Where can i buy an Amy Development board? If its only available to DIY is anyone out there willing to build me one? currently my surface mount skills are limited. name your price.
I have Amy PCBs and imp003’s. If I can get them working I’ll likely sell a few. At the moment I’m having some issues getting it to connect, which is likely a build issue. The imp003 is NOT particularly easy to build with a reflow gun. If I can’t get the issue sorted I’ll likely hand the boards off to my local manufacturer.
Ordered a few -003’s from Digikey the other day, and was quite surprised when they showed up shortly thereafter. (Said they wouldn’t be available 'til Feb). So, I needed some boards too, much sooner than I expected.
I sent off a board order to OshPark and am looking forward to getting first boards back in a couple weeks to try out. Have just got a new reflow oven working, so will give that a try to see I can get things soldered down properly. Will keep you posted.
Did you get a stencil or are you putting the solder down by hand? I’ve gotten pretty good at using my hot air station, but the 003 is a beast.
Stencils made me confident in trying qfn packages
Haven’t done the stencil yet, but plan to do so before the boards get back. I get mine from Ohararp.com/stencils/ … the kapton ones are pretty reasonably priced and I’ve always had good luck with them. Looking forward to the -003 challenge
I got mine from https://www.oshstencils.com/ I really love their system where I could see how it would look before ordering, that saved me from ordering something useless.
Looks nice … will give them a try. Didn’t realize they were doing stencils now too … maybe it’s fairly new. They don’t mention anything about resolution that I could find, but their gallery examples suggest precision laser cutting. And the prices are nice, especially for smaller stencils. (I think I was paying about $25 for Ohararp, but they charge for the whole 8.5" x 11" kapton sheet if I recall correctly). Thanks for the lead.
I ordered one from Pentalogix, where I get all my quick turn prototyping boards made. It isn’t cheap, but they can turn boards VERY quickly. Also, they have a deal where you can get a “free” copy of any version of Eagle CAD. Basically, they charge you for it, and then they give you 30% off every purchase until the purchase price is reached. You also get Viewmate and some time with one of their layout pros. Its not a bad deal if you use Eagle commercially or need to go past the “lite” limits.
I used to order stencils from Pentalogix, but switched to Ohararp … same quality and delivery at half the price, so that was an easy decision. Will be interesting to see how the Osh stencils compare.
Yes, have noted the Pentalogix PCB service, but haven’t used them yet. Haven’t had the need to upgrade Eagle yet or go for any quick turn boards, but they’re on the radar screen should that need arise.
One note on Pentalogix, Ask them for their design rules, and if you care, tell them to not make any adjustments to the board before fabricating it. They will make minor changes to accomodate the board house, and may or may not tell you depending on how minor it is. I’ve asked them to always reject my boards if any changes are needed.
@LarryJ - I’m also interested in sending the Amy design into OshPark but the gerbers include GL2.GP1 and GL3.GP2 files for the inner layers which are not recognized by their upload process and renaming them G2L and G3L inverts their intended pattern. How did you work around this?
Hi @senger - Yes, you are correct. What I did was to e-mail off the gerbers to them to check, and one of their support folks kindly responded by taking care of it for me, as well as forwarding me back an updated .zip with positives for the G2L and G3L layers. (They also added the GKO outline layer which was not included with the Dev Center bundle).
I’m sure they can take care of it for you, but if you’d like I can send you the revised .zip I submitted if you PM me.
(I just received boards back from the other day. On initial inspection they look very nice, but I haven’t had the chance to populate them or check things out in detail yet).
Yea @Hugo, got it on the routing comment. Haven’t actually layed out a board yet (just going with the Amy), but do note there’s only a few I/Os on the inner ring … probably only need to get to those for an I/O intensive design.
(On a related note, hadn’t heard of Elecrow, but followed up on them based on your comments on another thread. Sent them the Amy package and should have boards back in a couple of weeks. Will be interesting to see how those look, and have them lined up as a possible alternate supplier. Thanks for the lead).
I have also succeeded in building an Amy board. Placing the imp itself is a little scary since you can’t see anything to line up with so I just dropped it and left it alone. I used a board oven to bake it. Much to my surprise everything worked great.
There are a couple of things about the design that I’m not sure I can explain.
-
On the top layer there is no copper under the XTAL itself and on the first inner layer the copper is removed in a line around the area of the XTAL except that it is open to the rest of the ground plane at the left in an area near to the imp module. The caps on the in/out lines of the XTAL are connected to this region of the ground plane. What is the idea here?
-
I can see how 4-layers makes it easier to route the lines, power and ground plane but is there anything that really makes 4-layers a necessity?
-
I got the last seven imp003’s that Digi-Key had back at the beginning of Dec. They have more now but there seem to be two listings. One is LBWA1ZV1CD-716 and the other is LBWA1ZV1CD-TEMP. Is there a difference?
Nice @senger … glad you got it up and running. I had set my Amy PWBs aside for the holidays, but populated one last weekend. Happy to report I had similar success … it woke up and functioned right away. (Yea, that module is a bit scary at first. As you, I just lined it up carefully, tapped it down, and turned on the oven. All went well).
Can’t address your items 1 and 3, but it’s been reported the multi-layer is important for antenna performance as well as impedance control of the 50 ohm feed. And yes, it’s hard to imagine how to route out of the device without the add’l layers.
(Related to the antenna and impedance, I’m looking into this a bit more myself. Interestingly, OshPark uses an FR-408 board material, which has a slightly different dielectric constant than basic FR-4, so the antenna dimensions and impedance are probably off a bit. Some minor board mods are likely appropriate to optimize RF performance for that substrate. But for basic first order functionality it seems fine as is).
On 1, it’s common to void ground around crystals. I’m guessing it’s to ensure there are no current flows underneath that could cause jitter but it’s just something that’s always done
On 3, the -TEMP is from an initial production batch which had not yet passed all qualification tests, -716 is from later production. There’s no component or design differences, it’s just early vs later production.
As @larryj says, the 4 layers are for impedance control. You don’t need 4 layers to route out all the “high value” pins on the package - we ensured they were all on the outer ring.
Thanks @Hugo I think I understand the role of the inner ground plane in controlling impedance on the antenna trace until it reaches the no copper part of the board. @larryj mentioned that OshPark uses FR-408 while I see that Elecrow uses FR-4. How much difference is there between the behaviors of these materials? This is all new ground for me and I’m finding it very interesting.
@senger - didn’t mean to possibly open up a can of worms with this as RF/antenna design is a complex subject … (and I’m way beyond my experience zone here trying to offer any guidance, especially with regards to the antenna).
If you’re interested in the antenna, I came across this article which was interesting. (The antenna dimensions for .062" FR-4 match the Amy design).
I’m still trying to learn a bit more about this myself. What I know at this point is that the mat’l used by OshPark has a lower dielectric constant (3.7 vs. 4.5 for FR-4), which would have some small impact on feed and antenna dimensions. I traded a few e-mails with Silicon Labs that indicate the antenna length should increase about 10% due to the lower dielectric constant of the FR-408. Will let you know if/when I get more clarification regarding recommended antenna dimensions.