Skip to main content ↑ ↓
Topic: OLS won't trigger, signals always low? (Read 2193 times) previous topic - next topic

OLS won't trigger, signals always low?

I received this bug report via email:

Quote
Hello,
   Tried using sniffer for the first time today. All software installed
okay. I was trying to examine 7 signals off an AVR and could not get it to
trigger properly. Also, when I did get it to trigger all signals were low
(never seen a high). All logic is at 5 volts. Finally tried a simple test.
Wired a 4.7K  pull up to channel 1 and set the trigger to fire on low (Value
unchecked). Sniffer immediately triggered when I armed it (Input was high).
The trace showed channel 1 to be low for the whole sample period. I have
used analyzers at work for 30 years so I don't think it's "operator error".
Please advise.
Got a question? Please ask in the forum for the fastest answers.

Re: OLS won't trigger, signals always low?

Reply #1 –
The description of the setup doesn't contain enough information to point in a specific direction for help.

We'd have to know what firmware version/bitstream release is installed (was the OLS upgraded to
firmware v05 or v06 and the bitstream to release 2.04)? If so, is the latest SUMP Java client analyzer.jar
being used (it's part of the release 2.04 bitstream upgrade - but don't use the firmware in this package!)
and has the self-test been performed?

The OLS will work properly only if
- firmware v05 or v06
- bitstream release 2.04 and
- the latest SUMP Java client (no way to check this - only way is to remove all analyzer.jar files from the system and to install analyzer.jar and run_java_client.bat from the release 2.04 bitstream upgrade)
are being used!

For details on the channel (group) assignments for the outside and inside numbering scheme see attached pictures.

Test mode is described in more detail here: client Test Mode

Only bitstream release 2.04 in combination with the latest SUMP Java client (see above) support
- inside/outside numbering schemes (switching channel groups between the unbuffered I/O header and the buffered header)
- test mode and
- 16channel/8k mode (select by unchecking channel groups 2 and 3)

RLE is not supported at present - at least it does not work properly!
 

Re: OLS won't trigger, signals always low?

Reply #2 –
It always wondered me why there was an inside and outside numberingscheme? I'm not very experienced with logic analyzers (actually it is my first ;)) and dunno if this is common in LA land.

Nice drawing, Ipenguin.

Re: OLS won't trigger, signals always low?

Reply #3 –
Fantastic illustration. We need to include that somewhere prominent.
Got a question? Please ask in the forum for the fastest answers.

Re: OLS won't trigger, signals always low?

Reply #4 –
Sjaak & ian, thank you. I had prepared the pic for a more complete OLS guide ... neither the illustrations nor the text are complete, yet ... but I thought releasing this page might help to end some confusion about the OLS' probe headers, channels and the numbering scheme.
ian, consider it released under GPL ... I will make the OpenOffice (source) documents available to you and Jack once they are finished ... or on request "as is" earlier.

The main reasons for the inside and outside numbering scheme are:
- to be able to assign channel groups 0-1 either to the buffered or unbuffered header and channel groups 2-3 respectively to the other header
  (so users that will only need up to 16 channels can decide which header to use - for flexibility mainly)
- to be able to select one of the two headers as channel groups 0-1 for the 16x8K mode

They are called inside and outside numbering schemes because of the number markings next to the headers. The number lines closer to the FPGA depict the pin-channel assignment for the "inside numbering scheme", while the number lines closer to the headers (farther away from the FPGA) depict the pin-channel assignment for the "outside numbering scheme"! (Sorry, if you already knew this but I thought it was rather important to clearify this because it appears to me that there is still some confusion about it when looking at other threads).

Some LA history (quoted from Logic Analyzer on Wikipedia):

Quote
... When logic analyzers first came into use, it was common to attach several hundred "clips" to a digital system.
... Once the probes are connected, the user programs the analyzer with the names of each signal, and can group several signals into groups for easier manipulation. ...

On some of the more advanced LA systems you can essentially connect any physical channel to some signal on the traget system (correctly called SUT - system under test) - more or less randomly - and then sort the channels in a sequence of your preference, giving each channel an individual name and form groups of channels (i.e. 8 channel data bus, 16 channel address bus and a control bus). This helps greatly (in many cases it may be the only way) to analyze the captures and display them in a readable and understandable format.

With the current OLS client we can't yet assign individual names to channels and group channels ... however, the bitstream/hardware configuration option to swap channel groups 0-1 and 2-3 gives us the flexibility to select between 32x4k, 16x8k (maybe later 8x16/24k) sampling mode on both headers ... and in connection with future wing adapters for the unbuffered header to add special features/functions.

Most work will have to be put into the client software to implement more advanced features and protocol/state analysis. We have been looking at the sigrok project to replace the current SUMP Java client ... however, improving the OLS bitstream and releasing a "unified" firmware/bitstream/client package has the highest priority atm ...

[quote author="Sjaak"]
It always wondered me why there was an inside and outside numberingscheme? I'm not very experienced with logic analyzers (actually it is my first ;)) and dunno if this is common in LA land.[/quote]