HP42s Memory Upgrade Questions.

[Thomas Chrapkiewicz]

Hello:

I would like to upgrade a couple 42s's to 32k and would like to solicit others experience(s).

Is the 32kx8 SRAM from a dead HP48 or HP38 a satisfactory device or is the current consumption prohibitive?

Are there more modern SRAM ICs available that could substitute for the 62256?

Thanks for your assistance,
TomC
ps: Thanks again to Lyukas fine photo documentation.

[Lyuka]

Hi.

Quote:

---

Is the 32kx8 SRAM from a dead HP48 or HP38 a satisfactory device or is the current consumption prohibitive?

---

It depends, however, I might think, if it's low power SRAM or not won't be a matter so much,

as the total standby current that I have measured was not so small but around 9uA

which is much more than typical standby current of standard 62256 SRAM.

Quote:

---

Are there more modern SRAM ICs available that could substitute for the 62256?

---

There are modern 32Kx8 SRAM as Renesus's M5M5256DFP-70XG or M5M5256DVP-70XG

which have wide operating voltage range of 3V to 5.5V

and very low standby current of 50nA(typ) at Vdd=3.3V,

but they are not supplied in a classical relatively large SOIC28 package.

Regards, Lyuka

p.s. Thanks TomC. Good luck.

# typo collected : 7uA -> 9uA

Edited: 19 Oct 2007, 10:28 p.m. after one or more responses were posted

[Thomas Chrapkiewicz]

Lyuka:

Thank you for your helpful response. I realize that not all SRAM devices have the same 'idle' current drain - especially down in the nA range (7uA to 50nA is over a factor of 100!!!)

I will look into this much more closely.
Regards,
TomC

[Lyuka]

Hi.

"9uA" menssioned above is NOT the standby current of the SRAM but the TOTAL standby current drain from the batteries.

Generally, the typical standby current of the old 6264 SRAM will be in the range of 0.5~2uA.

That of the latest 62256 lowpower SRAM (used at Vdd=5V) will be 0.1~0.3uA.

[Randy]

Quote:

---

Is the 32kx8 SRAM from a dead HP48 or HP38 a satisfactory device or is the current consumption prohibitive?

---

They are CMOS static devices and as such the standby current is, in the scheme of things, insignificant. Active current is not a concern either...

Quote:

---

Are there more modern SRAM ICs available that could substitute for the 62256

---

What's wrong with the "older" devices? IMO, there is absolutely no reason not to use the older Fujitsu/Winbond/Sony/Toshiba 256kb devices found in the 48 and 38G's. I've used them all, along with some Cypress parts I bought early on. I found no difference in battery life regardless of which device was used.

The bigger and more important question is the one you should be asking yourself: Why do I need 32K in a 42S?

The thought of having to enter even 8k's worth of programs is enough to put me off the task. Loosing that much work would be very, very painful.

[Stefan Vorkoetter]

Quote:

---

What's wrong with the "older" devices?

---

Just that you can't get them any more.

Stefan

[Paul Brogger]

Quote:

---

The bigger and more important question is the one you should be asking yourself: Why do I need 32K in a 42S?

---

Wise counsel. Take heed!

The only reason I did it, lo so many years ago, was to see if I could. The answer turned out to be: "yes".

I've never come close to filling any keyboard-entry-only calculator with anything like 8K of meaningful programming. (I did fill my first 33s with I-don't-remember-how-many-thousand <Enter> commands, but just to test its program editor near its memory limit.)

You may have special needs, but do evaluate those carefully. Something like the PC-connectible 50G may be a far better choice, in the long run.

[Vieira, Luiz C. (Brazil)]

Hi, Stefan;

after Paul´s considerations, I'd add that you could try solving bigger matrices. I remember testing some years ago and I got these figures from this thread):

Quote:

---

maximum square matrices for a hacked, 32KRAM HP42S:

SIZE 0000

[62×62 Matrix], 1004 bytes free memory

[44×44 Cpx Matrix], 780 bytes free memory

Maximum size, nothing else in RAM: 3,967!)

---

Keying the matrix elements in is another story... any I/O resource is missed at this time.

What I did not test was what matrix operations could be performed with the remaining RAM as cache memory.

Cheers.

Luiz (Brasil)