NEW Keysight HD3

[msuthar]

So this Oscilloscope runs on Linux !

Any information about what is p700  here ?  edited later >>>> https://www.congatec.com/de/produkte/smarc/conga-smx8x/

in side update file     it mentioned   this file   tlo-instrument-weston-image-p700-quad.squashfs   


squashfs  and  filesystem = "vfat";  indicates they are  for file system for some live cd or os for scope.

[EEVblog]

So this Oscilloscope runs on Linux !

I didn't see an obvious serial boot output port. I probed the IDC connector and got nothing during boot.

[tv84]

Some quick peeks:

- "fitimage" parsing attached;
- the CPU board most certainly is the conga-SMX8X that @msuthar indicated above;
- they may be using "The SECO (SEcurity COntroller) is a HSM (Hardware Security Manager) that is available on the iMX 8 and iMX 8X only." to sign and secure the FW.
- the infiniiVisionCore app contains this table:

1 MHz / 16 Bits HD
5 MHz / 16 Bits HD
10 MHz / 16 Bits HD
20 MHz / 16 Bits HD
50 MHz / 16 Bits HD
100 MHz / 15 Bits HD
200 MHz / 14 Bits
350 MHz / 14 Bits
500 MHz / 14 Bits
800 MHz / 14 Bits
1 GHz / 14 Bits
3 GHz / 14 Bits
4 GHz / 14 Bits

[gslick]

So this Oscilloscope runs on Linux !

I didn't see an obvious serial boot output port. I probed the IDC connector and got nothing during boot.

ISL3243 RS-232 transceiver on the bottom side of the conga-SMX8X CPU module connected to the 6-pin connector on the CPU module.

You can buy a congatec 48000023 cab-RS232-Debug-ARM cable from Mouser.
Adapter cable for ARM console. MOLEX PicoBlade 6 circuit to two D-SUB 9 connectors.

Table 10 A35 and SCU Connector (X2) Pinout Description
Pin SoC Ball Description
1 SCU_GPIO0_01 SCU Debug: Transmit signal via ISL3243E RS-232 Transmitter/Receiver connected to UART4_TXD of the SoC
2 +VIN SMARC VDD_IN (+5 V)
3 GND Ground
4 UART2_TXD A35 Console: Transmit signal via ISL3243E RS-232 Transmitter/Receiver connected to UART2_TXD of the SoC
5 UART2_RXD A35 Console: Receive signal via ISL3243E RS-232 Transmitter/Receiver connected to UART2_RXD of the SoC
6 SCU_GPIO0_00 SCU Debug: Receive signal via ISL3243E RS-232 Transmitter/Receiver connected to UART4_RXD of the SoC

https://www.congatec.com/fileadmin/user_upload/Documents/Manual/SX8X.pdf

Also note that if only the outputs of the ISL3243E are probed without supplying valid RS-232 signal levels to any of the ISL3243E inputs, the ISL3243E may be in a low power mode with the outputs disabled.

Code: [Select]

The ISL3243E features an automatic powerdown function that
powers down the on-chip power supply and driver circuits. This
occurs when an attached peripheral device is shut off or the
RS-232 cable is removed, conserving system power
automatically without changes to the hardware or operating
system. It powers up again when a valid RS-232 voltage is
applied to any receiver input.

https://www.renesas.com/en/document/dst/isl3241e-isl3243e-datasheet

[ballsystemlord]

So, do you think the scope can be "hacked"?

[msuthar]

1.conga-SMX8-X/i-QXP-2G eMMC16  >  this is 2 GB Module

2.conga-SMX8-X/i-QXP-4G eMMC16  >  this is 4 GB Module

which one is used in HD3

Can any one confirm ?

[EEVblog]

1.conga-SMX8-X/i-QXP-2G eMMC16  >  this is 2 GB Module
2.conga-SMX8-X/i-QXP-4G eMMC16  >  this is 4 GB Module
which one is used in HD3
Can any one confirm ?

Doens't say directly on the sticker:
https://www.flickr.com/photos/eevblog/53984874289/in/dateposted-public/

[rteodor]

So, do you think the scope can be "hacked"?

I did not bothered to check if this particular iMX8 has System Controller on it. If it has and is well configured hacking it could be very hard.
(SC is a Cortex-M0+ security core that starts first and alone configures and starts other application cores. It's in charge of basically all the initial security and as little as I had to deal with it, its a stubborn bitch).

[tv84]

Can any one confirm ?

Very close but none of those. Probably a specific combination of parts, as Congatec also says that there could be 1,000s of combinations.

conga-SMX8-X/i-QXP-4G eMMC16 (P/N 051110)
conga-SMX8-X/i-QXP-2G eMMC16 (P/N 051113)

The P/N that Dave shows is 651100 rev B.2.  (an obvious difference is the absence of the 2nd GEth controller)

Some resources in the web about this type of boards here.

Datasheet, user's guide and full pin assignments attached.

So, do you think the scope can be "hacked"?

Rhetorical question.

[nctnico]

1.conga-SMX8-X/i-QXP-2G eMMC16  >  this is 2 GB Module
2.conga-SMX8-X/i-QXP-4G eMMC16  >  this is 4 GB Module
which one is used in HD3
Can any one confirm ?

Doens't say directly on the sticker:
https://www.flickr.com/photos/eevblog/53984874289/in/dateposted-public/

Just go to the Micron website and decode the number of the memory chip.

[porter]

credit for the lovely power supply perhaps goes to TDK

https://product.tdk.com/en/search/power/switching-power/ac-dc-converter/info?part_no=CUS400M-12

[jusaca]

credit for the lovely power supply perhaps goes to TDK
https://product.tdk.com/en/search/power/switching-power/ac-dc-converter/info?part_no=CUS400M-12

Is 2x MOPP standard for bench equipment...?
I mean, I'm not complaining...

[tv84]

Nice find. Totally agree with their approach: leave the necessary components to those who do it better (CPU daughter card, PSU, etc.) and use all the internal engineering expertise in improving their own ASICs. No need to reinvent the wheel.

[ballsystemlord]

I'm just surprised how many pieces of high end electronics equipment end up using really cheap capacitors.

[nctnico]

Nice find. Totally agree with their approach: leave the necessary components to those who do it better (CPU daughter card, PSU, etc.) and use all the internal engineering expertise in improving their own ASICs. No need to reinvent the wheel.

The power supply is a bit of specialist thing to design properly with all the approvals you need. But the CPU board is not difficult for Keysight to integrate into the mainbaord if Keysight wants to. I have a few IMX8 designs under my belt so I'm very aware of the skill levels required for doing this: NXP has excellent documentation and design support in place to guide a skilled EE through every step and avoid pitfalls. I have no doubt Keysight has the tools and ability to design the IMX8 into the mainboard succesfully. Using modules is always iffy and rather expensive. Modules are also subject to changes which can break software or even hardware compatibility. Worst case a module is discontinued without a last-time-buy. For volume production a full-custom design is typically cheaper from 100 to 150 units already. All in all I think the hardware of the HD3 Dave has is still in some kind of final prototype state awaiting further integration.

[tv84]

All in all I think the hardware of the HD3 Dave has is still in some kind of final prototype state awaiting further integration.

Don't think so. I bet on they having the possibility of upgrading the daugther board later on or, with this way, simply to have a better defined frontier between their SW and HW teams.

Also, for a company that can merge their own IMX8 design into their mainboard, I don't see any trouble in them designing their own daughter board should Congatec disappear with their offerings.

[nctnico]

All in all I think the hardware of the HD3 Dave has is still in some kind of final prototype state awaiting further integration.

Don't think so. I bet on they having the possibility of upgrading the daugther board later on or, with this way, simply to have a better defined frontier between their SW and HW teams.

That sounds great in theory but I have never seen a module manufacturer pull this off. Creating 100% compatible modules with different SoCs. There are always differences between the SoCs which lack overlap in features (which is kind of the point of having different types of SoCs). With a bit of luck you might be able to add features using extra peripheral chips. If you look at the Congatec module, you'll notice the TUSB8041. That is a USB3 hub chip to turn a single USB3 interface from the SoC into multiple ports. Likely Congatec tried to be compatible with a different module which has more USB3 ports. With the hub chip the module has the same number of ports, but not the same bandwidth as several dedicated USB3 host ports on a SoC would be able to provide.

Really, there is no upside to using SoC modules in volume products long term. No matter what promises manufacturers make. Been there, done that. Always ends up costing more time (=money) than expected down the road.

[EEVblog]

All in all I think the hardware of the HD3 Dave has is still in some kind of final prototype state awaiting further integration.

I greatly doubt that. They would have been working in this for years. No way my unit, just barely getting to me less than 24 hours before public release has a prototype module arrangement on it.
To change it you'd have to redo all the EMC compliance testing for one thing.

[EEVblog]

Really, there is no upside to using SoC modules in volume products long term. No matter what promises manufacturers make. Been there, done that. Always ends up costing more time (=money) than expected down the road.

Keysight also did this on the low cost 1000 series, although I think that would was their own design. But still, it was a processor (+FPGA's ASIC's) on a daughterboard design.
https://www.flickr.com/photos/eevblog/32798748040/in/album-72157677358745114

[AndyC_772]

I get the distinct impression that HD3 went from concept to production in a very short space of time. Too short, perhaps.

Of course Keysight's engineers could lay out a board with a CPU on it for themselves, but that's not something which adds a great deal of value to the product. It's not something which is "special", it just either works or it doesn't. So if someone else already has a board with the right combination of CPU and memory on it, it's likely quicker to just use that. Also, having the CPU and its memory on a different PCB means the two boards can use different geometry and stack-up, so there's a cost saving and possible yield improvement too.

What other clues?

The industrial design, for one thing. HD3 isn't pretty. It doesn't have a clear brand identity; remove the label and it could have been made by anyone. Nobody spent any time on it, or thought it was worth putting in that extra effort. Nobody loved it.

Round the back there's a Displayport connector, not HDMI. Either would work, but I know which cables I have lying around and they wouldn't fit the HD3. I wouldn't have called DP the obvious choice, but could it have been: "which one can we confirm right now, so we can get on with the rest of the schematic without further delay?", rather than having to wait for an HDMI licence?

And, of course, not forgetting the firmware. Missing features at release isn't what we've come to expect; in fact, the extent to which the firmware is fit for purpose has always been a defining characteristic of 'grown-up' scopes (as compared to their much, much cheaper Chinese competition).

Has Keysight been caught napping? Did they perhaps think they were ahead of the curve with a high resolution ADC on the way, only to realise too late that even a $400 scope comes with one these days?

[mikeselectricstuff]

Really, there is no upside to using SoC modules in volume products long term. No matter what promises manufacturers make. Been there, done that. Always ends up costing more time (=money) than expected down the road.

Keysight also did this on the low cost 1000 series, although I think that would was their own design. But still, it was a processor (+FPGA's ASIC's) on a daughterboard design.
https://www.flickr.com/photos/eevblog/32798748040/in/album-72157677358745114

In that particular case it was most likely because for cost reduction - the layer count and manufacturing tolerances would be higher for the processor & ASIC than the large main PCB. 

[mikeselectricstuff]

And, of course, not forgetting the firmware. Missing features at release isn't what we've come to expect; in fact, the extent to which the firmware is fit for purpose has always been a defining characteristic of 'grown-up' scopes (as compared to their much, much cheaper Chinese competition).

Depends what you call "release" - is it actually available to purchase yet?
I can think of another KS/Agilent case - the 34461A DMM only added capacitance measurement in a later firmware rev, though not clear if this was intended at the start as there was no legend on the front panel.
Hopefully a difference between this and the low-end makes is that they prefer to ship with  well-tested, solid functionality for those functions that are implemented rather than throw it all out & let the customer find the bugs. Dave said he had a schedule of several upcoming releases and the functionality to be added in each.

[nctnico]

Round the back there's a Displayport connector, not HDMI. Either would work, but I know which cables I have lying around and they wouldn't fit the HD3. I wouldn't have called DP the obvious choice, but could it have been: "which one can we confirm right now, so we can get on with the rest of the schematic without further delay?", rather than having to wait for an HDMI licence?

I think they choose Displayport because that is the newer standard for connecting displays. IIRC Displayport also supports higher resolutions compared to HDMI.

[AndyC_772]

That might be relevant if the scope supported a screen resolution that could be called "high" - but it doesn't.

Depends what you call "release" - is it actually available to purchase yet?

I assumed so. There's a promotion on for "free" memory which runs from Sept 4th until the end of the year, so presumably it's actually available to order.

https://www.mouser.co.uk/pdfDocs/HD3MaximizeYourMemoryforFreeFlyerC4006.pdf

[Zucca]

So, do you think the scope can be "hacked"?

of course! A ticking bomb will explode soon or later.