Re: [Discuss-gnuradio] Resampling with USRP

[Tom Rondeau]

On Tue, Oct 8, 2013 at 10:28 AM, Dincer Beken <address@hidden> wrote:

I just mixed up things I guess.

 

After searching a bit more, I saw that the taps below whre for the frac. Resampler. I again looked at python in the NONE case and saw the def design_filter function and implemented it in C++. Did not reached my ultimate goal yet but I think that this is the real solution for the taps in C++ code.

Yeah, I was wondering what you were talking about with those taps.

For the rational_resampler, you have to build the taps based on the resampling rates you are using. It's probably best to look at gr-filter/pythong/filter/rational_resampler.py. When the taps specified are None, it auto-generates a filter for you. You can probably walk through this process and figure out how to design one for your needs.

Two things. First, you want to build the filter with respect to the highest sampling rate of the filterbank. With this design, that number is 'interpolation' (which is why in rational_resampler they set Fs=1 and the rest is relative to interpolation rate).

Second, the gain is necessary. When you interpolate, you are spreading the energy over the interpolated signal. Setting the gain adjusts for this change. Are you doing this as a receiver? If so, you won't hurt your front-end since this is applied digitally. If you are transmitting, you'll drive your signal non-linear, but the values will be clipped in the FPGA before they get to the analog hardware, anyways.

Tom

 

 

Von: Dincer Beken [mailto:address@hidden]
Gesendet: Dienstag, 8. Oktober 2013 14:55
An: address@hidden
Cc: address@hidden
Betreff: Re: [Openlte-discuss] Resampling with USRP

 

Interpolator_taps.h in include file solves my issue I guess. Could not see it before.

 

For the LTE Part, I still am not sure if this is ok for the openLTE project as resampling taps.

 

 

Von: Dincer Beken
Gesendet: Dienstag, 8. Oktober 2013 14:32
An: Dincer Beken; address@hidden
Cc: address@hidden
Betreff: AW: Resampling with USRP

 

I found the folder: General interpolator Taps (over 3.6). That Is probably my answer…

Though I don’t know if I have to separately install it…

 

Von: Dincer Beken [mailto:address@hidden]
Gesendet: Dienstag, 8. Oktober 2013 10:00
An: address@hidden
Cc: address@hidden
Betreff: [Openlte-discuss] Resampling with USRP

 

Hi all,

 

I want to implement a low pass for the resampler. I am going to use not a polyphase filterbank, because I don’t know its details.

 

My USRP output is configured to 10Msps. I want to catch a 10Mhz LTE carrier. I want to use a normal resampler with decimation 1000 and interpolation 1536. I guess, need to fill in the taps part with an appropriate low pass filter (H(e^jw)).

 

 

Since I never implemented in practice a resampler, according to the DSP books, I need to have a low pass filter with gain of L(1536), if I want to use a resampler with non-integer rate (1,536). The low pass filter will have a gain of 1536.  

 

 

So I have to following questions:

 

1-      Can I use the rational_resampler_base_ccc ­or rational_resampler_base_ccf? I am tending more towards the ccc, sing H(e^jw) is usually complex. Or do I need to use a PFB filter?

2-      Is a gain with 1536 really necessary? I fear of destroying the Daughterboard.

3-      For the low pass filter, I want to use the gr::filter::firdes::low_pass (1). The sampling rate is L (15,36M). No idea about cutoff frequency and transition with. Can anyone please, please help me with those parameter..

a.       http://gnuradio.org/doc/doxygen/classgr_1_1filter_1_1firdes.html#a772eb5c542093d65518a6d721483aace

I would be very happy, if I could get some help. Especially from anyone who tried to implement openLTE on the USRP.

 

Thanks,

Dincer

 

_______________________________________________
Discuss-gnuradio mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/discuss-gnuradio