LM2917N F-V converter IC

[saiello]

Hi All, I've almost finished my project for a tachometer circuit. Testing went well until I appear to have hit upon a snag for which I need some help. It concerns the LM2917N frequency to voltage converter IC, which I can happily 'program' to give the correct output with a given frequency input ( 0-2500Hz ) using a frequency generator. The problem arose when I tried to vary the 'duty cycle' of the input frequency ( which is a square wave ), i.e. I adjusted the 'HI' time to make it a smaller percentage of the time 'LO'. Whatever frequency I input, if I adjusted the HI so that it was less than 0.1ms ( 0.0001s ), the voltage output from the LM2917N started to drop. I've tested the LM2917N with higher input frequencies with a Hi/Lo duty cycle of around 50% giving half-cycles much less than 0.1ms, and it still outputs a voltage ( although off the scale ). It is important that the LM2917N can accept low duty cycle input frequencies as it will be generated from a laser diode/phototransistor pair reflected off a rotating body where duty cycles can be as low as 1%. I've also tried testing using the laser diode/ phototransistor as the input in a 'real' test situation, i.e. reflecting off a rotating nut. With a given output voltage if I gradually block off the light recieved by the phototransistor, I start to get the same erroneous result, i.e a drop in output voltage, indicating that the reducing duty cycle is to blame. I've checked the datasheet for any issues concerning this but I can't seem to find anything relating to this. Has anyone had any experience of this? Any help much appreciated! 

Thanks,
Salvatore.

 

[audioguru2]

Hi Salvatore,
The datasheet talks about selecting R1, C1 and C2 values for a required response time.

 

[saiello]

Hi audioguru!
I remember you helping me out on my comparator problem!

 

[saiello]

Hi, I think I've found the problem! There's a formula in the datasheet for determining the maximum frequency that can be input: Fmax=I2/C1*Vcc=0.00018/(0.0000000047*7.57)=5060Hz. At this frequency, the half cycle pulse width would be 0.098ms which from my testing turns out to be just about the value that I determined was starting to give me problems. So the only solution I can see is to increase the pulse width at source, i.e. somehow delay the pulse generated by the phototransistor so that it remains a constant at around 0.1ms regardless of the input frequency or duty cycle. Anyone know how I might go about doing this? Maybe somehow using a capacitor on the input to the comparator to store the charge for a short time? I've attached a diagram of the input stage to give a clearer idea.

Thanks,
Salvatore.

View attachment 40293

 

[audioguru2]

hi Salvatore,
I thought the values of the resistor and capacitors of the IC will solve your problem.
A monostable "pulse-stretcher" can be made from a 555 or a couple of Cmos inverters.

 

[saiello]

Hi, If you can determine what values for R1/C1/C2 I need to solve the problem please let me know! Will the cmos inverters just invert the signal? If this is the case then it won't solve the problem as a