XE-216 Project Review

XCore Project reviews, ideas, videos and proposals.
Post Reply
RitchRock
XCore Addict
Posts: 186
Joined: Tue Jan 17, 2017 9:25 pm

XE-216 Project Review

Post by RitchRock »

Hello,

I had success prototyping my design with the xC-200-MC-AUDIO development board and some other EVMs. So, time to move to my first prototype PCB of my own design. I've read through all the relevant info I can find regarding PSU requirements & clocking, but wanted to share my schematics to see if there is anything I've missed. I included plots with just the relevant XE-216 components. Some questions/comments:

1. Any comment as to using MCLK_FSEL with a Mux VS using I2C to change the Si5351A every-time the sample rate needs to be changed? Any problem doing it either way?

2. It's been some time since I've done a bus powered USB design - does my protection look OK? How about the use of the Buck/Boost to make sure analog (not included in this schematic set) gets a steady 5v?

3. I measured current consumption of my design to be less than 250mA. I believe I read that as the complexity of my application grows, so does power consumption. Perhaps I can move to a processor with less cores, if I'm not using the XE-216 to it's full capabilities. Would that be pin compatible?

4. Did I miss anything with the PSU timing? It should follow the reference design pretty closely.

5. Regarding construction for this first prototype, I plan on using solder paste, a stencil and little toaster-reflow. It doesn't seem like I'll have any issues with these SMD packages working in this manner.

Thanks!
RitchRock
Attachments
[XE_216_breakout_PCB.PrjPcb] XE_216_breakout_PCB.PDF
(684.02 KiB) Downloaded 414 times
[XE_216_breakout_PCB.PrjPcb] XE_216_breakout_PCB.PDF
(684.02 KiB) Downloaded 414 times


User avatar
mon2
XCore Legend
Posts: 1913
Joined: Thu Jun 10, 2010 11:43 am
Contact:

Post by mon2 »

Hi.

1) what purpose does U4 serve ? VBUS is already @ 5v so this switcher can be removed. The regulator is a buck-boost so you are planning to support voltage drops on the Vbus line ? Openly, if the Vbus rail drops to where this regulator is needed then you have more severe issues with the host PC. Really don't believe you need this regulator to be present in the design unless you can cite a specific case ?

2) U8 is very expensive from TI and also low current @ 500 mA. Consider the lower cost AP3402 from Diodes Inc. and couple with Taiyo Yuden shielded inductors @ 3A rated.

https://www.diodes.com/products/power-m ... art/AP3402

3) U7 - same comments as above. AP3402 will work here as well.

4) Recommend to place either an USB load switch (to limit the current quickly) or at least a polymer in-line fuse on Vbus. Technically it appears your widget is to be USB powered ? If yes, then your entire solution will consume < 500 mA (max) so you could scale back the switching regulators to suit but the AP3402 are very nicely priced so go for the better stuff in case you end up being a higher powered device with an external power supply.

Polymer fuses are < $ 0.10 USD - check the one used on Arduino, etc.

5) If you apply the AP3402, then U6 can be removed from the design since there is an internal PG on the regulator. The PG remains to be open drain so a pull-up to 3v3 is still required to sequence the 1V0 rail.

NB: AP3402 will require the placement of external voltage divider resistors such that the junction of the 2 resistors = 0.6 volts. The values for this resistor divider will define the output voltage rail to dial up 3v3 or 1v0. It is vital that the resistor values be correct and soldered else downstream damage could occur. That is a small risk that needs to be considered vs. using a fixed regulator (ie. Semtech, etc. offer pin strapped voltage switching regulators).

6) On U11 and from memory, believe that CLK0 is a single and internal PLL while CLK1 & CLK2 share a PLL ? The reason for this discussion is that be sure that while you may alter the CLK0 or CLK2, that CLK1 = 24 Mhz remains to be stable as this is clocking the XMOS CPU. That last thing you want is to halt the XMOS CPU clock and end up in a dead-locked state !! If you have this all working already then no worries but the summary is that the CPU clock must keep running and without glitches else your code execution could also possible halt or lock up.

A simple remedy is to apply a static 24 Mhz clock to the CPU but believe this exact pinout is from an XMOS proven code ? So you may be already ok here.

7) U9 by Micrel (Microchip) is very expensive. You can replace with a lower cost and higher current, Toshiba TCR2EF series @ 2v5:

TCR2EF25,LM(CT

http://www.digikey.com/product-detail/e ... CR2EF25,LM(CT/TCR2EF25LM(CTCT-ND/4503323

8) U10 (Analog) is expensive, can replace with another POR such as p/n APX809. Use the PG from the 1v0 rail (with a local pull-up to 3v3) to enable this APX809. Once the 3v3 rail is stable, the 1v0 regulator will enable, once the 1v0 rail is stable, the PG will be pulled up to 3v3 and once this rail is 2v9 or higher, the APX809 will reset the CPU.

APX809-29SAG-7
http://www.mouser.com/ProductDetail/Dio ... Iw9g%3D%3D

9) Check some of my earlier posts but Socay has excellent ESD protection diodes for USB for < $ 0.10 USD. They mfr for many of the major brands. Max Echo has excellent ferrite beads for the filtering. The footprints are common across multiple vendors so you can start with local suppliers and source when the volume increases although many overseas suppliers are fine with small volume sales these days. Often you can buy a full tape & reel for a cost less than a few pieces through major distribution.

10) D2 may not light up due to the 1v0 rail. Check this before considering in the design. You may have to use a buffer with the input of the buffer to the PG of the 1v0 rail. Then when the 1v0 rails is stable, the PG will be high and the buffer will enable the D2 led. A single SOT-23 gate buffer should work here.

11) R8 & R9 could be merged into a single resistor network (1206 size with 4 x 0603 parts inside).

12) R1, R2 & R3 could be merged into a single resistor network (1206 size with 4 x 0603 parts inside).
RitchRock
XCore Addict
Posts: 186
Joined: Tue Jan 17, 2017 9:25 pm

Post by RitchRock »

Hi mon2,

Thank you very much for your reply. I read through your suggestions and appreciate your thoughtful consideration on lowering cost with regards to the expensive regulators. I have made a few high volume designs and like to use fixed regulators if possible. I'll check out the Semtech offerings, but understand I could use the AP3402.

1. I'm going back and forth in regards to using U4, but the purpose is simply to get a stable 5v from the USB Bus (Yes, my device is USB Bus Powered). USB VBUS can vary from about 4.75 to 5.25. This voltage will be used to bias an array of matched transducers and directly affects the resulting sensitivity. I want the sensitivity to be very stable from one device to the next, thus I Boost if under 5v and Buck if over. If you really don't think it's needed, I'm more than happy to get rid of it, I just want a stable, clean 5V. Open to suggestions.

2. I noticed the use of U11 in the XHRA-2HPA reference design and thought it would fit OK in this project. It seems in that design, it must get reprogrammed every-time the sample rate changes. In that project, CLK1 is labels 24M and directly tied to the XMOS processor. I will double check the shared PLL question in the datasheet.

Thanks again!

RitchRock
RitchRock
XCore Addict
Posts: 186
Joined: Tue Jan 17, 2017 9:25 pm

Post by RitchRock »

Is it true the only difference between the XE216-512-TQ128 and the XU216-512-TQ128 is that one has pins that can be dedicated for RGMII? If so I can also cut some cost there.
Redeye
XCore Addict
Posts: 131
Joined: Wed Aug 03, 2011 9:13 am

Post by Redeye »

Yes, the only difference between XE and XU is that the XE has the RGMII interface. If you're doing a low-volume prototype batch you might find the XU version hard to source (speaking from experience), but I believe (please check!) it's pin compatible with the XE variant (but not the XL variant) which is normally in stock at Digikey/Farnell and might save you waiting long lead times waiting for a tray of XU parts.
User avatar
Bianco
XCore Expert
Posts: 754
Joined: Thu Dec 10, 2009 6:56 pm
Contact:

Post by Bianco »

mon2 wrote:
2) U8 is very expensive from TI and also low current @ 500 mA. Consider the lower cost AP3402 from Diodes Inc. and couple with Taiyo Yuden shielded inductors @ 3A rated.

https://www.diodes.com/products/power-m ... art/AP3402
Nice one, will consider to use it in a project for which I had a TI TPS562210 in mind. Was looking for something small with leads, EN and PG, not necessarily cheap. Looks like a good option.

Thanks
RitchRock
XCore Addict
Posts: 186
Joined: Tue Jan 17, 2017 9:25 pm

Post by RitchRock »

After review, I think I'll just use an inexpensive, low noise LDO with a fixed 4V output for my low current consumption analog portion. This will get rid of U4. Something like the Microchip TC1015 should do the trick and is only .37 in single part qtys:

http://www.digikey.com/product-detail/e ... -ND/418620
RitchRock
XCore Addict
Posts: 186
Joined: Tue Jan 17, 2017 9:25 pm

Post by RitchRock »

Redeye wrote:Yes, the only difference between XE and XU is that the XE has the RGMII interface. If you're doing a low-volume prototype batch you might find the XU version hard to source (speaking from experience), but I believe (please check!) it's pin compatible with the XE variant (but not the XL variant) which is normally in stock at Digikey/Farnell and might save you waiting long lead times waiting for a tray of XU parts.
Good looking out Redeye. There are 185 XU parts in stock at Digikey, so I should be good for prototyping. I checked the pinouts and the XE parts and XU parts are the same on the datasheets.

http://www.digikey.com/product-detail/e ... ND/5148746
User avatar
mon2
XCore Legend
Posts: 1913
Joined: Thu Jun 10, 2010 11:43 am
Contact:

Post by mon2 »

trying to hook a brother up...don't forget about Future Electronics:

http://www.futureelectronics.com/en/Tec ... 60539&IM=0


Image
RitchRock
XCore Addict
Posts: 186
Joined: Tue Jan 17, 2017 9:25 pm

Post by RitchRock »

mon2 wrote:trying to hook a brother up...don't forget about Future Electronics:

http://www.futureelectronics.com/en/Tec ... 60539&IM=0


Image

Nice! Best price I've seen for proto qtys!!
Post Reply