I'm going to have to study this some more. I shipped thousands of linear regulators wherein the LM723 did not refuse to output below 6 volts. Apparently I need my beliefs adjusted.

Were you using the Dip ones with the built-in Zener? The metal can version that the OP is using doesn't have enough pins to have the Zener.

 

I'm going to have to study this some more. I shipped thousands of linear regulators wherein the LM723 did not refuse to output below 6 volts. Apparently I need my beliefs adjusted.

The datasheet says 2V min.

 

The datasheet says 2V min.

The guy who designed the 723 - was it Roger Widlar? - was a good deal cleverer than me and he thought a 6V zener was a good idea.

 

Thanks, #12 for your quick response.
The battery voltage will recover up to 12,6V after load is off. My opnion is max 0,2V inaccuracy is well enough for set point (0,5v a bit high). I already tested with a connection between pin 6 to wiper of the pot but with R much bigger than you proposed, say 390k or 470k but no help. But now I try with a 47k as you proposed and I tell you the result soon.

Hi

Two points -

1) I see no hysteresis anywhere (then again maybe I am going blind).

2) without Vz the base current of the transistor (in the on state) would be approx. 10V/R4. If we assume gain of 10 (the transistor is saturated after all) it needs approx 2mA so let R4 be 5k. The low voltage is 2V so 5k would give maybe (2-0.7)/5k = 0mA26, but what if the gain were in fact 50?

When I used a 12V 0,5A incandacent lamp as load the flickering occurs at transient stage. When no load is used or just a led as load relay operates as just a clik off and a clik on and no flickering.

 

Attention #12, pls. be informed
While using a 12V 0,5A incandacent lamp as load, I tested the circuit starting from 12k to 820k including 47k with other various resistors between pin 6 to wiper of the pot, still no help.

 

Hi

Post 12 shows added hysteresis which should clear the flicker problem. Post 12 simulation shows it working. Use post 12 circuit.

 

Since you are using the version of the '723 without the built-in Zener, you require the external one as I show here.

I have added some positive feedback to create Hysteresis. The lower R7; the wider the separation between cut-in and cut-out.

View attachment 92528

You put back the wrong-way relay logic that helps discharge the battery after it is already discharged. I showed you how to avoid that way back in post #4 using a PNP transistor in-place of the NPN.

Now I agree your proposal to use a PNP transistor ( in this case 2N2907A) rather than NPN transistor. I constructed the circuit you prosed as in #4 on a breadboard with dip package LM 723 and Vo ( pin 10) either full supply voltage (11,7V or 5v something while playing with pot R1. But the transistor is not saturating so the relay is not pulled in. What could be the reason?
When I connected the ralay to Vo it is pulled in, load is powered and no flickering any more. But, relay is always staying as pulled in even playing with pot R1. To relase it I had to put power off again.

 

Just jumping in quickly with an alternative suggestion. I built a very simple circuit to do the same job a few years back and used a 741 as a comparator driving a med power transistor to operate the relay. I think it was an mje3055 but a tip 41a or similar would be suitable. You will will need to add some hysteresis to the circuit by putting a 100k-1meg resistor between the o/p pin and the +input. You could put a 1meg preset in series with a small (47k?) resistor to adjust the amount of hysteresis so that you don't get relay flutter.
It is not essential to use a 741 or a dedicated comparator as nearly any of the op amps will work just fine. When I was using this sort of circuit to control the HV charge on some capacitors, I used whatever was available, TL071, TL081, 741 etc. and all worked just fine.

 

Hi

I am not sure about using a pnp. To turn off a pnp the output has to go virtually up to the rail and the datasheet says minimum differential
3V.

Have you considered using the npn version with zener, and swapping pins 2 and 3?

 

If we are going to change horses, there are many better ways (fewer parts count) to go. A '741 would be a big step backwards...

 

If we are going to change horses, there are many better ways (fewer parts count) to go. A '741 would be a big step backwards...

Why, if it serves the purpose? the lm723 is also very old in the tooth and he is currently using more parts than a simple op amp comparator would need

 

Here is an improved circuit with the PNP reversed relay logic. The problem in the circuit I posted in #4 is the min. Vdropout of of the '723. Just like you need a Zener in the base of the NPN, you also need one for the PNP, but it needs to be ~3V, and the current flows the other way. The Vf of a white LED is about 3V, so I show how to place one to offset the inability of Vout (pin10 in mine) to pull high enough to turn off the PNP.

I also show how to put the hysteresis in this circuit...

 

Why, if it serves the purpose? the lm723 is also very old in the tooth and he is currently using more parts than a simple op amp comparator would need

Can you do it with fewer parts than this?

 

Why, if it serves the purpose? the lm723 is also very old in the tooth and he is currently using more parts than a simple op amp comparator would need

Then again 723 does contain the voltage reference and can drive the relay directly

 

Can you do it with fewer parts than this?

View attachment 92589

changing horses? but with the op amp it is so easy to get adjustable and repeatable performance, the trigger point of most thyristors tends to vary from device to to device even of the same type and can be very temperature dependant. Of course if minimum component count is the prerequisite, why not put the relay coil in series with a zener diode, as soon as the voltage drops below the zener point, the relay coil will be de-energised. But again, probably temperature dependent and not the best way to do it.

 

...the trigger point of most thyristors tends to vary from device to to device even of the same type and can be very temperature dependant.

The TL431 is a "Adjustable Precision Shunt Regulator". It is uses a band-gap reference internally, so is likely an order of magnitude better than a Zener.

The exact trip point can be trimmed by playing with the R1/R2 voltage divider. The relay is pulled-in as the voltage at the Ref terminal crosses 2.495V.

 

The TL431 is a "Adjustable Precision Shunt Regulator". It is uses a band-gap reference internally, so is likely an order of magnitude better than a Zener.

The exact trip point can be trimmed by playing with the R1/R2 voltage divider. The relay is pulled-in as the voltage at the Ref terminal crosses 2.495V.

Sounds like a very useful device, thanks for the info, I'l probably find a use for it!

 

Hi again.
I never give up but demoralized sometimes and I let the subject cool for a while. I even considered to make a new circuit diagram by using LM 741 which is a single opamp and I had a couple of it. Your proposal at #33 seems a good one with very few components, also. But, I didn't want to change the horse just in the middle of the river. In the mean time, I prepared a home made voltage regulator giving stepless voltage output within my working range (10-15VDC) with use of LM 317. Thinking that maybe my circuit testing method by playing with the pot is not the correct way to test the circuit.
I also checked the pcb thoroughly for any cold solder joints.
I tried several other possibilities, still no use.
BUT fortunately, I think, found a solution to the hysteresis problem of the circuit by trial and error:
Please, refer to diagram at #12;
I added a resistor parallel to C1, between IN+ and FCOMP valued 56k which gives a differential voltage of 0,4V. I suppose 39 to 56k is good range to work with. As the value increase diff. volt. becomes less and goes towards hysteresis point. When the resistor has a value of 15K diff. V becomes almost or over 1,0V.
My questions about my solution are as follows:
1) What do you think about this solution, is circuit is reliable now, any damage to IC could occur or not?
2) Can MikeML make a simulation for this? How much mA will this with draw at standby case (relay dropped)?
3) I think I can make use of a PNP transistor if I reverse the connections of IN+ and IN-. Is this possible and any mA advantage at standby case for this alternative?
Hope to hear you soon.
Thanks in advance.

 

Some observations from my experience and without having looked into this deeply.

if you are getting some oscillation when the coil turns on then there is a good chance you are having power supply problems. Apply load > supply voltage drops making circuit disconnect load > supply voltage recovers and cycle repeats. Make sure you have plenty power supply, well filtered and decoupling capacitors. This is what scopes are for.

Regarding the zener diode, I have never seen an output pin that would not drop to zero but I have not looked at the circuit in detail. In any case, you might be able to use a voltage divider with two resistors. Make it a voltage divider that would give you 0.5 V at 6 V and 1V at 12 V. Unless the current demand is too great it should work.

 

Some observations from my experience and without having looked into this deeply.

if you are getting some oscillation when the coil turns on then there is a good chance you are having power supply problems. Apply load > supply voltage drops making circuit disconnect load > supply voltage recovers and cycle repeats. Make sure you have plenty power supply, well filtered and decoupling capacitors. This is what scopes are for.

Regarding the zener diode, I have never seen an output pin that would not drop to zero but I have not looked at the circuit in detail. In any case, you might be able to use a voltage divider with two resistors. Make it a voltage divider that would give you 0.5 V at 6 V and 1V at 12 V. Unless the current demand is too great it should work.

Regarding first paragraph, you seem to be fully right. Because, as load becomes bigger, the problem is starting, i.e. with a led as load no problem, with 0,5A lamp as the load the problem is there, with 10W lamp worse. So, I separated the circuit power and load power feed, the problem seems to be solved. Thanks for your idea.
Regarding 2. paragraph: You are right that output pin becomes never zero. But anyhow it is triggering the base of the pass transistor to put the relay on and off. Isn't that enough?
On the other hand, I couldn't under stand your proposal about voltage dividers. Can you say it schematically? Current demand from relay contacts are about 1-1,2A DC at 12V.