I am sure, it fails in DOS, Linux. windoz and ASM 386 machine code. ( its a hardware thingy)
THANKS FOR THAT ! MUT, yes, but my point was why 15,625? (Ans: 1E6 div 64 ! 6850)
Chips !: (those little details in electronics) UART's (serial ports to most or COM1)
Legacy:
that's easy,the 16550a (have datasheets , all versions 8250 (xt),16450, on up)
I have the full spec and schematics for all IBM PC's, from which all were cloned.
Lots of Legacy data I have.....
this problem, is the frequency of the master clock used for the PC.
1.8432mHZ. (IBM PC spec, for XT , AT and PS2) ( and its errors !)
the 16 bit DLM+DLL register sets the output rate. clock div by 16 div by DLM+DLL.
the correct load in to that reg for 9600, is 0012 (max speed on an IBM-XT, oops)
the 19.2kb is code 6 . (Msb = 0) LSB = 6.
others:
code , rate:
---------------
12 = 9600
11= 10473
10-= 11520
9= 12800
8= 14440
7= 16457
6 19.2k
as you see?, you will need to load a number between 8 and 7 and that will never happen.
not on legacy or on any clones ( heck the new PCs have dropped the port)
Realterm (tm) is cool, it writes it out and reads it back if it can come back , error.
Plug in new PCCard(pcmcia) and bingo , solved !
The old UART's could clock at 8mhz but IBM choice was 1.8mhz. (they had reasons)
Joe Campbell has a great book on that.
I worked at INtel when IBM was buying all our chips and even the broken ones, 8088.
500,000 a year was really over the top back then,.
NEW CARDS: (Uarts)
how ever, with a ADD-in CARDS, rS232, MANY have a very fast clock, (MOST)
The Dell is spec'd at 19.2k baud max , and if you try to run it above that, like
115k (code =1) baud like our MegaSQUIRT fans did , they found out this doesnt work.
(some do , many dont) they sell a USB Uart and that works perfect.
the data sheet for the 16C550a chip says , at 38400 (code 3)is max speed ! , the error will be beyond the EIA values of -2.5 and + 4%. ( says it plan as day)
the error is in the UArt and the crystal, as most xtls are very inaccurate and dirt cheap.
and veries from one clone mfg to the next, (sadly)
the add in cards, usb, or pccards , have a dual range UART, (cool) and allow speeds up to 1.5mbps and beyond, and accurately (what a concept, accuracy).
IBM:
the error at 56000 baud is 2.86% (ns16550d) and states that the next two higher are illegal. ( the dashes in the table ,mean , dont do that)
that considers an almost perfect crystal. (actually, so is 56k if it falls below -2.5%)
I say this ALL , because my last post , i got tons of email from developers wanting to know what I know about legacy PC's. (RS232 is not dead, just in the office !)
Many get confused by that,because to PC's running Laplink(tm) at 115.2k baud work great that is because they are both in error the same amount (2 wrongs can make a right) the "shift register" in side is happy. ( the net result is no framing errors)
yes, thAT windows software is nice, I have Visual C++ too, and can code like crazy.
But , I diverged:
the only issues (secrets) are filling out the address map,
function versas , index pointer code (3rd byte of 4)
once the this is done , the software is , really, a piece of cake.
refr:
15mpbs rates , no problem (ADD IN) even faster and fully downward compatible.
rs-232 pc-card pcmcia, Computers Networking, wii items on eBay.com
$15 and will do magic.
When i get the map done, i will give away the whole store, all for free.
Demand is low.
But those wanting to fix their cars , and want a better analytical tool, it is comming.
all you will need is 3 pieces of hardware.
cable, adaptor, and the above mentioned PCcard so that 15625 baud will be dead nutz right on.
best advice for anyone is to simply plug in a quality modern Addin RS232 card , pccard.
that legacy junk, is just that.
cheers.
PS:
one more issue, that is the PC talking to any modern equipment at high rates of speed.
the new equipment will be dead accurate but the PC will be way off in frequency. This is why people get so surprised.
Any thing over 38.4k can be an issue.
Best is to measure the error, or swap in quaility hardware.
