Signal measurements in AD620

Thread Starter

po210

Joined Dec 24, 2013
23
Hello
This thread is related with some problems with AD620 that I explained here,
https://forum.allaboutcircuits.com/threads/ad620-output-diamond-plot-and-some-questions.171686/

but i think this deserves an independent thread.
In following pictures ( sorry again for the quality), I measure with a single oscilloscope probe attaching the probe "gound" cable to the circuit ground (i.e,negative power supply cable) as shown with the green color,and attaching to the REF input to amplifier, shown with red color.By mistake, in the red part,i labeled the probe ground bad: the ground goes to REF and the other tip to amplifier output !

Next picture is with the signal generator ground connected to the same ground that oscilloscope and power supply have in common

IMG_20200826_193034.jpg

As you see, i have a constant output voltage if I measure relative to ground, ( green horiz line at scope) but a nice sinusoidal signal of about 2V with 1V on it if i measure against REF(red part)
But if i disconnect the generator from ground, I have this for the same measurement points:
IMG_20200826_193411.jpg

So then i can measure a signal in both points using the green or the red method( with much more noise,by the way)
I do not understand the signal offset in any of the two test, because I would expect a 1Vpp signal on top of a 3V offset.Common mode voltage?Offset from slightly different values in amplifier inputs?
Why in first test I can only measure the signal relative to the REF input?
Maybe then I'm putting a OV virtual ground on AD620 output?
In fact, that seems the way to go,because when measuring with REF as ground the signal is much better...
 

slackguy

Joined Feb 11, 2016
76
Did you follow advice? https://www.tek.com/document/application-note/three-facets-floating-measurement-solutions

> So, why is my circuit "working"?

I find the whole conversation convoluted. You've apparently used "sim software" to assert your circuit "doesn't work" and then use a chalk board to assert it does work and you want to know why: I see no photo showing you actually built the circuit, and you claim your OS shows your right. Not that I'm as qualified as the other regular posters to give you an answer.

Also you ignored some advice from the previous post, such as the OPAMP is not a simple NPN it is a (well i don't have a diagram of one) model dependent and you were told it had limits and your signal type was a subject you dodged.

Post a photo of the setup and OS testing it, not a drawing with an assertion the circuit does something and a simulator cut&paste. Including input signal type of course
 

Thread Starter

po210

Joined Dec 24, 2013
23
20200827_082902.jpg
At the left, AD620 with a gain pot. At the right, the reference voltage generation from a divider with a pot and a dual op amp. The left black and red crocodile clips are from generator,going to ad620 inputs
Here, the oscilloscope is meauring as I draw with green color.
 
Last edited:

ericgibbs

Joined Jan 29, 2010
15,713
hi,
Looking at the photo.
Where is the 3rd leg of the Gain trimmer connected too, it looks like it is inserted into pin#7 +Vcc.

Also you have a polarised Tantalum capacitor across the AD620 inputs.

I don't see any power rail decoupling on the board.?

You have not indicated what the External wiring/probes are connected too.
E
 

Attachments

Thread Starter

po210

Joined Dec 24, 2013
23
Hi Eric
I use only two legs of pot, just to have a variable resistor. The other is left unused.
The capacitor is ceramic, although certainly it looks tantalum in the photo.
The red and black clips at the left are the two clips of signal generator output.The two probes at the bottom are the oscilloscope ones.

With that configuration, this would be what I draw as green in the photos ( measure with the oscilloscope black clip to ground, and the other tip to amplifier output). That gives me a constant output when the generator,the oscilloscope and the supply share the ground rail, but i have a signal (not with the expected offset) when I disconnect the signal generator from that shared ground, as seen in the photos, comparing the green parts in both situations.

About the PS decoupling capacitors...to be honest, I have never seen much difference in protoboards. ( i tested that here,with no results)
 
Last edited:

Thread Starter

po210

Joined Dec 24, 2013
23
hi,
I LTSpiced your latest circuit, it works as expected, I still suspect that it is an external connection or component problem.?

E
Thanks for the effort Eric.
What happens in Spice if you remove R4/R5 resistors?.I remember from previous thread there is the need for a return path for the amplifier, but I do not see it in the ECG schematics I have read.
I will test with R4/R5 in my circuit check if output is as you simulate.
 

Thread Starter

po210

Joined Dec 24, 2013
23
With the R4/R5 resistors, I measure a 1V signal centered in 2.5V.
Withouth the resistors, the 1V signal is centered in 4.5V.
Both with the generator ground disconnected. So there is more noise in the output.

I don,t know if that is what is expected..Without the resistors, I would expect a 1V signal centered in 3V,

About the ECG circuit, the right leg derivation adds a common mode voltage Vcm to the inputs, which would be similar , I guess, to the effect of the two R4/5 resistors adding a voltage above ground. A minimun Vcm is needed to put the output signal in proper margins
In the next picture, i use the diamond plot tool to add a 4V common voltage in Vn
Screenshot_2020-08-27-18-26-39.pngScreenshot_2020-08-27-18-26-28.pngDoes it meant the right leg circuit has to add those 3 or 4V to the inputs?
Is that feasible? (Adding 4V to a person's body)
If not, then althought AD620 can be used with single suply, maybe this is not possible here, because the right leg circuit does not add 4V, but a correction signal centered on 0V.
 

ericgibbs

Joined Jan 29, 2010
15,713

Attachments

Last edited:

Thread Starter

po210

Joined Dec 24, 2013
23
Added an option, using the 9V, AD620 single supply.
CMV =4.5v
Vref =4.5v
Vdiff In =+/-30mV
Vout = +1.5v thru +7.5v
Thanks again for your inerest.

Yes, for this to work you must add some Vcm volts.
That's why I asked if that is feasible, for example, taking the right leg circuit (shown i the datasheet,or in your image) and adding 3 or 4 volts to its output with a summing amplifier,then applying that to ones body.
Maybe best thing i can do its to try
 

ericgibbs

Joined Jan 29, 2010
15,713
hi po,
Try this option.
V3 is the external pads input and R8 and R9 are the body resistance.

I assume this just an exercise project for an ECG and not for real World usage.??

E
 

Attachments

ericgibbs

Joined Jan 29, 2010
15,713
hi po,
This is the circuit with a human heart beat signal, I have reduced the AD620 for the simulation.

Attached the human1.txt file, download and change to human1.wav to play the sound on your computer.

E
 

Attachments

Thread Starter

po210

Joined Dec 24, 2013
23
Well, I tried
hi po,
This is the circuit with a human heart beat signal, I have reduced the AD620 for the simulation.

Attached the human1.txt file, download and change to human1.wav to play the sound on your computer.

E
Hi Eric,
After your effort, I feel obliged to answer to your last message so I tested your circuit and I think it works, although still the signal is not centered on Vref=4.5V but on 1.5V.
In fact, the OA at the left,that produces a centered offset for the two joined 470ohm resistors is not needed. I obtain the same signal if i join with a cable the Vref obtained with the previous follower (the one I used for obtaining Vref) with that same point
The signal I get is much clear when i use this derivation.
This is without it (85hz signal,1V/div)
20200902_200035.jpgAnd this is with the additional derivation from Vref to the common point to 470 esistors
20200902_200132.jpgI cant explain the incorrect offset,which should be 3 or 4.5 volts.
 

ericgibbs

Joined Jan 29, 2010
15,713
In fact, the OA at the left,that produces a centered offset for the two joined 470ohm resistors is not needed.
hi po,
Those two resistors are 470K not 470R, their purpose is the reduce the loading of the differential voltage signal input, the Vg OPA output is a low impedance.

Try this, short the two signal input 'pads' together, [not to 0V] so there is no diff signal input, what is the AD620 output voltage.?, it should be Vref.

E
 

Thread Starter

po210

Joined Dec 24, 2013
23
hi po,
Those two resistors are 470K not 470R, their purpose is the reduce the loading of the differential voltage signal input, the Vg OPA output is a low impedance.
V
Sorry,I meant 470K.

If I short the two signals, I get Vref exactly.
What I see is always an incorrect offset when the signal is applied. In fact,if I reverse the input signal clips from generator, the output signal offset appears "reflected" relative to Vref.


IMG_20200912_123821.jpgIMG_20200912_123852.jpgAlso to get a clear signal,at 9V power supply, I have to use a gain 80 at most, and Vref 5V at least( those are the real diamond diagram operational limits i think) Where that offset comes from?
 
Last edited:

ericgibbs

Joined Jan 29, 2010
15,713
hi po,
That shorting test shows that one of the inputs to the AD620 has a DC component, which maybe due to the signal driving source.
Please post a diagram showing the latest input connections.
E

Update:
This LTS sim shows the effect of having a -30mV DC Offset on one of the AD620 inputs, using a 3Vref
 

Attachments

Last edited:
Hello, the AD620 is an instrumentation amplifier and is specified for operation using bipolar supplies. It will not operate correctly with the -V supply tied to 0V (gnd). Specs on page 1: "Wide power supply range (+/- 2.3 V to +/- 18V)". Specs on page 5, Table 3: "Supply voltage: +/- 18 V". No where in the datasheet does is it stated that this device will operate on a single supply,
 
Top