Any suggestions for threshold component ?

[AugustineLeudar]

Hello all,
we have designed a relatively simple circuit that amplifies tiny electrical signals in nerves (these signals are known as action potentials) . We use an operational amplifier with a very high input impedance (10 gigohms minimum). The background noise is typically 20 mv even with a differential circuit but the action potential is clearly visible above this typically at about 160mv. Now I want a component that once the voltage rises above a certain level (say 80mv) it triggers an LED to go off - even better if the brightness of the LED could corrospond to the signal. Any suggestions what sort of component could act as such a threshold trigger ?

 

[albert22]

Did you considered a comparator like this?
https://www.ti.com/lit/ds/symlink/lm339-n.pdf
Its output is on off and probably you will need to amplify the signal a little more.
But it could give the impression of varying brightness depending on the characteristics of the signal.

 

[chuckey]

Another alternative to consider is a moving coil meter. These do not have any threshold but do have a finite "rise time" of the needle, so while it will flicker, you may find that its got the right effect. Most moving coil meters have a 100mV voltage for their full scale current, so a 1mA meter would be 100 ohms, a 100 micro amp meter, 1000 ohms. See if you can borrow one to try out.
Frank

 

[Darktrax]

Using the open input of a FET op-amp at 10GOhms is almost an amplifier where you might be tempted to calculate current as a few tens of million electrons instead of using pico-Amps!

Consider that measuring across skin resistance normally is in Meg-Ohms, so choosing a first amplifier to have approx 1GOhm impedance might dramatically lower the noise.

Measuring off the body, (aside from various safety considerations) using differential input high impedance amplifiers requires care to cancel common mode noise, but I am assuming this is all known to you. At some point in the amplifier instrumentation chain, there should be an op-amp follower, with no voltage gain, but with a considerable power gain to be able to drive your instrumentation. This is the point to tap off a voltage sample into a transconductance amplifier. You can get them ready-made, or you can contrive them with regular op-amps, but the idea is to get a current output proportional to a voltage input.

Achieve threshold by biasing the other input, and feed your LED. The LED brightness will vary in proportion to the current, which is in turn proportional to the sampled voltage waveform.

At least - that is the approach I would try..

 

[dick_freebird]

What do you know about the signal of interest, besides
its voltage? Other attributes such as duty cycle, dV/dt,
etc. can be useful discriminators.

If the event is very sparse, then with a low pass filter you
could obtain a baseline offset value, add some margin to
that (perhaps, programmable) and there's your threshold.

If your signal has a fast risetime and then a level of interest,
a high-pass filter followed by a one-shot could create a
gate for sampling and subsequent comparison.

Think about what distinguishes the waveform of interest
from the various ambient noise sources and turn that into
a tag, flag, threshold, whatever.

 

[AugustineLeudar]

Hi all,
thanks for the replies. I forgot to mention - the 20mv - 120 mv thing is before it goes into the operational amplifier - after that Im not sure what the voltage is but I assume considerably higher as its already amplified - normally we just go into a data aquisition board and then into labview - but here I would like to power an LED instead. I am not an electrician but the reason for such high input impedance is to match the input impedance of the plant (yes we're measuring electrical signals in plants) - apparently it shoud be as high as possible - I dont really get that - surely the high input impedance would block the tiny signal we're trying to measure - anyone can explain that to me Id be grateful ! SO yes normally we would use differential (DLR) circuits - but I am assuming out in the field there will not be o much noise so .... Anyway I was hoping to build these things without a daq or indeed any digital components - so what would be an alternative to the DAQ (ie op-amp follower ) ?

 

[Darktrax]

the reason for such high input impedance is to match the input impedance of the plant (yes we're measuring electrical signals in plants) - apparently it shoud be as high as possible - I dont really get that - surely the high input impedance would block the tiny signal we're trying to measure

Since I found out we all share plant DNA in our makeup, and that some static plant-like life is actually animal, I would not be too concerned about that.

Regarding the high impedance. Do not confuse that with "having to fight its way through a resistance". You cannot get any information from anything without also changing it a bit, but if you can take so little that you do not substantially load the thing to collapse the voltage significantly, then your sample is of a representative accuracy.

Nerve cells are not power stations! The voltage they have can only source a tiny current without collapse.
Connecting into an amplifier that will accept and amplify that tiny current without the amplifier input itself being an unacceptable load, means that the input appears as if it presents a very high impedance to the plant. The voltage available divided by the current gives a value in Giga-Ohms.

I would not be too eagar to use "as high as possible" as a design philsopy. I would go for "as high as needed, and not more". The advantages of finding the effective tradeoff are many. 500Meg-Ohm or 1 Giga-Ohm may still be high enough for the plant sample, and much easier to design with and find devices for, and cope with noise. at 10Giga-Ohm, you have to build it on glass or PTFE, and worry about surface contamination leakage. It gets harder to tell what is circuit and what is substrate.

 

[AugustineLeudar]

Since I found out we all share plant DNA in our makeup, and that some static plant-like life is actually animal, I would not be too concerned about that.

Regarding the high impedance. Do not confuse that with "having to fight its way through a resistance". You cannot get any information from anything without also changing it a bit, but if you can take so little that you do not substantially load the thing to collapse the voltage significantly, then your sample is of a representative accuracy.

Nerve cells are not power stations! The voltage they have can only source a tiny current without collapse.
Connecting into an amplifier that will accept and amplify that tiny current without the amplifier input itself being an unacceptable load, means that the input appears as if it presents a very high impedance to the plant. The voltage available divided by the current gives a value in Giga-Ohms.

I would not be too eagar to use "as high as possible" as a design philsopy. I would go for "as high as needed, and not more". The advantages of finding the effective tradeoff are many. 500Meg-Ohm or 1 Giga-Ohm may still be high enough for the plant sample, and much easier to design with and find devices for, and cope with noise. at 10Giga-Ohm, you have to build it on glass or PTFE, and worry about surface contamination leakage. It gets harder to tell what is circuit and what is substrate.

I can only go on what all the experts in the field of plant electrophysiology have told me - that less then 10 gigaohm is no good - we have some op amps that are 10 gigaohm - I think the cost 5 euros - and most of the electrometers built for action potential readings are at least 10 gigaohm - they seem to work ok - in fact the 5 euro one works as well as the expensive electrometer giving the same reading ! But thanks for the explanation of the need for input impedance - I think I understand a bit better now.

 

[FvM]

The discussion about required amplifier input impedance is a bit pointless unless knowing about the involved electrode type. I won't hold anything against 10 GOhm at this point.

Everyone how ever came across electropyhsiology research knows that there's a large choice of commercial lab equipment available since decades. I think it would be helpful to pick up some basic ideas for the instrument setup from the catalogs and instrument manuals.

 

[AugustineLeudar]

silver chloride electrodes

by the way - I am working in one of the worlds leading electrophysiology laboratories - we have no problem getting the signal into the computer with a DAq - As I say the op amps we have built work fine with the DAQ and give a reading that is basically the same as the electrometers, but noone has thought of using the signal from the op amps to light an LED before - thus this thread - all I need to know is how to light the LED from the op amp . The rest of the scientists here understand the input impedance thing much better than I do - I was just curious ..... the main pojnt of the thread however is how to get the signal from the op amp to light an LED - as in "is there some sort of simple gate/threshold component" ?

 

[FvM]

If everything is computer based, why don't you control the LED through LabView or a similar tool?

I was in fact referring to good old electrophysiology where signal processing has been performd by racks of slide-in modules. In this technique, there's no problem to tap an amplified signal using a T-connector, connect a trigger unit and light (or shut-off) a LED.

In the present case, we need to know about available auxilary signal outputs of your amplifier front end, and your options to assemble some electronic components.

 

[AugustineLeudar]

If everything is computer based, why don't you control the LED through LabView or a similar tool?

I was in fact referring to good old electrophysiology where signal processing has been performd by racks of slide-in modules. In this technique, there's no problem to tap an amplified signal using a T-connector, connect a trigger unit and light (or shut-off) a LED.

In the present case, we need to know about available auxilary signal outputs of your amplifier front end, and your options to assemble some electronic components.

ah ok - Ill find out and send you the details later. There are reasons that I do not want to use any digital processing but to be honest they would involve another 30 odd pages of discussion to explain ! Lets just say this is for a different application than straighforward data logging - in fact in this case the data is of no interest at all and having to use computers for this application would be a complete pain / nigh on impossible for many reasons - in this case I just want an LED to light up when an electronic signal passes through the plant. I am already happily reading the signal into the Data aquisition board after it has been apmlified by the op amp - everything works fine - no instead of reading the amplified signal into a Data aquisition board - I wan tit to light up an LED - ILL get the info you requested as soon as I can.