Process Duration Timer Blog 1 (The Timer)

RE: Process Duration Timer Blog 1 (The Timer)

mcb1 — Tue, 25 Jul 2017 06:53:57 GMT

I have to agree with [mention:4f57fc9d538949ad9eb336ddb9469bb8:e9ed411860ed4f2ba0265705b8793d05] about more DIY test equipment.

The schematic appears to be here.

https://forum.banggood.com/forum-topic-120330.html

The button seems to have a dual function and a long press allows some setting.

I was wondering what happens if you short and hold the reset line.

I suspect it will reset the display, but may simply stop the processor.

Cheers

Mark

RE: Process Duration Timer Blog 1 (The Timer)

dougw — Tue, 25 Jul 2017 04:48:40 GMT

Your crystal shorting solution is probably fine as long as you short the input side of the oscillator circuit, as Jon points out.

From your tests it seems the CPU maintains its state or operates all the way to zero hz, so it will continue when the clock increases. It probably doesn't matter too much in your application if the clock isn't perfectly accurate for the first few microseconds.

In the worst case you can add more complexity if it starts getting flaky.

RE: Process Duration Timer Blog 1 (The Timer)

DAB — Mon, 24 Jul 2017 19:06:40 GMT

Nice project John.

I agree with Shabaz.

Rather than shorting the crystal you should break the trace to the processor and install a simple AND gate to stop the clock signal.

If you short the crystal, you could cause damage.

Overall, it works great as is so great job.

DAB

RE: Process Duration Timer Blog 1 (The Timer)

jc2048 — Mon, 24 Jul 2017 12:19:19 GMT

You're doing DIY Test Equipment too!

The crystal oscillator is an amplifier (inside the processor) with the crystal as feedback. The crystal is very selective about the frequencies which it will pass (basically, the frequency at which it mechanically resonates and the odd harmonics of that frequency). So one pin (pin 5, XTAL1) is an input and the other (pin 4, XTAL2) is an output. I can't see which you've wired to. It's probably a bad idea to short the output to ground. CMOS can take a lot of abuse and you may get away with it but it won't necessarily be reliable over the long term.

Crystal oscillators start up through random noise being amplified, the crystal selecting only those frequencies that fall in its very narrow passband, and the oscillation steadily building. That can take quite a long time - if you look at it on a scope you'll see what I mean. Processors have internal timers to hold off processing at start-up to ensure the clock is running and stable before they operate. In this case you are bypassing that and hoping that the processor will pick up and execute code properly. Evidently it is doing that, but you shouldn't depend on it getting it right.

As Shabaz says, an external oscillator with a gating circuit would be cleaner as it allows for immediate execution from a standing start (in the old days, such circuits were used for single-stepping processors in order to watch code running slowly enough that you could see what the address and data buses were doing on LEDs). [The gating needs to be more than just an AND gate, though, so that you get full cycles and no danger of a short glitch at the start.]

The relay will load the oscillator a little, but the contact capacitance when the contact is open should be small, so it probably won't pull the crystal frequency very much. [It's probably not very precise as a clock, anyhow, as there's no way to trim the load capacitance. As a clock, you could see an error of anything up to a minute or two a month unless you're very lucky.]

If you are going to wire this up to an Arduino anyway, why not ditch the Atmel processor on this board and use the Atmel processor on the Arduino to do the clock timing and display? You seem to be jumping through hoops in order to avoid a little simple coding.

RE: Process Duration Timer Blog 1 (The Timer)

shabaz — Mon, 24 Jul 2017 05:00:54 GMT

It is such a cool thing to have a process timer for the bench : )

For some recent testing (I think it was the Capacitance-to-Digital evaluation board from TI) I needed to do some timing, and in the end used a Casio watch manually : ( A process timer is a really good idea.

RE: Process Duration Timer Blog 1 (The Timer)

shabaz — Mon, 24 Jul 2017 04:42:27 GMT

Hi John,

Clever idea, to freeze the clock as you have done : ) There are some other ways to do the freezing (e.g. use an external clock oscillator and gate it with (say) 'AND' logic, but that requires more circuitry than the relay. I guess there could be a risk of glitches during switching, causing an instruction to be incorrectly processed, but your testing doesn't seem to have identified any issue so far.

Your solution clearly works given the constraint of having a clock board with no source code, and I think there is no risk of damage by pulling the clock pin.