General Setup

The heart of the hardware installation is a Pentium II based personal computer with 400MHz CPU clock rate and 128MB RAM, mounted on a ASUS P2B ATX mother board with 100MHz bus clock, 4 PCI slots, 3 ISA slots and an AGP port. We have chosen that board mainly because it is one of the few which still has a sufficient number of ISA slots; we need the ISA slots for the GPIB card, the pulse card and the scope card, which we took over from an old installation and which are still based on that ancient standard. The computer is connected to the local network and to the internet (there is no remote control enabled; the system can only be used as a client) with a 10/100MBit ethernet card. It is running under Windows 98, since the seller of the cards provided his drivers for Windows 95 and since the program SeveNMR runs under Windows 3.11 for workgroups.

Most of the devices of our spectrometer have a GPIB port. A unique address has been given to each of them:

The phase is not controlled by GPIB since GPIB has quite a low data transfer rate. In order to control the amplitude of the PTS by the computer it has to be switched to remote mode. This can be done by turning the amplitude button down to the click. The option to connect the Le Croy oscilloscope to the computer is only used for high resolution fast fourier transforms, where we average over many pulses before we read it out.

A driver provided by the manufacturer helps with the data transfer between SeveNMR and the hardware. In our software environment it is called D16gpib. It is a copy of TPWGPIB, the 16 bit version of the driver obtainable from National Instruments.

For generating the pulses we used a MR3020 Pulse Programmer from s.m.i.s. . It works completely independent of the rest of the computer and is therefore not affected by jams on the data bus or in the CPU due to other applications. The price to pay for this independence is that one has to program the card as a little computer in a Forth language. Since Forth is quite messy to handle, the manufacturer provides the card with a compiler which makes a more comfortable PPL language available and translates it to Forth. But SeveNMR contains an extended compiler which is adapted to our situation, knows some more commands such as adjusting the phase or opening the amplifier gate automatically and allows a very primitive program structure. But we will come to the pulse program compiler soon. The data transfer to the card is done by the shared library MRCOMS.DLL provided by the factory. In order to install them properly we added the two entries [MRCOMS] and [SMISDEFS] to WIN.INI.

The MR3020 Pulse Programmer generates rectangular pulses on 16 possible channels with a maximum time resolution of 100ns, a shortest duration of 300ns and a maximum duration of 16000s. The pulse programmer is in principle nothing else than a device which enables us to switch single bits in the output on and off. A pulse program therefore basically consists in setting the corresponding bits in a word (1 word = 2 bytes), sending it to the output and waiting an appropriate delay. In addition to that the Forth language also contains some commands to handle conditions and loops. In our setup the channels, denoted by the bits representing them, are connected as follows:

We used an IMTEC PC SCOPE board T12840 for the data acquisition. Using a PC board for the data acquisition implies the advantage of being much faster than using an external scope. The IMTEC board has two 8-bit channels, a sampling rate of maximum 20MHz in the two-channel mode and it offers the choice between 9 analog input ranges between 128mV and 64V. The card is triggered by an external trigger line; the trigger should be high during one dwell time. The data transfer is again done by using a dynamically linked library which we downloaded from the manufacturer's page.