8250 vs 16C450 UARTs

181 views
Skip to first unread message

Lee Hart

unread,
Nov 23, 2020, 2:14:08 AM11/23/20
to se...@googlegroups.com
Hi all,

Has anyone tried a 16C450 or 16C550 in their H89's serial ports?

The H89 uses the National INS8250N (or N-B) UARTs. Some years ago, I bought some TI TL16C450N replacements to try. This chip is CMOS, and so uses less power. It also supports higher baud rates and has other features, but the stock software does not use any of them. The chips I bought worked fine in my H89; no difference from the 8250.

I used those chips up on my Z80-SIO kits, and bought another batch of them. But they don't work, either in the H89 or my Z80-SIO. Nuts; I figured it's another bad ebay buy.

But I just tried one in the serial port of my PC (replacing the stock 16C550). It worked! So I'm wondering... are these actually 16C550's mis-marked as 16C450's?

The 16C450 should be a drop-in replacement for the 8250... or is it? Is there some compatibility difference that prevents them from working in the H89?

Lee Hart

Terry Gulczynski

unread,
Nov 23, 2020, 8:28:58 AM11/23/20
to se...@googlegroups.com
Lee,

I have a few H89s with 16C550 UARTs installed on the Terminal port. I
know that the H89A model will handle the '550 without trouble. It's the
only way I could get the system clock over 8MHz and up to 12MHz in a
couple of cases. The 8250 seems to top out around 8MHz in an H89.

I also have an old style H19 upgraded to H88 and further to full H89
(non 'A' model) but I'm not certain if it has the '550 or the stock
8250. If you need to know, I can dig it out and open it up to see.

Note that, if you're referring to the 3-Port Serial adapter, I dunno -
haven't tried it.

Re: the '450 and '550 differences... At power up, they operate in an
identical fashion.


Terry
> --
> You received this message because you are subscribed to the Google
> Groups "SEBHC" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to sebhc+un...@googlegroups.com
> <mailto:sebhc+un...@googlegroups.com>.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/sebhc/CAF__8hdAMp2R3gnJPZNv56owKU8B%2B5kbCtOWbou8WaE22bbQew%40mail.gmail.com
> <https://groups.google.com/d/msgid/sebhc/CAF__8hdAMp2R3gnJPZNv56owKU8B%2B5kbCtOWbou8WaE22bbQew%40mail.gmail.com?utm_medium=email&utm_source=footer>.

Norberto Collado

unread,
Nov 23, 2020, 5:23:48 PM11/23/20
to se...@googlegroups.com
I see that we can buy the 16C550 in Mouse Electronics for about $5.00 in PLCC44 package.


Norberto
To unsubscribe from this group and stop receiving emails from it, send an email to sebhc+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/sebhc/59d38614-8326-9199-a15b-2cf41c9c5850%40cfl.rr.com.

Lee Hart

unread,
Nov 26, 2020, 2:42:31 PM11/26/20
to se...@googlegroups.com
The 8250 was used by the millions in PCs, so there are *lots* of clones. Many are still available new (though the surface-mount versions won't help much for our H89s or my Z80MCs).

I went "down the rabbit hole" looking at the data sheets for the various flavors (8250, -A, -B; 82C50, 16C450, 16C550, etc.) Newer versions are significantly faster (going from 2000nsec to 250nsec for a read or write cycle). It appears the 8250-B came out before the -A. Some manufacturer's -A versions are in fact CMOS, and significantly faster (like the 16C450).

The 16C550's have internal FIFOs, and there are pinout and functional differences. Since it sounds like they work in the H89, I guess Heath didn't use any of the pins that were changed, and Heath software doesn't "trip over" the new modes.

I found and ordered some 8250A's on ebay for my Z80MC. Will let you know if they work.

Lee

Lee Hart

unread,
Dec 4, 2020, 10:04:19 PM12/4/20
to se...@googlegroups.com
Thanks, Terry. Since the 16C550 worked in your H89's, I assume the simpler 16C450 should work the same. I ordered some 16C450's from a different vendor, as well as some NS8250A's from another.

The good news is that they *all* worked in my H89. So, I obviously got a bad batch of 16C450's. I don't know what they are good for except 40-pin thumbtacks.

I've also been reviewing datasheets. It appears the original National and Western Digital 8250 was slow (2 usec cycle time). It was upgraded to the 8250-B (with no A version in between). The -B fixed some minor internal bugs, and was a little faster (like 1.8 usec cycle time). Both of these are likely to be iffy at 4 MHz and up.

Then much later,  National introduced the 8250A. The -A was much faster (500 nsec). The 16C450 was basically the same thing, but CMOS.

G. Beat

unread,
Dec 12, 2020, 10:13:52 AM12/12/20
to SEBHC

Until Midnight Sunday, December 13,
Electronic Goldmine (Chaney Electronics) in Scottsdale, AZ is selling 
National NS8250N-B Universal Asynchronous Receiver/Transmitter for $2.49

National NS8250N-B UART in a 40 pin DIP package has independent receiver clock input. 
Features complete status report capabilities, line break generation and detection

Reply all
Reply to author
Forward
0 new messages