New Agilent scopes

[EEVblog]

Just a little bit of info - while there seems to be references to serial decode/trigger licenses for the 2k series in the firmware, it doesn't seem to work if you make a license. I guess they're disabled elsewhere.

Yes, I can confirm that trying to enable serial decode on the 2000 series does not work.

Dave.

[mikeselectricstuff]

(note the extra condition if (YYY) for serial in the 2000 case)

I don't know what that YYY is for sure yet and if you can enable it somehow (else than forcing the if by modifying the code of course)

I wonder if this could be FPGA size.  Perhaps serial decode needs the larger FPGA of the 3000, which they might decide to fit on the 2000 in future. From memory, the only main logic functionality changes between the 2000 and 3000 are decode and hardware frequency counter, and a frequency counter doesn't exactly need much in the way of FPGA resources.

[mikeselectricstuff]

These Windows CE scopes actually worry me.  I don't understand them.  I want the same functionality but based around low integration logic ICs or at least FPGAs. 

Just because you don't understand something doesn't mean it's not the right tool for the job. Would you prefer if it was all written in interpreted BASIC?
WinCE is specifically designed to run on resource-limited systems like PDAs from  ROM - it's a very different animal to desktop Windoze, although even the latter has an embedded flavour that can run from read-only media

There must be a Registry somewhere in there growing slowly.

Why? There is no reason for a system like this to be writing to 'disk' in a way that has a risk of messing things up (bugs notwithstanding..). The scopes are only running a single application, without all the hassles of umpteen different drivers etc. that a general-purpose PC has to tolerate.

Given the choice of an obscure system like the VxWorks used on earlier Agilent scopes and a more mainstream OS for which all sorts of 'investigative' tools and knowledge is available I know which I'd prefer...

[_Sin]

I wonder if this could be FPGA size.  Perhaps serial decode needs the larger FPGA of the 3000, which they might decide to fit on the 2000 in future. From memory, the only main logic functionality changes between the 2000 and 3000 are decode and hardware frequency counter, and a frequency counter doesn't exactly need much in the way of FPGA resources.

Could be. Has anyone established what the FPGA is in the different models? IIRC it was under a heat-sink in Dave's teardown.

[mikeselectricstuff]

I wonder if this could be FPGA size.  Perhaps serial decode needs the larger FPGA of the 3000, which they might decide to fit on the 2000 in future. From memory, the only main logic functionality changes between the 2000 and 3000 are decode and hardware frequency counter, and a frequency counter doesn't exactly need much in the way of FPGA resources.

Could be. Has anyone established what the FPGA is in the different models? IIRC it was under a heat-sink in Dave's teardown.

No it wan't - the FPGA types were identified in both reviews and  the 3000 had ISTR one with about twice the gate count.

[_Sin]

No it wan't - the FPGA types were identified in both reviews and  the 3000 had ISTR one with about twice the gate count.

Ah yes. I was thinking of the two/four custom chips.

Interesting there's a software check for some aspect of the serial decode capability on the 2k units (and licenses) though...

[Leo Bodnar]

Interesting there's a software check for some aspect of the serial decode capability on the 2k units (and licenses) though...

This is probably a check to see if larger FPGA is installed or not.  Software is usually written before final specs are nailed down.

[_Sin]

I would have expected the serial decode to have been hardware (part of the custom ASIC), rather than the FPGA, but I guess the workload can be shared.

Do we know if the ASICs are the same between the 2k and 3k series (other than the 3k clearly having 2 of everything)?

[EEVblog]

I wonder if this could be FPGA size.  Perhaps serial decode needs the larger FPGA of the 3000, which they might decide to fit on the 2000 in future.

From talking with the head Agilent scope guy, I was lead to believe the 2000 as-is is perfectly able to do serial decode (built into the Magazoom ASIC), it is just sitting there waiting to be enabled if market forces dictate it is required.
Agilent would have done their homework and built the capability in in the release version board.

Dave.

[EEVblog]

I would have expected the serial decode to have been hardware (part of the custom ASIC), rather than the FPGA, but I guess the workload can be shared.

Do we know if the ASICs are the same between the 2k and 3k series (other than the 3k clearly having 2 of everything)?

Yes, I asked and was told directly that it is exactly the same Megazoom ASIC. So the 2000 is quite capable of doing 1,000,000 wfm/s, serial decode, 4M sample memory etc. The 2000 is definitely software crippled to meet market price points.

Dave.

[_Sin]

Thanks for the update Dave... fingers crossed they enable that stuff sooner rather than later!

Mind you, the damn thing is pretty awesome as-is, and I've barely had a chance to use it in anger yet.

[Wim_L]

These Windows CE scopes actually worry me.  I don't understand them.  I want the same functionality but based around low integration logic ICs or at least FPGAs. 
There must be a Registry somewhere in there growing slowly.  And you don't get Win CE reinstall CDs with them!  

Not really a problem, except for boot times perhaps. They're running Windows, but on a singel hardware configuration, preloaded with only the necessary drivers and programs to make the device work, so it's expected to be more reliable than a desktop PC which might see frequent configuration changes.

One thing that might be a concern is safety on the network, if it's vulnerable enough to become part of a botnet. It seems unlikely someone would bother to write attack code specifically targeted to oscilloscopes (just not enough of them around compared to PCs to make it worth it unless you hav very specific targets like the Stuxnet worm), but the scopes might have some attack surface in common with mobile phones and end up being infected anyway. All depends on how exposed it is on the network.

[joelby]

They're running Windows, but on a singel hardware configuration, preloaded with only the necessary drivers and programs to make the device work, so it's expected to be more reliable than a desktop PC which might see frequent configuration changes.

Remember that it's Windows CE, which is an entirely different beast to your desktop's Windows.

[Rufus]

Been lurking here a while and thought I should join in and say thanks for the discussion and clues.

Couple of observations I don't think I saw mentioned. The 20110316 update changed some of the displayed names for options, some don't match the user guide any more.

When you upgrade BW or to MSO does the model name change to reflect the update? Maybe that is why 'that' tool was embedded in the update.

Someone asked about MSO pinout. Just measuring the pins on a 3k looking at the scope it looks like the bottom row is all ground, top left open, 16 channels then 3 more opens. The inputs measure 10k to ground. Probes (as someone said) will be 90k in series, possibly split and I imaging a bit of compensation capacitance. The User Guide has a bit about probe frequency response which gives you a clue about what must be inside. Would still be nice to see the inside of a pod.

[tnt]

Does it look like this :

http://cgi.ebay.com/HP-Agilent-54620-61801-54620-61601-54620-68701-POD-KIT-/330346126103

?

(those are insanely expensive for just cable and a 90k !)

[bandtank]

Yes. I've used the cables and that is what they look like. They are very high quality so it makes sense to me that they are expensive. The weaved wire is the best you can get. I actually know the guy who invented it.

[oPossum]

There are tiny wires woven into the fabric. Amazing!

The resistors are in the end of each probe cable, not the pod.

[oPossum]

2 ground wires beside each signal

[tnt]

Neat. Is that the actual cable shipped with your unit ?

On another matter, I'm wondering if the 1152A active probe is compatible with this scope. It's been obsoleted by Agilent officially and isn't listed in the compatible probes in the user manual but it should be a standard "autoprobe" interface ...

It's hard to find the newer probes ( 1156A / 1157A / 1158A / N2795A ) cheaply. While on the other hand you can grab a 1152A for ~ 300 $ which is a pretty good deal I think for a 2.5GHz active probe ...

[bandtank]

Yes, that is the probe that ships with the unit. A guy at Agilent invented that cable because he was interested in weaving as a hobby and realized one day it could work well. The probes have fantastic electrical characteristics. We use them at work on 1000+ channel logic analyzers

[Leo Bodnar]

You can imagine the process of making and QCing them.  Hence the cost.  Trawl eBay!
How do you test ground shield wires continuity during QC?  There are two GND wires between two signal ones. The only way I can think of is HF cross-talk between the signal wires.

When I got my cables with MSO-X 3024A I had a good look at them and actually realised how they are made.  You can see the wires going through the fabric without taking the unit apart.

oPossum, how did you remove the rivets?  

[tnt]

Trawl ebay ?

I did, and the cheapest was 250-300$ :p
There is sometime some at 150$, but when you read the description, it's just the pod, not the flying leads and the grabber. Getting those speparately brings you back to 250-300$

For the 2000X, you might has well buy the real option than trying to mess around.

[Leo Bodnar]

I have a question regarding pods.  Would it be possible to use standard 40 pin IDC cable for low frequency (e.g. 30MHz) digital designs?

It would be nice to have standard in-house debugging connector on prototype revision PCBs that you can just plug the IDC cable in and the other end goes into the scope.
I don't fancy mucking around with 18+ grippers plus 4 analogue probes plus head scratching of channel allocation.  It might be even possible to make a breakout PCB that has a high density 0.05" socket if prototype PCB space is a problem.
So...

[ MSO-X3000A ]~~~~< 40 pin IDC cable >~~~~~~[ 40-pin-0.1"  breakout-PCB 18-pin-0.05"]===[18-pin-0.05" development PCB ]

18 or 20 pin 0.05" will do.

You plug it into your design and boom!  All the digital channels are connected and ready for debugging in predetermined order.

I am not sure if you can feed TTL levels straight into the scope without 90k inline resistors.  They can be mounted on the breakout PCB in any case.

[oPossum]

The pictures I posted are from a broken cable assembly that I got off ebay many years ago. The other end of the cable is missing. I think I just drilled out the rivets.

80 conductor ATA (hard drive) cable may work for a DIY probe system. The 90k resistors would be required for proper threshold levels.

[tnt]

A lot of boards have "SAMTEC" connectors just for that. Seems to be sort of a standard to plug logic analyzers. Agilent has cable for that.
And for the analog channels, you also have "probe sockets" that you can put on your board and litterally "plug" the probe in it.

But as you said, you can probably roll your own, just putting 90k resistors arrays on your test boards and using a 80pin ATA cable that has extra ground wires should work fine.