Hi all, I'm trying to make USRP x310 retransmit on one of the outputs everything that it receives on one of the inputs, without sending the signal to the controlling PC. I've connected together two "RFNoC: Radio" blocks, one set to RX to another TX mode Expected behaviour was that the received signal would be copied inside FPGA and retransmitted through TX antenna. However, it doesn't work the way I intended - the easiest to check symptom is that LEDs under TX/RX and RX2 are not lighted up. Is it possible to do what I want with current RFNoC? (Attached is my flowgraph) Best Regards, Piotr

On 01.05.2015 11:23, Piotr Krysik via USRP-users wrote: > Hi all, > > I'm trying to make USRP x310 retransmit on one of the outputs everything > that it receives on one of the inputs, without sending the signal to the > controlling PC. > > I've connected together two "RFNoC: Radio" blocks, one set to RX to > another TX mode > > Expected behaviour was that the received signal would be copied inside > FPGA and retransmitted through TX antenna. > However, it doesn't work the way I intended - the easiest to check > symptom is that LEDs under TX/RX and RX2 are not lighted up. > > Is it possible to do what I want with current RFNoC? Currently, no. The radios in there were written pre-RFNoC, so they have some shortcomings. We're working on this, though, as this is something we definitely want to support. Cheers, Martin

W dniu 01.05.2015 o 20:58, Martin Braun via USRP-users pisze: > On 01.05.2015 11:23, Piotr Krysik via USRP-users wrote: >> Hi all, >> >> I'm trying to make USRP x310 retransmit on one of the outputs everything >> that it receives on one of the inputs, without sending the signal to the >> controlling PC. >> >> I've connected together two "RFNoC: Radio" blocks, one set to RX to >> another TX mode >> >> Expected behaviour was that the received signal would be copied inside >> FPGA and retransmitted through TX antenna. >> However, it doesn't work the way I intended - the easiest to check >> symptom is that LEDs under TX/RX and RX2 are not lighted up. >> >> Is it possible to do what I want with current RFNoC? > Currently, no. The radios in there were written pre-RFNoC, so they have > some shortcomings. We're working on this, though, as this is something > we definitely want to support. > > Cheers, > Martin > > If it's not supported currently by GNU Radio - maybe there is some lower level way to connect RX to TX in the FPGA? For example I've stumbled upon following lines in the uhd/fpga-src/usrp3/lib/radio/radio.v: // Digital Loopback TX -> RX (Pipeline immediately inside rx_frontend). wire [31:0] > rx_fe = loopback ? tx : rx; Maybe this is what I'm looking for? Or maybe there's another way - for example to use RFNoC to connect radio frontends that don't involve GNU Radio? Sorry if I'm asking silly questions, but I'm RFNoC novice and I'm trying to estimate what amount of work it would take to make USRP X3x0 stream samples from RX to TX inside FPGA. Best Regards, Piotr

Piotr, The loopback path you have identified is inward facing, in other words it forces the RX to listen to the TX digital data being sent to the DAC. (Its a very useful feature during testing both in simulation and the real world.) You could make a code change in that area and create a similar outward facing path. Personally I'm not sure I'd recommend that since you would be looping raw ADC data to the DAC without any DC offset or IQ balance corrections, I think it might be better if such a loop back path allowed data to pass through the "rx_frontend" and "tx_frontend" which means you are dealing with the "rx_corr_i/q" and "tx_fe_i/q" signals. If you also require digital tuning via the CORDIC then your loopback path would instead include the signals "sample_rx" and "sample_tx" and then you would also have the full signal processing capabilities of the DSP's included. Which you choose will depend I guess on what you are trying to achieve. -Ian On May 1, 2015, at 11:53 PM, Piotr Krysik via USRP-users <usrp-users@lists.ettus.com> wrote: > W dniu 01.05.2015 o 20:58, Martin Braun via USRP-users pisze: >> On 01.05.2015 11:23, Piotr Krysik via USRP-users wrote: >>> Hi all, >>> >>> I'm trying to make USRP x310 retransmit on one of the outputs everything >>> that it receives on one of the inputs, without sending the signal to the >>> controlling PC. >>> >>> I've connected together two "RFNoC: Radio" blocks, one set to RX to >>> another TX mode >>> >>> Expected behaviour was that the received signal would be copied inside >>> FPGA and retransmitted through TX antenna. >>> However, it doesn't work the way I intended - the easiest to check >>> symptom is that LEDs under TX/RX and RX2 are not lighted up. >>> >>> Is it possible to do what I want with current RFNoC? >> Currently, no. The radios in there were written pre-RFNoC, so they have >> some shortcomings. We're working on this, though, as this is something >> we definitely want to support. >> >> Cheers, >> Martin >> >> > If it's not supported currently by GNU Radio - maybe there is some lower > level way to connect RX to TX in the FPGA? > > For example I've stumbled upon following lines in the > uhd/fpga-src/usrp3/lib/radio/radio.v: > > // Digital Loopback TX -> RX (Pipeline immediately inside rx_frontend). > wire [31:0] > rx_fe = loopback ? tx : rx; > > Maybe this is what I'm looking for? Or maybe there's another way - for > example to use RFNoC to connect radio frontends that don't involve GNU > Radio? > > Sorry if I'm asking silly questions, but I'm RFNoC novice and I'm trying > to estimate what amount of work it would take to make USRP X3x0 stream > samples from RX to TX inside FPGA. > > Best Regards, > Piotr > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com

Many thanks for the answer Ian - it is very helpful. I will try to connect rx and tx in points highlited by you (by getting around DDC and DUC) as soon as I will get Xilinx tools to compile FPGA code (currently I have problems with Vivado). Meanwhile - do I understand it correctly that currently it is not yet possible (can't be easily done) to connect inside the FPGA radio fronend with RFNoC signal processing blocks? Best Regards, Piotr W dniu 02.05.2015 o 17:41, Ian Buckley pisze: > Piotr, > The loopback path you have identified is inward facing, in other words it forces the RX to listen to the TX digital data being sent to the DAC. (Its a very useful feature during testing both in simulation and the real world.) > You could make a code change in that area and create a similar outward facing path. > > Personally I'm not sure I'd recommend that since you would be looping raw ADC data to the DAC without any DC offset or IQ balance corrections, I think it might be better if such a loop back path allowed data to pass through the "rx_frontend" and "tx_frontend" which means you are dealing with the "rx_corr_i/q" and "tx_fe_i/q" signals. > > If you also require digital tuning via the CORDIC then your loopback path would instead include the signals "sample_rx" and "sample_tx" and then you would also have the full signal processing capabilities of the DSP's included. > > Which you choose will depend I guess on what you are trying to achieve. > -Ian > > On May 1, 2015, at 11:53 PM, Piotr Krysik via USRP-users <usrp-users@lists.ettus.com> wrote: > >> W dniu 01.05.2015 o 20:58, Martin Braun via USRP-users pisze: >>> On 01.05.2015 11:23, Piotr Krysik via USRP-users wrote: >>>> Hi all, >>>> >>>> I'm trying to make USRP x310 retransmit on one of the outputs everything >>>> that it receives on one of the inputs, without sending the signal to the >>>> controlling PC. >>>> >>>> I've connected together two "RFNoC: Radio" blocks, one set to RX to >>>> another TX mode >>>> >>>> Expected behaviour was that the received signal would be copied inside >>>> FPGA and retransmitted through TX antenna. >>>> However, it doesn't work the way I intended - the easiest to check >>>> symptom is that LEDs under TX/RX and RX2 are not lighted up. >>>> >>>> Is it possible to do what I want with current RFNoC? >>> Currently, no. The radios in there were written pre-RFNoC, so they have >>> some shortcomings. We're working on this, though, as this is something >>> we definitely want to support. >>> >>> Cheers, >>> Martin >>> >>> >> If it's not supported currently by GNU Radio - maybe there is some lower >> level way to connect RX to TX in the FPGA? >> >> For example I've stumbled upon following lines in the >> uhd/fpga-src/usrp3/lib/radio/radio.v: >> >> // Digital Loopback TX -> RX (Pipeline immediately inside rx_frontend). >> wire [31:0] > rx_fe = loopback ? tx : rx; >> >> Maybe this is what I'm looking for? Or maybe there's another way - for >> example to use RFNoC to connect radio frontends that don't involve GNU >> Radio? >> >> Sorry if I'm asking silly questions, but I'm RFNoC novice and I'm trying >> to estimate what amount of work it would take to make USRP X3x0 stream >> samples from RX to TX inside FPGA. >> >> Best Regards, >> Piotr >> >> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >

Hi all, After seeing https://github.com/EttusResearch/uhd/wiki/RFNoC:--Specification I will partially answer my question. It looks that radio blocks can be connected directly to other signal processing blocks in RFNoC. Still I dont know how to connect Rx2 port from radio0 to Tx/Rx port from radio1. Best Regards, Piotr W dniu 04.05.2015 o 12:48, Piotr Krysik via USRP-users pisze: > Many thanks for the answer Ian - it is very helpful. I will try to > connect rx and tx in points highlited by you (by getting around DDC and > DUC) as soon as I will get Xilinx tools to compile FPGA code (currently > I have problems with Vivado). > > Meanwhile - do I understand it correctly that currently it is not yet > possible (can't be easily done) to connect inside the FPGA radio fronend > with RFNoC signal processing blocks? > > Best Regards, > Piotr > > W dniu 02.05.2015 o 17:41, Ian Buckley pisze: >> Piotr, >> The loopback path you have identified is inward facing, in other words it forces the RX to listen to the TX digital data being sent to the DAC. (Its a very useful feature during testing both in simulation and the real world.) >> You could make a code change in that area and create a similar outward facing path. >> >> Personally I'm not sure I'd recommend that since you would be looping raw ADC data to the DAC without any DC offset or IQ balance corrections, I think it might be better if such a loop back path allowed data to pass through the "rx_frontend" and "tx_frontend" which means you are dealing with the "rx_corr_i/q" and "tx_fe_i/q" signals. >> >> If you also require digital tuning via the CORDIC then your loopback path would instead include the signals "sample_rx" and "sample_tx" and then you would also have the full signal processing capabilities of the DSP's included. >> >> Which you choose will depend I guess on what you are trying to achieve. >> -Ian >> >> On May 1, 2015, at 11:53 PM, Piotr Krysik via USRP-users <usrp-users@lists.ettus.com> wrote: >> >>> W dniu 01.05.2015 o 20:58, Martin Braun via USRP-users pisze: >>>> On 01.05.2015 11:23, Piotr Krysik via USRP-users wrote: >>>>> Hi all, >>>>> >>>>> I'm trying to make USRP x310 retransmit on one of the outputs everything >>>>> that it receives on one of the inputs, without sending the signal to the >>>>> controlling PC. >>>>> >>>>> I've connected together two "RFNoC: Radio" blocks, one set to RX to >>>>> another TX mode >>>>> >>>>> Expected behaviour was that the received signal would be copied inside >>>>> FPGA and retransmitted through TX antenna. >>>>> However, it doesn't work the way I intended - the easiest to check >>>>> symptom is that LEDs under TX/RX and RX2 are not lighted up. >>>>> >>>>> Is it possible to do what I want with current RFNoC? >>>> Currently, no. The radios in there were written pre-RFNoC, so they have >>>> some shortcomings. We're working on this, though, as this is something >>>> we definitely want to support. >>>> >>>> Cheers, >>>> Martin >>>> >>>> >>> If it's not supported currently by GNU Radio - maybe there is some lower >>> level way to connect RX to TX in the FPGA? >>> >>> For example I've stumbled upon following lines in the >>> uhd/fpga-src/usrp3/lib/radio/radio.v: >>> >>> // Digital Loopback TX -> RX (Pipeline immediately inside rx_frontend). >>> wire [31:0] > rx_fe = loopback ? tx : rx; >>> >>> Maybe this is what I'm looking for? Or maybe there's another way - for >>> example to use RFNoC to connect radio frontends that don't involve GNU >>> Radio? >>> >>> Sorry if I'm asking silly questions, but I'm RFNoC novice and I'm trying >>> to estimate what amount of work it would take to make USRP X3x0 stream >>> samples from RX to TX inside FPGA. >>> >>> Best Regards, >>> Piotr >>> >>> _______________________________________________ >>> USRP-users mailing list >>> USRP-users@lists.ettus.com >>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >

Hi Piotr, that is currently not possible -- this is not a GNU Radio limitation, but an issue of our current UHD/RFNoC implementation. However, we definitely want this to work, and are working towards making it work. Cheers, Martin On 04.05.2015 07:11, Piotr Krysik via USRP-users wrote: > Hi all, > > After seeing > https://github.com/EttusResearch/uhd/wiki/RFNoC:--Specification I will > partially answer my question. > It looks that radio blocks can be connected directly to other signal > processing blocks in RFNoC. Still I dont know how to connect Rx2 port > from radio0 to Tx/Rx port from radio1. > > Best Regards, > Piotr > > W dniu 04.05.2015 o 12:48, Piotr Krysik via USRP-users pisze: >> Many thanks for the answer Ian - it is very helpful. I will try to >> connect rx and tx in points highlited by you (by getting around DDC and >> DUC) as soon as I will get Xilinx tools to compile FPGA code (currently >> I have problems with Vivado). >> >> Meanwhile - do I understand it correctly that currently it is not yet >> possible (can't be easily done) to connect inside the FPGA radio fronend >> with RFNoC signal processing blocks? >> >> Best Regards, >> Piotr >> >> W dniu 02.05.2015 o 17:41, Ian Buckley pisze: >>> Piotr, >>> The loopback path you have identified is inward facing, in other words it forces the RX to listen to the TX digital data being sent to the DAC. (Its a very useful feature during testing both in simulation and the real world.) >>> You could make a code change in that area and create a similar outward facing path. >>> >>> Personally I'm not sure I'd recommend that since you would be looping raw ADC data to the DAC without any DC offset or IQ balance corrections, I think it might be better if such a loop back path allowed data to pass through the "rx_frontend" and "tx_frontend" which means you are dealing with the "rx_corr_i/q" and "tx_fe_i/q" signals. >>> >>> If you also require digital tuning via the CORDIC then your loopback path would instead include the signals "sample_rx" and "sample_tx" and then you would also have the full signal processing capabilities of the DSP's included. >>> >>> Which you choose will depend I guess on what you are trying to achieve. >>> -Ian >>> >>> On May 1, 2015, at 11:53 PM, Piotr Krysik via USRP-users <usrp-users@lists.ettus.com> wrote: >>> >>>> W dniu 01.05.2015 o 20:58, Martin Braun via USRP-users pisze: >>>>> On 01.05.2015 11:23, Piotr Krysik via USRP-users wrote: >>>>>> Hi all, >>>>>> >>>>>> I'm trying to make USRP x310 retransmit on one of the outputs everything >>>>>> that it receives on one of the inputs, without sending the signal to the >>>>>> controlling PC. >>>>>> >>>>>> I've connected together two "RFNoC: Radio" blocks, one set to RX to >>>>>> another TX mode >>>>>> >>>>>> Expected behaviour was that the received signal would be copied inside >>>>>> FPGA and retransmitted through TX antenna. >>>>>> However, it doesn't work the way I intended - the easiest to check >>>>>> symptom is that LEDs under TX/RX and RX2 are not lighted up. >>>>>> >>>>>> Is it possible to do what I want with current RFNoC? >>>>> Currently, no. The radios in there were written pre-RFNoC, so they have >>>>> some shortcomings. We're working on this, though, as this is something >>>>> we definitely want to support. >>>>> >>>>> Cheers, >>>>> Martin >>>>> >>>>> >>>> If it's not supported currently by GNU Radio - maybe there is some lower >>>> level way to connect RX to TX in the FPGA? >>>> >>>> For example I've stumbled upon following lines in the >>>> uhd/fpga-src/usrp3/lib/radio/radio.v: >>>> >>>> // Digital Loopback TX -> RX (Pipeline immediately inside rx_frontend). >>>> wire [31:0] > rx_fe = loopback ? tx : rx; >>>> >>>> Maybe this is what I'm looking for? Or maybe there's another way - for >>>> example to use RFNoC to connect radio frontends that don't involve GNU >>>> Radio? >>>> >>>> Sorry if I'm asking silly questions, but I'm RFNoC novice and I'm trying >>>> to estimate what amount of work it would take to make USRP X3x0 stream >>>> samples from RX to TX inside FPGA. >>>> >>>> Best Regards, >>>> Piotr >>>> >>>> _______________________________________________ >>>> USRP-users mailing list >>>> USRP-users@lists.ettus.com >>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> >> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> > > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >

Hi Martin, Do you maybe know what is the source of the problem? I remember that for example USRP N200 has a mechanism that turns off Tx channel when the device don't get any data to transmit from the controlling PC. After implementing a signal generator inside of FPGA it caused problems. Workaround was to transmit some data to the device. Is the problem of similar nature in this case? -- Piotr W dniu 04.05.2015 o 19:29, Martin Braun via USRP-users pisze: > Hi Piotr, > > that is currently not possible -- this is not a GNU Radio limitation, > but an issue of our current UHD/RFNoC implementation. However, we > definitely want this to work, and are working towards making it work. > > Cheers, > Martin > > > On 04.05.2015 07:11, Piotr Krysik via USRP-users wrote: >> Hi all, >> >> After seeing >> https://github.com/EttusResearch/uhd/wiki/RFNoC:--Specification I will >> partially answer my question. >> It looks that radio blocks can be connected directly to other signal >> processing blocks in RFNoC. Still I dont know how to connect Rx2 port >> from radio0 to Tx/Rx port from radio1. >> >> Best Regards, >> Piotr >> >> W dniu 04.05.2015 o 12:48, Piotr Krysik via USRP-users pisze: >>> Many thanks for the answer Ian - it is very helpful. I will try to >>> connect rx and tx in points highlited by you (by getting around DDC and >>> DUC) as soon as I will get Xilinx tools to compile FPGA code (currently >>> I have problems with Vivado). >>> >>> Meanwhile - do I understand it correctly that currently it is not yet >>> possible (can't be easily done) to connect inside the FPGA radio fronend >>> with RFNoC signal processing blocks? >>> >>> Best Regards, >>> Piotr >>> >>> W dniu 02.05.2015 o 17:41, Ian Buckley pisze: >>>> Piotr, >>>> The loopback path you have identified is inward facing, in other words it forces the RX to listen to the TX digital data being sent to the DAC. (Its a very useful feature during testing both in simulation and the real world.) >>>> You could make a code change in that area and create a similar outward facing path. >>>> >>>> Personally I'm not sure I'd recommend that since you would be looping raw ADC data to the DAC without any DC offset or IQ balance corrections, I think it might be better if such a loop back path allowed data to pass through the "rx_frontend" and "tx_frontend" which means you are dealing with the "rx_corr_i/q" and "tx_fe_i/q" signals. >>>> >>>> If you also require digital tuning via the CORDIC then your loopback path would instead include the signals "sample_rx" and "sample_tx" and then you would also have the full signal processing capabilities of the DSP's included. >>>> >>>> Which you choose will depend I guess on what you are trying to achieve. >>>> -Ian >>>> >>>> On May 1, 2015, at 11:53 PM, Piotr Krysik via USRP-users <usrp-users@lists.ettus.com> wrote: >>>> >>>>> W dniu 01.05.2015 o 20:58, Martin Braun via USRP-users pisze: >>>>>> On 01.05.2015 11:23, Piotr Krysik via USRP-users wrote: >>>>>>> Hi all, >>>>>>> >>>>>>> I'm trying to make USRP x310 retransmit on one of the outputs everything >>>>>>> that it receives on one of the inputs, without sending the signal to the >>>>>>> controlling PC. >>>>>>> >>>>>>> I've connected together two "RFNoC: Radio" blocks, one set to RX to >>>>>>> another TX mode >>>>>>> >>>>>>> Expected behaviour was that the received signal would be copied inside >>>>>>> FPGA and retransmitted through TX antenna. >>>>>>> However, it doesn't work the way I intended - the easiest to check >>>>>>> symptom is that LEDs under TX/RX and RX2 are not lighted up. >>>>>>> >>>>>>> Is it possible to do what I want with current RFNoC? >>>>>> Currently, no. The radios in there were written pre-RFNoC, so they have >>>>>> some shortcomings. We're working on this, though, as this is something >>>>>> we definitely want to support. >>>>>> >>>>>> Cheers, >>>>>> Martin >>>>>> >>>>>> >>>>> If it's not supported currently by GNU Radio - maybe there is some lower >>>>> level way to connect RX to TX in the FPGA? >>>>> >>>>> For example I've stumbled upon following lines in the >>>>> uhd/fpga-src/usrp3/lib/radio/radio.v: >>>>> >>>>> // Digital Loopback TX -> RX (Pipeline immediately inside rx_frontend). >>>>> wire [31:0] > rx_fe = loopback ? tx : rx; >>>>> >>>>> Maybe this is what I'm looking for? Or maybe there's another way - for >>>>> example to use RFNoC to connect radio frontends that don't involve GNU >>>>> Radio? >>>>> >>>>> Sorry if I'm asking silly questions, but I'm RFNoC novice and I'm trying >>>>> to estimate what amount of work it would take to make USRP X3x0 stream >>>>> samples from RX to TX inside FPGA. >>>>> >>>>> Best Regards, >>>>> Piotr >>>>> >>>>> _______________________________________________ >>>>> USRP-users mailing list >>>>> USRP-users@lists.ettus.com >>>>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >>> _______________________________________________ >>> USRP-users mailing list >>> USRP-users@lists.ettus.com >>> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >>> >> >> _______________________________________________ >> USRP-users mailing list >> USRP-users@lists.ettus.com >> http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >> > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com >

On 04.05.2015 12:51, Piotr Krysik via USRP-users wrote: > Hi Martin, > > Do you maybe know what is the source of the problem? > > I remember that for example USRP N200 has a mechanism that turns off Tx > channel when the device don't get any data to transmit from the > controlling PC. After implementing a signal generator inside of FPGA it > caused problems. Workaround was to transmit some data to the device. > > Is the problem of similar nature in this case? No, this is something else entirely. In essence, the radios used predate RFNoC, so they behave differently than the other RFNoC blocks. We need to change that, which we will do together with several other changes. That should fix this issue. Cheers, Martin

W dniu 05.05.2015 o 20:36, Martin Braun via USRP-users pisze: > On 04.05.2015 12:51, Piotr Krysik via USRP-users wrote: >> Hi Martin, >> >> Do you maybe know what is the source of the problem? >> >> I remember that for example USRP N200 has a mechanism that turns off Tx >> channel when the device don't get any data to transmit from the >> controlling PC. After implementing a signal generator inside of FPGA it >> caused problems. Workaround was to transmit some data to the device. >> >> Is the problem of similar nature in this case? > No, this is something else entirely. In essence, the radios used predate > RFNoC, so they behave differently than the other RFNoC blocks. We need > to change that, which we will do together with several other changes. > That should fix this issue. > > Cheers, > Martin > > _______________________________________________ > USRP-users mailing list > USRP-users@lists.ettus.com > http://lists.ettus.com/mailman/listinfo/usrp-users_lists.ettus.com > Hi all, I've hacked the FPGA code according to Ian's advice. In what I've done DDC output (sample_rx) is feeded to DUC input. With simple gnuradio-companion flowgraph that receives signal from the device and sends any signal that will be replaced by DDC output - it works. Many thanks Ian for pointing this signals to me. The hardest part was getting Xilinx tools to compile the FPGA code - the outputted error messages tend to be rather cryptic (which is only one of their multiple drawbacks btw.). Best Regards, Piotr