Is the customer always right?

Lancie1 said:

That is a good example. I worked on the flight computers for the NASA Saturn V rocket. The requirement was that every circuit had 3 parallel paths. That made it complex, but there was no way to fix it after it left Earth.

Click to expand...

Ours are completely inaccessable once we start a hazardous operation. If you can't get to it, you might not have anything left to get back to...

Mark Snodgrass said:

Stupid question, how do you know if one of your redundant systems has failed?

Click to expand...

We are moving our redundant systems from the GE redundant processor hardware to the Triconex TMR system. It uses three concurrently running processors that use voting logic to control, and critical I/O is also redundant; if a card faults, it swaps to the redundant one within a scan. All the processors and I/O cards give a fault indication, both locally and through software, and all are hot-swappable.

Speed costs money... how fast do you want to go?

+1 more.

dploof23 said:

+1

Wow, that's impressive.

Click to expand...

Depends. I once did a system for making quartz glass that was also everything redundant and somewhat complicated as a result. However... each batch was a 1-week process time. Considering maintenance, the customer planned on 50 batches a year. So... each run was 2% of the annual production, assuming that all runs were successful which, as it turns out, was not a given although it was the most likely outcome.

Easy calculation to determine how much the redundancy cost them and whether or not it was too much.

Lancie1 said:

That is a good example. I worked on the flight computers for the NASA Saturn V rocket. The requirement was that every circuit had 3 parallel paths. That made it complex, but there was no way to fix it after it left Earth.

Click to expand...

You didn't wire the 'oxygen stirrers' on Apollo 13 did you lancie.

I recently had a project where the customer initially wanted redundant CPUs.
I think that they must have had an incident where it took a long time to pinpoint the problem, get an electrician, find a spare CPU, install the spare CPU and load the programs, and restart the process.

What I did was simply to make 2 quotes, with and without redundancy. In the case of the 'without redundancy' I included the price of a spare CPU with the programs pre-loaded. And I made a manual how to replace the CPU and restart the process, so that they should be able to get up and running again in the shortest possible time.
In the end they chose 'without redundancy'.

But of course, there are cases where redundancy does make sense.
I think that for very big plants and/or very costly products, there is a reasonable relation between risc reduction and expense.