HP-71B 4Kb (CMT32k FrontPort) Memory Module "Battery" - FRAM

[KimH]

As I prepare a module for FRAM71B to be the Port 1 "Take Over" module I had a thought.

While you open the the 4K Module to build in the OD Switch as described in the FRAM Users Manual P.21, could you have the "Donor" module to be modified to also have a built in "Battery" - doubling the functionality.

This would be neat when you move a small LEX or such from one machine to another

SMD Capacitors offer high capacity Caps with 100uF or more in very small packages.

I would solder a SMD Cap 10-100uF across the Red/Black circles in the attached picture.

BUT (this is the question) how long can the module be "Out" while moving it from one Device to another?

Using Q=I*T with Q=C*V and W=0,5*Q*V hence W=0,5*C*V*V I can't solve for T unless I know I (in uA). And probably some other insights like how the discharge curve look like

I can't seem to find the I for the 4k Memory Modules anywhere, I hope 10uA which would give 15-20 Seconds of time.



Didier, thanks for the help showing the PICT, not a link to the PICT! All it takes is to have dl=0 replaced with RAW=1 at the end of the link

Yes, I know the picture is a ROM, but please humor me...

[Hans Brueggemann]

from my experience, 10 µA is absolutely on the safe side, you probably get away with just 1 µA.
so, basically you just adjust the size of the cap to the allowable delta_V for, say, 10 sec. of "shuttling" time. with Vnom = 4.8 V and Vmin = 3.6 V you get (4.8 - 3.6) V x C > 10 x 10 µAs. thus C > 83 µF.

one caveat, but i'm just guessing here: with digital inputs left unterminated during shuttling, chances are that parts of the module's input logic might go into linear operation and thus increase the current drain from the capacitor.
as to what your HP-71B's opinion on the possibility of file chain corruption is... well, you know the answer!

[KimH]

Thank you Hans, i’’l try it and report back

(01-22-2021 05:43 PM)Hans Brueggemann Wrote:  from my experience, 10 µA is absolutely on the safe side, you probably get away with just 1 µA.
so, basically you just adjust the size of the cap to the allowable delta_V for, say, 10 sec. of "shuttling" time. with Vnom = 4.8 V and Vmin = 3.6 V you get (4.8 - 3.6) V x C > 10 x 10 µAs. thus C > 83 µF.

one caveat, but i'm just guessing here: with digital inputs left unterminated during shuttling, chances are that parts of the module's input logic might go into linear operation and thus increase the current drain from the capacitor.
as to what your HP-71B's opinion on the possibility of file chain corruption is... well, you know the answer!

[Dave Frederickson]

(01-22-2021 04:46 PM)KimH Wrote:  As I prepare a module for FRAM71B to be the Port 1 "Take Over" module I had a thought.

Consider a 3-D printed "take over" module.
https://www.shapeways.com/product/2AW7Y2...rt-housing?

[Didier Lachieze]

(01-22-2021 04:46 PM)KimH Wrote:  How the heck can I post a Picture rather than a link, it is possible, I have seen it

Just replace dl=0 by raw=1 at the end of the Dropbox url:

[Garth Wilson]

Note that the piezoelectric effect on MLCCs is huge, and the capacitance drops off rapidly above approximately 20% of the WVDC. The dielectric type has an effect too, Y5V being considerably worse than X7R. A Y5V will lose over 85% of its capacitance at its WVDC, whereas the X7R only loses 70%. There's an article on this in EDN magazine, Apr 12, 2007, starting on page 77. (Would it be a no-no to post it here?) So for a power backup capacitor for the module, you'll want to get the largest capacitor that will fit mechanically, a 25V (5 times 5V) X7R capacitor with the highest capacitance that will fit.

[cruff]

(01-22-2021 08:24 PM)Garth Wilson Wrote:  Note that the piezoelectric effect on MLCCs is huge, and the capacitance drops off rapidly above approximately 20% of the WVDC. The dielectric type has an effect too, Y5V being considerably worse than X7R. A Y5V will lose over 85% of its capacitance at its WVDC, whereas the X7R only loses 70%.

Surely at the working voltage, the capacitors have their rated values, within their stated tolerance?

[Garth Wilson]

(01-23-2021 01:00 AM)cruff Wrote:  

(01-22-2021 08:24 PM)Garth Wilson Wrote:  Note that the piezoelectric effect on MLCCs is huge, and the capacitance drops off rapidly above approximately 20% of the WVDC. The dielectric type has an effect too, Y5V being considerably worse than X7R. A Y5V will lose over 85% of its capacitance at its WVDC, whereas the X7R only loses 70%.

Surely at the working voltage, the capacitors have their rated values, within their stated tolerance?

Nope. Their rated values start falling off at 20% of their rated voltage. This is for MLCCs. It does not apply to other types like aluminum electrolytic. Would it be ok to post scans of the EDN article here?

[cruff]

(01-23-2021 01:10 AM)Garth Wilson Wrote:  Nope. Their rated values start falling off at 20% of their rated voltage. This is for MLCCs. It does not apply to other types like aluminum electrolytic. Would it be ok to post scans of the EDN article here?

Well, hopefully it is well called out in the MLCC data sheets! I'd like to see the scans, I vote yes for posting!

[Sylvain Cote]

(01-23-2021 01:10 AM)Garth Wilson Wrote:  Would it be ok to post scans of the EDN article here?

Google: "EDN magazine, Apr 12, 2007" then follow the link 'EDN Magazine Issue 2007-04-12 - Internet Archive'

[Garth Wilson]

(01-23-2021 01:50 AM)Sylvain Cote Wrote:  

(01-23-2021 01:10 AM)Garth Wilson Wrote:  Would it be ok to post scans of the EDN article here?

Google: "EDN magazine, Apr 12, 2007" then follow the link 'EDN Magazine Issue 2007-04-12 - Internet Archive'

Ok, well, I won't use Google, but duckduckgo.com, and here it is:
https://archive.org/details/EDN20070412/...7/mode/2up
If you don't want to read the whole article, look at least at figure 7.

[KimH]

This is VERY helpful, thanks! Learned something new - again, I was sure I would..

I have bought some 10uF & 22uF of the X5R 25V version (off course the versions I have are 6,3V )

As Dave from the EEVBlog would say - "For those playing along at home" - the way to read the MLCC is:
X = Min temp -55°C
5 = Max Temp 85°C
R = Tolerance +/- 15% (Temperature Driven) - another reason why X5V sucks (+20% to -85%)

Once I have the 25V versions, I'll try both 6,3 and 25V and report here

(01-22-2021 08:24 PM)Garth Wilson Wrote:  Note that the piezoelectric effect on MLCCs is huge, and the capacitance drops off rapidly above approximately 20% of the WVDC. The dielectric type has an effect too, Y5V being considerably worse than X7R. A Y5V will lose over 85% of its capacitance at its WVDC, whereas the X7R only loses 70%. There's an article on this in EDN magazine, Apr 12, 2007, starting on page 77. (Would it be a no-no to post it here?) So for a power backup capacitor for the module, you'll want to get the largest capacitor that will fit mechanically, a 25V (5 times 5V) X7R capacitor with the highest capacitance that will fit.

[KimH]

I got my shipment of 10uF 25v - X5R - SMD 1210 - MLCC. - they are tiny! About 3*2*2mm

So out with the Soldering iron, mounted two short wires on the CAP and placed it in the "slit" which the Module mold has in the handle-part, wires through 1mm drilled holes.

Placed two files in the FREEPORT'ed module, now IRAM - LEXCAT (Basic) and PEEKUTI2 (That's what I call my custom LEX-File) something I could run which used both filetypes to validate that the content remained intact.

Two TITANs next to each other and got out my stopwatch.

First try, moving module from machine 1 to 2, in a few seconds, ran LEXCAT on the PEEKUTI2 everything works.

Then, 30 Seconds, 1 Minute and so on, every time testing that the content was intact, no Memory Lost, obviously that is part of the test.

My last test was 20 minutes, still all good. Amazing!

I am happy

I will add 10 minutes at a time over the course of today and report back when the 4k Module actually fails to retain it's content or lose it's wits in the memory chain.

[KimH]

I had to try...

Took one of my precious 32k CMT Front-Port modules and cracked it open.

CMT actually built in 2 soldering pads, which fits with a SMD CAP - this time a 22uF 25v X5R

The profile of the 1206 I had in this value is a bit too tall, so a smaller would be better to get the lid back on. Only a datasheet can help here. The 1206 is referencing "footprint" only

Again, 20 minutes without being plugged in was no problem at all.