speedy 33 background noise

[pagauthier]

Hello,

We are finishing a project with the Speedy-33. The analog outputs are connected to a small amplifier for headphone listening. When connecting headphones directly to the Speedy-33 output or to the headphone amplifier, a background noise is heard : it sound like a tone. As you can imagine the amplifier jsut make it more audible. But the tone is present even without the amplifier ... so it is clearly from the speedy-33 output.

The speedy-33 is running in standalone mode ... as the USB connection tends to make a huge amount of noise.

We tried to listen to the Speedy-33 in "full bypass" (the labview patch is the mere connection of the input to the output) ... in that case only, their is no noise. As soon as we put any processing (exemple: we tested a single biquad between the outs and the ins) ... we get the famous background tone.

Any information about Speedy-33 background noise ... caused by the DSP or any other component on the board? Any advice?

Many thanks,

[Eric_Cunningham]

Hi pagauthier,

Unfortunately, there is no information available on the noise specifications for the Speedy-33.  However, there are some great resources about noise considerations:

https://www.ni.com/en/shop/data-acquisition/measurement-fundamentals/analog-fundamentals/dithering--...
https://www.ni.com/en/shop/data-acquisition/measurement-fundamentals/field-wiring-and-noise-consider...

If you are able to determine the frequency of the noise, that might help locate the source.  You'll also want to make sure there are no grounding issues with your signal.

Eric C
Applications Engineer
National Instruments

[ypasco]

HI !

I'm working with Pagauthier.

The noise is a "sine" at about 370 to 400Hz when using 36kHz of sampling frequency. The frequency is changing with the sampling frequency

Yann

[Eric_Cunningham]

You could try generating a tone to offset the noise that the board produces.  Use the Tones and Noise Waveform.vi to produce an output signal that offsets the noise.

Eric C
Applications Engineer
National Instruments

[johnsold]

Yann,

When you indicated that the frequency changes with the sampling frequency, it suggested aliasing. The interfering signal is at a higher frequency than the Nyquist limit. If your sampling frequency is 36 kHz, then the Nyquist limit is half that or 18 kHz. It also suggests that the interfering signal may be about 400 Hz away from a harmonic of the Nyquist frequency.

Is there anything in your system at or near those frequencies? Do you have a switching power supply? What is its frequency? Do you have an anti-aliasing filter in your signal path?

Lynn

[ypasco]

Thanks to all !

The main trouble is that when connecting directly the inputs to the outputs, the noise is not there. When a process is added between the input and the output (like a simple biquad as mentioned by PaGauthier), the board is generating a "sine" background noise.

In our application, the speedy is running in standalone mode using a DC power supply. The power suply was changed twice to make sure that it was not a noise coming from it (we used an analog regulated one, and a switching one with the same results)

The inputs are feeded with simple electret preamp using transistors or opamps (we have two preamps because we were thinking that is was radio frequencies interferences). The bandwidth of the preamp is quite large but the capsule itself is limited to 20kHz... and once again, when there is no process between the inputs dans the outputs, no background noise is added.

I have a simple question : Is there an aliasing filter onboard for each input ? If yes, I hope that designers did not simply added those filters at the nyquist frequency... In most cases, 80 to 90 % of the nyquist frequency is better... This could explain a 400Hz of aliasing : but couldn't explain that the added noise is a simple sine.

PaGautier and I will test the board next week to make it clear about aliasing filters.


Thank you for your help.

Yann

[Eriquito]

Hi Yann,

The SPEEDY-33 does not have an anti-aliasing filter. Please see this site (partly sponsored by National Instruments) for more information.

CONNEXIONS - DSP Laboratory: Aliasing


Message Edited by Eriquito on 05-19-2008 04:11 PM

[ypasco]

Thank you for the link.

So i still do not understand why the "sine noise" is not there when using a simple input to output connexion.

This must be more than aliasing because it would also appears in the last simple case.

I will try some new tests tomorow, it is really annoying.

Yann

[The Specialist]

Hi jpasco,

I like johnsold's response.  Are you sure that is is not an aliasing problem (i.e. high frequency noise)?  Maybe you should toss an analog filter before the signal to see if it takes care the the noise (a simple RC lowpass).  At the very least it might help us understand the problem.  Also how are you sure that there is sine noise at those frequencies.  I assume it is thru FFT, but can you explain further how you detect it.

Plus pictures might help.

[ypasco]

Hi All !

I added an aliasing filter after the mic preamp. same trouble !

The aliasing filter is a LR4 at 16kHz, the sampling frequency is 36kHz, buffer size 32 I think. And I have a 30mVRMS sine at 360~400Hz at the input of the Speedy33.

This sine is nearly 0 in amplitude if there is no process between the input and the output of the Labview VI. (direct through).

It seems to clearly be a speedy33 problem.

I will try some other tests today.

Yann