When a technology is architected to be robust, and has proven to be so in countless deployments over tens of years, it can be very easy to take it for granted.
A prime example is MIL-STD-1553B. It just works, right? Well: yes and no. In theory, you buy the right parts, drop them on to a PCB, write some software — and you’re good to go. After all: the 1553 standard is designed to deal with a host of signal integrity issues — so why would you need to check your design that closely?
I’m often asked why anyone would need to test 1553 with a variable voltage test card? After all — the voltage outputs from the components that you purchase are all fixed voltage outputs. These are all parts that are suggested by the manufacturer to be used. All the designer does is place them on the board. But: I’d argue that the kind of testing that the variable voltage output allows is essential to verifying a design.
The key is to ensure that the design has an acceptably low risk of failure: after all, designers will typically want to know that they are well within the limits that are required to meet the Mil Standards for their product. This is a part of the RT Validation testing. In the RT Validation Test plan, there is a test where the Manchester amplitude is adjusted down from 6v p-p down to 0.1v p-p. The voltage is varied down to a very low level to measure what the input threshold is for the receive section of the card.
For the RT Validation to be compliant to MIL-STD-1553B, the requirement is for a device to respond normally down to 0.86v p-p, and not to respond at all below 0.2v p-p. Voltage threshold testing is an important test that can indicate at a very early stage how well the design is likely to do in other physical layer tests.
If the design is good and meets the proper input threshold, this will help to meet other criteria that play into the input signal detection and decoding.
Critical check
The PCB copper traces have resistance and will create losses. Depending on the layout, it can be critical to check that there have not been any issues created by the routing on the board and through various connectors. The transmission line helps in this — but there will still be losses that must be accounted for.
A lab or SIL doesn't typically have the same wire lengths as the intended real vehicle. The only thing that can be done here is to simulate those lengths by time and amplitude variation. This is the reason for the requirements in the MIL-STD-1553B and tested to in the RT Validation testing; these were created with the worst case in mind. Then, the qualification testing verifies that those requirements are met under worst case conditions of temperature, shock, vibration and other parameters that can be varied.
If the requirement is to create a more realistic SIL, adjusting the transmit voltage of a transmitting RT is one more level of realism that can be used, although this is not a very typical requirement for a SIL: this might be something that would only be used for an extreme case. All of that particular simulated RT’s amplitude responses could be set to the amplitude for that location in a vehicle, irrespective of the lab use.
The farther away from the BC that the device is, the lower the amplitude of the command or the response from the RT transmitting. This is simply IR losses, as might be expected in various lengths of copper wire. The backbone of a bus that runs typically the length of a vehicle will have the most IR losses that reduce the signal amplitude. The amplitude is also influenced by the various stubs that are attached, their lengths and where they are located relative to any other individual RT or BC. Every stub that attaches a device to the bus will create some amount of reflection back into the bus: however minimal, it has an effect.
In summary: as long as a designer has verified that the device meets the MIL-STD-1553B requirements for the minimal amplitude detection, that is one very critical design step closer to ensure a successful design. The golden rule applies to 1553 as it does to so many other things: test, test and test again. And, yes: that includes testing with variable voltage.
