741 op-amp supply question

[Martin G7MRV]

I dont do a lot of audio stuff really, but i'll bear that in mind with these.

One further question - regarding the supply voltages and the devices ability to get the output 'close' to the supply rails, how exactly is that specified in the data sheets?

My mock-up with two 741s works (the first acting as the battery level detector, its output being used to pull down the seconds GND pin, turning the device on. The second is the astable providing the pulses) but I can see the cumulative effect of the output of the 741 not getting close enough to the supply rails, in that the 1sts output only goes low to about 2v (with the supply at 19v for the threshold of detection), and as such the 2nd now has an output that only goes down to about 3.5v. Since im driving an optocoupler, the result is that it is always slightly on!

Looking at the LM358s datasheet (since i'll be moving the circuit to this device), I can see specs for 'output voltage - high limit' and 'low limit'. For the high at Vcc 30v it specifies 28v. For the low, it only specifies Vcc 5v but a limit of 5mV. So, does that mean that the maximum output voltage will be Vcc -2v (so for my 19v threshold input an output of 17v), yet the minimum output voltage almost 0v?

If thats the case then the cumulative error will still be way lower than the turn-on voltage for the optos LED, and so it should sort my problem!

Diagram here https://g7mrv.blogspot.com/2019/03/3...y-circuit.html

I suspect with a device that can reach closer to the rails this effect will be eliminated. If not, i'll have to add to the component count yet again!

[kalee20]

Originally Posted by Martin G7MRV

For the high at Vcc 30v it specifies 28v. For the low, it only specifies Vcc 5v but a limit of 5mV. So, does that mean that the maximum output voltage will be Vcc -2v (so for my 19v threshold input an output of 17v), yet the minimum output voltage almost 0v?

Yes! The LM358 has a really good pull-down (NPN transistor, though I didn't realise it was QUITE that good!), and an average pull-up.

The input stage, too - works right down to the negative supply rail (and even a few mV below!), though can't cope near the positive supply rail.

[julie_m]

That's because the LM358 has PNP inputs. The 741 and 1458 (dual version of 741) have NPN inputs, which can work nearer to the positive rail than the negative rail.

[Martin G7MRV]

Smashing, thats something else demystified for me!

Looks like the 358 will get me down below the 1.2v that the PC357 optocoupler turns on at. Just need to find a low cost surface mount 100k preset now!

It might be due to how maths heavy it was made at college, but op-amps have always baffled me!

[Martin G7MRV]

oh...

Just realised the flaw in my fiendish master plan! In my circuit I use the 1st op-amp output to pull down the ground of the 2nd op-amp in order to turn it on...

...something I clearly cant do with a dual op-amp that shares the supply pins!

Hmmm. Looks like i'll still need two devices, unless I can work out a way around it

[Trigon.]

Originally Posted by Martin G7MRV

For the high at Vcc 30v it specifies 28v. For the low, it only specifies Vcc 5v but a limit of 5mV. So, does that mean that the maximum output voltage will be Vcc -2v (so for my 19v threshold input an output of 17v), yet the minimum output voltage almost 0v?

Technically yes, but it's rather misleading - there's very little current available from a 358 below 0.6v output...



Cheers

[julie_m]

This is how I'd do it. The op-amp has its own Zener-stabilised supply via R6 and ZD1; a portion of which is tapped off via R2, VR1 and R1 to set a precise reference. R4 and R3 divide down the unstabilised 24V supply. As long as this is higher than the voltage set with VR1, the output will be low. As soon as the 24V supply drops below the set point, the op-amp output will go to +12V. D1 then conducts and R5 (which should be at least 5 * R2) shifts the reference voltage even higher to ensure a clean snap action.

Reduce R6 if the op-amp is driving an LED directly via a resistor.

[kalee20]

Originally Posted by Radio Wrangler

Treat the 741 as the default starting point for thinking about opamps.

It was the first really popular one. Others came before, but had inconveniences that the 741 brushed away...

You had to design things so that inputs and outputs stayed comfortably within the rails, but the little opamp got on with it and did the job.

It was a bit noisy ... but you could still do a lot with it.

It has been bettered in every single parameter by newer designs, but not in all by any one...

Absolutely agree. I believe it was a Bob Widlar design. It was a breakthrough - short-circuit proof, no additional components needed to 'tame' it. As an easy-to-use gain block, it offered incredible value for money. The 8-pin DIL package pinout became the de facto standard for nearly all the subsequent designs, to this day. It was hugely popular. And for a reason.

Having said all that, I still have to specify it!

Originally Posted by Argus25

The 741 has a low slew rate (by modern standards) and makes for a great trouble free "comparator" that is virtually fool proof where the slew rate is not important and it has intrinsic RF immunity too.

Agree to a point - with significant RF input noise, the input stages can get overloaded and start rectifying the input signal. You might get asymmetric currents flowing through input resistors. Meanwhile, the internal compensation capacitor (which virtually defines slew rate) stops the output swinging around. The net result is that you seem to get unusual DC offsets and things don't swing quite at the level you expect them to.

Me? I usually follow my comparators, when it matters, with a latch (74HC74 type of thing), or maybe a thyristor. I hate the trick of hysteresis by a bit of positive feedback. It makes setting the trip point so difficult!

[julie_m]

Originally Posted by kalee20

Me? I usually follow my comparators, when it matters, with a latch (74HC74 type of thing), or maybe a thyristor. I hate the trick of hysteresis by a bit of positive feedback. It makes setting the trip point so difficult!

That's why I used D1: to keep the trip point exactly where you would expect it to be until the output goes high. There will be no positive feedback while the output is low.

[mhennessy]

The 741 was by David Fullagar (Fairchild). Bob did the LM301.

The main complication with hysteresis in an op-amp circuit is variability in supply rails and the drop-out voltage of the output stage when sat at whichever rail it's at, which in turn complicates the maths for the threshold voltages. For most applications where only a modest amount of hysteresis is required, it's normally not an issue. It's rare (and often risky) to have no hysteresis at all.

Agree the LM358 is a handy thing to have around - I buy them in tubes of 50! Glad someone else spotted the output limitation (post #26) - the internal schematic explains that. Not a problem if driving a load that is returned to the negative rail, which is often the case for single-rail circuits. Working down to 3V is pretty handy too...

I'm guessing this is a big battery, so power consumption is not an issue, but for what it's worth, the last time I wanted a simple low-battery detector circuit, I came up with this, which draws about 200uA (thanks to using low-power LEDs). That's less than many op-amps (the 741 typically draws about 2mA).

https://www.markhennessy.co.uk/dvm_r...tery_indicator

[llama]

To interrupt the 2kHz tone you could maybe make the single 741 be its own LF relaxation oscillator. Squegging I think it's sometimes called. So when it triggers "on" it sends an RC coupled level to the input to trigger it "off" which stays that way until the C charges and lets it all commence again.
Graham

[llama]

Sorry, not "squegging" - that's when an oscillator stops for an interval and then restarts, only to stop again after the interval.
What I describe above could be considered a VLF oscillator whose operating conditions depend upon its DC inputs.
Graham

[Martin G7MRV]

llama,

the 2kHz oscillator is an integral part of the radio, not part of my circuit. It is controlled by switching a 3v line within the radio, which is part of the local/remote system. By circuit just switches that 3v line, hence the use of an optocoupler, which makes the interfacing really easy. But it also has the advantage that the same circuit can be readily adapted to other equipment, or other indication techniques, such as the pulsed noise method used on other radios in the Clansman range for low battery indication.

I have now redesigned everything fully to use a single LM358 dual op-amp, in SOIC/0603 SMT, and in fact have PCBs on order!

My final task is to play with component values to get the detection threshold, pulse rate and duty cycle just right, and current consumption to minimum.

mhennessy - typically a 4Ah NiCd battery, although some use the small 1Ah, so not a great worry over current consumption, although I do with to reduce it as much as possible. Currently my circuit takes some 5-12mA, depending on detected state.

[mhennessy]

That sounds like quite a lot - I wonder what's responsible for that much? Looking at your most recent blog entry, I'm struggling to see where 12mA could be going.

I'd be tempted to reduce the current for the opto-coupler LED - last time I played around with a "jellybean" 4n25 type, I found it was perfectly reliable at 2mA.

The resistor feeding the base of the transistor is rather low, taking a milliamp or two - you could up that to 100k with no issues, I'd have thought. Likewise, you could probably increase the two 10k resistors in the oscillator and save half a milliamp maybe.

Those changes might add up to around 5mA, perhaps. Other than that, perhaps the chosen values of RA and RB are a bit on the low side?

If low current was really important, replacing the zener with something else would help. I see there's about 1.2mA of operating current in that - which is a bit on the low side, to be honest. The ubiquitous TL431 works well at 1mA, and is one of the more useful parts that can be recycled from a switched-mode power supply. Hardly worth worrying about here, but mentioned for completeness...

Either way, it's a neat little circuit

[Martin G7MRV]

I have already tried bumping up various of the 10k resistors to 100k, but oddly the current seemed to have increased?! That said, that was on the mock up with the 2x 741, as my 358s havent arrived yet. But yes, I think a fair bit of it is in the drive to the opto, so I should be able to save some there.

My tests were fairly quickly done, so may not be tremendously accurate, i'll be doing more careful measurements once the 358s turn up and I can swap out the 741s.

I'd be happy I think if I can get the current down to less than 10mA during an active alarm state.

The SMT PCBs using 0603/SOIC come out at around 2.5cm x 1.5cm, which will comfortably fit in the only free space inside this radio!

[llama]

Originally Posted by Martin G7MRV

llama,

the 2kHz oscillator is an integral part of the radio, not part of my circuit..

I know, that's why I corrected my incorrect use of "squegging".

I'm just suggesting that adding a few components (a resistor, capacitor, maybe a diode) to the original 741 single op-amp circuit could make it self-pulsing.

Still, your SMC version is likely to be smaller than just the capacitor I'm suggesting!
Graham

[Martin G7MRV]

Ah, sorry, misunderstood! I see what your meaning now.

[Martin G7MRV]

Ive had chance to make some more accurate current measurements, but still on the 2x 741 mock-up.

At 26v supply, so above the detection threshold, the circuit draws 3.52mA/5.79mA. At 19v supply, below the detection threshold, 2.76mA/4.20mA. The higher figure in each represents the astables 'high' output state. This is with the new 8k2 series resistor for the optocouplers LED, giving an Iled of about 2mA. The opto does seem to work perfectly at this.

These figures of course include the fact that the detector op-amp cannot go all the way to ground, and so there is always some drive to the optocouplers control transistor, and hence a bit of leakage! With the opto removed from circuit, the circuit seems to actually use less current when the astable output is high! - 26v supply 3.47mA/2.95mA, 19v supply 2.53mA/1.98mA, again the second figure represents the astable 'high' output condition! Of course, that high state value now doesnt include the LED current for the optocoupler.

This is with all the original 10k resistors swapped for 100k, including the transistors base resistor. This is well below my original <10mA target, and I think it will be a little lower again once its using the LM358s.

[Martin G7MRV]

A bit of an update on this, now that ive received the LM358s (should anyone be at all interested!)-

Having swapped the two 741s on the breadboard for a 358, I was somewhat dismayed to find it didnt work! (this was worrying as Iive had 10x SMT PCBs made up!)

Lots of checking showed the circuit was correct, and I was starting to think I had duff chips (they have come from china cheap!), but -

Testing showed that at least one op-amp was working, as the sensing part of the circuit operated correctly. I then made a rather odd discovery - it would all work, but only if I had the tip of my multimeter probe on pin 5!!!

How very odd! I thought this might be a poor connection issue then, so checked, lifted and reseated everything, still not working, then I found it would also work if i touched pin 5, or the resistor leg going to it!

I also found that when working, whereas the 741 combo drew a few mA, the 358 drew 650uA!!!

I havent changed the values yet to find out, but im beginning to think that the 100k feedback resistors that go to pin 5 are now too high in value! I'll try dropping these back down to 10k later. My reasoning is that my finger, or the multimeter impedance, is effectively in parallel with the 100k resistor from pin 5 to ground, so lowering the effective resistance.

[Martin G7MRV]

Absolutely at a loss here!

Swapped the feedback resistors from 100k to 10k and it still will only work if I touch pin5! Im not touching anything else at the time, I can touch it with the tip of my insulated long nose pliers and it will work, but yet not a length of wire In both cases im not directly touching the pin! I tried adding a capacitor to ground, that just stopped it entirely! Changed the IC, just the same (unless both chips are faulty in the same way!)

Would anyone be willing to knock up a copy and see if you get the same weird results? Diagram here https://g7mrv.blogspot.com/2019/03/temp-test-post.html