CO60472 Delay Line

[+Philsan]

What is the purpose of CO60472 IC?

 

Why it is so different than the other ICs (taller, asymmetric legs)?

[Bryan]

A 64Kx1 DRAM doesn't have 16-pins for 65536 addresses. Instead, it uses multiplexed row and column addresses (you can even address it faster by only updating the column for nearby addresses). However, this means the Atari has to perform multiple DRAM operations per 6502 cycle. The Freddie chip does all this by having a 4X clock, but on older machines it was done with a delay line to activate RAS and CAS at certain intervals within a cycle.

 

The part is in a strange package because it contains a piezoelectric element that the waves travel through at a fixed speed. On the Atari part, the unused delay line taps were cut off.

[+Stephen]

Wow - that sounds like a slightly more sophisticated model of the old mercury delay lines used for memory.

[Bryan]

Similar idea, but in that case the delay line was the memory. Since a wave put into the mercury would reach the other end after a known delay, you could get your data back by checking the sensor exactly when the wave should be there. Then you put that pulse back in the other end. Pretty crazy stuff.

[ClausB]

60s programmable calculator with acoustic wire delay line memory:

http://www.vintagecalculators.com/html/monroe_epic_3000.html

[Rybags]

If you look at the 1064 Ram module schematics, it seems to use the /RAS line fed 4 times into an LS04 inverter which has a net result of giving you back what you started with but should give a propogation delay equal to about half a machine cycle. This is then fed to the multiplexors to select R/C. Though R/C to the Ram chips themselves comes direct from the Ras/Cas on the PBI.

 

/wonders - could such a setup be used inside the computer to replace a dead delay line.

Edited May 26, 2016 by Rybags

[Bryan]

/wonders - could such a setup be used inside the computer to replace a dead delay line.

Oh sure. Stacked gates are often used for delays, but they'll vary more with temperature and from sample to sample. You'd be better off using high-speed gates with precision R/C delays in between.

[+Philsan]

Thanks for replies.

 

The unanswered question is why it is so different than the other Atari IC?

It is taller and has asymmetric legs.

[Bryan]

Thanks for replies.

 

The unanswered question is why it is so different than the other Atari IC?

It is taller and has asymmetric legs.

I added that info to the end of my first post.

 

The delay line element itself is bulky. It's literally a miniature version of this:

 

https://upload.wikimedia.org/wikipedia/commons/thumb/3/33/Ultrasonicdelayline.jpg/220px-Ultrasonicdelayline.jpg

 

The part is probably someone's standard part with multiple delay taps and the legs Atari didn't need were cut off for some reason (but they still work if you solder a little wire to them).

[ClausB]

Back in the day I put an oscilloscope on the delay line and drew this diagram.

 

[manterola]

Can an Atari delay line be reproduced with modern components? and more important... can be done easily and conveniently? Maybe it is a good project to learn a bit of KiCAD and build something based on SMT delay line standard and cheap components.

 

 

[Kyle22]

IIRC, Freddie has a number of inline inverters that perform the delay.

 

[1050]

7 hours ago, manterola said:

Can an Atari delay line be reproduced with modern components? and more important... can be done easily and conveniently? Maybe it is a good project to learn a bit of KiCAD and build something based on SMT delay line standard and cheap components.

 

 

 

[Rybags]

Aren't such delay lines off-the-shelf components to begin with?

 

The difference with ours being the unused legs are removed and it's stamped with an Atari part # rather than an OEM one.

[TGB1718]

On 5/26/2016 at 4:07 AM, Bryan said:

Oh sure. Stacked gates are often used for delays,

When I was training to be an electronics engineer, I remember the term "Race Hazards", something that happens when 2 or more signals

take different paths through a logic system, they often used extra buffer chips in the faster circuit to "Slow it Down" so both signals meet at

the destination at same time.

[+DrVenkman]

5 hours ago, Rybags said:

Aren't such delay lines off-the-shelf components to begin with?

 

The difference with ours being the unused legs are removed and it's stamped with an Atari part # rather than an OEM one.

Isn’t that the question? ARE they? And if so, has anyone sourced a suitable modern equivalent with a compatible pin-out, or wired up an adapter of sorts?

 

Or alternately, is this just one of those parts that fails rarely enough that there hasn’t been a need to investigate such measures?

[Rybags]

I suspect if they're just a piezo type element then they should be extremely reliable so long as they don't suffer impact or huge power surge.

[manterola]

I have two XL with failed delay line (and cd4051 problems, keyboard). Best is the unique source of delay lines... until it is not.   So I was basically trying to understand the same, how special this IC is, and it looks like it is not special at all, and it might be possible to reproduce with new delay lines ICs or even it was suggested (?) that the Atari IC might be  just a standard one with some legs removed.

There is also a circuit to produce the delays with an LS14 plus discrete components (thank you @1050).

 

 

[_The Doctor__]

Wonders if the some dead delay lines could be hit with acetone to reveal some info, some over the years were kinda rough as if sandblasted others were smooth... as Atari cut costs and suppliers changed it's possible a number of them were just blacked out and remarked with the Atari PN...

[TGB1718]

Which IC is this by the way, maybe try to measure the delay and see what could be used to replace it

 

Edit: never mind, found another thread on the subject with more detail (u29)

Edited June 2, 2020 by TGB1718
Update

[ChildOfCv]

10 hours ago, Rybags said:

Aren't such delay lines off-the-shelf components to begin with?

 

The difference with ours being the unused legs are removed and it's stamped with an Atari part # rather than an OEM one.

There are certainly delay lines with multiple outputs, but none of them have the same timings as the Atari ones.  They used a special order for them.

You can find datasheets for the EP82xx line of delay lines, and all 5 outputs will be a multiple of the same delay value.  In the Atari case, they have non-linear delays.  EP8212 does not even appear on the datasheet.

 

But they do show the basic schematic of the IC on the datasheet, so with the appropriate selection of inductors and capacitors it should be possible to build a working replacement.

ep82xx.pdf

Edited June 2, 2020 by ChildOfCv

[_The Doctor__]

It's a funny world, many special orders end up out there in other items or even duplicated.

[ChildOfCv]

In fact, looking at the schematic, it matches the pinout of a 74HC14 hex inverter.  Just tie some inductors and capacitors and a resistor across the pins like shown in the schematic above, and you'll have a workalike.  All that's left is to calculate the proper values.

 

Here's a Kicad design for an overlay that could be soldered on top of a hex inverter to create the delay line.

 

delay line.zip

Edited June 2, 2020 by ChildOfCv

[ChildOfCv]

Well, first attempt at guessing usable values.  And a picture of the hex inverter overlay

 

delay line.pdf

[Kyle22]

Someone has done this already. I just can't remember who. It was from Poland or Eastern Europe IIRC.

I saw it a few years ago.