[query] - part of electronic circuit to check

Thread Starter

Rafał Kotrys

Joined Apr 22, 2016
12
Hi Folks,

I'd like to make timegrapher to measure accuracy of watch. One element of project is sound sensor to collect "tic" "tac" sound from watch. I found ready to work project and I tried to figure out each part of it. Unfortunately I do not understand what is the marked with circle part of the circuit and why I have feedback from op amp positive with capacitor inside. Because I'm not very experienced in electronics I tried to match this element with:
-sallen key filter, band-pass filter, low or high pass filter
-kind of amplifier

but I couldn't find similiar example. Could anybody show me what kind of circuit is that and what is the purpose of having it in this place?

upload_2016-4-22_19-12-39.png
 

Jony130

Joined Feb 17, 2009
5,487
-sallen key filter, band-pass filter, low or high pass filter
-kind of amplifier
Jak dla mnie żadne z powyższych. Tu TL robi w roli komparatora ale układ jest kiepski i może się wzbudzić a tranzystor T1 nigdy nie będzie zatkany. Dodatkowo TL072 kiepsko działa przy tak niskim napięciu zasilania.
Try to find a new circuit or redesign this one.
 

Thread Starter

Rafał Kotrys

Joined Apr 22, 2016
12
I managed to find ready to work project which prefilters and amplifies microphone signal, but it gives on output analogue signal. I'd like to combine somehow previously provided method as it seems to look reliable (I don't know if it's the best, so I'd appriciate Your help in here) to use key transitior. I'd like to make also possibility to join the headphones and give opportunity to listen to signal from the microphone. May such solution work?

Here is the circuit:

upload_2016-4-22_21-57-30.png

Thanks for help
 

Dodgydave

Joined Jun 22, 2012
11,285
Your first circuit is an ac amplifier gain approx 120, with an integrater on the output to flash a led, your new circuit is again an ac amplifier with a gain upto 100, are you trying to pick up a ticking noise from a clock or watch?
 

grahamed

Joined Jul 23, 2012
100
Hi

The circuit is an OPA acting as a simple comparator but with a slight twist.

The rule of the comparator (for any OPA come to that is) - if V+ > V- then Vo is high and if V- > V+ then Vo is low.

A problem occurs with slow moving inputs, for a period the two voltages are approx. equal and Vo is not well determined.

The capacitor provides positive feedback. Say V- > V+ (Vo low) and V- is moving negative. At some time V- becomes just less than V+ and Vo starts to rise. The rising voltage is fed back to V+ making V+ rise i.e. higher than V- so Vo rises more, and so on.

Unlike the Schmitt trigger this circuit does not exhibit hysteresis.

(where V+ is voltage at non-inverting input and V- is voltage at inverting input. and Vo is output voltage).

OK?
 

Thread Starter

Rafał Kotrys

Joined Apr 22, 2016
12
Your first circuit is an ac amplifier gain approx 120, with an integrater on the output to flash a led, your new circuit is again an ac amplifier with a gain upto 100, are you trying to pick up a ticking noise from a clock or watch?
PD6 is output to uC which will count the signal and compare it with x-tal frequency to measure accuracy of watch mechanism.

What I'd like to do is this:

upload_2016-4-22_23-29-33.png
as I was told in 1st response circuit which i 1stly showed was wrongly designed therefore I tried to use projected solutions which are sold and validated (2nd circuit)

Could You please help me and validate what exactly should be change or which part should I redesign to make it work. Also here is the link to full project (unfortunatelly in German and I had to use Google translator)
http://mc.mikrocontroller.com/de/Zeitwaage.php
It seems for guy who designed it everything worked correctly, but since I got info from Jony130 I got confused:(
 

RichardO

Joined May 4, 2013
2,270
Keep in mind that the clock frequency of the circuit measuring the tick frequency must be very accurate. A typical uC crystal is nowhere near accurate enough.

To measure an error of one minute out of a month would be about 60 minutes x 24 hours x 31 days = about 1 part in 45,000. An inexpensive crystal is typically 100 parts per million = 1 part in 10,000.
 

Thread Starter

Rafał Kotrys

Joined Apr 22, 2016
12
RichardO. I'm going to use external crystal 8MHz. According to datasheets should meet requirements of uncertainty 0,01ms. I was rather wondering if You could help me to identify mistakes and how to fix them or at least where could I read how to fix them. If You are not happy to provide me this help, for whatever reason, let's close this topic:)

Regards,
Kuba
 

RichardO

Joined May 4, 2013
2,270
I'm going to use external crystal 8MHz. According to datasheets should meet requirements of uncertainty 0,01ms.
.01ms out of what total period? For example, if the 0.01ms is out of 1 second that is very inaccurate -- only one part in 100.

What is the part number for the crystal you are looking at? Keep in mind that no matter how accurate the crystal, you will have to calibrate it in _your_ circuit to get the exact frequency specified for the crystal. This requires using a known frequency standard as a reference. Here in the U.S. that would be the radio station WWV or WWVB.

You are better off looking at crystal oscillator. It is already calibrated and ready to use. The more accurate the calibration and stability of that calibration, the more expensive the oscillator.

Here is an example of an inexpensive 10 MHz TCXO (Temperature Compensated Crystal Oscillator):
http://www.mouser.com/ProductDetail...=sGAEpiMZZMt8oz/HeiymAOqzUXC3weGFrHcwaz7vvYI=

Another frequency standard that is readily available is from a GPS receiver such as this one:
http://www.digikey.com/product-search/en?mpart=FTS250-010.0M&v=222

This one is quite expensive ($215.00 !) but you may be able to find a much cheaper version.
 
Top