Topics

Mar1e gets (some) address LEDs #Homebrew #Electronics

John
 

Hi Lee and the group,

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

The black wire exiting stage left is for the 'scope's earth clip.

Any ideas, anyone?

Thanks,

John

Paul Schmidt
 

John, when I built the first of my two identical 'original configuration' ELFs (I have the original and my duplicate is (I think) still on display at Bletchley Park, both I (with mine) and the Bletchley volunteers (with theirs) noticed a slight glitchiness when making changes to memory. I my case, I initially attributed it to just clumsiness on my part, or brain fatigue during tedious program entry. When multiple guys at Bletchley contacted me about experiencing the same issue, I played around and found an alternate sequence of toggle switch movements that seemed to completely cure the problem. I think my online-published ELF documentation still shows the original sequence:

- MP up
- Press IN repeatedly until the byte immediately before the desired byte is displayed
- Set 8 toggles for correct byte contents
- MP down
- Press IN once
- MP up

In my correspondence with Bletchley, I have a record of the improved sequence, but offhand I don't recall what it is. My computer's boot drive took a dump over this last weekend, and although all normal data was automatically backed up, my OUTLOOK emails were not (oops!), so at the moment I cannot check to see what I told them. A data recovery service is working to recover those stored emails, but it might be a couple weeks before I (hopefully) have access to them again.

Maybe somebody else on this forum has the answer in the meantime.

Paul Schmidt

-----Original Message-----
From: John
Sent: Apr 10, 2019 12:30 PM
To: cosmacelf@groups.io
Subject: [cosmacelf] Mar1e gets (some) address LEDs

Hi Lee and the group,

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

The black wire exiting stage left is for the 'scope's earth clip.

Any ideas, anyone?

Thanks,

John


Paul Schmidt
ocleide@...


David Madole
 

Paul’s design looks like it lets the /WE input on the RAM float during the short time that the MP switch is between breaking one side of the throw and making the other. I suppose this is fine if the /WE input on the RAM floats high but that’s probably not a good assumption, especially if you are not using the same RAM he did.

 

For the simplest fix, I would suggest tying /WE of the RAM high through a pull-up resistor (22K should be fine) to avoid this. Once you do that, you don’t really need a double-pole switch for MP either, but it won’t hurt.

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of John
Sent: Wednesday, April 10, 2019 1:30 PM
To: cosmacelf@groups.io
Subject: [cosmacelf] Mar1e gets (some) address LEDs

 

Hi Lee and the group,

 

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

 

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

 

The black wire exiting stage left is for the 'scope's earth clip.

 

Any ideas, anyone?

Thanks,

 

John

_._,_._,_

 

David Madole
 

John,

 

Another thought that came to me… why do you need a DIP switch to hold /WE high when that is exactly what the MP switch appears to do?

 

It seems like you could have just flipped MP to power off?

 

There may be more going on here than meets the eye and a schematic may be useful if further discussion is needed.

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of David Madole
Sent: Wednesday, April 10, 2019 3:15 PM
To: cosmacelf@groups.io
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs

 

Paul’s design looks like it lets the /WE input on the RAM float during the short time that the MP switch is between breaking one side of the throw and making the other. I suppose this is fine if the /WE input on the RAM floats high but that’s probably not a good assumption, especially if you are not using the same RAM he did.

 

For the simplest fix, I would suggest tying /WE of the RAM high through a pull-up resistor (22K should be fine) to avoid this. Once you do that, you don’t really need a double-pole switch for MP either, but it won’t hurt.

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of John
Sent: Wednesday, April 10, 2019 1:30 PM
To: cosmacelf@groups.io
Subject: [cosmacelf] Mar1e gets (some) address LEDs

 

Hi Lee and the group,

 

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

 

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

 

The black wire exiting stage left is for the 'scope's earth clip.

 

Any ideas, anyone?

Thanks,

 

John

Paul Schmidt
 

For what it is worth (and I am gratified that folks are using the ELF documentation I prepared), there is no "Paul's Design" in regard to the ELF schematic. What I show is just the same as the magazine article schematic, merely cleaned up and made neater, with more descriptive comments and such. I did not change anything that I can recall in regard to the way "MP" works.....

Paul

-----Original Message-----
From: David Madole
Sent: Apr 10, 2019 2:15 PM
To: "cosmacelf@groups.io"
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs

Paul’s design looks like it lets the /WE input on the RAM float during the short time that the MP switch is between breaking one side of the throw and making the other. I suppose this is fine if the /WE input on the RAM floats high but that’s probably not a good assumption, especially if you are not using the same RAM he did.

 

For the simplest fix, I would suggest tying /WE of the RAM high through a pull-up resistor (22K should be fine) to avoid this. Once you do that, you don’t really need a double-pole switch for MP either, but it won’t hurt.

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of John
Sent: Wednesday, April 10, 2019 1:30 PM
To: cosmacelf@groups.io
Subject: [cosmacelf] Mar1e gets (some) address LEDs

 

Hi Lee and the group,

 

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

 

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

 

The black wire exiting stage left is for the 'scope's earth clip.

 

Any ideas, anyone?

Thanks,

 

John


Paul Schmidt
ocleide@...


David Madole
 

Ah ok, I was just replying to what John called it. I took a quick look at the schematic but I really didn’t read the history or other documentation with it to understand that.

 

It’s been 40 years since I’ve seen the magazine article so I have no idea any more what the original design looked like :)

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of Paul Schmidt
Sent: Wednesday, April 10, 2019 3:36 PM
To: cosmacelf@groups.io; cosmacelf@groups.io
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs

 

For what it is worth (and I am gratified that folks are using the ELF documentation I prepared), there is no "Paul's Design" in regard to the ELF schematic. What I show is just the same as the magazine article schematic, merely cleaned up and made neater, with more descriptive comments and such. I did not change anything that I can recall in regard to the way "MP" works.....

 

Paul

-----Original Message-----
From: David Madole
Sent: Apr 10, 2019 2:15 PM
To: "cosmacelf@groups.io"
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs


Paul’s design looks like it lets the /WE input on the RAM float during the short time that the MP switch is between breaking one side of the throw and making the other. I suppose this is fine if the /WE input on the RAM floats high but that’s probably not a good assumption, especially if you are not using the same RAM he did.

 

For the simplest fix, I would suggest tying /WE of the RAM high through a pull-up resistor (22K should be fine) to avoid this. Once you do that, you don’t really need a double-pole switch for MP either, but it won’t hurt.

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of John
Sent: Wednesday, April 10, 2019 1:30 PM
To: cosmacelf@groups.io
Subject: [cosmacelf] Mar1e gets (some) address LEDs

 

Hi Lee and the group,

 

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

 

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

 

The black wire exiting stage left is for the 'scope's earth clip.

 

Any ideas, anyone?

Thanks,

 

John

 

Paul Schmidt
ocleide@...

 

John
 

Thanks again for all the help. I can confirm;

There is no difference between the original circuit diagram and the one I called "Paul's design" 
The DIP switch I added does exactly the same as leaving the MP switch in the active condition (why didn't I spot that?)
Adding the 22k pull-up between the switch wiper and +5v has fixed the problem. I can now flick the MP switch up and down without any unwanted writes to RAM.

Adding the address LEDs was worthwhile too. I can see where I am. I will sleep well tonight.

Next job is to wire the other DIP switches to A8, 9 and 10, make the new back apron with D connector cut-outs (maybe tomorrow), add the MAX232 and some ports.

I love this machine, and I only found out about it by chance while planning a SC/MP-based project. 

John,

Nottingham

On Wed, 10 Apr 2019 at 20:47, David Madole <david@...> wrote:

Ah ok, I was just replying to what John called it. I took a quick look at the schematic but I really didn’t read the history or other documentation with it to understand that.

 

It’s been 40 years since I’ve seen the magazine article so I have no idea any more what the original design looked like :)

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of Paul Schmidt
Sent: Wednesday, April 10, 2019 3:36 PM
To: cosmacelf@groups.io; cosmacelf@groups.io
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs

 

For what it is worth (and I am gratified that folks are using the ELF documentation I prepared), there is no "Paul's Design" in regard to the ELF schematic. What I show is just the same as the magazine article schematic, merely cleaned up and made neater, with more descriptive comments and such. I did not change anything that I can recall in regard to the way "MP" works.....

 

Paul

-----Original Message-----
From: David Madole
Sent: Apr 10, 2019 2:15 PM
To: "cosmacelf@groups.io"
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs


Paul’s design looks like it lets the /WE input on the RAM float during the short time that the MP switch is between breaking one side of the throw and making the other. I suppose this is fine if the /WE input on the RAM floats high but that’s probably not a good assumption, especially if you are not using the same RAM he did.

 

For the simplest fix, I would suggest tying /WE of the RAM high through a pull-up resistor (22K should be fine) to avoid this. Once you do that, you don’t really need a double-pole switch for MP either, but it won’t hurt.

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of John
Sent: Wednesday, April 10, 2019 1:30 PM
To: cosmacelf@groups.io
Subject: [cosmacelf] Mar1e gets (some) address LEDs

 

Hi Lee and the group,

 

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

 

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

 

The black wire exiting stage left is for the 'scope's earth clip.

 

Any ideas, anyone?

Thanks,

 

John

 

Paul Schmidt
ocleide@...

 

David Madole
 

The 22K shouldn't go to the wiper, if it's wired like Paul's diagram. It should go to the terminal that connects to the RAM.

Maybe your circuit is a little different though.

The MAX232 is not "period correct"... don't do it! Just kidding. Kind of. 

David

On April 10, 2019 5:08:23 PM John <easydogxray@...> wrote:

Thanks again for all the help. I can confirm;

There is no difference between the original circuit diagram and the one I called "Paul's design" 
The DIP switch I added does exactly the same as leaving the MP switch in the active condition (why didn't I spot that?)
Adding the 22k pull-up between the switch wiper and +5v has fixed the problem. I can now flick the MP switch up and down without any unwanted writes to RAM.

Adding the address LEDs was worthwhile too. I can see where I am. I will sleep well tonight.

Next job is to wire the other DIP switches to A8, 9 and 10, make the new back apron with D connector cut-outs (maybe tomorrow), add the MAX232 and some ports.

I love this machine, and I only found out about it by chance while planning a SC/MP-based project. 

John,

Nottingham

On Wed, 10 Apr 2019 at 20:47, David Madole <david@...> wrote:

Ah ok, I was just replying to what John called it. I took a quick look at the schematic but I really didn’t read the history or other documentation with it to understand that.

 

It’s been 40 years since I’ve seen the magazine article so I have no idea any more what the original design looked like :)

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of Paul Schmidt
Sent: Wednesday, April 10, 2019 3:36 PM
To: cosmacelf@groups.io; cosmacelf@groups.io
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs

 

For what it is worth (and I am gratified that folks are using the ELF documentation I prepared), there is no "Paul's Design" in regard to the ELF schematic. What I show is just the same as the magazine article schematic, merely cleaned up and made neater, with more descriptive comments and such. I did not change anything that I can recall in regard to the way "MP" works.....

 

Paul

-----Original Message-----
From: David Madole
Sent: Apr 10, 2019 2:15 PM
To: "cosmacelf@groups.io"
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs


Paul’s design looks like it lets the /WE input on the RAM float during the short time that the MP switch is between breaking one side of the throw and making the other. I suppose this is fine if the /WE input on the RAM floats high but that’s probably not a good assumption, especially if you are not using the same RAM he did.

 

For the simplest fix, I would suggest tying /WE of the RAM high through a pull-up resistor (22K should be fine) to avoid this. Once you do that, you don’t really need a double-pole switch for MP either, but it won’t hurt.

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of John
Sent: Wednesday, April 10, 2019 1:30 PM
To: cosmacelf@groups.io
Subject: [cosmacelf] Mar1e gets (some) address LEDs

 

Hi Lee and the group,

 

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

 

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

 

The black wire exiting stage left is for the 'scope's earth clip.

 

Any ideas, anyone?

Thanks,

 

John

 

Paul Schmidt
ocleide@...

 


Lee Hart
 

Paul Schmidt wrote:
there is no "Paul's Design" in regard to the ELF schematic. What I show is just
the same as the magazine article schematic, merely cleaned up and made neater,
with more descriptive comments and such. I did not change anything that I can recall in regard
to the way "MP" works.
That's correct. Paul's schematic is the same as the original PE Elf article. Both leave the RAM's /WE pin "floating" during the time the switch transitions between WriteEnable and WriteProtect.

The 1802 Membership Card includes a pullup resistor on RAM /WE, and I've never had a problem with phantom writes. Perhaps adding one will fix your problem? It's an easy thing to try!

--
Fools ignore complexity. Pragmatists suffer it. The wise avoid it.
Geniuses remove it. -- Alan Perlis, "Epigrams on Programming"
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com

John
 

Hello Paul,

The wiper of the switch does go to the active low write enable pin(s) on the RAM in both the original PE diagram and Paul's diagram. The poles are connected to +5v (write protect) and pin 35 on the 1802 (active low memory write).

Not much of my Elf is "period correct", having a 6116 and laser-cut and engraved panels, but the general layout started like that of the original (and quickly changed). I might one day build an unexpanded original with authentic parts but that's not a priority. Having fun with it is the priority to justify the time I'm spending on it. An alphanumeric LCD is a probable addition, and a hex keypad. I fancy building an AFSK oscillator too, 1275/1445Hz, so it can talk acoustically to an RTTY app on my Android tablet at 50bps. I love that warbling sound!

John

On Wed, 10 Apr 2019 at 22:20, David Madole <david@...> wrote:
The 22K shouldn't go to the wiper, if it's wired like Paul's diagram. It should go to the terminal that connects to the RAM.

Maybe your circuit is a little different though.

The MAX232 is not "period correct"... don't do it! Just kidding. Kind of. 

David

On April 10, 2019 5:08:23 PM John <easydogxray@...> wrote:

Thanks again for all the help. I can confirm;

There is no difference between the original circuit diagram and the one I called "Paul's design" 
The DIP switch I added does exactly the same as leaving the MP switch in the active condition (why didn't I spot that?)
Adding the 22k pull-up between the switch wiper and +5v has fixed the problem. I can now flick the MP switch up and down without any unwanted writes to RAM.

Adding the address LEDs was worthwhile too. I can see where I am. I will sleep well tonight.

Next job is to wire the other DIP switches to A8, 9 and 10, make the new back apron with D connector cut-outs (maybe tomorrow), add the MAX232 and some ports.

I love this machine, and I only found out about it by chance while planning a SC/MP-based project. 

John,

Nottingham

On Wed, 10 Apr 2019 at 20:47, David Madole <david@...> wrote:

Ah ok, I was just replying to what John called it. I took a quick look at the schematic but I really didn’t read the history or other documentation with it to understand that.

 

It’s been 40 years since I’ve seen the magazine article so I have no idea any more what the original design looked like :)

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of Paul Schmidt
Sent: Wednesday, April 10, 2019 3:36 PM
To: cosmacelf@groups.io; cosmacelf@groups.io
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs

 

For what it is worth (and I am gratified that folks are using the ELF documentation I prepared), there is no "Paul's Design" in regard to the ELF schematic. What I show is just the same as the magazine article schematic, merely cleaned up and made neater, with more descriptive comments and such. I did not change anything that I can recall in regard to the way "MP" works.....

 

Paul

-----Original Message-----
From: David Madole
Sent: Apr 10, 2019 2:15 PM
To: "cosmacelf@groups.io"
Subject: Re: [cosmacelf] Mar1e gets (some) address LEDs


Paul’s design looks like it lets the /WE input on the RAM float during the short time that the MP switch is between breaking one side of the throw and making the other. I suppose this is fine if the /WE input on the RAM floats high but that’s probably not a good assumption, especially if you are not using the same RAM he did.

 

For the simplest fix, I would suggest tying /WE of the RAM high through a pull-up resistor (22K should be fine) to avoid this. Once you do that, you don’t really need a double-pole switch for MP either, but it won’t hurt.

 

David

 

 

From: cosmacelf@groups.io <cosmacelf@groups.io> On Behalf Of John
Sent: Wednesday, April 10, 2019 1:30 PM
To: cosmacelf@groups.io
Subject: [cosmacelf] Mar1e gets (some) address LEDs

 

Hi Lee and the group,

 

I've made a board with eight LED and eight 3k3 resistors which plugs into the second 6116 socket to show the lower eight bits of the address bus. It confirms my fears that, when I step through a program with MP active in order to edit a location, switching MP off writes the byte from the switches to the current location. Pressing IN (as I'm supposed to) then writes the byte from the switches to the next location (as it's supposed to). My circuit closely follow's Paul Schmidt's design. 

 

I don't yet know whether this happens every time or just sometimes. I'll try it repeatedly later tonight. I think I can get away with it by setting the current byte for the current location on the switches (as shown by the TIL311s), switching MP off, setting the next byte on the switches and pressing IN. I don't think this should be necessary though. Something is wrong.

 

I'll attach a new photo of Mar1e. The 2114 sockets are gone, the 6116 sockets are on the right and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted. The other three switches will be used on the 6116's A8, 9 and 10 inputs, giving me effectively eight blocks of 256 bytes of RAM which will be maintained when the power is off. I've also relocated the 1MHz can oscillator to give me more space at the top. I'll make a new back apron with holes for a 9-way D connector (RS-232 via a MAX232) and a 25-way one (for parallel interface). 

 

The black wire exiting stage left is for the 'scope's earth clip.

 

Any ideas, anyone?

Thanks,

 

John

 

Paul Schmidt
ocleide@...

 


Lee Hart
 

John wrote:
Not much of my Elf is "period correct", having a 6116 and laser-cut and
engraved panels, but the general layout started like that of the
original (and quickly changed). I might one day build an unexpanded
original with authentic parts but that's not a priority.
The original PE Elf wasn't a commercial product; it was hand-wired on perfboard. As you look through the series of articles, it keeps changing as Weisbecker added features. So there is no "right" way to build it.

The 2114 and 6116 both existed in the 1970's. They were just more expensive than the 2101's in the original Elf. Still, it cost about $80 for all the parts. That was a lot in 1976!

Having fun with it is the priority to justify the time I'm spending
on it.
That's the name-of-the-game for all of us! :-)

I fancy building an AFSK oscillator too, 1275/1445Hz, so it can talk
acoustically to an RTTY app on my Android tablet at 50bps. I love that
warbling sound!
The terminal for my first Elf was a Baudot teletype! I wrote my own driver to get it to work with a monitor and then Tiny BASIC.

--
Fools ignore complexity. Pragmatists suffer it. The wise avoid it.
Geniuses remove it. -- Alan Perlis, "Epigrams on Programming"
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com

David Madole
 

John wrote:
I fancy building an AFSK oscillator too, 1275/1445Hz, so it can talk
acoustically to an RTTY app on my Android tablet at 50bps. I love that
warbling sound!
Try just putting a speaker on Q and doing the tones in software.

David

John
 

Hi All,

Earlier I wrote "and I've added a 4-way DIP switch. The left-most switch holds the active-low MW pin of the battery-backed RAM chip high for switching off and on, that way the RAM contents don't get corrupted."

I was wrong (forgotten already!) The left-most switch was used to pull the active-low chip select high, it was nothing to do with write enable. While running the chip select was held low, then I took it high for power down. I've disconnected the switch now and so far I haven't had a problem with memory loss. 

I've made a new back apron for Mar1e today, with a 9-way D socket for serial and a 25-way D socket for parallel and analog interfaces, none of which I've added yet. I'm eyeing-up the remaining space on the board for a 74HCT373, a MAX232 (no 1488/1489 for me as I don't want to have to add two more power rails) and whatever else I can fit. I think the two D sockets might be too close together, depending on the with of the covers. I've found a Psion Organiser XP (two line display) and two comms links, both with 25-way D connectors. Somewhere I have one where I swapped the connector for 9-way, but I haven't found it. Somewhere, too, I have a Psion LZ64 and an original LZ which I made into a 64 by adding more memory. Four lines would be good.

John

On Thu, 11 Apr 2019 at 17:52, David Madole <david@...> wrote:
John wrote:
> I fancy building an AFSK oscillator too, 1275/1445Hz, so it can talk
> acoustically to an RTTY app on my Android tablet at 50bps. I love that
> warbling sound!

Try just putting a speaker on Q and doing the tones in software.

David




Lee Hart
 

David Madole wrote:
John wrote:
I fancy building an AFSK oscillator too, 1275/1445Hz, so it can talk
acoustically to an RTTY app on my Android tablet at 50bps. I love that
warbling sound!
Try just putting a speaker on Q and doing the tones in software.
Certainly! 1802 cassette tape storage programs did exactly that. :-)

--
Fools ignore complexity. Pragmatists suffer it. The wise avoid it.
Geniuses remove it. -- Alan Perlis, "Epigrams on Programming"
--
Lee Hart, 814 8th Ave N, Sartell MN 56377, www.sunrise-ev.com