On Sun, 1 Jun 1997, Neil A. Carson wrote:

> I don't know about the HENSA driver, but I have one of the devices
> (low-rate though they were) buried here. Basically AFAIR you waggle one
> pin up and down in software (D1 or D2?), which then causes the data to
> be sent in bit by bit on

Yup, that's exactly it.  There are quite a few of these serial ADC chips
around, and you can make an entire sampler with basically just one of them
and an amplifier (which is how the Oak Recorder manages to fit in a dongle
and run off the power from the parallel port).

> the busy line. The key to this is very accurate timing, which might be
> tough on a preemptive system like NetBSD? (not really knowing)? A number of

I think it would need a kernel-mode driver.  Since the only thing that
regulates the sample rate is the speed at which you clock the ADC, and you
have to do that in software, user-space (even at realtime priority) would
probably be too unpredictable.  Also of course this is quite a CPU hog,
because you have to spin in a very tight loop pulsing the data lines and
fielding the bits as they come back.  The Oak software used to turn off
all interrupts and the video DMA to get the speed it needed, though this
was on A400 vintage hardware. 

p.