• Please review our updated Terms and Rules here

Looking for the part number for a PT 48kra Data Delay

Corey986

Administrator
Joined
Dec 7, 2011
Messages
821
Location
East Coast USA
I just picked up a 48kra, but it's missing the data delay that plugs into u50. I can't seem to find a valid part number in the manuals on line or by checking any of the pics (google searches can't find the part number in the pics).

Anyone know the part number or at least the delay specs so I can find a replacement.

Thanks,
Corey
 
The problem is DDU-4 7884 isn't a part number I have found with google. DDU-4 is a series from Data Delay, but 7884 isn't a listed part with specs for delay time so if someone knows the specs or a replacement part number that would be cool. I'm guessing 7884 may not be the listed part, but something else like a date code and the part number may be on the side or something.

Thanks,
Corey
 
Corey,

I took a look at all three 16KRA boards here, it has DDU-4-7672, with the dashes, this would be the part number. On the side of the component, it has 1178, this would be the datecode. The components are not unique, we used them in many of the mainframe designs of the day. I just did a preliminary search online without any lunk. But I would suspect some of the surplus parts suppliers may carry this. Maybe also some of the Parts Dealers [aka Part Sharks, as I like to call them] cause they like to hoard just about every component out there. But it will take some perseverance to search all of these places - I have about 100 bookmarks just for component suppliers. I saw your post on the Altair group, I would think that one of these parts suppliers would have then for less than $25each.

Dan
 
According to data delay, they are custom parts. So I don't think a surplus place would have it. The big question is how close do we have to be to the original spec.

For the 48kra, which according to pics (mine is missing) is DDU-4-7784
Tap 1=48ns
Tap 2=90ns
Tap 3=143ns
Tap 4=320ns
Tap 5=360ns

Not quite a normal progression.
Maybe a 10 step 50ns data delay could work using taps 1,2,3,6,7. If the timing allows for that.

They are about $6-7 each.

Cheers,
Corey
 
According to data delay, they are custom parts. So I don't think a surplus place would have it.

I suppose they meant they are discontinued and since you wanted them to remake these parts, they kept insisting it was custom. I doubt a small company such as PT would be ordering custom parts versus say a military contractor. When I looked at their DDU4 datasheet here http://www.datadelay.com/datasheets/ddu4.pdf , you can see the part number list is somewhat segmented, they must have discontinued the older parts since then.

The big question is how close do we have to be to the original spec.

Most of these parts always had either a 5% or 10% tolerance. But then on each Tap, there was also some additional tolerance. You can see in the datasheet for their current devices, they still have this rating. You'll see them specify the range and tolerance for each Tap, eg 50nsec +/-5ns, that's 10%

PT would have to make their design to accommodate this tolerance. Their circuit would still be expected to work if any Tap was between 45nsec up to 55nsec delay [this is +/-5nsec]. Otherwise, they would have been pulling their hairs out every time another delay line had varied in tolerance. One batch might have been -3nsec/Tap and then the next batch might have been +4nsec/Tap, and so on.

For the 48kra, which according to pics (mine is missing) is DDU-4-7784
Tap 1=48ns
Tap 2=90ns
Tap 3=143ns
Tap 4=320ns
Tap 5=360ns

Not quite a normal progression.
Maybe a 10 step 50ns data delay could work using taps 1,2,3,6,7. If the timing allows for that.

I suppose these measurements are from your working board ?
I haven't checked my board yet. Also, I haven't compared the 48KRA schematic to the 16KRA intensely just yet. From your measurements, you can see how the tolerance spreads out between the Taps . But Tap4 looks like it has a much larger delay, more than 50nsec, about 150nsec delay - this is quite odd.

Did you mean to write 220ns instead ? and then Tap 5 would be 260nsec then ?
This doesn't appear to be normal progression. As multi-tap delay lines normally have an equal distribution in delay. I can be mistaken, it's been a long while since we designed with this.

Did the folks at DataDelay confirm your measurements of the delay time per Tap with their old component specifications?

You mentioned Tap 6 and 7, I suppose you're looking at the DDU7 parts, DDU-7J-50 ? Mine has only 5 Taps just like the 48KRA.

I just noticed that Maxim carries the same pin compatible delay line too http://datasheets.maximintegrated.com/en/ds/DS1005.pdf

Do the folks at DataDelay still have a datasheet on file there for either of these old parts from our old memory boards which they could scan for us ??

I could take some measurements on my 16KRA if it would help.



They are about $6-7 each.

Cheers,
Corey

Now that's a much better price :)
I suppose you were asking about the DDU-7J-50.
Was there a minimum quantity for this ?

Dan
 
Actually I'm missing the part. They had to pull the paper order for me from PT to look up the specs. It was a special order back in the day for PT according to them. It was never an off the shelf part. I think that is why many 48kra are no longer around. If the delay went so did the card.

I was looking at part 3D7220-50 which is 14 pin and should be easy to make an adapter. I'm going to see if they have any at surplus stores next week. The only place I found with them had a 10 unit minimum order. I'd get that many as a "life time supply" if I knew they would work.

I guess I won't know till I try. Also yes, the 320 is not a typo. If you look at the 48kra timing diagram, while not easy to measure exact timing from, you can see tap 4 and 5 are almost on each other in timing. I should also ask data delay about the units in my early 16kra and 32kra. Maybe they are close to the 48kra. If they are I guess I could use 10 of the known to work spec ones because I have 4 16kra cards and one 32kra. Good to have spare parts.

I'll take a look at the maxim parts as well later. I didn't follow your link yet.

Cheers,
Corey
 
ok, Took a look at that timing diagram in the manual. Your're not going to like what I found. But that timing diagram is a very rough estimation. It's not the first time I found a 40yr old typo in a manual [or software]. Basically I measured the read timing cycle with a ruler and then calculated the timing marks from the ratios. Below is all of the measurements.

Actually I'm missing the part. They had to pull the paper order for me from PT to look up the specs. It was a special order back in the day for PT according to them. It was never an off the shelf part. I think that is why many 48kra are no longer around. If the delay went so did the card.

If they pulled the specs off the PT order, can they share it with us. I mean are they withholding this information unless we buy it ?

I was looking at part 3D7220-50 which is 14 pin and should be easy to make an adapter. I'm going to see if they have any at surplus stores next week. The only place I found with them had a 10 unit minimum order. I'd get that many as a "life time supply" if I knew they would work.

oh ok, cool
Which place is this ? I wonder if I have them on my list too.
As for the adapter you might want to use one of the old style 16pin DIP breadboards - the kind with the little u-shaped pins which hold the component leads [I have a pic around here somewhere] - fold the pins on the IC, hot glue it on there, and wire the pins to match the connections on the circuit board for U50. I think it will still have a low profile to fit in the S-100 slot.

I guess I won't know till I try. Also yes, the 320 is not a typo. If you look at the 48kra timing diagram, while not easy to measure exact timing from, you can see tap 4 and 5 are almost on each other in timing. I should also ask data delay about the units in my early 16kra and 32kra. Maybe they are close to the 48kra. If they are I guess I could use 10 of the known to work spec ones because I have 4 16kra cards and one 32kra. Good to have spare parts.

yes, I also see that Tap4 and Tap5 are very close, I show this below in the measurements. I haven't looked closely at the Timing cycles on the 16KRA just yet.


I'll take a look at the maxim parts as well later. I didn't follow your link yet.

Cheers,
Corey

Timing Diagram Measurements:

I scaled the page with the read timing cycle to 6in length, for 489nsec, between the start timing mark and the end timing mark
So for the read timing cycle; 6in = 489 nsec

The Timing marks for the Phase2 clock at the top for 210ns and 289ns are what's already messed up.
From measuring these timing marks, I get the actual time.
I'm only using 1/16th of an inch for the resolution - 1/16th inch = 5nsec
And I get a discrepancy already from using their timing marks.
The Timing diagram is not much help if they can't draw it correctly.

Phase2(210ns) = 2.75in ; actual time = 224.1ns (+14ns)
Phase2(289ns) = 3.625in ; actual time = 295.4ns (+6.4ns)

The Delay Line is fed with signal CAE, so as the input, this can be considered Tap0 with respect to the start of the read timing cycle. And the delay time for CAE on the JK flip-flop from Phase2 clock is rather long, 40.7nsec, for a Schottky TTL part (U49). It's not outrageous, but something else to verify. Q5 may still have some significant delay even though it's wired as a current buffer.

Measuring from the start timing mark in the read cycle, these are the timing values.
Tap0: 0.5in = 40.7ns
Tap1: 1.1875in = 96.7ns--------> +56ns
Tap2: 1.75in = 142.6ns-------> +50ns
Tap3: 2.1875in = 178.2ns-------> +35ns
Tap4: 4.25in = 346.3ns-------> +168ns
Tap5: 4.3125in = 351.4ns-------> +5ns <-----very short, about the same as a delay on a logic gate


Suffice it to say we should really borrow the actual part from someone who has a working 48KRA and take actual measurements. Because the timing marks on that diagram indicates a very unique delay line. It would help if we can verify this ourselves. Because, as it stands right now, that one part you mentioned and the one I mentioned won't work in here.

All you need is the part, not the whole board, just pop it in a breadboard. Add power, pump it with a oscillator, which is not faster than the delay timing, so for this - nothing faster than say 1Mhz, just to make it easy. Then just measure the delay taps with a o'scope, putting the first scope probe on the oscillator output, with the trigger set to chan.1. Then take the 2nd probe on chan.2 and measure the delay on each tap. If you have a decent o'scope, it should have onscreen cursors to measure the delay time. That's the proper way to get this done unless that company can provide us with a real datasheet. I can do this with work with my delay line, since the 16KRA has a different part#, we might as well get it recorded before they all go bad.

Which part is on the 32KRA again ? same at the 48KRA or the 16KRA ?

And after that is all done, it would be really nice to make a printout of the actual timing diagram using a Logic Analyzer to use as reference.
Dan
 
ok, Took a look at that timing diagram in the manual. Your're not going to like what I found. But that timing diagram is a very rough estimation. It's not the first time I found a 40yr old typo in a manual [or software]. Basically I measured the read timing cycle with a ruler and then calculated the timing marks from the ratios. Below is all of the measurements.

That's why I didn't attempt to make measurements off the diagram. Figured it was good for using a scope or analyzer for shape not exact timing.

If they pulled the specs off the PT order, can they share it with us. I mean are they withholding this information unless we buy it ?

They shared the numbers I gave above, I don't think they had a data sheet.

oh ok, cool
Which place is this ? I wonder if I have them on my list too.

I was going to hit HALTD in Santa Clara and Excess Solutions in Milpitas. Excess Solutions has a section of data delays. I know they don't have the correct one, but maybe they have a 10 tap 50ns step to 500.

As for the adapter you might want to use one of the old style 16pin DIP breadboards - the kind with the little u-shaped pins which hold the component leads [I have a pic around here somewhere] - fold the pins on the IC, hot glue it on there, and wire the pins to match the connections on the circuit board for U50. I think it will still have a low profile to fit in the S-100 slot.

Figured I'd have to do something like that.

Suffice it to say we should really borrow the actual part from someone who has a working 48KRA and take actual measurements. Because the timing marks on that diagram indicates a very unique delay line. It would help if we can verify this ourselves. Because, as it stands right now, that one part you mentioned and the one I mentioned won't work in here.

That would be cool if someone is reading this threads and can measure theirs.
Which part is on the 32KRA again ? same at the 48KRA or the 16KRA ?

On the early 32KRA like mine that look like the typical 16KRA, the part is the same. While I haven't seen one there was a 32KRA-1 which according to documentation is based on the 48kra design. It might have the same as the 48kra.

Thanks for all the info...
Cheers,
Corey
 
That's why I didn't attempt to make measurements off the diagram. Figured it was good for using a scope or analyzer for shape not exact timing.

They shared the numbers I gave above, I don't think they had a data sheet.

I was going to hit HALTD in Santa Clara and Excess Solutions in Milpitas. Excess Solutions has a section of data delays. I know they don't have the correct one, but maybe they have a 10 tap 50ns step to 500.

Figured I'd have to do something like that.

That would be cool if someone is reading this threads and can measure theirs.

On the early 32KRA like mine that look like the typical 16KRA, the part is the same. While I haven't seen one there was a 32KRA-1 which according to documentation is based on the 48kra design. It might have the same as the 48kra.

Thanks for all the info...
Cheers,
Corey

Oh ok, those timing values they gave you looked to be actual measurements rather than from a datasheet. That's why I was asking.
So then the 50,50,50,150,50 delay timing would work using the 10tap delay line.
I remember HALTD, but didn't know about Excess Solutions, I'll have to update my list - thanks.
But some of the regular suppliers should have this part too, just have to look for 5V level parts.
I'll go ahead and measure mine too.

Dan
 
Back
Top