• Please review our updated Terms and Rules here

IC profiler

D

Druid6900

Veteran Member
Joined
May 7, 2006
Messages
3,809
Location
Hamilton, Ontario, Canada
Again, I am looking for something that no longer, or never did, exist.
I seem to recall this REALLY cool, REALLY expensive device that you would plug into the CPU socket of a known working motherboard, plug the CPU into the device and it would profile that motherboard. In the future, with the same make and model of motherboard, you would do the same and it would tell you were the problem was.
We have POST cards now for the same sort of thing and, if you are fortunate, they will, do, roughly, the same.
However, what I am looking (dreaming?) for is a device that you would put a known good chip into (let's say a MOS 6526) and it would profile and store the workings of that chip. In the future, you would drop in a chip (let's say from China) and it would analyze and say "Yes, I am a MOS 6526" or "No, I'm NOT a MOS 6526, but, (with enough different profiles stored) I AM a MOS 6532A".
Would such a device be possible with modern technology?

Thanks,
Druid
 
For modern CPUs? Just coming out with various versions for the various sockets involved would be a chore--and some sort of "wedge" might make it impossible to use the heatsink. So possible? Maybe. Practical? Probably not.

I think that the concept died out around the time of the 80186.
 
No, no, the motherboard analyzer was just something I remember from long ago and, you're right, it would be useless beyond the DIP CPU carrier without a bunch of HS&F adapter heads.
I want to profile, store and test against chips that are beyond the capabilities of the simple TTL/CMOS chip tester we have now.
I want to be able to build the database the chips are compared against whether it's in circuit, out of circuit or through circuit.
 
I have a logic chip analyser that does something like what you describe - an ABI Electronics DDS- 40XP. It has a list of known parts, all the 74 series ICs, DIL package RAM ICs, EPROMs etc in its ROM module. You can do in circuit testing, or parts checking via its ZIF socket. The more interesting feature is its custom version of BASIC you could use to design your own tests for known parts or design tests for parts not already in its ROM. Don't think it could handle CPUs though.

In terms of modern equipment, does this look closer to what you mean:

https://www.abielectronics.co.uk/Products/RevEngSchematicLearningSystem.php
 
My Xeltek EPROM programmer has a similar feature. I've got some Fujitsu house-numbered TTL parts and it did pretty well finding the 74LSxxx equivalent.
 
OK, maybe I'm not explaining myself correctly.
This has nothing to do with CPUs.
So, I buy a bunch of MOS 6526, MOS 6532A and MOS 6522 ICs, for example.
All of them are suspect because the source doesn't have the computers, diagnostic or harnesses to test them.
What we have to do is, one by one, put these chips into the appropriate computer, run the diagnostics and cook them for about 15 minutes each to see if a) they really are the chips they say they are, that they work correctly and that they work both cold and hot. Before they are used or sold, they go through the same process for 24 hours.
As you can imagine, this takes a while but is necessary to weed out the defects and have them replaced or refunded.
If I had a chip tester that would test these types of chips based on the characteristics of a known good sample for a go/no go check and be able to tell if it's an authentic chip or a re-stamped fake it would speed up the process and identify the fake chip (which may be useful after all) it would be beneficial in the long run.
Now, for the learning stage, it doesn't matter if it works it from a ZIF socket on the device, an adapter under the chip in-circuit or magic as long as it stores the profile in a database and lets you label it with a display name.
When testing a suspect chip, it would be nice if it would confirm the chip as being legit or confirm that it's not and offer to see if these is a match in the database.
 
Again, sure. Note that devices of this nature use programmable pin drivers, as especially with earlier devices, you don't know where Vcc (and any other supply rails) go. Not hard to do, but something to keep in mind.
 
Writing code to test a chip while complex is not that hard to do.

Writing code to "learn" a chip borders on impossible.
For most LSI parts you need to know something about the internals to get it to do anything.

A simple example: 8255 parallel interface.
Toggle every pin and record the results. You got nothing. Without setting the internal state
every pin acts like an input.
If however if you know it's supposed to be a 8255 you would use groups of pins to set the register content. The you could then use known bit patterns in to check for expected bit patterns out.
A few hundred tests. Repeatable results.

It's a lot easier if you know what the part is supposed to be and can examine for the expected behavior.

And i didn't even bring up timing..

In a past life I did board testing. we used "bed of nails" for circuit testing. But used downloaded test code for function tests.

joe
 
Druid6900 said:
Again, I am looking for something that no longer, or never did, exist.
I seem to recall this REALLY cool, REALLY expensive device that you would plug into the CPU socket of a known working motherboard, plug the CPU into the device and it would profile that motherboard. In the future, with the same make and model of motherboard, you would do the same and it would tell you were the problem was.
We have POST cards now for the same sort of thing and, if you are fortunate, they will, do, roughly, the same.
However, what I am looking (dreaming?) for is a device that you would put a known good chip into (let's say a MOS 6526) and it would profile and store the workings of that chip. In the future, you would drop in a chip (let's say from China) and it would analyze and say "Yes, I am a MOS 6526" or "No, I'm NOT a MOS 6526, but, (with enough different profiles stored) I AM a MOS 6532A".
Would such a device be possible with modern technology?

Thanks,
Druid
Click to expand...
Druid, are you still around? I need a radial alignment on a Tandon TM-100-1 that works perfectly, but it is aligned only to itself (I believe the stepper motor shaft slipped). I can't DM you because I don't have 10 posts yet or some such BS. Let me know if you still fix these things please.
 
We used to have a Schlumberger 635 tester at work that did something similar. It had a library of lots of devices and all you did was clip on to test them. It could even do them in circuit and had the ability to learn that certain pins would not be testable in certain circuits (ie for hard pullup resistors etc).

Never really worked reliably enough to be of huge use.
 
dave2521 said:
Druid, are you still around? I need a radial alignment on a Tandon TM-100-1 that works perfectly, but it is aligned only to itself (I believe the stepper motor shaft slipped). I can't DM you because I don't have 10 posts yet or some such BS. Let me know if you still fix these things please.
Click to expand...

If you have a formatted disk from an aligned machine, you can use that. Just measure the output from 5B and slightly adjust the head for maximum output.
Works for me
 
Druid6900 said:
I seem to recall this REALLY cool, REALLY expensive device that you would plug into the CPU socket of a known working motherboard, plug the CPU into the device and it would profile that motherboard.
Click to expand...
Sounds maybe like a "signature analyzer", like HP5004/HP5005/HP5006.
 
Gary C said:
If you have a formatted disk from an aligned machine, you can use that. Just measure the output from 5B and slightly adjust the head for maximum output.
Works for me
Click to expand...
Hi Gary I used my "Golden Standard" NIB Tandon to format up a clean new disk - and was able to align 4 of the drives by using my O-scope setup exactly as one would if using a Dysan or Dymek alignment disk. They all seem to work well and track with my known-good standard drive. That NIB Tandon so far has read every disk I have put in it, so I keep it safe for rainy days. On another note....I will pay a really STUPID amount of money for a good Dymek, Dysan or Accurite disk to have on hand...Really stupid, crazy amount of money.
 
dave2521 said:
Hi Gary I used my "Golden Standard" NIB Tandon to format up a clean new disk - and was able to align 4 of the drives by using my O-scope setup exactly as one would if using a Dysan or Dymek alignment disk. They all seem to work well and track with my known-good standard drive. That NIB Tandon so far has read every disk I have put in it, so I keep it safe for rainy days. On another note....I will pay a really STUPID amount of money for a good Dymek, Dysan or Accurite disk to have on hand...Really stupid, crazy amount of money.
Click to expand...

What we need is something like a greaseweazle to write the special tracks to a disk. Surely it would be possible.
 
Bruce Tomlin said:
Sounds maybe like a "signature analyzer", like HP5004/HP5005/HP5006.
Click to expand...
That triggered an old idea. I have this project: Debugger-LPT. It is a card that enables you to run a computer step by step. As it is controlled by a computer, there is no problem storing all steps in one or another way. Just do it for a working computer and then use the debugger on a faulty one. The moment the program the computer is running deviates from the good one, you stop everything and then you must find out what caused the difference.
 
Would using greaseweasle to copy an alignment disk work?
It is possible the special tracks are on each side from the desired head position, as I recall you don't align for maximum signal, but a waveform.

I think I have a 5.25 Dysan one possibly for 360k somewhere, and haven't aligned a floppy in many years.
Way way back I used to align the Persci 8" monsters, which were fast but furious and could be dangerous to the unwary.
 
Greaseweazle has nothing to do with it--you could even use a signal generator. You need to get the tracks offset by a calibrated amount from center. The idea, if you're using an analog disk is that when you're centered on the track, you get a symnetrical output from both sides of center. Think of it in the same sort of idea of ILS in aircraft landing.

The DAA uses a similar system, but with ID address headers offset from center. If you have a program that does repeated READ ID operations, when you're centered, only the ID address headers closest to the centerline are readable.

You can readily appreciate that this is mostly a mechanical problem.

Back in the day, we made our own 100 tpi alignment diskettes using a drive mounted on a 3/4" aluminum plate with the positioner being driven through a (IIRC) 100:1 precision reduction. Something like this.

At Tandon, I saw a hard disk servo writer that used laser interferometric positioning with a cylinder supplied from a bottle of nitrogen.
 
Last edited:
mark0x01 said:
Would using greaseweasle to copy an alignment disk work?
It is possible the special tracks are on each side from the desired head position, as I recall you don't align for maximum signal, but a waveform.

I think I have a 5.25 Dysan one possibly for 360k somewhere, and haven't aligned a floppy in many years.
Way way back I used to align the Persci 8" monsters, which were fast but furious and could be dangerous to the unwary.
Click to expand...
I meant 'like' not actually

But yes, of course how to position the head :(
 
You must log in or register to reply here.
Back
Top