hi,
If you want such low level thresholds you must use a stable 5V ref at the top end of the divider chain, also some power rail decoupling for starters.
E

Actually, doesn't any noise on the 5V scale down ratiometrically? So, if you're using a voltage divider to scale 5V down to 50mV, that's 1% of the original voltage.

Correspondingly, if your 5V had 50mV of noise on it, that scale down to 1%, yielding 0.5mV of noise at the inverting output. Of course l know that there are other noise sources that will factor in and be significant when dealing with such small signals, but l would've thought that reasonable noise levels at the top of the voltage divider would scale down and still be reasonable at the inverting input.

Don't get me wrong, I'd never argue against better supply filtering, decoupling, etc. but I wouldn't have thought that the voltage divider section in particular needed extra attention.

Please correct me if I'm wrong - I'm always learning, and small signal noise considerations certainly aren't my strong suit.

 

hi,
Its the Vref that must be stable.
If the TC's 5V rail is not stable, then the Vthreshold he has set will change as the 5V changes.
Regarding the 'noise', it could cause false triggering when the Vin is close to the threshold.

E

 

For the small Vref around 1mV ~100mV, you may use R1=2K, R2=200 Ω, R2 may call it VR1, or R1 = 1K, R2 = 100Ω, and in parallel with a 0.1uF with R2, if you want to get a lower output voltage, at least adds -Vcc to -1V or use rail to rail op amp.

Edit:
When you want to get the voltage as mV then you better use the Linear power supply.

 

hi,
If you want such low level thresholds you must use a stable 5V ref at the top end of the divider chain, also some power rail decoupling for starters.
E

well, I want only the signals with amplitudes above 50 mV. I have done some calculations to what to expect as an output. Also I did observed the signals on the oscilloscope to get some ideas about the signals I have.

But did you have that connected when you were adjusting the pot and getting the offset in post #1?

yes, before I build the final board, I am prototyping. Thats when I hit the low level threshold stone.

Maybe it is better to amplify the signal I monitor? 3-5 times amplifications will bring me to the save zone?
yes, I did observe whats going on on the oscilloscope,

 

hi dude,
This LTSpice sim of you circuit, shows a threshold of approx 170mV
E

 

hi dude,
This LTSpice sim of you circuit, shows a threshold of approx 170mV
E

Hi!, does it means that this is the lowest possible stable threshold for LM311N?

 

hi,
No, its a level that you stated in an earlier post, for demo only.
E

 

For the small Vref around 1mV ~100mV, you may use R1=2K, R2=200 Ω, R2 may call it VR1, or R1 = 1K, R2 = 100Ω, and in parallel with a 0.1uF with R2, if you want to get a lower output voltage, at least adds -Vcc to -1V or use rail to rail op amp.

Edit:
When you want to get the voltage as mV then you better use the Linear power supply.

Well as I know the Vref, I can try this now as you suggested.

hi,
No, its a level that you stated in an earlier post, for demo only.
E

yes, I was using variable R2 to play with the threshold voltage, Vref. As I know more or less the value of the latter, I can set the voltage divider.

BTW, using dual power allowed me to set the Vref to about 40 mV, but it is noisy. I will try to compile some oscillograms if someone is interested. I will set the design value of Vref to about 50 mV.

 

hi dude,
Have you added a stable Vref divider supply and also power rail decoupling.?
I would spilt the 20K into say two 10k's in series, connect a 100nf or 1uF at the junction to 0V.
E

BTW:
I would consider Scott's point regarding using a 2k at top end of the divider [replace the 22k], the lower end pot will give you better control.

 

Just to update the status of this project:

I have amplified the input signal (about 10 times) and used a dual power supply (for the comparator and the op-amp). That allows me to set the Vref to 250 mV which is already at the safe range (above 100 mV). I've addaed some pulse shaping before - a differentiator before and integrator after the op-amp in order to reduce the tail of the input pulses and eventually to differentiate possible pile-ups during high count rates. LIke this I could use the material I had and avoid looking for rail-to-rail. I am working in really noisy EMF environment but the circuit looks stable. Now its time to count the pulses I guess I need counter IC and a beagle bone .

 

hi dude,
Pleased to hear you are making progress, lets know the spec for counter etc.
E

 

I need to count the output from 9 detectors using a Beagle Bone Black(BBB).

The max count rate are pretty moderate: 5 kHz.

The issue to solve is how to count them simulataneously (for example every minute) with the BBB?

I guess I would need a counter IC that will be able to count up to 5 000 and then a multiplexer to collect that data by the BBB?

 

Then a reciprocal counter possibly make's sense.

Questions -

1) Resolution of measurement required ?
2) Latency limitations, eg. time measurement takes ?

Some useful basics -

https://www.dropbox.com/s/dmnyb69ntfgd3o3/Counters.zip?dl=0

Regards, Dana.

 

What' the purpose of counter, just counter as CD4518 or divider or for time delay?