完善资料让更多小伙伴认识你,还能领取20积分哦, 立即完善>
电子发烧友论坛扫一扫,分享给好友
|
我在下面构建了这个互斥锁电路,我想知道如何量化两个授权引脚都很高的漏洞概率。
我读到了这篇名为“为什么你不能建立一个仲裁者”的论文,它似乎表明一个仲裁者只能建立起来,所以我想知道如何降低毛刺概率。 http://dspace.mit.edu/bitstream/handle/1721.1/3346/P-2217-29812795.pdf?sequence=1 我可以将它们连在一起。 拉下电阻会有帮助吗? 除了总线仲裁者之外,我找不到有关Xilinx和仲裁者的任何参考资料。 以上来自于谷歌翻译 以下为原文 I built this mutex circuit below and I'm wondering how to quantify the probability of it glitching where both grant pins would be high. I read this paper called "Why You Can't Build an Arbiter" which seems to state an arbiter can only probablistically be built so I'm wondering how to decrease the glitch probability. http://dspace.mit.edu/bitstream/handle/1721.1/3346/P-2217-29812795.pdf?sequence=1 I could chain these together. Would pull down resistors help? I couldn't find any references about Xilinx and arbitors except bus arbitors. |
|
相关推荐
4个回答
|
|
|
在我看来,由于逻辑和路由延迟,这样的电路(具有组合反馈环路)是灾难(或至少是fustration)的一个配方。
如果您可以容忍授权信号的时钟周期延迟,那么使用不同的实现同步生成这些延迟会好得多。 BT 以上来自于谷歌翻译 以下为原文 In my opinion, circuits like this (with combinatorial feedback loops) are a recipe for disaster (or at least fustration) because of logic and routing delays. If you can tolerate a clock cycle latency on the grant signals, it would be much better to generate these synchronously with a different implementation. bt
|
|
|
|
|
|
我正在观察路由时序问题。
我做了一个测试电路,正在测试这个故障,它工作正常。 我在原理图中添加了一些IO引脚,这改变了门之间的布线要大得多,现在它有很大的问题。 在这种情况下,任何特定约束是否有用? 我假设有一种方法可以使用锁存器或FF执行此操作,除非以某种优化方式内置到系统中。 以上来自于谷歌翻译 以下为原文 I'm observing the routing timing issue. I made a test circuit that was testing for this glitch and it was working fine. I added some more IO pins to the schematic, which changed the routing between the gates to be much larger and it has big problems now. Are any particular constraints useful in this type of situation? I assume there's a way to do this with latches or FFs unless those built-in to the system in some optimized way. |
|
|
|
|
|
我通过将两条请求线连接到Spartan板上的50mhz和133mhz时钟来测试这个电路。
我通过一个AND门将两个输出连接到一个SR锁存器,看看它是否会出现足够的毛刺以使其跳闸。 它没有。 然后,我将两个从授权引脚中分出的网络分开,一个进入和门,一个进入LED IO引脚,然后开始跳转SR锁存器。 我必须串联3个互斥电路,以便SR锁存器不再跳闸。 当连接到其他引脚时,我得到了相同的结果,似乎连接到IO引脚使得该电路表现得更糟。 我把MAXDELAY约束放在MUTEX电路中的反馈网上,所以现在它们都放在平面图中,我不希望布局再引起这个问题了。 消息由clemahieu于06-06-2009 12:31 AM编辑 以上来自于谷歌翻译 以下为原文 I was testing this circuit out by attaching the two request lines to the 50mhz and 133mhz clocks on the Spartan board. I piped the two outputs through an AND gate and then to an SR latch to see if it would ever glitch enough to trip it. It didn't. I then split both of the nets coming out of the grant pins, one going to the and gate and one going to an LED IO pin and then it started to trip the SR latch. I had to add 3 of the mutex circuits in series in order for the SR latch to not trip again. I got the same results when connecting to other pins, it seems something about connecting to IO pins makes this circuit behave much worse. I put MAXDELAY constraints on the feedback nets within the MUTEX circuit so now they're all placed close together in the floorplan, I wouldn't expect the layout to be causing this issue anymore. Message Edited by clemahieu on 06-06-2009 12:31 AM |
|
|
|
|
|
单独放置解决这个问题可能没有用,因为它也与路由有关(当你连接I / O时会改变)。
通过MAXDELAY控制延迟可能有所帮助,但你真的想要像MAXSKEW这样的东西来限制差异。 即便如此,我怀疑你会再遇到这个电路的问题。 当你试图在FPGA中实现一些异步(带有组合反馈)电路时,这就是野兽的本质 - 这些工具通常是为同步方法设计的。 我会把它重写为一个简单的同步状态机。 BT 以上来自于谷歌翻译 以下为原文 Solving this with placement alone may not help as it also relates to routing (which changed when you connected your I/O). Controlling the delay via MAXDELAY may help but you really want something like MAXSKEW to limit the differences. Even then, I suspect you will have problems with this circuit again. That is the nature of the beast when you try to implement some of the asynchronous (with combinatorial feedback) circuits in an FPGA - the tools were generally designed for synchronous methodologies. I would rewrite it as a simple synchronous state machine. bt
|
|
|
|
|
只有小组成员才能发言,加入小组>>
3118 浏览 7 评论
3407 浏览 4 评论
Spartan 3-AN时钟和VHDL让ISE合成时出现错误该怎么办?
2874 浏览 9 评论
3966 浏览 0 评论
如何在RTL或xilinx spartan fpga的约束文件中插入1.56ns延迟缓冲区?
3057 浏览 15 评论
请问vc707的电源线是如何连接的,我这边可能出现了缺失元件的情况导致无法供电
1326浏览 1评论
求一块XILINX开发板KC705,VC707,KC105和KCU1500
1168浏览 1评论
/9
小黑屋| 手机版| Archiver| 电子发烧友 ( 湘ICP备2023018690号 )
GMT+8, 2025-12-3 01:26 , Processed in 0.673241 second(s), Total 79, Slave 62 queries .
Powered by 电子发烧友网
© 2015 bbs.elecfans.com
关注我们的微信
下载发烧友APP
电子发烧友观察
版权所有 © 湖南华秋数字科技有限公司
电子发烧友 (电路图) 湘公网安备 43011202000918 号 电信与信息服务业务经营许可证:合字B2-20210191

淘帖
625
