Amplitude varying on changing the input frequency

[Badri.S]

Hi,

I using NI-5752 Digitizer for measuring high frequency analog input. NI-5752 is connected to FLEX RIO 7966R module and the input to the NI-5752 is provided using SMB-2145 terminal output box.

Now I'm feeding the 0.5Vpp @ different frequencies from tektronix function generator directly and measuring using the digitizer. On changing the frequency measured amplitude is getting changed.

I used the example VI NI 5752 Configure ADC(Host).vi to take this measurements.

Input Voltage @ 0.5Vpp
Frequency Range Input	Measured Output
	Gain @ 0 db	Gain @ 6db
	Vpp	Vpp
100 KHz	1.393	2(Saturated)
200 KHz	0.703	1.4
300 KHz	0.508	1.015
400 KHz	0.423	0.843
500 KHz	0.378	0.753
1 MHZ	0.306	0.61
1.1MHz	0.301	0.601
1.2 MHZ	0.297	0.593
1.3 MHz	0.294	0.587
1.4 MHz	0.292	0.583
1.5 MHz	0.29	0.577

 I need to know the reason for amplitude changes in changing frequencies, what could be the reason for this problem?

-Sibana

[Henrik_Volkers]

Maybe not so well matched source-cable-input impedances?

How long is your cable?

[Badri.S]

Hi,

Cable I'm using is 1 meter long.

-Sibana

[DFGray]

What is the output impedance of your function generator?  If you are outputting 0.5Vppk and getting 1.3Vppk, at low frequency, that tells me there is a major impedance mismatch somewhere.  Your terminal block has a fairly standard 50Ω input impedance.  You may want to verify that your cable and function generator are the same.  75Ω and high impedances are common and could cause your issue.

[Badri.S]

Hi,

I verified this impedance mismatch and terminated the cable with 50Ohm. I'm using the function generator output also 50 Ohm termination. Still I'm facing this issue.

-Sibana

[etweb]

Hi,

we have a similar problem with our NI 5752 on a 7965R. We did a linear frequency sweep and white noise using a BNC 645 with selectabe output termination straight into the Adapter module via one of the boxes.

We tried both the SMB 2145 (50 Ohms) and the SCB-68 box (100 Ohms?)

benath the plot for the SCB at 100 Ohms, one can see the strong nonlinearity, the frequency response function is only flat from about 1 MHz to about 8 MHz

When we did the same thing on our other DAQ systems we got the results we expected. 

We think this might possibly be related to a filter applied in the adapter module

[etweb]

I could  only attach 3 plots, so here the plots for SMB with 100 Ohms (should be missmatch) and to compare the plots we got from a nother DAQ (sweep is from 10Khz to 25Mhz)

This other DAQ card shows an almost completely flat freq. response from 0.2 MHz on

[etweb]

deleted

[Nathan-P · Accepted solution]

All,

For the NI 5752 the strange frequency response is expected behavior.  The ADC in the NI 5752 is the TI AFE5801 (http://www.ti.com/lit/ds/slos591d/slos591d.pdf), and if you look at Figure 19 (p.13) you will see that the ADC's frequency response is only flat after about 1 Mhz.  I've verified this with our R&D department and they said that they are unaware of any settings combination that will yield a flat frequency response at these low frequencies.  

If a flat frequency response is needed for these lower frequency ranges, then the NI 5751 is much better suited since it has a different ADC.  (http://www.ni.com/pdf/manuals/375602b.pdf)  The frequency response for this device is very flat in the lower frequencies (see Figure 11 on p.15).

Therefore the two devices, the NI 5752 and the NI 5751 are designed for different frequency measurement ranges.  I hope this helps.

[TUjan]

All,

The problem of the frequency response described here also occured in my application. Furthermore the pulse response showed severe 'ringing'.

In my application the cause of the problem seemed the internal AC-coupling of the AFE5801 ADC chip (see fig 39, p 18 of AFE5801 datasheet)

By default this coupling is set to AC coupling, but it is possible to set it to DC coupling.

You have to set this in the 'general-purpose register map' of the AFE5801, see page 20 and 21 of the datasheet.

Sending data to this register can be done using the SPI bus connected to the ADC chips. This cannot be done directly in Labview, but needs extra code.

In the menu Help/NI example finder of Labview you can find the 'NI5752 configure ADC' project (under Hardware_input_and Output / FlexRIO / IO_modules / NI5752).

This example code contains the tools to communicate via the SPI bus to the ADC chips.

Use this code to send the setting you want to the corresponding adress of the register map of the AFE5801.

In the case of AC/DC coupling:

set bit [1] on adress 7 to '1' for DC-coupled mode.

in the host code:

SPI Addr write : 7

SPI Write Data : 2 (or binary 10)

Doing this in my application, the frequency response was now flat downto ~ 40 kHz (lower frequencies are cut-off by the AC coupling outside the ADC chip), there was no more peaking of the response at 50 kHz..

Furthermore, the pulse response is good.

Regards.