It is the same as the diagram in post 49 where I replaced r3 (100K)with a 2M trimpot

That capacitor, C1, may be causing a problem. ... It might cause some distortion to a square wave input. So try removing it from the circuit and see if there is any improvement.
... That is, replace C1 with a wire segment, and observe any different results.

 

You can try that but change C to 1uf

I used 2 100V caps (all i have here) in total they make up 1005 nF
With R set at 164K,( any higher does not seem to make any difference) the alarm LED is flashing
Running at 5 l/min, it keeps flashing same speed, but blowing air into it, stops the alarm.

With only power on (no sensor), It is now behaving erratically eg 5 flashes , stop 1/2 sec, 15 flashes stop 1/2 s. 6 flashes stop 1 sec maybe its a lot more flashes.. flashing too fast to count really, but you get my drift

 

Understand. The problem is the frequency is too low to accurately decyfer between 5 and 4hz. There are circuits that can effectively multiple the frequency of the pulse output that might be worth a try.
Best bet may be to find a sensor with a higher output frequency.

 

It suggests to me that the pulse train from the sensor isn't even. (Have you looked at it on a scope?) How about increasing the value of the integrating capacitor (on the LM2917 circuit) - that's the cap with the resistor across it. It won't alert you quite so quickly, but probably soon enough!

 

That capacitor, C1, may be causing a problem. ... It might cause some distortion to a square wave input. So try removing it from the circuit and see if there is any improvement.
... That is, replace C1 with a wire segment, and observe any different results.

Cutting out C1 and replacing with wire, gives the opposite effect as before, Now the alarm is off when powered up and the sensor not running, with the sensor running the alarm in on . Either way, it seems not to make much if any, diff in sensor speed. The mind boggles..

An other thing I noticed yesterday already, is that R4 gets quite hot,,, too hot to touch after 10minuts. R5 is set to 168K

 

It suggests to me that the pulse train from the sensor isn't even. (Have you looked at it on a scope?) How about increasing the value of the integrating capacitor (on the LM2917 circuit) - that's the cap with the resistor across it. It won't alert you quite so quickly, but probably soon enough!

Hello Ian , nice to hear from you again.
I'll try to borrow a scope tomorrow (actually later today), but my meter shows a pretty consistent frequency plus or minis 0.5Hz.
I presume you are talking about C2 in diagram on post #72 . where it states 1uF ?? I changed this to 100n as SG told me to do 2 months ago.

 

C2 on post #72 or C1 on post #78. It's the filtering capacitor. I'd try it MUCH bigger. 10uF would only have a time constant of 2 seconds. 47uF might be nearer the mark. How soon do you need to know that there is a problem?
The fundamental difference between the two circuits is that the 555 will alarm when any single period is too long, the 2917 will only alarm when the average period is too long. C2 will cause it to take the average over a longer period of time.
From the datasheet
"C2 acts as a filter to smooth the pulses of current and does not affect the output voltage. However, the size of C2 determines both the output response time for changes in frequency and the amount of output voltage ripple"

 

. How soon do you need to know that there is a problem?
The fundamental difference between the two circuits is that the 555 will alarm when any single period is too long, the 2917 will only alarm when the average period is too long. C2 will cause it to take the average over a longer period of time.
From the datasheet
"C2 acts as a filter to smooth the pulses of current and does not affect the output voltage. However, the size of C2 determines both the output response time for changes in frequency and the amount of output voltage ripple"

I like to know when there is a problem within 15 seconds, as if the pump runs dry for longer than 30 seconds
(speed related), the vanes will disintegrate.
Re C1 post 78, I don't have a 10 nor 47uF tantalum , but i have plenty electrolytic caps. Useful ?

 

Nothing wrong with good old aluminium electrolytics!
(Personally, I don't like tantalums, and haven't used one in the last 20 years)

 

Nothing wrong with good old aluminium electrolytics!
(Personally, I don't like tantalums, and haven't used one in the last 20 years)

I soldered a electro 47uF 50v instead of the .1uF. The alarm does not shutdown at any resistance or sensor speed
i put the .1u back. it shuts down alarm but only at high sensor speed.
it's the sensor speed that must be the problem. The water flow is too slow to reach the frequency.
SG suggested a frequency multiplier to get it to around 100Hz . or where to find a simple one? Any other ideas?

 

I borrowed an oscilloscope , attached a is a pic . The sensor is running at about 5 l/min and alarm on on.

 

So what is the frequency there? Looks like about 10Hz to me

 

sorry I Don't know how to use scopes. I just fiddled with it until i got a pretty picture

 

What goes around comes around.
This latest version was first suggested by drc_567 back in post #2, a missing pulse detector with a few extras and a few options.
Verified on the bench at 5Hz. The circuits of Q2 and Q3 provide a steady alarm condition when activated below 5Hz.
If the sensor stops turning and the output is high the alarm will sound immediately. If the output is low the alarm will sound after about 1 sec by the IC2 555 circuit.
SG

 

Hello SG, Yes,,, Just like the planet, round and round............
Man, you're on the ball, I forgot this altogether.
You've spent a lot of time thinking about & drawing this circuit, thank you so much..

I take it that this design works, as you say " Verified on the bench at 5Hz "
I'll have to draw it out in easyeda and order boards from JLCPCB.
Time will be around 2 weeks, minimum until i get one ready
No more involvement with 2907/2917 due to frequency problems.?
(I searched for frequency multipliers, but could not find anything applicable for square wave at 5Hz)

 

I tested a couple frequency multipliers and wasn't happy with the results so decided to start from scratch.
C2 and Rv1 are the frequency determining components based around 5hz.
I'm still surprised the frequency is that low.

 

Please post your redraw for review

 

Certainly. will do
Q, Are C2 -3-& 4 multilayer Monolithic ceramics or also electo like C1

 

All just ordinary electrolytic.

 

Added green LED to the output of Q2 to aid in adjusting Rv1.