I need some help.

After several years retired and many more years away from cobbling together TTL and field
programmable parts,
I'm trying to get my hands dirty again on a project.

I am trying to design a replacement UART module for a radio designed about 25 years ago.  The
original
module used a Phillips SCN2661A UART.  That one is not manufactured or available anymore.  The
closest
available replacement is the SCN68661 which I can only get in small quantities in the 68661B
version.  "B" is different than "A"
in the Baud Rates provided.

I thought it might be feasible to use a PIC which includes  a UART to build a replacement.  It
looked like I could get everything I needed
into one PIC.  But then I looked at the instruction cycle times of 200 nSec.  My requirement is to
handle Reads and Writes of the CPU to
UART Data and Control Registers in a fixed period of 550nSec.  The PIC just can't do it.

So I would like to place all the registers in a chip or chips outside the PIC.  The CPU Bus would
fill and empty the registers.  The PIC
would asynchronously move the register data to and from the UART.  The only register file chip I
know about is the TTL 'LS170 which is only
4 bits by 4 bytes (and may not be available).  I would have to use eight of them plus some glue
logic and buffering.

So I'm looking for some kind of modern register file or small static RAM which can be addressed and
written at one end and addressed and read
at the other end.  It would need at least 10 Bytes by 8 Bits.  Setup and Hold times need to be on
the order of 100nSec or less.

I appreciate all the time that the experienced contributors to this newsgroup take with answering
requests like mine.

Thank You All,
Ken Fowler
K...@Comcast.net

On Mon, 05 Feb 2007 22:54:12 +0000, Ken Fowler wrote:
...
> After several years retired and many more years away from cobbling
> together TTL and field programmable parts,
> I'm trying to get my hands dirty again on a project.
> 
> I am trying to design a replacement UART module for a radio designed
> about 25 years ago.  The original
> module used a Phillips SCN2661A UART.  That one is not manufactured or
> available anymore.  The closest
> available replacement is the SCN68661 which I can only get in small
> quantities in the 68661B version.  "B" is different than "A"
> in the Baud Rates provided.

What's the problem with this one? The quantities? Available baud rates?
Clocking? 
 
> I thought it might be feasible to use a PIC which includes  a UART to
> build a replacement.  It looked like I could get everything I needed
> into one PIC.  But then I looked at the instruction cycle times of 200
> nSec.  My requirement is to handle Reads and Writes of the CPU to UART
> Data and Control Registers in a fixed period of 550nSec.  The PIC just
> can't do it.

If it's got a UART, it won't be dependent on the uP's clock rate - the
UART does its own clocking, and just presents a byte of character data
when it's "ready".

> So I'm looking for some kind of modern register file or small static RAM
> which can be addressed and written at one end and addressed and read at
> the other end.  It would need at least 10 Bytes by 8 Bits.  Setup and
> Hold times need to be on the order of 100nSec or less.

If you use a proper UART, you should be able to just write a buffer
to do this. Or, if you insist on doing it in hardware (which would be
way overkill) you could google for actual FIFO chips; I haven't checked
lately, so I really don't know if anybody's making them any more, but
I doubt you need this anyway.

Have Fun!
Rich

On 2007-02-05, Ken Fowler <k...@yahoo.com> wrote:
> I need some help.
>
> After several years retired and many more years away from cobbling together TTL and field
> programmable parts,
> I'm trying to get my hands dirty again on a project.
>
> I am trying to design a replacement UART module for a radio designed about 25 years ago.  The
> original
> module used a Phillips SCN2661A UART.  That one is not manufactured or available anymore.  The
> closest
> available replacement is the SCN68661 which I can only get in small quantities in the 68661B
> version.  "B" is different than "A"
> in the Baud Rates provided.
>
> I thought it might be feasible to use a PIC which includes  a UART to build a replacement.  It
> looked like I could get everything I needed
> into one PIC.  But then I looked at the instruction cycle times of 200 nSec.  My requirement is to
> handle Reads and Writes of the CPU to
> UART Data and Control Registers in a fixed period of 550nSec.  The PIC just can't do it.

Try an ATTiny2313, at 20Mhz instruction times are 50nsec, 
it'll be tight, but may be possible.

Otherwise maybe a small fpga.

Bye.
   Jasen

On  6-Feb-2007, jasen <j...@free.net.nz> wrote:

> > I thought it might be feasible to use a PIC which includes  a UART to build a replacement.  It
> > looked like I could get everything I needed
> > into one PIC.  But then I looked at the instruction cycle times of 200 nSec.  My requirement is
> > to
> > handle Reads and Writes of the CPU to
> > UART Data and Control Registers in a fixed period of 550nSec.  The PIC just can't do it.
>
> Try an ATTiny2313, at 20Mhz instruction times are 50nsec,
> it'll be tight, but may be possible.
>
> Otherwise maybe a small fpga.
>
> Bye.
>    Jasen

Thanks Jasen...

With the PIC, I first assumed 50 nSec instruction time,  and I wrote some assembly that would just
barely do the job.  It was about 10 instructions and used indirection.  One of the slowdowns on the
PIC is the bank selecting.  I had to do some careful pipelining.  I'll go look at the ATTiny part.

-ken-