wrong UNT

[daniel_schwarz]

Hello,
i've got a problem with untalking my device. It seems like, each time the controller addresses my devices, it sends UNL and UNT at first. UNL is send correctly as B'11000000' (inverted). But UNT is sent as B'10111111' (inverted). This would mean, that UNT addresses TAD 0 ?!?!. Does anyone have an idea why this happens ? At first i thought, that this is a failure of my NI card, but another INES card does it the same way. Very curious. May there be a difference between 488.1 and 488.2 ???

[ScottieB]

Is this actually causing problems, or are you just confused when you look at the lines?  The values you're getting for UNT do seem incorrect.  UNL should be 0011 1111 (DIO8 to DIO1) and UNT should be 0101 1111 (DIO8 to DIO1) where '1' represents a voltage potential near 0 V and the '0' is higher (3.5 V or 5V or something like that).  If you're measuring this with a logic analyzer that doesn't understand GPIB, you need to realize that the data is only required to be correct while DAV is asserted (grounded).

There's no difference between 488.1 and 488.2 in this regard.  Again, is this actually causing problems?  I'm guessing that if 2 different cards are doing this it's a problem with your measurement rather than the cards themselves.

Scott B.
GPIB Software
National Instruments

Message Edited by ScottieB on 02-27-2006 10:13 AM

[daniel_schwarz]

I've implemented GPIB in software on my PIC-Controller, so it isn't very difficult to adapt to the incorrect UNT. I've measured the values for UNT and UNL with my DSO while triggering on falling edge of DAV, and the value for UNL seems to be correct. Paradoxically UNT is sent correct if i make a scan for instruments with MAX. Unfortunately the controller isn't able to find my device, maybe because of my adaption of UNT. UNT is also sent correct if i use ibic in 488.2 mode (using set 488.2 0), and communication between CIC an my device functions well. Very curious.

[GPIB_Guru]

If it works in 488.2 ibic and it works in MAX, how are you sending it so that it doesn't work? 488.1 ibic? Board-level or device level? What exact string are you sending?

[daniel_schwarz]

At first i initiate my gpib controller with ibdev 0 3 0 10 1 0
Then i send for example ibwrt "*IDN?". The wrong UNT appears also when i get the identification string from my PIC using ibrd. Very mysterious is, the when using ibrd the UNT appears as 11011111 (low active), and when using ibwrt UNT appears as 10111111 (low active). 

[GPIB_Guru]

---

@daniel_schwarz wrote:
ibdev 0 3 0 10 1 0
ibrd the UNT appears as 11011111 (low active)
ibwrt UNT appears as 10111111 (low active).

---

Mystery solved (hopefully). What you are seeing is not an UNT command but a real talk command. The NI-488.2 driver does not send UNT when doing device-level calls. The IEEE 488.1 standard states that all devices shall interpret a talk address that is not their own as an UNT (in 488.1 lingo, this is OTA, or other talk address). Therefore, when NI-488 makes a device level call, it sends the following command bytes: MTA(talker) UNL MLA(listener) (I am not 100% sure about the order, but those are the three command bytes that are sent across the bus).

In the example you listed above, you are talking to a device at address 3, so when you try to do an ibrd, the talker will be address 3 and the listener will be address 0 (the controller), so the command bytes will be (using logic, not active levels) 0x43 0x3F 0x20 (MTA3 UNL MLA0). When you do an ibwrt, the talker will be address 0 (the controller) and the listener will be address 3 and the command bytes will be 0x40 0x3F 0x23 (MTA0 UNL MLA3).

The two bytes you are showing up there are 0x20 and 0x40, which are MLA0 and MTA0 respectively and are part of the addressing scheme.

Does this make sense?

Message Edited by GPIB Guru on 02-28-2006 07:55 AM

[daniel_schwarz]

Thank you very much !!! This sounds really plausible.