November 3, 2011
about a year ago
Have you seen or compiled a list of uSD cards and their access times? Or is such a list doomed to be instantly wrong, due to specifications being “subject to change without notice”?
Excellent - I can’t wait to try it when I get back to my shop in April. In the meantime, stuck here in Paris, sigh.
The Windows WAV Trigger utilities (WT-InitMaker_100_20140102.exe, WT-Flasher_101_20140209.exe, and WT-Remote_022_20140209.exe) won’t open COM ports higher than 9 (not string truncation - see below). No worries, we can simply renumber the FTDI’s COM port down to the COM1-COM9 range.
These steps (based on this comment) will show all devices which have ever been added as COM ports, and will allow removal of old associations to COM Ports no longer used.
(comment moved down)
Love the board in general but I noticed this clicking and rebooting behavior with an 8ohm speaker connected at powerup. With speaker disconnected, there’s definitely 2.25V driven across the speaker pins at powerup. I’m wondering if the reset behavior is related to no series input capacitor for the LM4990:
the TI PCM5100 DAC datasheet states that its output is ground-centered, due to its internal -3.3V charge pump, and it mutes to ground
the TI LM4990 amp datasheet shows a 0.39uF non-polarized (ceramic) coupling capacitor in all the examples, and never indicates that the cap is optional. Internally, the amp is DC coupled, so 0V at the amp input will produce approximately ½ * 5V measured across the output pins. So the speaker will see a steady state DC voltage (and hence wasted power) under some conditions, and clipping under others:
if the DAC goes into “mute” (0 VDC), this drives the amp’s input to its bottom rail, again putting 2.5V across the speaker
Played WAV samples which have a DC offset will bias the speaker. Such offsets are easily removed in, say, Audacity.
All DAC excursions driving the amp’s input below ground will result in clipping distortion. Biasing the DAC data so that its output remains in the 0V-3.3V range works, but removes half the DAC’s effective output voltage, reducing dynamic range, and raising the noise floor.
Best workaround: Solder in a 0.39uF SMT ceramic cap in series with R9 22K (adjacent to the speaker pins), like a pair of leaned-together tombstones.
Easier but less-good workarounds which produce reduced power and/or reduced voltage range:
try speakers with higher resistance (16, 32, 40 ohms) until the reset problem goes away. This gets you some functionality at reduced power.
insert a non-polarized 10uF 10V capacitor in series with the speaker to AC-couple it (or use two back-to-back series 25uF caps to create a 10uF “non-polarized” cap). This solves half the problem, but is better than nothing.
WT firmware hack: when the amplifier is enabled, never mute the DAC, have I2S run continuously so the DAC doesn’t automute, then “center” the DAC at 2.5V (LM4990’s guesstimated midway input voltage) to produce 0VDC across the speaker output pins. Consequences: A) this will put a DC bias on the headphone/stereo output. So, if amp is on, don’t use headphones. B) this approach consumes more power, due to driving the DAC continuously. C) the playback code should levelshift on the fly, too.
Ultimately, adding the amp input capacitor is the best solution, as far as I can tell.
Thanks Jamie. I second the request for true one-shot behavior. Changing the “edge” behavior not to loop would settle the issue that 518170 and I are having. For my application, the WT is being built into an existing, unmodifiable, sculpture with rocker switches. True one-shot behavior so desirable, I considered using external quad one-shots.
about 3 years ago
Using a plain resistor divider, you lose the temperature compensation which the 2nd device provides “for free”. The temperature effect on “zero” is 1% of full scale per 10 degrees C. I’d recommend testing in a variety of temperatures to characterize its performance.
This might not be important in your application, in which case, you’re right.
about 3 years ago
According to the photos and the datasheet, it’s definitely quadrature: two square waves, 90 degrees out of phase. These two signals give direction of rotation, as well as the usual speed.
It’s a kind of gray code, since only one bit changes at a time, but because there are only two bits it only counts 0, 1, 3, 2, 200 times per revolution.
No public wish lists :(
Forgot your password?
No account? Register one!