使用PIC32MZ I2S模块怎么检测到勘误表

更新，似乎帧同步控制有点关闭这些芯片。虽然在SPIxCON寄存器中没有正确设置SSN，但在和声库中有一个错误，我可以得到从属模式。它被设置为0，禁用PIN SS的字时钟使用，并设置为GPIO。无论如何，它工作，我可以切换采样率为一段时间，然后它基本上失去它，它再次成为噪音箱。看起来是在帧同步控制中可能会受到热的影响。我在96MHz运行这个漂亮的光，MaqAdioOothiHiRes运行在192MHz。我运行应用程序大约10分钟左右，它开始拧紧。我把电源从电源上拔了回来，问题马上就来了。所以可能是热相关的，谢谢，戈登



      以下为原文

    Gang,
 
Update seems like the Frame Sync Control is a bit off on these chips. I am able to get SLAVE mode working though there is an error in the Harmony Libraries as SSEN is not set correctly in the SPIxCON register. It is set to 0 which disables the word clock usage of the pin SS and sets that to GPIO.
 
Anyway it works and I can switch sample rates for awhile then it just basically losses it and it becomes a noise box again. Looks to be something in the Frame Sync Control that is effected maybe with heat. I am running this pretty light at 96Mhz, the mac_audio_hi_res runs at 192Mhz.
 
I ran the application for about 10 minutes or so and it started to screw up. I removed the product from power plugged it back in and the problem was immediate. So probably heat related.
 
Thanks,
Gordon

非常糟糕的消息。我们正在等待Microchip的响应。我找不到PIC32 MZ的任何替代方案：-USB HSS与PHY-I2S中的BID一起使用50MHz时钟-I2S，外部参考时钟为JAREKCPS。我的技术支持案例还没有答案。



      以下为原文

    Very bad news for me.sad:
We are waiting for the response from Microchip.
 
I can not find any alternatives for PIC32MZ with:
- USB HS with buid-in PHY
- I2S with 50MHz clock
- I2S with external reference clock
 
Regards
JarekC
 
PS. I stil haven't any answer for my technical support case.
 

Jarek，对我来说，这完全是浪费了几个月的工作！！！！我发了一封电子邮件给我当地的分部，我们一直在和公司和编写代码的工程师一起工作。我几乎所有的电子邮件都是与公司相关的电子邮件。首先要知道SPI速度有一个ELAX处理器限制到25MHz的勘误表。它在勘误表数据表中，但没有别的。但实际上我认为这可能是我所看到的总体问题。我用MaqAdioooHiHiRes的代码，在他们的评估板上设置了所有的测试，并显示了问题。我只打算使用这个部分，直到我没有运行的样本率（24.566MHz），只要你不改变采样率。我船上有49152MHz和45.1584MHz的时钟，所以你会尝试让这个工作到768千赫采样率？如果你PM我，我可以为其他可能对你有用的部分提出建议。我真的很失望Microchip处理这个问题的方式。在一次电话会议上，一位工程师真的不尊重我，我不得不说。真的？？？？嘿，这是你的产品，你的软件，它不起作用。无论如何，它看起来不像微芯片对商业音频产品感兴趣，所以不要期望从他们那里得到任何回应。谢谢，戈登。



      以下为原文

    Jarek,
 
Well for me it's been a total waste of months of work!!!
 
I sent an email to my local branch, we have been working with corporate and the engineers who wrote the code. I pretty much have all the emails for anyone at corporate related to this.
 
First do know that there is an errata for the SPI speed that limits EFx processors to 25MHz. It's in the errata data sheet, but nothing else. But actually I think this maybe the overall problem with what I am seeing. I laid out all the testing on their evaluation board with unaltered code from the mac_audio_hi_res and showed the problems.
 
I was only going to use the part to 384Khz which I had running (24.576Mhz) as long as you don't change sample rates. I have both 49.152Mhz and 45.1584MHz clocks on board.
 
So you were going to try and get this to work to 768KHz sample rate? If you PM me I can make suggestions for other parts that may work for you.
 
I am actually really frustrated with the way Microchip handled this. In a conference call one of the engineers was really disrespectful of me and what I had to say. Really??? Hey it's your product, your software and it doesn't work. Anyway it does not look like Microchip is interested in commercial audio products so don't expect any response back from them.
 
Thanks,
Gordon

嗨，你在PIC32 MX项目中使用了和声吗？在新的和声2.06“释放笔记”：MaqAddiooHiHiRes演示：不改变采样率在音频播放，但它意味着什么重放？当I2S IE启用时，当I2S启用且FIFO不为空时，当AppDATA活动界面AdTeNeSeTeTe= APPUSUBSSPACKERY PARBACKIX立体声？关于JAREKC



      以下为原文

    Hi,
Did you use Harmony in the project with PIC32MX ?
 
In new Harmony 2.06 "Release note":

mac_audio_hi_res Demonstration:
•Do not change the sample rate during audio playback
 
But what does it mean during playback?
- when I2S ie enabled
- when I2S is enabled and FIFO is not empty
- when appData.activeInterfaceAlternateSetting = APP_USB_SPEAKER_PLAYBACK_STEREO?
 
Regards
JarekC

Jarek，好的，他们最终测试了它，并发现它失败并杀死了应用程序。好的，这里有……所以我有BT DEV工具包，包括ECx和EFX版本的PIC32 MZ。在我的产品上工作了四个月（其中3个试图解决采样率变化的问题）之后，我认为最好的办法是通过硬件和软件不受影响地把它吸引到微芯片上。我既有TEKTRONIX I2S模块，也有棱镜DSTITEII和几种USB协议。分析器。我可以在链中的任何地方验证数据。当然，如果你使用音频MIDI（应用程序/实用程序）设置采样率，然后断电并上电DEV板，这意味着MACO将在插入到设定速率后立即设置采样率，然后运行iTunes，然后iTunes W。Microchip的开发团队是如何在18个月前写下这一点的。所以他们知道这个问题。如果你使用Audirva、AAMRA、鲁恩、位完美、分贝、J河、纯音乐或任何与采样率匹配的DAW软件，那么输出通常会被完全损坏或者应用程序崩溃并重新启动并重新枚举。我把所有的数据和测试结果发送给我当地的FAE。然后我在Chandler和Microchip召开了一个电话会议，他们说这是唯一可行的方法。我说，我们一直使用PIC32 MX（PIC32 MX270和PIC32 MX470）这样的方式，在过去的2年中，有超过100K的客户。在那一刻，我被称为说谎者，有问题的工程师离开了电话会议。他们显然认为MX和MZ以相同的方式工作，而它们不这样。事实上，MX和MZ寄存器之间找不到任何差异。当然，假设在2X的频率（勘误表指出当前25MHz最大）时，新的EF系列出来，并有一些修复的I2C速度。但是坦率地说…如果I2S在改变采样率时不起作用。这对我来说是一个要求，对于任何商业音频回放或录制设备来说，正如我在其他帖子中所说的，我花了20个多小时转换应用程序，使得PIC32 MZ1024EF处理器是一个奴隶，而我的ESS技术DAC芯片就是主人。我真的很高兴当这工作了最初的10分钟，然后似乎PIC32 MZ丢失了位时钟和字时钟的边缘，声音变得模糊，然后完全死亡。这可能是在测试中错过的，因为MaqAdioooHiSyRes是以一个采样率测试的。如果你使用PIC32 MZ的SPI部分，那么你可能是很好的，因为你没有改变时钟速度，并且可能保持固定。我仍然认为，这应该写在所有PIC32 MZ处理器的勘误表数据表中。我和其他公司有过相同的经历。我并不真的想回到XMOS公司，但看起来我可能不得不这么做。谢谢，戈登。



      以下为原文

    Jarek,
 
Ok so they finnally tested this and sees that it fails and kills the application. Ok here goes...
 
So I have the BT Dev kit with both the ECx and EFx versions of the PIC32MZ. After working on my product for four months (3 of which trying to fix the sample rate change problem) I figure the best way to bring this to the attention of Microchip was via their hardware and their software untouched.
 
I have both a Tektronix I2S module and Prism dScopeIII and several USB protocol analyzers. I can verify data at pretty much anywhere in the chain.
 
Sure enough if you set the sample rate using Audio Midi (Applications/Utilities) and then power down and power up the Dev board, meaning that macOS will set the sample rate immediately when plugged into the set rate and run iTunes then iTunes will resample what ever the track is to that sample rate.
 
This is how the development team at Microchip wrote this over 18 months ago. So they knew about this problem then.
 
If you use Audirvana, Amarra, Roon, Bit Perfect, Decibel, J River, Pure Music or any of the DAW software which matches the sample rate to the track sample rate then the output is usually totally corrupted or the application crashes and reboots and re-enumerates.
 
I sent all this data and test results to my local FAE. I then had a conference call with Microchip in Chandler and they stated that this is the only way it works. I said that we have been using the PIC32MX (PIC32MX270 and PIC32MX470) that way for the last 2 years, have over 100K customers. I was called a liar at that point and the engineer in question left the conference call. They obviously thought that the MX and MZ worked the same way and they don't.
 
The fact is, I cannot find any differences between the MX and MZ registers. Sure it is suppose to work at 2x the frequency (errata of course points out currently 25MHZ max) when the new EF series comes out and some fixes to the I2C speeds. But frankly... if the I2S doesn't work when changing sample rate on the fly. Then it's pretty much dead to me as this is a requirement for any commercial Audio playback or record device.
 
As I said in the other posts I did spend more than 20 hours converting the application so that the PIC32MZ1024EF processor was a SLAVE and my ESS Technologies DAC chip was the master. I was really happy when this worked for the first 10 minutes, then it seemed as though the PIC32MZ was missing the edge of the bit clock and the word clock and the sound got fuzzy and then died completely.
 
This was probably missed in testing because the mac_audio_hi_res was tested at one sample rate. If your using the SPI portion of the PIC32MZ then you are probably fine as you are not changing the CLOCK speed and probably keep that fixed.
 
Oh well, boat load of time wasted. I am still arguing that this should be written up in the errata data sheet for all PIC32MZ processors. I have been in contact with other companies experiencing the same issue.
 
I didn't really want to go back to XMOS but it looks as though I might have to.
 
Thanks,
Gordon