Why PCIe-1x and not PCIe-4x? ACQ132CPCI runs a local bus up to 128MB/s, so at a peak rate of 256MB/s, PCIe 1x is a good match. Adding more lanes wouldn't add value. PCIe 1x also features a cable that is much lighter, more flexible and cheaper than the higher order cables.
Is it scaleable? Definately. Remember each card can have its own dedicated PCIe on cable link, contrast to the regular bus expander scenario where there is one link per chassis. A system with ACQ196CPCI generally tends to have far fewer cards than an equivalent 3U PXI system, but at the same time it tends to have many more simultaneous channels per crate.
How many PCIe uplinks per host? D-TACQ recommends a maximum of 4 uplinks per host computer. This can be accomodated using dual-port host adapters in 2 PCIe slots on the host. This isn't a hard [slot-defined] limit, more rule of thumb that the streaming data rate will start to become too high to handle economically; eg when streaming to disk. It can be cheaper to add more host computers, with reasonable disk subsystems, that to keep uprating a single "monster" system. The modularity of the RTM-T one-card-one-cable design encourages cost effective data farming in this way.
What are the advantages of a non-transparent bridge ? A typical bus expander system projects the cards in the downstream rack into the bus and address space of the host operating system. This can be helpful - it allows standard device drivers to run on the host, and the bus expander is transparent. However that scheme does have the weakness that the host and downstream systems are tightly coupled together - all the cards share the same address space, with consequences for reliability as the system grows, and, resetting the host resets the downstream system as well. In contrast RTM-T offers a non-transparent bridge. RTM-T appears as a single, limited address device in the host address space; there is little sharing of memory space, resulting in a simpler system, and the downstream crate can be powered off independently of the upstream host, while the host can be reset without resetting the downstream system. This improves the maintenance and availability of both systems.
What is the purpose of the Gigabit Ethernet port? The Gigabit Ethernet port replaces the 100T Ethernet port found on ACQ196CPCI-RTM1. While it does represent an upgrade, the data rate is about 2..3X improvement, it will not show the same high performance as a native PCI MAC as used on ACQ132CPCI; the Gigabit Ethernet port is recommended primarily as a control interface. The Gigabit Ethernet also enables the standalone AWG function using AO32CPCI. The Gigabit Ethernet port also features an IEEE1588-enabled PHY, and, in the future, RTM-T will support precision timing using the standard PTP protocol.
Is this really a retrofit? Yes. All ACQ196CPCI and ACQ132CPCI cards are supported, although some may need a small modification to allow correct operation with RTM-T. AO32CPCI cards must have a rev2 artwork or later.
Is ACQ216CPCI supported by RTM-T? Unfortunately, no, the local bus to the RTM is not compatible. A digitizer with the throughput of ACQ216CPCI would be a good candidate in the future for a PCIe-4x link. Please contact D-TACQ for details.
How do I use the SFP port ? D-TACQ is developing a low-cost PCIe HBA to allow a fiber optic uplink to a host computer.
Example test setup with PCIe on cable to host PC.
RTM-T Interface Detail
D-TACQ Solutions Ltd,
James Watt Building, Scottish Enterprise Technology Park,
East Kilbride, G75 0QD, SCOTLAND UK
tel: +44 1355 272511