YouTube2Text

What Happens When Routing Over Splits In Power Planes — Transcript

by Altium Academy · 2,619 words · 382 segments · language en · Watch on YouTube

Full transcript

  1. 0:00Hello everybody, welcome back to Altium
  2. 0:02Academy. I'm your host Zach Peterson and
  3. 0:04today we're going to be looking at
  4. 0:07routing over splits in power planes.
  5. 0:10Should you do it? We're going to answer
  6. 0:11that question in this video. Now, Burt
  7. 0:14Simonovich recently put out a paper in
  8. 0:16Signal Integrity Journal that examined
  9. 0:18this and I wanted to examine the same
  10. 0:20thing in Altium Designer and Simbeor.
  11. 0:23What I'm going to do in this video is
  12. 0:24show you how to set up a test board
  13. 0:25properly in order to do a simulation of
  14. 0:28this and then we'll look at what happens
  15. 0:30when we actually route over a split
  16. 0:32between two large power rails with
  17. 0:34stripline routing. Make sure to hop into
  18. 0:36Altium Designer and follow along and
  19. 0:38let's get started.
  20. 0:50Now, the situation that we want to look
  21. 0:51at in this video involves stripline
  22. 0:53routing and it typically arises when you
  23. 0:55have six-layer boards. Typically, when
  24. 0:57you have a six-layer board that's
  25. 0:59designed for high speed, you'll have
  26. 1:01signal on the top layer and then you'll
  27. 1:03have
  28. 1:04symmetric signal on the bottom layer.
  29. 1:07Then on layers two and layer five
  30. 1:12you will typically have ground
  31. 1:14and then another layer with ground.
  32. 1:16Now, one of the reasons that you might
  33. 1:17use a six-layer board is because one of
  34. 1:20your internal layers can also be sig.
  35. 1:22So, we have some signal routing in this
  36. 1:24layer, but then you can typically have a
  37. 1:27power layer here as well. Now, one thing
  38. 1:30that often happens in these situations
  39. 1:32with power layers is you may have two
  40. 1:34different rails. One routed with V1 and
  41. 1:38then the other routed with voltage V2.
  42. 1:40Now, as you're routing your signal layer
  43. 1:42around in this internal part of the
  44. 1:43stackup, you may end up routing over one
  45. 1:46of these splits in this power layer.
  46. 1:49Does this split actually matter? Well,
  47. 1:51that's the situation that we want to
  48. 1:52look at in this video and we're going to
  49. 1:54run some simulations to determine
  50. 1:56whether or not it does actually matter.
  51. 1:58I think the intuition here is that
  52. 2:00because you have a ground plane above
  53. 2:03this signal layer, then the signal layer
  54. 2:05is still going to be strongly coupled to
  55. 2:07the ground plane, and you may expect
  56. 2:10that the impedance deviation here in
  57. 2:12this region where you have the split
  58. 2:14between your power rails might be
  59. 2:16negligible. And as a result, you can
  60. 2:19probably ignore this. Well, I think this
  61. 2:21is one of those things that require some
  62. 2:23context. At certain frequencies, it may
  63. 2:25not matter, and at other frequencies, it
  64. 2:27might actually matter. So, this is one
  65. 2:29of those things that you really should
  66. 2:30simulate. The reason is that because
  67. 2:33here in this region where you have the
  68. 2:35solid power plane, you have the
  69. 2:37impedance determined by layer two and
  70. 2:40layer four. And so, you would then put
  71. 2:42layer two and layer four as your
  72. 2:43references into the impedance
  73. 2:44calculator, such as if you're using the
  74. 2:47layer stack manager in Altium Designer.
  75. 2:48However, in this region, your reference
  76. 2:50is no longer layer four. Your reference
  77. 2:53is actually layer two in this region,
  78. 2:55and then layer five down here. So, you
  79. 2:58have slightly more distance from the
  80. 3:00signal to the bottom reference layer,
  81. 3:03which could then create your impedance
  82. 3:04deviation. Will this actually matter?
  83. 3:06Well, it depends on the input impedance
  84. 3:08looking over this direction through the
  85. 3:11signal layer. Now, this is something
  86. 3:13that we can actually simulate in an
  87. 3:15electromagnetic simulator. And what we
  88. 3:17want to do is look at the S parameters.
  89. 3:19So, since we're dealing with an input
  90. 3:21impedance deviation right here in this
  91. 3:23region, we would then want to calculate
  92. 3:25our S11 value for the traces that are
  93. 3:28being routed over this region. Now,
  94. 3:30another question that you might ask is,
  95. 3:32is there going to be a difference for
  96. 3:34single-ended signals versus differential
  97. 3:36signals? So, for single-ended signals,
  98. 3:39the only reference you have are these
  99. 3:41plane layers above and below the signal
  100. 3:43layer. But, if you have a differential
  101. 3:45signal, the differential signal can
  102. 3:47provide the return path, and it
  103. 3:48functions as the reference for one of
  104. 3:50the other traces. So, is that going to
  105. 3:52produce a difference? We're going to
  106. 3:54look at that situation as well. We'll
  107. 3:55look at single-ended S parameters and
  108. 3:59we'll look at
  109. 4:01differential S parameters when we do our
  110. 4:03simulation. Now, let's jump into Altium
  111. 4:05Designer. We'll take a look at how to
  112. 4:07set up a test board for this type of
  113. 4:09situation and then we'll throw that into
  114. 4:11a simulator and we'll get our S
  115. 4:12parameters and we'll see what the
  116. 4:14difference is between routing over solid
  117. 4:16power versus routing over split power.
  118. 4:18So, now I'm inside of Altium Designer
  119. 4:20and what I want to do is show you how to
  120. 4:22set up a test board in order to
  121. 4:24accurately perform simulations for this
  122. 4:26power plane split. So, here inside of
  123. 4:27Altium Designer, what I've done is just
  124. 4:29created kind of a simple model where I
  125. 4:31have two different sizes of gaps for our
  126. 4:33power plane split and then I've created
  127. 4:35an identical set of interconnects over a
  128. 4:37solid ground and power plane. Now, if we
  129. 4:40look at the stack up, you can see here
  130. 4:42where our power ground and signal planes
  131. 4:44are located. Here on the top and bottom
  132. 4:46layers, we have uniform ground. Here,
  133. 4:49layer two is our power layer, that's
  134. 4:51where our split is going to be and then
  135. 4:53layer three is our signal layer, so
  136. 4:54that's where our signals are being
  137. 4:56routed. In this example simulation model
  138. 4:58that I've created, I've placed two
  139. 5:00different size gaps. So, here I have a
  140. 5:02400 mil gap and here I have a 200 mil
  141. 5:04gap. And then if you look at the
  142. 5:06schematics here, I've defined
  143. 5:08single-ended and differential traces
  144. 5:10that we can use to simulate what happens
  145. 5:12for our different types of signals when
  146. 5:13they route over these gaps. Now, on each
  147. 5:15end of these nets, what I've done is
  148. 5:17I've created these dummy components and
  149. 5:18these dummy components are basically
  150. 5:20just a pad. That's going to make it very
  151. 5:22easy to assign a port on each end of
  152. 5:24these interconnects when we then import
  153. 5:26this into the simulation tool. Now, one
  154. 5:28thing I'd like to do before we jump into
  155. 5:30simulation is to just discuss these via
  156. 5:32transitions. Now, you can see here I've
  157. 5:34placed some stitching vias around these
  158. 5:36signal vias and then you can see that
  159. 5:38there's kind of a default anti-pad
  160. 5:40around these signal vias. Depending on
  161. 5:42the frequency range that you're working
  162. 5:43in, you would actually want to design
  163. 5:46these vias so that they have a 50 ohm
  164. 5:48impedance specifically in the frequency
  165. 5:51range that you care about. For this
  166. 5:53demo, we're not going to worry about
  167. 5:54that because that's a whole 'nother
  168. 5:56level of complexity, but in general, if
  169. 5:58you were going to create a simulation
  170. 5:59model and whether you were going to use
  171. 6:01ANSYS or use Symbeor or some other
  172. 6:03simulator, you would want to make sure
  173. 6:05that that via transition provides
  174. 6:07correct matching. If you can't do that,
  175. 6:09you would then need to relocate the pad
  176. 6:12for this port to an internal layer and
  177. 6:14eliminate those vias. Now, the other
  178. 6:16thing that you might want to do, again,
  179. 6:17depending on the frequency range you're
  180. 6:19working in, is move the location of this
  181. 6:22split closer to your input port. So, for
  182. 6:24lower frequencies, you can have this
  183. 6:26somewhat farther away, whereas at higher
  184. 6:28frequencies, you would want to move this
  185. 6:31gap closer to the input. So, to do that
  186. 6:34is pretty simple. You're, of course,
  187. 6:35just dragging the gap here from these
  188. 6:37polygons closer to your input side.
  189. 6:40Then, you would just size it so that you
  190. 6:41have the correct gap size. Here, I'm
  191. 6:42going to leave it here and just in the
  192. 6:44middle, and we're going to see what
  193. 6:45happens when we simulate the S
  194. 6:47parameters. We're going to look at how
  195. 6:48the different size gaps influence our S
  196. 6:51parameters in different frequency
  197. 6:53ranges. For this set of simulations,
  198. 6:55we're going to use Symbeor. To get your
  199. 6:56board into Symbeor, you would just do an
  200. 6:59ODB++ export, and then you can re-import
  201. 7:01the board back into Symbeor and start
  202. 7:03selecting the nets that you want to
  203. 7:05simulate. So, now I've brought the board
  204. 7:06up in Symbeor, and now I can start
  205. 7:08selecting the ports that I want to
  206. 7:10simulate. Here, when I'm in Symbeor,
  207. 7:12what I can immediately do is just run an
  208. 7:14ERC, and I can see what the expected
  209. 7:16impedance deviation is along this
  210. 7:19section of interconnect. So, you can see
  211. 7:20here for the differential pair with the
  212. 7:22200 mil gap that we have a 6 ohm
  213. 7:25deviation in the differential impedance.
  214. 7:28Here, if we go back to the single-ended
  215. 7:30port and I run the same, you can see
  216. 7:33that I get an 8 ohm deviation. So, it's
  217. 7:35actually much more significant when you
  218. 7:37have the single-ended signal versus the
  219. 7:40differential signal. For the
  220. 7:41single-ended signal, it's a almost 20%
  221. 7:44deviation in the impedance, whereas it's
  222. 7:46only a 6% impedance for the differential
  223. 7:48pair. So, that should underscore the
  224. 7:50importance of having differential pairs
  225. 7:53and sometimes having them spaced closer
  226. 7:55together. When they're closer together
  227. 7:57and you route over this kind of split,
  228. 7:59you're going to have a smaller deviation
  229. 8:01in the differential impedance. That's
  230. 8:03because one trace always references the
  231. 8:05other trace. Next, let's start selecting
  232. 8:07our ports here and then we can set up a
  233. 8:08quick S-parameter simulation. So, here
  234. 8:10I'm going to select our 200-mil gap, our
  235. 8:12400-mil gap, and then our uniform plane.
  236. 8:15And I'm going to run a fast SI
  237. 8:17simulation to get the S-parameters. So,
  238. 8:19now we have our return loss data pulled
  239. 8:21up and the return loss data pretty
  240. 8:23clearly shows what happens when you have
  241. 8:25that split in the plane, especially when
  242. 8:27it's a reasonably large split. Here,
  243. 8:29when we look at really low frequencies,
  244. 8:31we see that all the S-parameters pretty
  245. 8:33much overlap with each other. And this
  246. 8:35is when we're sub 100 MHz or right
  247. 8:39around 100 MHz. Once we start to get
  248. 8:41into the gigahertz range, we really
  249. 8:43start to see the deviation in the
  250. 8:44S-parameters really quickly. Here, the
  251. 8:46blue curve is for S33 and that's for
  252. 8:50this middle trace right here. So, that's
  253. 8:52where we have the uniform plane. Then
  254. 8:53here, the red is S11, so that is for our
  255. 8:57top left corner here. That's where we
  256. 8:59have our smaller gap. And then here, the
  257. 9:01brown or the orange curve, that is our
  258. 9:04trace down here in the lower right,
  259. 9:06which is our 400-mil gap. So, this is
  260. 9:08where we start to see a pretty big
  261. 9:09deviation in the S-parameters in the 1
  262. 9:12to about 7 or 8 GHz range. And you can
  263. 9:16see here that we do get pretty high
  264. 9:18return loss compared to the trace over
  265. 9:21uniform planes. However, is this too
  266. 9:24much reflection? Well, as we can pretty
  267. 9:26clearly see in this case, the return
  268. 9:28loss does get pretty high, but it
  269. 9:30doesn't jump up above -10 dB. In a lot
  270. 9:33of cases, this would probably still be
  271. 9:35acceptable even if we're routing over
  272. 9:38these large gaps. Now, once you get
  273. 9:40around 10 GHz and higher frequencies, we
  274. 9:43see that the S parameters all start to
  275. 9:44overlap each other a little bit. So,
  276. 9:46this is because the dominant return loss
  277. 9:49mechanism switches from the gap in the
  278. 9:52power plane to the loss or the impedance
  279. 9:54deviation that's created by the loss in
  280. 9:57these interconnects. Now, remember, in
  281. 9:59Altium Designer, it's determining a
  282. 10:01lossless impedance, but the real
  283. 10:02interconnects have dielectric loss
  284. 10:05tangent as well as skin effect that
  285. 10:07create deviations in the impedance, and
  286. 10:09that's what's creating these reflections
  287. 10:10at much higher frequencies. So, our
  288. 10:12conclusion is this, at really low
  289. 10:14frequencies, sub 1 GHz, you're really
  290. 10:16not going to notice a difference. Once
  291. 10:18you get into the low GHz range, you do
  292. 10:20start to see a large difference.
  293. 10:22However, with the small spacing that we
  294. 10:24have between ground and signal, it
  295. 10:26doesn't create so much loss that the
  296. 10:28interconnect may not still be usable.
  297. 10:30Once we get to really high frequencies,
  298. 10:32the split in the plane becomes less
  299. 10:35important, and you have other loss
  300. 10:36mechanisms that are creating the
  301. 10:37impedance deviation that you see in
  302. 10:40these S parameter plots. Now, let's take
  303. 10:41a look at the differential S parameters
  304. 10:44for all of these differential pairs in
  305. 10:45this test board. We have all of the
  306. 10:47differential S parameters prepared. Now,
  307. 10:49what you can see here is that all of the
  308. 10:52differential S parameters are pretty
  309. 10:53similar. We have the same ordering here
  310. 10:56for these curves by color. The gapless
  311. 10:58and the small gap and the large gap are
  312. 11:01all really similar all the way up to
  313. 11:03really high frequencies. This can be
  314. 11:05explained in a couple of ways. First of
  315. 11:07all, as we saw earlier, the impedance
  316. 11:09deviation when routing over that split
  317. 11:11in the power plane was smaller for the
  318. 11:13differential pairs. Remember, it was
  319. 11:14only 6%, whereas for the single-ended
  320. 11:16traces, it was almost 20%. Next, we have
  321. 11:19the same change in the dominance of the
  322. 11:22return loss from low to high frequencies
  323. 11:25as we had in the single-ended case. So,
  324. 11:27just because we're in differential
  325. 11:28pairs, doesn't mean that the loss
  326. 11:30tangent and the skin effect go away.
  327. 11:33They also will dominate and create an
  328. 11:35impedance deviation seen looking into
  329. 11:37the interconnect. And so that's what's
  330. 11:39creating this big deviation at higher
  331. 11:41frequencies and then creating this
  332. 11:42higher return loss rising just above -10
  333. 11:45dB. So this channel would still be
  334. 11:47appropriate all the way up to about 20
  335. 11:49GHz, but to really perfect this we would
  336. 11:52want to do what I stated earlier both
  337. 11:54for the single-ended transition and the
  338. 11:56differential transition. And that would
  339. 11:58be to optimize these vias so that we get
  340. 12:01a really flat near 50 ohm single-ended
  341. 12:04impedance or 100 ohm differential
  342. 12:06impedance within the bandwidth that we
  343. 12:08care about. That's going to help
  344. 12:09eliminate some of this discrepancy and
  345. 12:11then it's going to give us a little bit
  346. 12:12more accurate view into what's happening
  347. 12:15with our system. Now as I've shown in
  348. 12:16some other videos, we really don't care
  349. 12:18about the via impedance until we get
  350. 12:20above a couple of GHz. So if we were
  351. 12:23only looking from for example
  352. 12:25uh up to 2.5 GHz here, we could get a
  353. 12:28really clear view of what's going on in
  354. 12:30both sets of curves. Here if we just
  355. 12:32zoom in to the 2.5 GHz range for our
  356. 12:36single-ended uh traces, we see here the
  357. 12:38really big impedance deviation that we
  358. 12:40get when there's a gap versus no gap.
  359. 12:42And it's a it's a little over 20 dB. For
  360. 12:45the differential pairs, we have a much
  361. 12:46smaller deviation between the return
  362. 12:48loss curves, so it's a little bit more
  363. 12:50difficult to conclude that the split in
  364. 12:53the power layer is the sole cause of all
  365. 12:55of this behavior. To really nail that
  366. 12:57down, we would definitely want to
  367. 12:59optimize those vias or we could bypass
  368. 13:01the vias completely and just move those
  369. 13:03pads into the inner layer and then
  370. 13:06assign our port directly to those pads
  371. 13:08on the inner layer. Make sure to take a
  372. 13:09look at the blog that's linked in the
  373. 13:11description to learn more about this and
  374. 13:12I'll discuss this further in that blog.
  375. 13:14Thanks for watching everybody. Make sure
  376. 13:15to hit that subscribe button, hit that
  377. 13:17like button, leave your comments and
  378. 13:18questions in the comment section, and of
  379. 13:20course don't forget to call your
  380. 13:21fabricator, folks. We'll see you next
  381. 13:23time.
  382. 13:31Mhm.

About this transcript

This page contains the full transcript of What Happens When Routing Over Splits In Power Planes by Altium Academy, generated from the public captions YouTube serves with the video. The transcript has 2,619 words across 382 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.

What you can do with it

Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.

Free YouTube transcript tool

YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.