Signature analysis: Difference between revisions

From TekWiki
Jump to navigation Jump to search
No edit summary
No edit summary
 
(6 intermediate revisions by the same user not shown)
Line 3: Line 3:
The idea is to describe "known good" states of signals in the system – typically when in some diagnostic / troubleshooting mode – as a four-digit "fingerprint" of that signal, when referenced to a known clock and defined start/stop conditions.  The concept was introduced in [https://doc.xdevs.com/doc/HP_Agilent_Keysight/journals/1977-05.pdf Hewlett Packard Journal May 1977] along with their first SA instrument, the HP 5004A.
The idea is to describe "known good" states of signals in the system – typically when in some diagnostic / troubleshooting mode – as a four-digit "fingerprint" of that signal, when referenced to a known clock and defined start/stop conditions.  The concept was introduced in [https://doc.xdevs.com/doc/HP_Agilent_Keysight/journals/1977-05.pdf Hewlett Packard Journal May 1977] along with their first SA instrument, the HP 5004A.


The signature algorithm is a 16-bit [[wikipedia:cyclic redundancy check|cyclic redundancy check]] that can be efficiently implemented in hardware using a [[wikipedia:linear-feedback shift register]].  HP selected the polynomial x<sup>16</sup>+x<sup>12</sup>+x<sup>9</sup>+x<sup>7</sup>+1 instead of a standard CRC-16 or CCITT-16 polynomial, in order toWe would ''scatter the missed errors as much as possible'' and to ''avoid selecting feedback taps that are evenly spaced or four or eight bits apart because the types of instruments [...] that we will most frequently be testing tend to repeat patterns at four and eight-bit intervals''.
The signature algorithm is a 16-bit [[wikipedia:cyclic redundancy check|cyclic redundancy check]] that can be efficiently implemented in hardware using a [[wikipedia:linear-feedback shift register|linear-feedback shift register]].  HP selected the polynomial x<sup>16</sup>+x<sup>12</sup>+x<sup>9</sup>+x<sup>7</sup>+1 instead of a standard CRC-16 or CCITT-16 polynomial, in order to ''scatter the missed errors as much as possible'' and to ''avoid selecting feedback taps that are evenly spaced or four or eight bits apart because the types of instruments [...] that we will most frequently be testing tend to repeat patterns at four and eight-bit intervals''.


They also chose a particular "funny hex" format to display the resulting 16-bit number (four hexadecimal digits) to be read easily on seven-segment displays, as 0–9,A,C,F,H,P,U instead of 0–9,A,B,C,D,E,F. (Presumably, including "H" and "P" in the display was an extra benefit.)   
They also chose a particular "funny hex" format to display the resulting 16-bit number (four hexadecimal digits) to be read easily on seven-segment displays, as 0–9,A,C,F,H,P,U instead of 0–9,A,B,C,D,E,F. (Presumably, including "H" and "P" in the display was an extra benefit.)   
Line 9: Line 9:
Signature analysis was only popular for a fairly short time in the late 1970s and early 1980s before increasing integration made it ineffective.   
Signature analysis was only popular for a fairly short time in the late 1970s and early 1980s before increasing integration made it ineffective.   
Some arcade game manufacturers used signature analysis to allow component-level repair by field technicians.
Some arcade game manufacturers used signature analysis to allow component-level repair by field technicians.
It is also possible to use signature analysis in a limited way in equipment that has not been designed for it, by inserting a "free run" or "NOP" adapter between the CPU and its socket.  This adapter is wired to present a "no operation" opcode to the CPU, which then loops through all its addresses, generating specific and constant signatures on address lines, allowing these lines in the circuit to be tested or identified (see reference below).


''to be expanded''
''to be expanded''
Line 19: Line 21:
* It made it possible for anyone, without prior knowledge or experience with digital electronics, to troubleshoot the new digital circuitry that was rapidly appearing in electronic instruments.
* It made it possible for anyone, without prior knowledge or experience with digital electronics, to troubleshoot the new digital circuitry that was rapidly appearing in electronic instruments.
* HP freely shared the concept so others could incorporate Signature Analysis in their products.
* HP freely shared the concept so others could incorporate Signature Analysis in their products.
Because Signature Analysis was developed by HP, Tektronix would never use it. This was the word from the very top at Tek.
Because Signature Analysis was developed by HP, Tektronix would never use it. This was the word from the very top at Tek. [...]
 
[...]


In 1978 Tek was developing [[7854|a radically new analog/digital scope]] that had a microprocessor, A/D converter, digital memory, programming capability, and a GPIB interface (also originally developed by HP as the HPIB). All of this complex digital circuitry was going to make these scopes virtually impossible to be repaired in the Field Office Service Centers.
In 1978 Tek was developing [[7854|a radically new analog/digital scope]] that had a microprocessor, A/D converter, digital memory, programming capability, and a GPIB interface (also originally developed by HP as the HPIB). All of this complex digital circuitry was going to make these scopes virtually impossible to be repaired in the Field Office Service Centers.
Line 38: Line 38:


==Signature Analyzers==
==Signature Analyzers==
* [https://literature.cdn.keysight.com/litweb/pdf/05004-90001.pdf HP 5004A]
* [https://literature.cdn.keysight.com/litweb/pdf/05004-90001.pdf HP 5004A] (10 MHz)
* HP 5005A/B
* [https://www.keysight.com/us/en/assets/9018-02650/user-manuals/9018-02650.pdf HP 5005A Signature Multimeter] (20 MHz) / [https://www.keysight.com/us/en/assets/9018-05951/user-manuals/9018-05951.pdf 5005B] (25 MHz)
* HP 5006A
* [https://www.keysight.com/us/en/assets/9018-05700/user-manuals/9018-05700.pdf HP 5006A] (25 MHz)
* [[308|Tektronix 308]]
* [[308|Tektronix 308]] (20 MHz)
* [[SA501|Tektronix SA501]]
* [[SA501|Tektronix SA501]] (40 MHz)
* [https://devkits.handheldmuseum.com/Atari-CatBox/index.htm Atari CAT BOX arcade tester] / [http://gamearchive.askey.org/General/Test_Equipment/Atari/CatBox.pdf Schematics]
* [https://devkits.handheldmuseum.com/Atari-CatBox/index.htm Atari CAT BOX arcade tester] / [http://gamearchive.askey.org/General/Test_Equipment/Atari/CatBox.pdf Schematics]
* [https://hc-ddr.hucki.net/wiki/doku.php/elektronik/signaturanalyse robotron 31020] (East Germany)
* [https://www.flippers.com/pdfs/KK-Signature2.pdf Kurz-Kasch Signature II] (15 MHz) (1979)
* [https://hc-ddr.hucki.net/wiki/doku.php/elektronik/signaturanalyse robotron 31020] (10 MHz) (East Germany 1987)
* [https://www.eevblog.com/forum/testgear/signature-analyzer-project/ Signature analyzer project] @ EEVBlog Forum
* [https://www.eevblog.com/forum/testgear/signature-analyzer-project/ Signature analyzer project] @ EEVBlog Forum
* [https://sigrok.org/wiki/Protocol_decoder:Signature Signature Analysis support in sigrok]
* [https://sigrok.org/wiki/Protocol_decoder:Signature Signature Analysis support in sigrok]
Line 50: Line 51:
==Tektronix products designed for signature analysis==
==Tektronix products designed for signature analysis==
* [[468]]
* [[468]]
* [[1240]], [[1241]]
* [[5D10]]
* [[5D10]]
* [[7854]]
* [[7854]]
Line 58: Line 60:


==Links==
==Links==
* [https://hpmemoryproject.org/an/pdf/an_222.pdf HP Application Note 222 A Designer's Guide to Signature Analysis]
* [https://hpmemoryproject.org/an/pdf/an_222.pdf HP Application Note 222, ''A Designer's Guide to Signature Analysis'']
** [https://www.hpmemoryproject.org/an/pdf/an_222-1.pdf AN-222-1, ''Implementing Signature Analysis for Production Testing with the HP 3060A Board Test System'']
** [https://www.hpmemoryproject.org/an/pdf/an_222-2.pdf AN-222-2, ''Application Articles on Signature Analysis''], 1979
** [https://www.hpmemoryproject.org/an/pdf/an_222-3.pdf AN-222-3, ''A Manager's Guide to Signature Analysis''], 1980
** [https://www.hpmemoryproject.org/an/pdf/an_222-4.pdf AN-222-4, ''Guidelines for Signature Analysis – Understanding the Signature Measurement'']
** AN-222-5, ''Increasing Productivity in Manufacturing and Service with a Logic Troubleshooting System'', 1982
** [https://www.hpmemoryproject.org/an/pdf/an_222-6.pdf AN-222-6, ''Troubleshooting with Composite Signatures''], 1983
 
* [http://gamearchive.askey.org/General/Test_Equipment/HP/SigAnalNotes.pdf Troubleshooting Microprocessors with "Free Run" Fixtures] /  [http://gamearchive.askey.org/Video_Games/Manufacturers/Atari/signatures/6502NOP.txt A 6502 NOP adapter]
{{PatentLinks|Signature analysis}}
{{PatentLinks|Signature analysis}}



Latest revision as of 05:58, 14 October 2024

Signature Analysis is a method to simplify (field) troubleshooting of complex digital systems, developed at Hewlett-Packard in the early 1970s.

The idea is to describe "known good" states of signals in the system – typically when in some diagnostic / troubleshooting mode – as a four-digit "fingerprint" of that signal, when referenced to a known clock and defined start/stop conditions. The concept was introduced in Hewlett Packard Journal May 1977 along with their first SA instrument, the HP 5004A.

The signature algorithm is a 16-bit cyclic redundancy check that can be efficiently implemented in hardware using a linear-feedback shift register. HP selected the polynomial x16+x12+x9+x7+1 instead of a standard CRC-16 or CCITT-16 polynomial, in order to scatter the missed errors as much as possible and to avoid selecting feedback taps that are evenly spaced or four or eight bits apart because the types of instruments [...] that we will most frequently be testing tend to repeat patterns at four and eight-bit intervals.

They also chose a particular "funny hex" format to display the resulting 16-bit number (four hexadecimal digits) to be read easily on seven-segment displays, as 0–9,A,C,F,H,P,U instead of 0–9,A,B,C,D,E,F. (Presumably, including "H" and "P" in the display was an extra benefit.)

Signature analysis was only popular for a fairly short time in the late 1970s and early 1980s before increasing integration made it ineffective. Some arcade game manufacturers used signature analysis to allow component-level repair by field technicians.

It is also possible to use signature analysis in a limited way in equipment that has not been designed for it, by inserting a "free run" or "NOP" adapter between the CPU and its socket. This adapter is wired to present a "no operation" opcode to the CPU, which then loops through all its addresses, generating specific and constant signatures on address lines, allowing these lines in the circuit to be tested or identified (see reference below).

to be expanded

History

Dennis Tillman says:

Signature Analysis was a very interesting troubleshooting concept that HP developed. It had several important benefits:

  • It was easy and inexpensive to incorporate this concept into products under development.
  • It made it possible for anyone, without prior knowledge or experience with digital electronics, to troubleshoot the new digital circuitry that was rapidly appearing in electronic instruments.
  • HP freely shared the concept so others could incorporate Signature Analysis in their products.

Because Signature Analysis was developed by HP, Tektronix would never use it. This was the word from the very top at Tek. [...]

In 1978 Tek was developing a radically new analog/digital scope that had a microprocessor, A/D converter, digital memory, programming capability, and a GPIB interface (also originally developed by HP as the HPIB). All of this complex digital circuitry was going to make these scopes virtually impossible to be repaired in the Field Office Service Centers.

Tom Rousseau, the 7854 Project Manager, realized this was going to be a huge problem for the Service Centers. When Tom read the May 1977 Hewlett Packard Journal which was entirely devoted to Signature Analysis, Tom realized this concept would make it much easier for the Service Centers to troubleshoot the 7854.

But there was a problem: Howard Vollum was adamantly opposed to using anything from HP. Tom eventually got Howard to budge a little when Tom made it clear how difficult this scope would be to troubleshoot in the service centers without a massive training program for all the field technicians.

Howard reluctantly agreed to allow Tom to incorporate Signature Analysis in the 7854 provided Tek did not have to buy the HP 5004A Signature Analyzer. Unfortunately there were no other signature analyzers.

As a result Tek developed their own 067-1090-00 Signature Analyzer single-wide plugin for the TM500 mainframe. It was later renamed to the SA501.

Signature Analyzers

Tektronix products designed for signature analysis

Links

Patents that may apply to Signature analysis

Page Title Inventors Filing date Grant date Links
Patent US 3976864A Apparatus and method for testing digital circuits Gary B. Gordon George A. Haag Jan R. Hofland Daniel I. Kolody 1974-09-03 1976-08-24
Patent US 4192451A Digital diagnostic system employing signature analysis Stephen Swerling Steven R. Smith Thomas W. Wiesmann 1978-05-30 1980-03-11
Patent US 4527272A Signature analysis using random probing and signature memory Michael G. Reiney 1982-12-06 1985-07-02
Patent US 4752928A Transaction analyzer David D. Chapman Donald C. Kirkpatrick 1985-05-06 1988-06-21