Mechanical stuff

National fiesta last Friday, everything was shut, so did a bit of machining.

Wasn’t happy with the clap hinge on the prototype, slightly agricultural.

It was really uncomfortable to hold in one hand, so I machined down the arms to make the brackets flush.

The clearance required for the hex fixing bolts was tight, so need to move the fixing hole a bit to the right.

The sharp bracket angle was annoying, so rounded them off. I didn’t have any dome headed hex bolts

A lot more comfortable to hold.

Got a few quotes for making a real protoype box, ouch, about the same as a month’s rent for my Spanish appartment. But finally got some money for an illegal naughty but race winning modification I did to a slot car controller, over 18 months ago.

Delivery, 1 to 2 weeks. 3 to 4 weeks, arghh

The end is in sight.

Advertisements

The first real test, run over night, free running jam mode

It works.

I connected the clap sensor on the slate to a spare pin on the TCG microprocessor,(Red phono lead) and told the TCG to freeze the display when the clap arm was closed. Closing the clapper many times, it seems as if the slate is running slightly fast, maybe  1/4 to half a frame after 8 hours. About what was expected.

The software that does nanotweaks to compensate for any basic crystal oscillator offset was not enabled. Workshop temperature was about 23C. The whole thing was run on a 5V bench power supply, since the slate display was on all the time, I didn’t think the internal 4 AA batteries would last overnight.

Pretty good for a less than 5€ TCXO.

 

EDIT:

12 hours in, still under 1 frame fast, that’s about 1 part per million accuracy

Boring Battery Test

I hadn’t done any battery life tests on the slate, somewhat amiss, so I added a couple lines of code that turned on the display at full brightness for 1 second every minute in Jam mode. I think this maybe a reasonable test, and it’s easily repeatable.

The batteries are Panasonic NiMh 2100mAh, almost new, with less than 10 recharge cycles.

I ran the test, first for 3 hours, 180 “slates”, switched it off overnight, then left it running until the battery volts dropped to 4.8V, (1.2V/cell) measured when the display was off. The slate electronics will work down to about 4.1V, when the switched mode supply for the display doesn’t switch any more.

Time        no load        full load
0                5.36            5.20
30 min     5.30        5.12
1hour        5.25        5.12
1hr30      5.22        5.06
2hr            5.20        5.03
3hr            5.16        4.95

Over night rest

0            5.19        5.02
1hr        5.14        4.98

Off to the street cafe for breakfast, more later.

2hr        5.11        4.96
3hr        5.10        4.93
4hr        5.08        4.92
5hr        5.07        4.91
6hr        5.07        4.89
7hr        5.06        4.86
8hr        5.05        4.85
9hr        5.05        4.85

Getting boring.

10hr        5.05        4.86

Take Two, 5 seconds on each minute, duty cycle 8.3%

Time        No load        Full load
Start        5.33        5.18
30min    5.27        5.13
1hr          5.20        5.06
2hr         5.11        4.96
3hr         5.04        4.91
4hr         5.02        4.88

bedtime

Re start   5.09        4.93
1hr        5.01        4.86
2hr        4.99        4.85
3hr        4.97        4.82
4hr        4.96        4.81

etc.

10hr        4.84        4.69

Conclusions

The software hasn’t crashed, and it looks as if 18 hour battery life  might be attainable

Drift

A quick drift test.

Scope’ pics of a free running Slate generating a test square wave against a TC generator locked to a Rubidium 10MHz clock

Image

Start

Image

4 hours later

About 0.17 frames by the looks of it. Workshop temperature was a constant 27C

Done a few mods

Mainly pics
Left hand side

Right side

Clapper

Battery and USB port

64×128 LCD. Pity I crushed the fragile connector when assembling the box for the first time.

Full frontal

In colour

Babyslate preliminary specifications

I’ve put a basic specification sheet here

The basics are there, still a bit of fine tuning to do, need to change some of the metalwork to protect the switches for when the slate gets dropped on a concrete floor and that sort of thing. I’m stuck in limbo until the PCB factory restarts in September, after their summer break, then a few days hours to check the re-mapping of the hardware. Done quite a lot on the battery/power side to minimise power consumption in the new hardware, so it’s going through the old code and looking for Gotcha’s.

Testing a TC slate

Spent a couple of days adding a LED display to my Time Code generator, added a display driver to the software. Then it went pear shaped. I must remember to make sure I’m programming the micro with the correct hex file, not one from another project. Shit happens.

I put a cheap temperature compensated 12MHz crystal oscillator on the slate instead of the normal 50cent 12MHz crystal, it measured about 1PPM fast, not bad, as my workshop is warm, about 28 C this time of year.

The time code is set up for 25fps, and the camera is in frame mode not PAL. . Here’s a pretty dreadful workbench video, the firewire died on my normal camera:

I ran out of patience, and stopped videoing at 6 hours. It looks pretty OK to me with no obvious drift.

Next thing to do is to make a small environment chamber with a Peltier heat pump, so I can drop the temperature of the slate, to below 0C, and see how it drifts. I went dumpster diving and got a nice thick sheet of polystyrene that should do nicely for the thermal insulation.

The PCB firm I use is on holiday until 1st of September, I really want to get my hands on the next set of boards.