PID Without a PhD, Finally

Most swept-sine measurement systems (I have one that I compile in with any serious control code in my embedded stuff) will run a measurement unattended. Just leave a yellow card with "test in progress" on the thing and go have a nice lunch.

There's formal methods for extracting parameters from a step response, but if your accuracy need isn't super-stringent then eyeballing it with RC networks isn't a bad way to go.

Yes on the nonlinearity, although if it's not going to batter the power supply too much then a simple slow PWM to the heating element will take care of that right off, or use the pass transistor for the heating element (which is easy if you're talking about incubating eggs, but challenging if you're melting tungsten).

These days there's probably a lower-budget way to do that with a two- channel O-scope that can record to disk plus some number-crunching software like Scilab.

It was exactly what I was going to suggest -- stick a piezo accelerometer to the table, give it a rap with the knuckles light enough so the thing doesn't lose lock, and look at the response.

Maybe, but I paid $350 for mine, which is hard to beat.

Cheers

Phil Hobbs

Goodgawd -- starting at $300 on eBay. Yeeks!

I actually had the predecessor to the 35665A available to me when I started building control analyzer code into my software, on the theory that if you're tapping into the exact same signals as you're doing the controlling with, then (at least from the processor's perspective) your plant measurements will be exact.

It's worked consistently and well for me, so I'm going to stick with it.

Absolutely. About 80% of my control stuff is analogue, though, so the DSA is great.

Cheers

Phil Hobbs

Does the version you have show the phase difference between excitation and system output? At FLIR we had the predecessor, which was slow and clunky and we lived in fear of the hard drive going out, and we had something similar to those that only showed the magnitude difference.

I actually used it in my first big digital control project, for a variety of reasons. Later, I ended up inheriting the analog boards that my digital board replaced, and used it to maintain those, too.

Part of the reason that I like my in-built thingie is because it's designed so that I can suck the response off into a file. That gets put on a computer with Scilab, and I have scripts that let me generate closed- loop Bode plots from the model of the control rule used by the processor. Since it's all digital, that model is damn near exact, meaning that for well-behaved plants the tuning often only took one or two iterations.

(Plants with significant nonlinearities took more work, partially because the apparent response changed with the controller used, and partially because one had to make tradeoffs involved with dodging and/or embracing the nonlinearities.)

Yup. So did the 35660A and 3561 I had at IBM BITD.

If you have time and interest (and some shelf space!), try one out. The only bug I know about is that mine gets the noise BW of its flattop window wrong. Uniform and von Hann are fine.

You can also get a ROM for $50 or so that will turn on all the options, e.g. various sources and math settings.

I keep several of my boat anchors in a nice HP rack that I bought from ecnerwal in 2009 or so.

(Hi, Lawrence, how's tricks?)

Cheers

Phil Hobbs

Hi guys, I was wandering around, 1/2 thinking of this, and realized (You'll all say Duh...) that its the response of my drive that I need to measure. A piezo stack, driving a Al flex mount. (7025 Al? 7075? some stiff alloy) with a grating attached... (grating feed back into the laser.... that determines the frequency, to some extent) The laser is much faster than my piezo stack.

I've got a split photodiode somewhere. George H.