magnetic memory

Hello to all.
can you help me on the following problem?.
I would like to understand if it is possible to create a very simple memory using a core inductor and a Hall sensor in on-off mode or in high-low state mode (it doesn't have to be linear mode). there are lots of hall effect sensors, maybe you already know which type is right to use.
I have read that non-polarized bistable relays use a core as follows:
nickel, aluminum, titanium or niobium
(55 -85% Co, 10-12% Ni).
when no more current flows in the coil in the core a magnetic field remains strong enough to keep the relay contacts attracted.
I wonder if there are core inductors that use the same material or if there are inductors that do not use this material but that maintain a sufficiently strong residual magnetic field even when no current is flowing in the coil.
naturally in the second case if the magnetic field is less intense then sufficiently sensitive hall effect sensors are needed.

LOL... Have we come so far.... All we used back then was magnetic memory...

In your case... keeping an electro magnetic switch open / closed whist power down would b a feat in itself.
Thinking purely on solenoids, they reverse using springs.. I suppose a mechanical "latch" to hold it in place, whilst unpowered, could be done..

A lot of faff though.. Remind me why you cant use a simple 8 pin micro??

I believe that part of the operation of a remanence relay relies on the fact that once the armature pulls in, there is a minimal air gap in the magnetic circuit, as the moving armature closes the gap when activated.

A hall sensor would really have to be in a gap in a core to sense the field?

The last remanence relays I saw in use were GPO 3000 series, like this:



I believe the large metal section at the end of the coil is the remanence alloy slug.

Original "core memory" used in early computers relied on remanence, but that had to sense data by switching all elements to a fixed state and sensing a pulse from any that were not already in that state (called destructive read); any read cycle had to immediately write the data back to prevent it being lost.

If you have never seen that, these are photos of one I have; 32K x 16 bit, I believe, so 512,768 individual ferrite cores in total...

Magnetically biassed bistable types are far more common and more reliable; you may be able to sense one of those by putting a hall sensor against where an armature gap may be when the armature is pulled to the opposite position, as there will be a field from the bias magnet across the open gap.

I still think you are far better off using an off-the-shelf bistable relay, with one contact for sense and another contact operating a slave conventional relay to do the external power switching of whatever it is..

arivel said:

Hello to all.
can you help me on the following problem?.
I would like to understand if it is possible to create a very simple memory using a core inductor and a Hall sensor in on-off mode or in high-low state mode (it doesn't have to be linear mode). there are lots of hall effect sensors, maybe you already know which type is right to use.
I have read that non-polarized bistable relays use a core as follows:
nickel, aluminum, titanium or niobium
(55 -85% Co, 10-12% Ni).
when no more current flows in the coil in the core a magnetic field remains strong enough to keep the relay contacts attracted.
I wonder if there are core inductors that use the same material or if there are inductors that do not use this material but that maintain a sufficiently strong residual magnetic field even when no current is flowing in the coil.
naturally in the second case if the magnetic field is less intense then sufficiently sensitive hall effect sensors are needed.

Click to expand...

yes, it is possible but highly impractical because many lower power, more reliable, easier to use, more widely available options exist with clear technical support and datasheets.

Sorry for the thread drift...
but your pictures of the Minic computer make me go all nostalgic for the PDP11 series.

Front panels with lamps and switches, you could see what the thing was doing (or not doing).
Numbers in octal (base 8).
2.5 megabyte disc drives the size of a filing cabinet drawer, and twice as heavy.

JimB

arivel said:

if it is possible to create a very simple memory using a core inductor

Click to expand...

Yes.



ak

JimB said:

your pictures of the Minic computer make me go all nostalgic for the PDP11 series.

Click to expand...

>OT<
Yes, I owned a PDP-8E and an PDP-11/03 many years ago, but had to let them go when I first moved away from my parents & in to a bedsit.

No hard drive, just an ASR-33 & paper tape with the PDP-8 and a dual 8" floppy unit with the 11..

(The MINIC panel is a spare - a customer has some old NC machines that run on MINIC systems with extra boards, but they never use the operator panels, they were just dumped on a shelf with other spares; I swapped one for something more useful to them).

>/OT<

thanks for the replies.
I believe there is a misunderstanding.
i don't want to use relays but passive fixed inductors.
I think I have solved the problem. just search for inductors with
a high value of BR (H).
then I have to find the suitable hall effect sensor.

Ian Rogers said:

LOL... Have we come so far.... All we used back then was magnetic memory...

In your case... keeping an electro magnetic switch open / closed whist power down would b a feat in itself.
Thinking purely on solenoids, they reverse using springs.. I suppose a mechanical "latch" to hold it in place, whilst unpowered, could be done..

A lot of faff though.. Remind me why you cant use a simple 8 pin micro??

Click to expand...

it's just a matter of principle.

gophert said:

yes, it is possible but highly impractical because many lower power, more reliable, easier to use, more widely available options exist with clear technical support and datasheets.

Click to expand...

can you give me some examples in which there is no programming to do?

arivel said:

can you give me some examples in which there is no programming to do?

Click to expand...

This is an example parallel access FRAM; a Ferroelectric memory device.

Set all the address lines to a fixed level such as all low (0V) and data can be stored or read back on the DQ pins, without a CPU or any programming.
Pulse CE and WE low to store data, or pulse (or set) CE and OE low to enable the stored data to be output.

You could add a switch to control one or more address inputs to allow different memory locations to be used.

The original magnetic core memories did not use hall sensors. They relied on the fact that if they were magnetized on one direction (Let's call it +) and you passed a current through the write wire in the direction that would magnetized in the same direction there was no significant change of flux so there was no output pulse to the read wire. But if you passed a current pulse through the write wire that CHANGED the direction of magnetization there was a change of flux in the core so there was a voltage pulse seen on the read wire. To write to a core you just pulsed current in the + or - direction (To write a zero or one to the core.)
Reading was more complicated. You had to pulse the write wire in the direction that would write a zero. If you got a pulse on the read wire you knew that the core contained a one state. But you had now changed it to a zero so you had to re write it back to a one state. In practice there was two write wires though each core and the cores were arranged as an X / Y matrix. One write wire was through cores on the X axis and the other wire through the Y axis. The current through each write wire was only half the current required to flip the magnetic state of a core. This was used to simplify addressing groups of cores.

Les.

arivel said:

I believe there is a misunderstanding.
i don't want to use relays but passive fixed inductors.

Click to expand...

I think we understand what you are trying to do.
What we don't understand is why on earth you want to do it like that?

JimB

rjenkinsgb said:

This is an example parallel access FRAM; a Ferroelectric memory device.

https://www.tme.eu/Document/d198f7d7f7707d6c6f99533a9aa41126/FM22L16-DTE.pdf

Set all the address lines to a fixed level such as all low (0V) and data can be stored or read back on the DQ pins, without a CPU or any programming.
Pulse CE and WE low to store data, or pulse (or set) CE and OE low to enable the stored data to be output.

You could add a switch to control one or more address inputs to allow different memory locations to be used.

Click to expand...

good .
I have to go and check the compatibility between this memory and the other integrated circuits.
I have to memorize 6 bits that come from two binary counters (cd40193) in cascade.
something is needed that senses the change of state of the 40193 outputs and gives a pulse to the memory which has to write the data. how would you do it?

Why do you need to memorize the cd40193 output? Is the device powering down? Is it battery powered? Could you use a small backup battery for memory on a latch that stores the state of the 6-bits?

yes the device turns off.
I don't like the backup battery

no idea ?

And

arivel said:

no idea ?

Click to expand...

"If it was easy as fishin' you could be a musician..."

if it was easy, everyone would be doing it. Try it your way listed above.
Alternatively, a mechanically latching relay can preserve a state for each of your 6-bits without power. Tie one end of the relay's switched terminals to ground (through a resistor), connect the other switched terminal to the power supply rail. The output will be logic low when the relay switch is open, high when closed. A short pulse on the relay coil should toggle the mechanical switch on the relay.

As you seem to be rejecting all the suggestions for solutions it looks like you will only be satisfied with your hall effect sensor idea. It will be interesting to see the results of your experiments.

Les.

If "No idea ?" relates to this -

arivel said:

something is needed that senses the change of state of the 40193 outputs and gives a pulse to the memory which has to write the data. how would you do it?

Click to expand...

You need a simple sequencing circuit - an edge from the counter input passing through a continuously clocked shift reg, or just a couple of cascaded 555s; or anything to generate the appropriate pulses in the correct order.

Create something yourself and people will help you refine it!

I'm sorry to see that the phrase "no idea?" has been interpreted as a negative criticism-provocation.
it didn't want to be that

arivel said:

I'm sorry to see that the phrase "no idea?" has been interpreted as a negative criticism-provocation.
it didn't want to be that

Click to expand...

Well, we gave you many ideas but your feedback on each was, essentially, "it is not a my magnetic memory idea so I want you to give me another idea" which, we assume is your magnetic memory idea.

I think you need to understand
- that low power latching circuits draw so little power that a single alkaline battery can last years.
- Arduino and other microcontrollers can write to static memory and it is well worth your time to learn if you plan to do more projects (and may be with learning if you want more than a slightly stable and somewhat unreliable non-microcontroller memory solution.