Re: Commodore joystick ports

From: john/lori (
Date: 2007-04-11 08:40:44

Scott McDonnell wrote:

I hope you won't mind if I offer some comments.

>>>At 3mA, the 1nF cap would charge in one clock cycle, at 10uA it
>>>would take 256 clock cycles.
>>>(5V would equal 1 and 200mV would equal 255)

The PRG says use a max RC of 4.7E-4 or 470K with the built in 1nF,
assuming that's for a count of 255, that works out to 1843 ohms for
a count of 1 or about 2.7ma with 5V across it so your 3ma sounds in
the ball park.

Jim Brain says 2-5us to discharge the cap which suggests ~2K on
resistance for an RC of 2us (I don't know the exact conditions though) 
but that means the cap will never discharge because that puts the 
voltage at 2.6V  (1.8K pulling up, 2K pulling down)

In times past John Iannetta reported a short circuit sink current of
45ma on the CIA IO pins that'd work out to 111 ohms which would let
the cap discharge to .28V with 1.8K pulling up.

Jeri told me what she knew about it years ago, but I'm not sure I
remember correctly (ie that there was nothing special about the
discharge transistor)
Perhaps she'd care to comment.

You're schematic is not what I was envisioning from your description.

Your bias network (330k in series with 56K) gives you .72V at the base.
But if the input is at 0V it's 330k in series with 35.9K or .49V at the
base, not enough to turn the transistor on.
Elsewhere you said Hfe of 300 typical for the 2N3904, but the
Fairchild datasheet I have says 300 max. Using 300, the 270 ohm
emitter resistor will look like 270 x 300 or 81K so the base current
will be a significant fraction of the current in the bias network.
(it'll be worse 'cause the Hfe will be something less is my guess)

>>So why don't you have the 1nF cap in your simulation?
> Because it wasn't necessary and the refresh rate of the DMMS would 
> not be able to track the current visually.
>>What's the simulation do when it is there?
> The global climate changes and cats begin their take over of Earth! :)
> J/K... it charges the cap in a given amount of time based on the current
> supplied and as determined by the RC formula.

As the cap charges, raising the emitter voltage, the base will lose

> If I add in the cap, an o'scope, and a switch to discharge the cap, 
> you can watch the charge time change depending on the current as given 
> by the RC formula. No magic. The DMMs just show nothing useful because 
> they are not fast enough.
> The charge rate of a cap is predictable given a certain current. So all 
> that was required in the simulation was to show the current ranges. The 
> rest should be predictable based on established science (to some
> degree.)

You're not providing a certain current though, it'll change as the cap

I'll stop there.

Don't give up.

> The 270 ohms was added since it is in series with the cap and I wanted
> to know the exact current that the cap would see. It doesn't make too 
> much difference if I remove it.
> I was still going on the 2.5V charge threshold. I need to adjust it for 
> the 2V that Jim determined through experimentation. It might not make
> much
> of a difference, though.
> Scott McDonnell
>        Message was sent through the cbm-hackers mailing list

       Message was sent through the cbm-hackers mailing list

Archive generated by hypermail pre-2.1.8.