WARP Project Forums - Wireless Open-Access Research Platform

You are not logged in.

#1 2010-Aug-16 03:16:28

chan
Member
Registered: 2010-Jul-13
Posts: 5

Radio DC attenuation

In warplab_siso_example_Comm.m, there is a up-conversion to 5MHz. Can you explain why this is done. In the example, you use nsamp = 8 which makes the raised cosine pulse to occupy a bandwidth of 5MHz (-2.5 to 2.5) in the baseband. After upconverting, the pulse will occupy a bandwidth from 2.5 to 7.5 MHz. I want a pulse which occupies the entire 20MHz in the passband. So, I tried to make nsamp=2 which will make the pulse to occupy -10MHz to 10MHz in the baseband  which is almost within the specifications of WARPLAB  of -9.5 to 9.5 MHz. But I cannot up-convert  here as it will go out of band. Can you please tell me if upconversion is absolutely necessary. Otherwise, can you suggest any other way to generate a pulse which occupies the entire 20MHz bandwidth in the passband.

Thanks

Offline

 

#2 2010-Aug-16 10:17:01

sgupta
Administrator
Registered: 2007-Feb-26
Posts: 104

Re: Radio DC attenuation

The upconversion is necessary cause the radio will null out the data at DC with its own high pass filters. WARPLab is a narrowband transmission so if you don't upconvert you will lose some data. If you want to use all 20MHz perhaps a wideband system like OFDM might work better. Then you can avoid using the subcarriers at DC and still use the rest of the spectrum.

Offline

 

#3 2010-Aug-16 11:15:10

chan
Member
Registered: 2010-Jul-13
Posts: 5

Re: Radio DC attenuation

Thanks for the quick reply. Can you let me know what is the cut-off frequency of the highpass filter. Also, when you say "radio  will null out the data at DC with its own high pass filters" is it at the baseband stage (or at DAC) or RF stage. I thought, at baseband, the allowed frequency range is -9.5 to 9.5 MHz. I don't mind if I loose some data at DC (like if the radio attenuates between -100KHz to 100KHz) but would like 20MHz (-10MHz to 10MHz) bandwidth occupancy at 2.4G band.

Will a method where I take nsamp = 4 (this makes the pulse to occupy -5 to 5 MHz). Then multiply the data with cos(2*pi*5e6*t) (This will shift the centre frequency to +5MHz and -5MHz thus occupying -10 to 10 MHz). Will this be fine or do I still have attenuation. 

Also, in warplab_siso_example_TxRx_BandwidthChange.m example sinusoids of 0-20MHz are used and the LPFs can be selected with cutoff frequencies of 12,18 or 24MHz. Is this different from warplab_siso_example_Comm.m example. Here, does a cut-off frequency of 12 MHz mean, frequencies in the range -12 to 12 MHz are accepted or does it mean frequencies in the range -6 to 6MHz are accepted. Can you also please point me to some documentation where I get a clearer picture of these.

Thanks

Offline

 

#4 2010-Aug-16 11:42:22

murphpo
Administrator
From: Mango Communications
Registered: 2006-Jul-03
Posts: 5159

Re: Radio DC attenuation

The MAX2829 Tx/Rx paths both include low pass filters whose corner frequencies are configured via calls to the radio controller driver. The effective bandwidth is 2x the corner frequency.

The Rx path also includes a high pass filter with a cutoff at very low frequency. This cutoff is also configurable, based on MAX2829 register settings and the state of the RXHP pin. The state of RXHP depends on your choice of manual or automatic gain control; see the WARPLab C code for details. It is critical the RXHP pin be high whenever gain settings are changed (big variations in DC levels occur during gain changes; the higher cutoff frequency when RXHP=1 helps null these out).

I would recommend spending some time with the MAX2829 datasheet. All of the filter options set by the radio controller driver are implemented by this chip.

Offline

 

Board footer