From: Wolfgang Moser (womo_at_news.trikaliotis.net)
Date: 2008-11-09 10:01:21
Hi Jim,
Jim Brain schrieb:
>>
>> I told him about the construction of the XA1541 cable and
>> the long history about finding and measureing the right
>> transistor types to serve the IEC bus in the best way.
> It'd be nice to see a condensed version of this.
hmmmm, currently I'm working on my REU compatibility
document again -- writing test programs to find out
about crude hardware implementation details.
Please have a look into my photo gallery, I opened
a 30 day guest access to the XA1541 transistor
measurments section (the photo gallery service
doesn't allow for unlimited guest access):
http://sd.1und1.de/mc/CCzixING6dtrkCN6eBVWbR9nmEbBtT
For example check out picture:
XA-Tr_IEC-Clk_attack_90Ohm-11Drv_2x.png
There you can see the falling times of different
transistors and also two diode types with a simulated
load of eleven IEC bus devices (90 Ohms pullup load).
Nearly any driver solution fails in this setup.
The 1N4001 diode is very slow speaking in reaction
time and, more notable, it is not able to fall below
a voltage of 1.8V. Even in setups with not such a high
load, silicon diodes like the 1N4001 are not able to
fall below the 0.8V...1.2V voltage level when driven
_behind_ a LPT port. This is not low enough to trigger
the 74LS14/74LS04 input drivers of the 1541.
Compare the picture above with:
XA-Tr_IEC-Clk_attack_3drives_2x.png
There you can see that the 1N5819 Schottky diode is
the really fastest driver method, but it also has got
a voltage offset. Luckily this is near enough the 0.8V
trigger level to successfully drive the 1541 input
circuits.
Any other transistor based solution showed in this
picture always creates a low voltage level that is low
enough. But, the BC547 and the 2N3903 are not fast
enough, this can be seen, when looking onto the rising
times:
XA-Tr_IEC-Clk_release_3drives_2x.png
Within this picture, the BC547 didn't even start to
react! The BSV52 was "our" (The Star Commander beta
test team) favorite driver solution for the XA1541 for
several years along with the replacement types BSX20
and PH2N2369A.
Later Klaus Stock from the german Forum-64 proposed the
BS170 MOSFET transistor and this one revealed as a true
champion. Fast on one side and rather strong, when
getting the 11-drive chart again:
XA-Tr_IEC-Clk_attack_90Ohm-11Drv_2x.png
Although I didn't make new measurements, I suspect that
the MOSFETs 2N7002 and 2N7000 have got the very same
characteristics, because their datasheets look very
similar and because it is the same creation process
(vertical D-MOS MOSFET).
Other MOSFET technologies like the HEXFETs may even be
more promising.
> The next board revision of uIEC will use inverting open collector
> buffers, like the rest of the drives. MOSFETs and such are nice, but
> it's harder to stuff a board with them (as I noted to Shadowolf) and a
> buffer IC will accomplish the same thing.
In a board solution that offers a power source, true
TTL or HCT buffers/drivers are the very, very best
option. I think so, too.
To make your design work the best, you should also
spend some 74LS14 / 74HCT14 input drivers with
integrated Schmitt triggers, this is what all the
cheap cable solutions are missing.
The ATmega controllers don't offer input pins with
Schmitt trigger funtion, do they?
Womo
Message was sent through the cbm-hackers mailing list
Archive generated by hypermail pre-2.1.8.