Hi ,
This is Mangesh Jadhav from INDIA. I have developed ECG system based on ADS1298 IC. It works properly whenever I give inputs from ECG Simulator. all 12 leads are properly placed in graph. But as soon as I start taking ECG on human body then waveforms are properly captured but all are shifted with some small voltages. so there is no noise in waveforms but waveform itself is shifted somewhere else.
So what will be the problem ? I have also grounded the RL to system gnd. Then I have also tried without RL .So there is no effect of RL on waveforms. Is it the offsets from body and if its a offset from body then it will be different for different bodies. So how to remove offsets or is there any adaptive method to remove the offsets from different bodies.
waiting for reply.
Hi Mangesh,
Sounds like you are seeing DC baseline drift coming from the front end and electrodes. Our design recommendation is to include a high pass filter in the digital domain with a 0.05Hz cut off frequency to help remove this slower drift. Some designs have AC coupled their ECG inputs using ~1uF capacitor to also help remove the DC component.
Regards,
Tony Calabria
Hi Tony,
Thanks a lot. I was also thinking of the same. But the waveform is not constantly shifted throughout the graph. It goes up and down or sometimes it becomes stable at constant
shift. so it is the baseline drift. So I will implement HPF in software. One more thing I have found out that we have facility to short the ADC inputs. so what I will do , I will short
the ADC input and then measure the noise or offset from body and I will subtract that noise or offset from every reading. I will do this several time in ECG capturing so that I can track change in offset or noise.
So will it be all right ? Please let me know your comments.
There are internal register settings that you can use to run an internal short. This will allow you to determine the offset error from the ADC for your digital processing. Depending on the application, offset error may or may not be a concern. Measuring the offset from the system/body will not be a concern if you implement a 0.05Hz digital high pass filter.
I have just checked the input short method to observe offset from body. But it gives ADC offset only. Because I think that if we short the inputs then voltages from LA and RA(Lead I)will be fed as Common mode voltage and ADC will remove that voltage.Please correct me if wrong.
I have also attached few ECG images for your reference. First one is on body and second one is from simulator .
As you can see that on body waveform is shifted and its not constant through out the graph. But on simulator it is properly aligned and remains the same through out the graph, So is it offsets from body ?Also its changing as graph progresses. So please refer to graphs and let me know the exact issue. Because I am badly stuck at this point.Because waveform is captured properly but there is variable offset from body. Waiting for reply.
Mangesh Jadhav
Please find the ECG images at your email
Please find the images from Mr. Mangesh in the attached zip file.
Thanks & Regards,
Shridhar.
Hi Shridhar,
The images did not post. Please send them to me in a separate email.
Thanks,
Tony
I received the attachments from Shridhar and was able to take a look at what you are seeing. Looks like you have quite a bit of drift once you go away from using the simulator for usage with a patient. All of our testing has been done with a simulator but I have seen similar behavior from others when they either do not have the input properly biased through using the RLD, pull up/down resistors or have baseline drift coming from the changing electrode/patient connection properties. A digital high pass filter at 0.05Hz cutoff should help with the baseline drift but the input biasing must be done in the analog domain. Are you using a RLD electrode for bias or pull up/pull down resistors at the front end to bias the inputs within operating range? Is there any way you can post or have Shridhar send to me the front end portion of your schematic so that I can see how you have it connected?