Red Pitaya - TANGO Controls

# 7 years ago
Manu	Hello all, We are testing the Red Pitaya board (125-14) and experiment some troubles. As far as I know, within our community, MAX IV has already tested this board (and is may be using it). We are using their classical scope application with a trigger at 100 Hz and a decimation factor set to 1 (meaning a sampling frequency of 125 MHz) Our problem is that as soon as we try to transfer more than 1000 points, the time required by the data transfer from the RP board is sometimes so long that we miss triggers! Is there anybody (at MAX IV or elsewhere) who experimented a similar problem? Thank's in advance for your answers Emmanuel

# 7 years ago

Hello all,

We are testing the Red Pitaya board (125-14) and experiment some troubles.
As far as I know, within our community, MAX IV has already tested this board
(and is may be using it).

We are using their classical scope application with a trigger at 100 Hz
and a decimation factor set to 1 (meaning a sampling frequency of 125 MHz)
Our problem is that as soon as we try to transfer more than 1000 points, the time required by the data transfer from the
RP board is sometimes so long that we miss triggers!

Is there anybody (at MAX IV or elsewhere) who experimented a similar problem?

Thank's in advance for your answers

Emmanuel

# 7 years ago
hardion	Hi Manu, You are right, our Red Pitaya boards should be in production soon for the PSS. I will ask my colleague if they find the same limitation. /Vincent Vincent Hardion Control System MAX IV Laboratory

# 7 years ago
Manu	Hi Vincent, We have investigated more in this problem and finally succeed to minimise it. Our use case for the RedPitaya board is to acquire signal on their RF inputs with a trigger frequency set to 10 Hz. This means that you have 100 mS to transfer the read data from the RedPitaya to your host. Obviously, if you configure your board to read few points at each trigger, you won't have any problem. But as soon as we wanted to get several hundreds of points, we suffer from missed triggers. Therefore, we have modified the supplied SCPI server running on the board in order to set the TCP_NODELAY socket option to true. On the Tango class on the other side, we also set the socket option TCP_NODELAY to true. On top of that, we have configured the board to do the transfer in binary format. After theses changes, we are now able to get up to 5000 points at each trigger with a reasonable missed trigger (1 for 3000 trigs) Have a nice week-end Emmanuel

# 7 years ago

Manu

Hi Vincent,

We have investigated more in this problem and finally succeed to minimise it.
Our use case for the RedPitaya board is to acquire signal on their RF inputs with a trigger frequency set to 10 Hz.
This means that you have 100 mS to transfer the read data from the RedPitaya to your host. Obviously, if you configure
your board to read few points at each trigger, you won't have any problem. But as soon as we wanted to get several hundreds
of points, we suffer from missed triggers.
Therefore, we have modified the supplied SCPI server running on the board in order to set the TCP_NODELAY socket option to true.
On the Tango class on the other side, we also set the socket option TCP_NODELAY to true.
On top of that, we have configured the board to do the transfer in binary format.

After theses changes, we are now able to get up to 5000 points at each trigger with a reasonable missed trigger (1 for 3000 trigs)

Have a nice week-end

Emmanuel