[Freetel-codec2] CHIP

[James Cloos]

Given the $9 cost and its battery and audio capabilities , NextThingCo's
CHIP looks like a useful option for a codec2 transceiver.

From some simple tests, it looks like c2enc and c2dec would each use

about fifteen to twenty percent of the cpu when used real-time.

I'm still wating on batteries, so I don't know how much talk time one
would get per Ah; I should be able to do some tests next week or so.

(I'll have to simulate radio costs by using its 802.11 radio; I don't
have anything else available.)

By default the T/R/R/S socket has video out on the Sleeve, but a simple
hardware mod (scrape a tiny trace and add a solder blob nearby) will
switch the sleeve to Microphone In; doing that would allow the use of
any T/R/R/S headset. Else one can use a USB headset.

The price point seems like a good fit.

Has anyone else done any c2+chip testing?

-JimC

[Bruce Perens]

There is also Raspberry Pi Zero, which is supposed to cost $5.

If you want good battery life you will need to modify David's present
software to be interrupt-driven, to have some sort of hardware squelch that
interrupts when there's a signal to be handled, and to spend most of its
time halted. This is the conventional means of getting battery efficiency.

SM1000 currently busy-loops. This makes it run somewhat hotter and it used
more power than otherwise, but mostly nobody cares. The main economy would
come from hardware squelch.

Post by James Cloos
Given the $9 cost and its battery and audio capabilities , NextThingCo's
CHIP looks like a useful option for a codec2 transceiver.

From some simple tests, it looks like c2enc and c2dec would each use

about fifteen to twenty percent of the cpu when used real-time.
I'm still wating on batteries, so I don't know how much talk time one
would get per Ah; I should be able to do some tests next week or so.
(I'll have to simulate radio costs by using its 802.11 radio; I don't
have anything else available.)
By default the T/R/R/S socket has video out on the Sleeve, but a simple
hardware mod (scrape a tiny trace and add a solder blob nearby) will
switch the sleeve to Microphone In; doing that would allow the use of
any T/R/R/S headset. Else one can use a USB headset.
The price point seems like a good fit.
Has anyone else done any c2+chip testing?
-JimC
--
------------------------------------------------------------------------------
Site24x7 APM Insight: Get Deep Visibility into Application Performance
APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month
Monitor end-to-end web transactions and take corrective actions now
Troubleshoot faster and improve end-user experience. Signup Now!
http://pubads.g.doubleclick.net/gampad/clk?id=267308311&iu=/4140
_______________________________________________
Freetel-codec2 mailing list
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

[Stuart Longland]

Post by Bruce Perens
There is also Raspberry Pi Zero, which is supposed to cost $5.

Downside with the Pi Zero is zero audio hardware. You'd have to
interface an I²S chip yourself.

Post by Bruce Perens
If you want good battery life you will need to modify David's present
software to be interrupt-driven, to have some sort of hardware squelch
that interrupts when there's a signal to be handled, and to spend most
of its time halted. This is the conventional means of getting battery
efficiency.
SM1000 currently busy-loops. This makes it run somewhat hotter and it
used more power than otherwise, but mostly nobody cares. The main
economy would come from hardware squelch.

From what I've seen, the ADC/DAC drivers seem to operate using DMA and
interrupts. That would seem to be most of where the real-time work is
being done.

I'm slowly moving more things to state machines so one could possibly
determine when the CPU can be put to sleep from that. I think
power-saving support is a long way off, and even without that, the
SM1000 seems to sip power.

[David Rowe]

Post by Stuart Longland
I'm slowly moving more things to state machines so one could possibly
determine when the CPU can be put to sleep from that. I think
power-saving support is a long way off, and even without that, the
SM1000 seems to sip power.

Stuart is correct. I calculated 24 hours on the AH capacity of 2 AA
batteries (if you had a suitable DC-DC converter). It uses a trivial
amount of power compared to the HF radios it is connected to.

The "busy loop" observation is misleading. It's "busy" doing DSP 80% of
the time. No power savings there, nor are they required.

- David

[glen english]

I reckon that is pretty good. As David says, not much compared to what
it is connected to.
I imagine as long as it is not a thermal issue, there is no problem.

Any scope to drop the the power supply it runs from? As the core has the
internal series pass 1.2V regulator. These days I run my STM32 micros
from 2.5V or 1.8V. There are a few caveats on flash memory prefetch etc
and FLASH write #bytes wide etc.

A little optimization of the code would probably yield 20-30%.

glen

Post by Stuart Longland
I'm slowly moving more things to state machines so one could possibly
determine when the CPU can be put to sleep from that. I think
power-saving support is a long way off, and even without that, the
SM1000 seems to sip power.