be a one-off.

 

If I tie the unused inputs to ground, would that be suitable?

Better to tie non-inverting to ground or the non-inverting input of the other amp, and tie the inverting input to the output.

 

Decoupling is not a suggestion. It is a universal requirement, which is why it is not mentioned on most datasheets.

0.1 uF and 100 nF are the same thing, so I guess either one will work. For future reference, if the opamp is making more than a few mA output current or is in a full-bandwidth audio circuit, parallel a 0.1 uF ceramic with a 10 uF electrolytic. If the opamp is running on +/- supplies, decouple each supply pin to GND separately.

ak

 

Decoupling is not a suggestion. It is a universal requirement, which is why it is not mentioned on most datasheets.

0.1 uF and 100 nF are the same thing, so I guess either one will work. For future reference, if the opamp is making more than a few mA output current or is in a full-bandwidth audio circuit, parallel a 0.1 uF ceramic with a 10 uF electrolytic. If the opamp is running on +/- supplies, decouple each supply pin to GND separately.

ak

Thank you, I am still learning. Everyday for everyone is a school day

 

Yes, I see the diagram you posted is using a photo diode but I have loads of the GL5528 LDR's which I purchased through AliExpress. They seem to be OK and lots of sellers there are selling them.

The circuit includes a high input impedance voltage follower to buffer the tiny photo current of the PD - you can probably leave that bit out if you use a LDR.

 

Being an "analog designer" myself I appreciate the design approach taken here. However, there are simpler ways to achieve this.

First, if it were me, I'd use a PIC and forget all the analog stuff. The LDR/resistor divider becomes an input to the internal A/D; this allows one to set the hysteresis to anything you want with a few lines of code.

Output; the key here which I think a couple of people have tried to say is you do not want to remain anywhere in the linear range of the transistor for any appreciable amount of time. This is where you will encounter your greatest I²R losses. Think of the transistor (no matter if it's a BJT, MOSFET, IGBT - whatever) as a "switch"; you want it fully on and fully off - get through the "in between" as fast as possible. Also, and I think even "pro's" overlook this sometimes; there are multiple reasons for a device to get "hot"; one of them is dropping too much voltage across the device. If at all possible, "match" the supply voltage to the LED/driver string as close as possible. You want to drop as much voltage across the string such that everything is sharing current. Not that it's being done here, but I've seen "designs" where they have 10 volts of diode and 50 volts of supply... not good.
Remember, I said "if it were me" I'd use a PIC, only because I've been using them for years, have the tools and can program them, and they cost about 50¢. That all being said, the very simplest approach...

Check out On Semi's line of constant current regulators like the NSI50150ADT4G (http://www.onsemi.com/pub/Collateral/NSI50150AD-D.PDF). These are specially designed for driving LED strings, are very inexpensive and butt-simple to use. They come in all varieties of current/voltage handling; well worth becoming familiar with. Look on page 7, Figure 12. It's demonstrating using PWM for dimming; your LDR/resistor can easily be configured to turn this device on/off with none of the heating problems you see (providing you watch voltages as I mentioned before).

Good luck!
MIke Tripoli

 

As I said yesterday, I have designed a circuit to switch on and off a 3528 SMD 12v LED striplight.
I have chucked the idea of using transistors and have come up with this:
View attachment 124106

This is using an LM358 and an IRF530 MOSFET.
I have tested the LDR against the light levels and have noticed that the LDR is about 60k ohms when I want the striplight to turn on so I have tried to design the circuit as close to this theshold as possible.
The R8 100k resistor controls the hysteresis so that the striplight does not flicker when the light levels are close to the switch on/off theshold. If the hysteresis is too close, I can change this value, but I will test it and replace the R8 later if necessary.
VR2 is an trimmer to adjust the theshold switch on/off levels/sensitivity of the theshold values.

I hope it's ok, but I will be anxiously waiting for your thoughts and comments.[/QUOTE

Since R6 and R7 are only used as a reference voltage, you can set their values much higher to reduce drain on the battery. IE 600K each or more for 6 volts.

 

Decoupling is not a suggestion. It is a universal requirement, which is why it is not mentioned on most datasheets.
ak

But why are unused pins of op-amps rarely mentioned on data sheets?
If you look at most data sheets for the LM358, (I haven't looked at other op-amp data sheets), these data sheets do not appear to mention what to do with unused pins.

 

Being an analogue guy I'd simply keep the two transistors of your original circuit but use a MOSFET for the output. THe big problem I see is that you have no current regulation in the LED strip. I take it that this regulates the current itself, in which case a MOSFET with hysteresis resistor should be fine and not need a huge gate drive. If the strip needs a current limit then you need to add a resistor in series with the source and use a bipolar NPN clamp (another 2N2222 would work) to limit the current. A value of 2.2 ohms will limit the current to about 300mA but can be trimmed b y varying the resistor value.

 

But why are unused pins of op-amps rarely mentioned on data sheets?

For the same reason they are not mentioned on the datasheets of CMOS digital logic, Pentium processors, and CK722 transistor; another universal law: Except for the ULN2003 and its kind (parts that have input conditioning built-in), never ever leave input pins unconnected.

ak