Hi Ed,
A RAM chip can be viewed as a linear array of storage cells. The index
(or location) of a specific cell is called the 'address'. The size of
the cell is the word width, which is most commonly a multiple of 8 bits
(one bit is able to hold the information '0' or '1').
In case of the Polysix patch RAM the word width is 4 bits; so at each
location within the RAM array 4bits of information are stored.
When the CPU wants to write (store) something inside the RAM:
1. The CPU applies the address and a "chip select" signal to the
RAM chip.
2. The CPU applies the data word to be stored to the RAM chip
3. The CPU pulses the 'write' line (called "Write Enable" or "Write" or
similar). With the trailing edge of the write pulse the data is
latched (stored) into the RAM chip.
When the CPU wants to read something form the RAM:
1. The CPU applies the address and a "chip select" signal to the
RAM chip.
2. The CPU pulses the 'read' line. With the leading edge of the read
pulse the RAM retrieves the data word and outputs it. With the
trailing edge the CPU adopts the data word.
Usually a scope has two channels, so my description will rely on that.
1. Connect one probe (e.g. CH 2) to the chip select signal. This is
pin 8 of the RAM chip IC32. You migth want to solder a short piece
of wire to this pin (when the unit is switched off of course) to be
able to attach the probe hook. Don't forget to connect the GND lug
to a GND reference point on the circuit board. When I remember
correctly there is a GND test point on KLM-367.
The chip select (CS) signal is (as well the read and write signals)
active low, this means it is at some voltage value (> 3V) for most
of the time. Only when it is active during a read or write cycle it
becomes (nearly) 0V.
2. set the vertical scale to 2V/div
3. Set the scope trigger to CH2, falling edge
4. Press a PROGRAM button on the Polysix and watch the oscilloscope
triggering. You may have to adjust the trigger level to a value in
the signal range (about at one division).
If your scope is of the analog type you just see a short dash of the
beam. Set the trigger to "normal" (opposed to "auto") and press a
PROGRAM button repeatedly. Each time you press the PROGRAM button the
PORGRAMMER CPU goes out to the RAM and reads 2x16 data words from the
patch RAM.
One patch is composed by 16 bytes of data, since the RAM is only
4bits wide it has to access it twice to obtain a full byte.
If you are a lucky owner of a digital storage oscilloscope the
signals are recorded and steadily displayed by the scope, which is
much easier to readout then.
5. Set the horizontal scale to a setting that the CS pulse spans about
the scope screen width.
6. Now line per line probe with the other channel the RAM address lines
(pins 1..7, 16, and 17; refer to KLM-367 schematics in the service
manual).
To be able to see the 2nd trace you have to set the scope to show
both channels, usually this is called "dual" as opposed to "CH1"/
"CH2" or similar in the vertical section.
Set CH1 to 1V/div or 2V/div and separate the two traces by means of
their associated "vertical position" knobs. Leave the trigger source
at CH2.
You should see a clean trace for each address line. It is clean when
it has a stable level during the active CS pulse. Furthermore the
level must be < 0.4V indicating "low" or 0 or > 2.4V, indicating
"high" or 1.
If you don't get clean signals then something in the circuitry around
the RAM (e.g. address latches IC26, IC34; or a PCB trace) could be
faulty. If you get clean signals on all address lines most probably the
RAM chip itself is the reason for the fault.
Additionally (or even before) you might want to check whether the
address lines do not have shorts to one another (which again may be
caused by faulty circuitry, PCB, or the RAM chip).
Be careful with the probes not to make any shorts between neigbouring
pins of ICs...
I hope this helps!
Johannes
On 25.11.2013 11:10, djangowes@yahoo.fr wrote:
>
>
> Hello Johannes,
>
>
> thank you.
>
> I've tested continuity between IC32 pin 15 and IC 22 pin 22 and it's OK.
>
> I've played quite a lot the synth (what a sound!) last night to check if
> other parameters don't work but the synth is fine and all controls
> respond. Saving a patch takes into account all parameters.
>
>
> About the RAM accesses, I think I need a few informations. I have a
> scope that I use to calibrate delays or flangers that i build. Checking
> RAM is another story.
>
> So in a few words, how am I going to test RAM using my scope? Which
> signal do I have to expect?
>
> sorry for these questions...
>
>
> thanks!
>
>
> Ed.
>
>
>
> ---In PolySix@yahoogroups.com, <johau@...> wrote:
>
> Apparently the addressing of the patch RAM has a fault. This could be
> a long term subsequent fault from the leaked battery.
>
> There could be a number of reasons causing wrong adressing; the RAM
> chip itself is a candidate, I had a few chips failing.
> The high address line is an I/O port line of the 8048 MPU, which is
> also used as a selector for the analog control voltages (P21). So if
> the analog controls work this signal is OK. It could be interrupted
> (e.g. by corrosion). You can check this with a DVM:
> IC32 (RAM chip) pin 15 to IC 22 (CPU) pin 22.
>
> With an oscilloscope you could easily check RAM accesses, trigger on
> the chip select (pin 8) and check one (address) line after the other
> while writing or reading to / from the RAM (e.g. recalling a patch).
>
> Good luck!
>
> Johannes
>
>
> On 23.11.2013 14:57, djangowes@... <mailto:djangowes@...> wrote:
> >
> >
> > Hello,
> >
> >
> > forgot to add some things : my P6 had the battery issue. It has been
> > repaired before I've had it, 4 years ago. I've changed all
> electrolytics
> > in the power supply section. It has worked perfectly during these
> 4 years.
> >
> > I've tried to save a new preset, bank A, preset 5. The P6 saved it but
> > it saved it also on bank A,preset 1. So it seems that half of the
> memory
> > does not work. Could it come from the battery, or one of the memory
> > component has failed (where to check?) ? I'm a bit confused as I don't
> > know how the memory saving process works. If someone has some
> knowledge
> > about this, that would be really helpfull.
> >
> >
> > thanks,
> >
> >
> > Ed.
> >
> >
> >
> > ---In PolySix@yahoogroups.com <mailto:PolySix@yahoogroups.com>,
> <djangowes@...> wrote:
> >
> > Hello,
> >
> > Today my P6 has begun to behave strangely : considering each bank
> > A,B,C,D the first 4 presets (1,2,3,4) have been "duplicated" to the 4
> > last presets locations (5,6,7,8).
> > Otherwise, the synth seem to work fine.
> > Maybe someone had this issue in the past... or maybe some
> P6-specialists
> > in here can help me and tell me where to check in my favorite
> > synthesizer. KLM367 maybe, where the "brain" is...
> >
> > thanks for any kind of help.
> >
> > Ed.
> >
> >
> >
>
>
>
>