在数据输入引脚上使用IODELAY2是否适用于SPI？

让我们说我构建自己的SPI模块，我需要做什么才能使它工作。
如果我使用带有相位检测器的IODELAY2原语，我可以在闪存的开头放一些训练模式来进行一些bitlip操作吗？
谢谢你的帮助



以下为原文

Lets say that I build my own SPI module, what would I need to make it work.
 
If I use a IODELAY2 primitive with a phase detector, could I put some kind of training pattern on the beginning of the flash memory to do some bitslip operation?
 
Thanks for your help

忽略电缆延迟时间......我怀疑电缆SPI从端的100Mhz时钟信号质量太差，无法正常工作。
可能上升/下降沿将具有被从器解释为多个时钟边沿的反射。
麦克风



以下为原文

Ignoring the cable delay times ... I suspect the signal quality of the 100Mhz clock at the SPI slave end of the cable will be too poor for proper operation. Likely the rising/falling edges will have reflections that get interpretted by the slave as multiple clock edges.
 
Mike

1）我不会使用LVDS，因为我使用的存储芯片不通过LVDS进行通信。
我也没有足够的引脚使其通过LVDS工作。
我的客人我可以将它降低到50MHz甚至25MHz，以确保良好的信号质量。
HDMI线缆与外部干扰非常隔离。您认为带有同轴线的定制HDMI线缆会比标准HDMI线缆更好吗？
2）对于阅读部分。
（味噌）。
出于同样的原因，我无法折叠原始时钟;
即缺少额外的针脚。
但是，我正在谈论闪存，我控制着我放在它上面。
它只会写一次。
所以我总是可以在我的内存芯片上放一些训练模式，因为它比我需要的大。
你怎么看？



以下为原文

1) I will not be using a LVDS since the memory chip I use does not communicate via LVDS. I also don't have enough pin left to make it work via LVDS.
I guest I can slow it down to 50MHz or even 25MHz go ensure good signal quality. HDMI cables are very well isolated from external interference. Do you think a custom HDMI cable with coax wires would be better than a standard HDMI cable?
 
 
2) For the reading part. (MISO). I cannot make a fold back of the original clock for the same reason above; i.e. lack of additionnal pins. However, I am talking to a Flash memory and I control what I put on it. It will be written only once. So I can always put some training pattern on my memory chip because it is larger than I need.
 
what do you think?

bareil76写道：
1）我不会使用LVDS，因为我使用的存储芯片不通过LVDS进行通信。
我也没有足够的引脚使其通过LVDS工作。
我的客人我可以将它降低到50MHz甚至25MHz，以确保良好的信号质量。
HDMI线缆与外部干扰非常隔离。您认为带有同轴线的定制HDMI线缆会比标准HDMI线缆更好吗？
2）对于阅读部分。
（味噌）。
出于同样的原因，我无法折叠原始时钟;
即缺少额外的针脚。
但是，我正在谈论闪存，我控制着我放在它上面。
它只会写一次。
所以我总是可以在我的内存芯片上放一些训练模式，因为它比我需要的大。
你怎么看？
你如何将这个内存芯片连接到电缆？
我假设你正在建造一块板子。
那么为什么不把LVDS接收器放在电路板上呢？
----------------------------是的，我这样做是为了谋生。



以下为原文

bareil76 wrote:
1) I will not be using a LVDS since the memory chip I use does not communicate via LVDS. I also don't have enough pin left to make it work via LVDS.
I guest I can slow it down to 50MHz or even 25MHz go ensure good signal quality. HDMI cables are very well isolated from external interference. Do you think a custom HDMI cable with coax wires would be better than a standard HDMI cable?
 
 
2) For the reading part. (MISO). I cannot make a fold back of the original clock for the same reason above; i.e. lack of additionnal pins. However, I am talking to a Flash memory and I control what I put on it. It will be written only once. So I can always put some training pattern on my memory chip because it is larger than I need.
 
what do you think?

How are you connecting this memory chip to the cable? I assume you're building a board. So why not put LVDS receivers on the board? 
 
 
----------------------------Yes, I do this for a living.

我只有3根线用于闪存（N25Q512A13G1240E）。
我已经将我需要的第4根线与其他电子设备的另一根信号复用。
我不能通过差分对发送时钟....相信我，我会喜欢！
B计划是将时钟降至最小，只读取序列号。
该序列号将对应于可以以标准方式访问的另一个永久闪存驱动器上的另一个静态文件。
我会尝试制定计划A工作并让你发布。
所以总结一下
1）我需要检查50MHz的信号质量，并尝试不同的电缆类型（我们已经做了宽带HDMI线缆），如果可以，那么......
2）我需要用训练模式实现一点滑动。
3）我还需要让相位检测器工作以至少将MISO的上升沿与时钟对齐。
我错过了什么吗？



以下为原文

I have only 3 wires left for the Flash memory (N25Q512A13G1240E). I am already multiplexing the 4th wire that I need with another signal used for the rest of the electronics. I cannot send the clock via a differential pair.... believe me I would love to! 
 
Plan B is to reduce the clock to the minimum possible and only read a serial number. That serial number would correspond to another file static on another permanent flash drive that can be accessed in a standard manner.
 
I will try to make plan A work and keep you posted. So to summarize it
 
1) I will need to check the quality of the signal at say 50MHz and try different cables types (we already do custo HDMI cables), if this is ok then...
2) I will need to implement a bit slip with a training pattern.
3) I will also need to get a phase detector working to at least align the rise edge of the MISO with the clock.
 
Am I missing anything?
 

我终于成功了。
实际上我很容易。
我使用带有IODELAY2的相位检测器将数据对准我的时钟。
从那时起，我正在阅读的数据是几个时钟关闭。
所以我所做的就是制作一个我首先阅读的训练模式来设置一个bitlip。
我需要再读1个字节......就是这样。



以下为原文

I finally succeeded. I was pretty easy actually. I used a phase detector with a IODELAY2 to align the data in to my clock.
 
From that point the data I was reading was couple of clocks off. So what I did is make a training pattern that I read first to setup a bitslip. I needed to read 1 more byte... that is all.