This version (09 Oct 2018 15:11) was approved by email@example.com.The Previously approved version (24 Feb 2018 18:47) is available.
Whether you want to understand the changes between revisions, or just understand how to probe the PCB, this is where all the information should be.
Removing the case
The plastic case comes off quite easily, with the removal of two black Phillips screws on the bottom of the case. This is a picture of the pre-production Rev B boards. The production version may be different. It will for sure be CE and FCC certified (already passed).
Removing the screws will allow you to take the top off the case, and expose the PCB.
If you want to remove the PCB, and place it on the table, we recommend that you attach Cylindrical Bumpers (also known as feet), on the PCB to protect the components on the bottom of the PCB. These are not included in the design, and must be purchased separately (as we don't expect too many people wanting to do this).
Is in the concept/noodling stage, we do not have schematics, and do not have layout, don't have this back yet.
Why do a Rev C?
our “low-risk” OTG changes, caused problems, since we put the VBUS monitoring (R88) on the wrong side of the fuse, and the inclusion of the DC choke. While the DC choke reduces noise, and there is no DC difference between PGDN, and GND, on certain hosts, with certain operating systems, there can be a 250mV AC difference between PGND and GND. Since the analog comparators inside the Microchip USB3320 will be referenced to GND (quiet), it appears to toggle between host mode and device mode. A temp workaround was to change this value from the recommended USB spec
of 1k to 10k.
We also had issue on ESD
testing, and will be adding an ESD
protection to the RF connector.
As a cost optimization (and feature enhancement), we may redo some of the power section (replacing the NCP339AFCT2G, ADM1177, with the LTC4413
) - this is still TBD based on lab work to be done. One of the usability issues - is that the 2 NCP339AFCT2G devices are designed to be break before make - which causes the power supply to reset if you plug the ADALM-PLUTO into a PC (which causes it to boot), and then eventually a power supply (which causes the supply to disappear for a small time, and then re-appear). We don't think this is a huge issue, but it would be enhance the usability. It would mean however, that we loose the ability to read USB
voltage/current (which we currently do in the ADM1177
to pin out the 2nd channel with pads/footprints.
Why do a Rev B?
RF performance was bad. (EVM was poor, ~-32dBm; goal was -45dBm). This was due to the Rakon oscillator having no PPSR1)
, and us powering the oscillator directly from a 1.8V digital rail (which was noisy). A LDO was dead bugged
onto rev A, and it fixed the performance problem. We tried a few L/C & ferrites, but it was not as good as the LDO, so we added an ultralow noise LDO, the ADM7160ACPZN1.8
to the design.
RF performance was bad on certain PCs, even with with the added LDO (above). This was tracked down to to noise on the power supply. We added a power choke (DLW5BSM801TQ2L), and this went away, and we got much better results independent of platform.
Power supply noise was a little higher than we would have liked, and was effecting RF performance, so sync'ed the multiple switchers together to reduce input ripple. This also allowed us to reduce input capacitance on the entire unit.
were barely passing FCC, (due to noise @ 480MHz, which was also causing some RF issues at 480 MHz
), so added DLW21HN900SQ2L (90Ω choke) on the USB
swapped QSPI flash and DDR3L devices, to something not on the obsolete list.
based on early end user feedback, added some test points to unused pins to make extending the device a little easier.
based on early end user feedback, added On-The-Go support for USB
. This was some minor power circuit, and monitoring VBUS, so was looked at as “low risk”
Like all things, PlutoSDR is a integration/copy/paste of previous known working platforms. There were a few different AD936x /Zynq platforms that were worked on that we borrowed things from:
FMCOMMS5 FMC Card
, which is where we borrowed the (buggy) oscillator design, since the AD9363 can not be used with a crystal.
PicoZed SDR (which is were most of the Zynq and USB