Did you find out where the high-frequency noise came from? If not, it will most likely comply with Murphy's Law and come back when you least need it.

The signal you see is aliasing, in fact. With a pulse frequency of 14.3 Hz in the pCLAMP Membrane Test (first image), the sampling rate is about 4.5 kHz. Using a filter frequency of 3 kHz, you are in violation of the Nyquist theorem, for which one may be punished with severe aliasing. And you have been.

The Nyquist theorem says that you need to sample with at least twice the highest frequency present in your signal to reliably avoid aliasing. And that only goes for ideal filters, which don't exist. So it is common practice to oversample 5 to 10 times, depending on the signal you are looking at.

Provided the amplifier telegraphs its filter setting (which seems to be the case in your setup), pCLAMP will bark at you if you undersample in protocol acquisition. But it will let you run into aliasing in the Membrane Test. And that is what happened here.

For me it looks like 50 Hz noise. So it could come from some pluged device. Most likely pump or stimulator. You can try to ground them. You can also cover electric wires with foil, sometimes it helps.

Hi Olga, As you see I have tried to unplug electronic devices and see if it persists. It does. Also in cell model mode, I did not see any 50 Hz noise.

Have you tried to wash the pipette's holder?

I cleaned it properly. Did not make a difference

Hi, Arjun!

The following measures might help:

1) Wrap in aluminum foil all the wires that enter the Faraday cage form outside of it. Ground the foil to the cage.

2) Ground the inlet and the outlet of the perfusing chamber (if it is possible)

3) Do you use any type of temperature control device which utilizes the thermocouple-sensing element? Make sure that metal is not in direct contact with the recording medium (only through some layer of insulator).

Hi, Arjun: Either a ground is missing in your system (which in the cell model does exist), or you have electrical loops. These arise from having a ground on every electrical outlet. Try "floating" your system by using a single ground, which can be done isolating every other with a three-to-two prong adapter.

Unfortunately, none of your noise pictures shows the time axis. But even without that, this clearly does not look like line frequency hum. The first two images are classic examples of an aliased signal. You pick up high-frequency noise, and digitizing it at a too low frequency creates an apparent signal with components at lower frequencies.

The high-frequency noise can come from a variety of sources, but you should be able to shield your headstage with a grounded piece of aluminum foil. Try moving that to different locations for troubleshooting and to help locate the direction of the noise source. A faulty bath ground may be the reason you see the issue with the pipette in the bath. What amplifier do you use, and what is the output filter frequency?

Thank you all.

Hello Jan,

I am using BVC 700A.filter frequency is 3KHz.

Thank you all for the kind feedback. I somehow managed to make it. Still some issues with scaling, but I think I can manage it

Did you find out where the high-frequency noise came from? If not, it will most likely comply with Murphy's Law and come back when you least need it.

The signal you see is aliasing, in fact. With a pulse frequency of 14.3 Hz in the pCLAMP Membrane Test (first image), the sampling rate is about 4.5 kHz. Using a filter frequency of 3 kHz, you are in violation of the Nyquist theorem, for which one may be punished with severe aliasing. And you have been.

The Nyquist theorem says that you need to sample with at least twice the highest frequency present in your signal to reliably avoid aliasing. And that only goes for ideal filters, which don't exist. So it is common practice to oversample 5 to 10 times, depending on the signal you are looking at.

Provided the amplifier telegraphs its filter setting (which seems to be the case in your setup), pCLAMP will bark at you if you undersample in protocol acquisition. But it will let you run into aliasing in the Membrane Test. And that is what happened here.

Good eye, Jan Dolzer! I did not notice the pulsing frequency in those pictures. Thanks for your very useful answer. But, does it not appear that the sampling frequency is 20KHz? At least in those images where it can be seen, it reads 20000 Hz.

That is the sampling frequency of the Protocol that is loaded. The sampling frequency of the Membrane Test is determined by the pulse frequency. Each pulse has a fixed number of samples (500, IIRC), and thus changing the pulse frequency affects both pulse duration and sampling rate.

The danger of introducing aliasing this way is one of the reasons we implemented it differently in SutterPatch Software. I don't want to make this an off-topic commercial post, though, and will therefore refrain from discussing our implementation in detail. Feel free to contact me offline.

Dear Jan,

Thank you very much for spending your precious time to answer my problem. Very useful information. I have reduced the noise to an extend by changing the reference electrode. Importantly I also had to play with the capacitance and the gain. It looks decent for the moment. I never knew the information that you have shared. I will contact you in person for further details. Thank you very much.

Ah, yes. You can get a similar noise with capacitance overcompensation