GPIB chain, must all instruments be ON?

[DIPLover]

Hi guys,

I was happily using my prologix usb-GPIB controller to get data from a Keithley 196, and images from my Tek TDS784.

I had a single GPIB cable running to both instruments and the controller plugged into the cable, at the back of the K196.

Yesterday I decided I'd like to do sweeps and plots with my HP8903B so I daisy chained it to the tek scope.

GPIB chain is now :
K196 with controller piggybacked <-> Tek 784 <-> 8903

In the past, If I wanted to play with just the K196, I turned it on and talked to it. Same with the Tek.
Now with the daisy chain, I can't talk to the Tek scope unless the 8903 is turned ON. Otherwise I get no response from the controller. As soon as I turn on the 8903 I receive the missing responses so the command at least got to the controller.

It's the same for the Keithley, it needs the 8903 to be ON, otherwise I get no responses. Having the Tek ON at the same time as the K196 makes no difference.

Is this expected behaviour? Should the last instrument in a chain be ON for communication to work? Should they ALL be ON normally?

[oPossum]

The Prologix adapter does not have GPIB drivers so it really should only be used with one instrument.

Proper GPIB drivers are designed to not load the bus when they are unpowered so everything should still work. My experience is that in practice if there are enough devices on the bus then most of them have to be powered up for it to work. Some instruments have drivers that load the bus when unpowered so they always have to be on.

[KE5FX]

IIRC, the IEEE 488 standard calls for no more than 15 device loads on each bus, with at least two-thirds of them powered on.  So this is indeed a consideration.

In practice, some instruments are much better than others in that respect.  While I haven't used the 8903, the 8656/8657 signal generators are major offenders.  Any one of those models will load down the bus if powered off.  In my experience it doesn't matter how many other instruments are powered on, and it doesn't matter if you're using a Prologix adapter or an NI interface.  Meanwhile, the 8662A/8663A signal generators can be left powered off with no problems regardless of the state of any other instruments on the bus, along with most of the other HPAK equipment I've used (which is a lot).

It sounds like the 8903 may have used the same GPIB interface hardware as the 8656/8657.  If you can't leave it powered up, then nothing else can easily be done except to give it a separate bus controller.

[DIPLover]

Good to know that having a number of devices on is part of the specification.

For now, since the 8903 is last in the chain, I'll just disconnect it until I want to use it.

The prologix adapter works very well on linux by the way and is very easy to use in programs. I really don't want to deal with NI VISA (or anything else by National Instruments really).

What do you guys use, Agilent 82357B?  I see lots of chinese... let's call them "clones"... on eBay for very cheap. Complete with "Functional Test Ceryificate". Do they work?

[oPossum]

I designed and made my own because there wasn't any low cost USB to GPIB available with proper GPIB drivers and USB MCU.

[windsmurf]

http://www.compendiumarcana.com/gpib
I got a 404...

[oPossum]

Still working on the firmware, so the web page does not exist yet. Sorry.

[DIPLover]

Well it may not be “proper”, but the prologix has been working perfectly for me these past 5 years...

[TheSteve]

Good to know that having a number of devices on is part of the specification.

For now, since the 8903 is last in the chain, I'll just disconnect it until I want to use it.

The prologix adapter works very well on linux by the way and is very easy to use in programs. I really don't want to deal with NI VISA (or anything else by National Instruments really).

What do you guys use, Agilent 82357B?  I see lots of chinese... let's call them "clones"... on eBay for very cheap. Complete with "Functional Test Ceryificate". Do they work?

My favorite(by far) is the Keysight E5810A - regular price isn't great but good deals do pop up if you're patient.

[Kleinstein]

Ideally GPIB should work with some instruments turned off. However it does not work with all instruments.

It does not make much difference were in the chain the instruments are - they are all wired in parallel. It may be an issue at higher speed if there are lots of cables. 3 cables should be OK in essentially any configuration including a triangle  .

[MarkL]

To be more exact about this, IEEE 488 does allow less than 2/3 devices powered on as long as the signal lines are not degraded.  In other words, "it depends".

From 488.1-2003:

8.2.4 Devices powered off and on

A system will operate without adversely affecting normal data transfer with at least two-thirds of the devices powered on. A system will operate correctly with any number of devices powered off, provided all of those devices powered off do not degrade the specified high state condition, that is, that the voltage on each signal line with all its output drivers passive false should exceed +2.5 V with respect to the logic ground at each device.

Unless special precautions are taken (that is, use of special driver circuits beyond the scope of this standard), powering a device to on while the system is running may cause faulty operation.

Using an old NI PCI card, I routinely have only 1 device powered on out of 11 in a random topology and everything works fine.

[macboy]

GPIB data and control (handshaking, attention, interrupt, etc) lines all float high and are driven low, using open-collector type drivers. (There are some exceptions, with certain "newer" instruments that support high speed signalling and use Tri-state drivers which drive the data bus high and low). The high signal level is created entirely through a termination network on each connected instrument that is nominally a 3.1K pull-up (to 5 V) and 6.2k pull-down resistor to ground. This creates a nominally 3.3 V signal on the line when floating. The signal levels are TTL, so a "high" is anything over 2.0 V. A low should be below 0.8 V.  (It's worth noting that GPIB uses negative logic on all lines, so > 2.0 VDC is interpreted as 0/false, not 1/true).

If you have multiple unpowered devices, then the pull-up resistors are not pulling the bus to 5 V, but dragging it down to 0 V instead. This results in the floating signal level falling below the logic high threshold. This problem is made worse by the lack of proper GPIB drivers (with TTL level inputs) on the prologix adapter.

You may be able to work around the issue by installing extra pull-ups on the prologix controller. Maybe around 1 K, connected to 5V. Add these to the 8 data lines, and all 8 control signals (DAV, NRFD, NDAC, ATN, IFC, REN, SRQ, EOI).  This is pins 1 to 11 and 13 to 17 on the GPIB connector.

[mark03]

What do you guys use, Agilent 82357B?  I see lots of chinese... let's call them "clones"... on eBay for very cheap. Complete with "Functional Test Ceryificate". Do they work?

My favorite(by far) is the Keysight E5810A - regular price isn't great but good deals do pop up if you're patient.

Going even older, don't forget the E2050A.  I got mine for about the same price as a Prologix or 82357B (I think I paid $100).  I find the LAN adapters very convenient, as I can easily access my instruments from any computer in the house, over the wireless network, etc.  Best of all, the python-vxi11 project allows working directly from python without installing any VISA crud.