New 12 April 2000

Z88 Internal Memory Modifications and Power Consumption

I had for some time been interested in replacing the original internal RAM chip with a new one larger and/or with less power consumption. When I got hold of a 128K static RAM chip I decided to go ahead. Not being confident at unsoldering multi pin chips I simply cut off all the legs of the chip as close to the chip as possibly. I could then carefully unsolder the legs one by one. Apart from the unsoldering technique I followed Don Waltermans instructions at Z88 Forever!

I wanted to know what the power consumption was before doing the modification in order to have something to compare with afterwards.

Some tests performed by Vic Gerhardi. The figures are not easily comparable to other measurements because they are taken when running off batteries. The current is dependant on the voltage. Lower voltage is compensated by higher current and the other way around. Note one interesting point. There are two types of 128K RAM cards of which one takes more current. It is described in the service manual also at Z88 Forever!

Installed memory                 Power consumption in mA on batteries
SLOT 0                 SLOT 1    OFF   ON    Running BBC BASIC
32K RAM, 128K (EP)ROM            0.40  88.5   97.0
32K RAM, 128K (EP)ROM  128K RAM  0.85  97.6  105.0
32K RAM, 128K (EP)ROM  128K RAM  0.65  75.0
32K RAM, 128K (EP)ROM  512K RAM  1.20  97.0  140.0
32K RAM, 128K (EP)ROM  1M RAM    0.75  95.5  135.0

Matthew Soar made tests with an external power supply at a defined voltage. This test illustrates another interesting point. RAM cards placed in slot 3 takes more power.

Installed memory                         Power consumption in mA at 6.5V external power
SLOT 0                SLOT 1   SLOT 3    OFF  ON    Fast   Running BBC BASIC
                                                    Typing 10 FOR I=1 TO 1E4 
                                                           20 A=1 
                                                           30 NEXT I   
32K RAM, 128K (EP)ROM                    0.34  64.2  70.0  86.5
32K RAM, 128K (EP)ROM 128K RAM           0.82  70.8  80.0  95.0 
32K RAM, 128K (EP)ROM          128K RAM  0.72 134.5 155.0 178.0

I decided to base my tests on Matthew Soar's methods. As I went along fast typing was excluded because it is not so easy to type in a standardised way to make measurements comparable.

First two lines are the original chip readings. Then with a new RAM chip. For some reason I changed the contrast of the screen and noticed a big difference in power consumption that's why there are readings for lightest, darkest and normal screen. Even using darkest or lightest screen does not make measurements comparable because how dark or light the screen can get is adjustable by some trimmer potentiometer inside the Z88 (Jack Lawrie's tip). Adjusting to "normal" a couple of times got readings within 1 or 2 mA. As there are two different types of LCD screens used in the Z88 that could also make a difference regarding power consumption. The older ones like mine have an Epson EG2001S-AS-1 and the newer an Optrex DMF690N.

Even what is on the screen makes a difference. Being in Index with the dark window banners takes more current then being in BASIC with no program running and relatively few characters and dark pixels on screen.

Installed memory                                                         Power consumption in mA at 6.50V external power
SLOT 0 RAM                     SLOT 0 EPROM          SLOT1  SCREEN       OFF  ON   Running BBC BASIC
                                                      RAM   Epson                  10 FOR I=1 to 1E4
                                                            EG2001S-AS-1           20 A=1
                                                                                   30 NEXT I
32K  NEC        D42832C-12L    128K NEC D27C1000D-20        Normal       0.36 64.7 94.7 
32K  NEC        D42832C-12L    128K NEC D27C1000D-20  512K  Normal       0.71 71.1 105.4

128K Mitsubishi M5M51008AP-70  128K NEC D27C1000D-20        Lightest          52.8
128K Mitsubishi M5M51008AP-70  128K NEC D27C1000D-20        Darkest           58.4
128K Mitsubishi M5M51008AP-70  128K NEC D27C1000D-20        Normal       0.24 53.3 77.3 
128K Mitsubishi M5M51008AP-70  128K NEC D27C1000D-20  512K  Lightest          58.3 85.6
128K Mitsubishi M5M51008AP-70  128K NEC D27C1000D-20  512K  Darkest           63.9 91.4
128K Mitsubishi M5M51008AP-70  128K NEC D27C1000D-20  512K  Normal       0.57 58.9 86.2 

32K  Toshiba    TC55257PL-12   128K NEC D27C1000D-20        Normal,Index 0.26 51.4 75.5 
32K  Sony       CXK58257AP-10L 128K NEC D27C1000D-20        Normal,Index 0.26 49.8 72.6 
32K  Hyundai    HY62256ALP-10  128K NEC D27C1000D-20        Normal,Index 0.26 46.2 68.3 
32K  Hyundai    HY62256ALP-70  128K NEC D27C1000D-20        Normal,Index 0.24 56.9 80.9 
32K  Hitachi    HM62256LP-10   128K NEC D27C1000D-20        Normal,Index 0.29 52.7 77.0 
32K  Samsung    KM62256BLP-10  128K NEC D27C1000D-20        Normal,Index 0.33 50.6 73.4 
32K  NEC        D43256C-12L    128K NEC D27C1000D-20        Normal,Index 0.33 54.0 77.7 

32K  Hyundai    HY62256ALP-10  128K NEC D27C1000D-20        Normal,Index 0.25 46.4 68.6 

32K  Hyundai    HY62256ALP-10  128K ST  M27C1001,1993       Normal,Index 0.24 45.6 66.2 
32K  Hyundai    HY62256ALP-10  128K ST  M27C1001,1995       Normal,Index 0.27 44.4 63.3 

32K  Hyundai    HY62256ALP-10  512K TI  27C040-10           Normal,Index 0.25 47.1 69.0 
32K  Hyundai    HY62256ALP-10  512K AMD 27C040-12           Normal,Index 0.26 46.1 67.0 
32K  Hyundai    HY62256ALP-10  512K ST  M27C4001-12,1996    Normal,Index 0.27 44.8 64.5 
32K  Hyundai    HY62256ALP-10  512K ST  M27C4001-10,1996    Normal,Index 0.26 45.1 64.8 
32K  Hyundai    HY62256ALP-10  512K ST  M27C4001-12,1998    Normal,Index 0.28 45.0 64.4 

32K  Hyundai    HY62256ALP-10  128K NEC D27C1000D-20        Normal,Index 0.24 46.6 68.6

Reading data sheets to compare chip power consumption proved difficult as some stated mA or mW without voltage and some mA or mW per MHz and some without. Remained to do some tests. A week after installing the new RAM I had gathered a number of different 32K RAM chips, 128 and 512K EPROMS and prepaired a JEDEC to ROM pinout socket converter for the EPROMS.

The OFF current was most difficult to measure as it jumped up and down so an average function had to be used to get a stable reading. Even so it showed a slow drift of maybe 0.02 or 0.03 mA in half a minute but not every time. Also the ON figures had a similar drift. The running BASIC figures seems fairly stable.

Conclusions

Changing the original internal pseudostatic 32K RAM to a static 128K RAM gave a larger capacity AND lower power consumption.

RAM chips are more different than EPROMs.

To make more diferentiated measurements, a special test program is required that could execute loops in only RAM to test RAM in active state, long sequences or eternal loops in only EPROM without RAM access to test EPROM active state and turning the screen on and off to see the screen contribution.

Back to Cambridge Computer Z88