The M8195A can operate in different modes: ‘Single Channel’, ‘Single Channel with 
markers’, ‘Dual Channel’, ‘Dual Channel Duplicate’, ‘Dual Channel with markers’, or 
‘Four Channel’ 
 
There are two different memory modes available: ‘Internal’ and ‘Extended’. The 
memory mode is configurable for each channel. 
 
The Sample Rate of all four Digital to Analog Converters (DAC) is selectable between 
53.76 GSa/s … 65 GSa/s. The internal waveform memory always operates at the 
sample rate. The extended waveform memory can operate at sample rate 53.76 
GSa/s … 65 GSa/s or at one half of the sample rate 26.88 GSa/s … 32.5 GSa/s or at 
one fourth of the sample rate 13.44 GSa/s … 16.25 GSa/s. The speed of operation of 
the extended memory is adjustable using the parameter ‘Sample Rate Divider 
(Extended Memory)’ which can be changed by the user. Possible values are 1, 2, and 
4. The Sample Rate Divider is identical for all channels that are sourced from 
extended memory. In case the Sample Rate Divider is adjusted to two or four, the FIR 
filters are used as interpolation filters by factors of two or four. The interpolation is 
necessary as the DAC always operates in the range 53.76 GSa/s … 65 GSa/s. 
   
Each channel has a programmable FIR Filter.  
The number of filter coefficients depends on the Sample Rate Divider; 16, 32, or 
64 filter coefficients are available if the Sample Rate Divider is set to 1, 2 or, 4 
respectively.   
In case the Sample Rate Divider is changed, the FIR filter coefficients of each 
channel sourced from extended memory are loaded to operate as a by one or by 
two or by four interpolation filter. 
Figure 4 depicts how the FIR filters are used as interpolation filters. If the sample 
rate divider is set to two, the sample value ‘0’ is inserted between each sample 
that is read from extended memory. If the sample rate divider is set to four, three 
consecutive times the sample value of ‘0’ is inserted between each sample that 
is read from extended memory. 
 
 
There are two sets of filter coefficients for each channel. One set is currently 
used for data generation. The other set can be reconfigured in parallel with new 
coefficients. After reconfiguration, the entire reconfigured set can be used 
simultaneously for data generation. This allows reconfiguration during data 
transmission without generating distortions at the output signal. By pressing the 
corresponding ‘Send To Instrument’ button of the SFP or by sending the 
corresponding API command, the new set of filter coefficients is applied.