dsox2000/3000 serial decode

[georges80]

Having recently purchased an msox2024a I can say it's an impressive scope (especially at agilent's ebay store pricing).

Anyhow, I've played a bit with the serial decode and can't imagine using it long term with scope probes hanging all over a project.

Given I have the msox version and it comes with the typical excellent logic analyser cables/clips it seems daft to have to use the analog probes for serial decoding.

So, to that end I have designed and sent off to fab a panel that includes the layout below. Essentially it is just an adapter (with 4 bnc vertical pcb mount females) that will take 2/3/4 bnc cables (not the scope probes) and properly terminate them for interface to the standard agilent 20x2 logic analyser female connector.

The termination in conjunction with the impedance of the analyser probe tips creates a divide by 10, so compatible with the divide by 10 scope probes and it means you don't have to keep changing the channel settings each time you switch from the analog scope probes and the analyser.

Should make probing/decoding serial channels on boards much easier than with the lumbering analog probes.



I should receive boards in a week or so and will post up some pics of the 'real' thing. BNC's and 20x2 shrouded header should be in my hands today.

cheers,
george.

[Arcticas]

Is there any reason why you did not fill the bottom side with a continuous ground plane?

[georges80]

Seemed to be no need for serial decodes that are running at a few hundred kbits per sec max at least in my projects.

cheers,
george.

[PlainName]

So, to that end I have designed and sent off to fab a panel that includes the layout below.

Great idea, and nice job! 

[Arcticas]

Seemed to be no need for serial decodes that are running at a few hundred kbits per sec max at least in my projects.

Fair enough. But it would have been very easy to add a plane on the bottom side. This improves the ground impedance a lot and thus improves signal integrity also at low speed. By the way, it is not the frequency that counts but the edges.

Also the production gets more ecological as less copper has to be removed.

So I would always use a solid ground plane, if there is no important reason against it.

[Hydrawerk]

MSOX2024a cannot do serial decoding on digital channels.

[georges80]

I wouldn't just add ground planes to everything expecting them to always 'help'. Without an impedance match they don't necessarily help - in this case the scope inputs are nominal 1 meg, so trying to get 50 ohms or whatever on the PCB won't help improve anything. The logic analyser probe connector is certainly not 50 ohms either.

To get an impedance match requires knowing the spacing from the trace to the ground plane AND the Er of the PCB material and then you can control the trace width (and spacing for differential pairs). I've gone through this with PCIe differential pairs and DDR3 clock/control differential pairs etc - it requires working with the PCB house to control layer spacing and FR4/PrePreg Er.

I've had ground planes introduce noise into circuitry and in some designs cutting the ground plane under certain traces reduces noise dramatically. So, just saying it isn't necessarily good to just flood ground plane everywhere just because there's open space.

I won't be worrying about 'excess' copper being etched - that copper (in solution) is recovered/recycled/reused. I'm not making these boards in my kitchen...

cheers,
george.

[georges80]

MSOX2024a cannot do serial decoding on digital channels.

Yes, exactly why I've built this board. To allow me to use the logic analyser probes with the ANALOG channels of the scope...

cheers,
george.

[Hydrawerk]

Oh, OK, now I understand it.

[georges80]

Protoboards arrived today, so cut out the adapter. I had the parts on hand waiting:



Soldered up and plugged the msox LA cable in, all nice. I used a polarised socket so no chance of plugging in the LA cable incorrectly:



Test fit with 2 coax cables, all good to go:



This will definitely make it much nicer to hook things up for serial decode.

cheers,
george.

[georges80]

And a few pictures etc of it in use to prove it works





and a capture of the signature read across the AVR ISP (spi) bus. Signature 0x1e, 0x93, 0x0d (attiny861)



cheers,
george.

[PlainName]

Nice!

But... I thought this was going to plug into the digital port so you get 8 channels of digital data, 4 of which you can also route to the analog ports for decoding.

[georges80]

Nope, I just need to be able to decode SPI/I2C etc. So, it's just a way to get nice LA style probes into the analog ports for the hardware decoder functions. I'd rather clip the LA probes onto serial buses versus the ungainly scope probes.

AFAIK you can't use the digital ports AND do serial decode at the same time anyway...

Straight from the dsox2000 manual:

"Digital channels and serial decode cannot be on at the same time."

cheers,
george.

[Rigby]

That is a very odd limitation on the scope's part.

[Hypernova]

It's a price tier thing, on my 3000X there are no such restrictions.

[AndyC_772]

My money would be on it being a silicon bug in the ASIC. Surely nobody would deliberately design an instrument with serial decoding, and digital inputs, but not the ability to use them together...?

As restrictions go, that one's just plain weird, and has to come from a hardware bug IMHO.

[Hydrawerk]

That is a very odd limitation on the scope's part.

Well, Agilent wants us to buy the MSOX3000.