Forums

Strange fm component in the ECG

Started by paramonte November 16, 2007
Dear all,
We are acquiring the ECG with a high resolution system, sampling frequency
2400Hz. The system is an oversampling sigma delta, 24 bits.

In some ECG´s (from normal subjects and patients) we get, added to the
ECG, a  frequency modulated version of the ECG itself, e.g. a sinus wave
with constant amplitude and whose instant frequency is proportional to the
ECG signal value in that particular time instant. This component is not
visible in the acquired ECG, and we can see it if we perform some advanced
spectral analysis. Moreover the carrier frequency of this component is
around 800Hz which looks way to high to be due to some physiological
mechanism .

Where the hell is this component coming from? Sigma-delta behavior? 
Interference? (we are doing the acquisition in a  shielded room). 

Your input is welcome.

Arnaldo


On Fri, 16 Nov 2007 08:47:44 -0600, paramonte wrote:

> Dear all, > We are acquiring the ECG with a high resolution system, sampling frequency > > 2400Hz. The system is an oversampling sigma delta, 24 bits. > > In some ECG´s (from normal subjects and patients) we get, added to the > > ECG, a frequency modulated version of the ECG itself, e.g. a sinus wave > > with constant amplitude and whose instant frequency is proportional to the > > ECG signal value in that particular time instant. This component is not > > visible in the acquired ECG, and we can see it if we perform some advanced > > spectral analysis. Moreover the carrier frequency of this component is > > around 800Hz which looks way to high to be due to some physiological > > mechanism . > > Where the hell is this component coming from? Sigma-delta behavior? > > Interference? (we are doing the acquisition in a shielded room). > > Your input is welcome. > > Arnaldo
My first guess would be that something in your signal conditioning chain between the contacts and the ADC is oscillating -- then the FM would be caused by the bias point of the unintended oscillator changing with signal from the patient, hence changing it's frequency of oscillation. I wouldn't rule out the ADC, but as I understand it a sigma-delta may make a tone at certain unfortunate input levels, but the tone would always be a subharmonic of the 1-bit sampling frequency and it would come and go, not frequency modulate. -- Tim Wescott Control systems and communications consulting http://www.wescottdesign.com Need to learn how to apply control theory in your embedded system? "Applied Control Theory for Embedded Systems" by Tim Wescott Elsevier/Newnes, http://www.wescottdesign.com/actfes/actfes.html
Dear Tim Wescott


Yes, we came across those tones sometimes, which makes me think if
sigma-delta is the best option for high-resolution biomedical signal
acquisition.

As for the frequency modulated component, your coment seems to be
quite logical, although I can=B4t see how to test this assumption.

Thank you for your input.
Arnaldo

On Nov 16, 11:44 am, palmar <runge_k...@hotmail.com> wrote:
> Dear Tim Wescott > > Yes, we came across those tones sometimes, which makes me think if > sigma-delta is the best option for high-resolution biomedical signal > acquisition. > > As for the frequency modulated component, your coment seems to be > quite logical, although I can=B4t see how to test this assumption. > > Thank you for your input. > Arnaldo
what do you mean by frequency modulated, the ECG is shifted up onto an 800 Hz carrier and the carrier is FMed? is it randmomly FMed or with some pattern?? is there a switching power supply in your system that may be causing higher frequency inputs that get aliased into your band? Mark
"paramonte" <vuela@netvisao.pt> wrote in
news:_5SdnWBhDKcNMKDanZ2dnUVZ_tajnZ2d@giganews.com: 

> Dear all, > We are acquiring the ECG with a high resolution system, sampling > frequency 2400Hz. The system is an oversampling sigma delta, 24 bits. > > In some ECG&acute;s (from normal subjects and patients) we get, added to > the ECG, a frequency modulated version of the ECG itself, e.g. a > sinus wave with constant amplitude and whose instant frequency is > proportional to the ECG signal value in that particular time instant. > This component is not visible in the acquired ECG, and we can see it > if we perform some advanced spectral analysis. Moreover the carrier > frequency of this component is around 800Hz which looks way to high to > be due to some physiological mechanism . > > Where the hell is this component coming from? Sigma-delta behavior? > Interference? (we are doing the acquisition in a shielded room). > > Your input is welcome. > > Arnaldo > > >
Sometimes you can see such oscillations when something in some circuit isn't bypassed correctly. Instead of plugging directly into the DAC, try buffering with a one op-amp voltage follower and see what happens. -- Scott Reverse name to reply
On Nov 16, 10:44 am, palmar <runge_k...@hotmail.com> wrote:
> Dear Tim Wescott > > Yes, we came across those tones sometimes, which makes me think if > sigma-delta is the best option for high-resolution biomedical signal > acquisition. > > As for the frequency modulated component, your coment seems to be > quite logical, although I can=B4t see how to test this assumption. > > Thank you for your input. > Arnaldo
Maybe this is too simple, but could you possibly "test this assumption" by checking for the modulated 800 Hz carrier under different input conditions, such as starting off with a DC input and, step by step, increasing the complexity of the signal and of the cable/ contact arrangement connected to the ADC? If your ADC board is generating the carrier, then I think that the carrier might still exist in some form even with no ECG input.