Record 64 bits at up to 8MHz on 4 inputs (on nRF52832)

[Cedric Honnet]

Hi all!

Is there a way to record 64 bits at up to 8MHz on 4 inputs? (on nRF52832)

The closest seems to be with 3 SPI and EasyDMA, would it work? (is there a way to do better?)

 

Otherwise, it seems possible on nRF52840 with QSPI and EasyDMA:

http://infocenter.nordicsemi.com/index.jsp?topic=%2Fcom.nordic.infocenter.nrf52840.ps%2Fqspi.html

Is it true?

 

Thanks for reading this!

Cedric ;)

[haakonsh]

Hey Cedric,

We need the specs for your communications. 

Cheers,

Håkon.

[Cedric Honnet]

Hey Håkon!
Thanks a lot for this quick answer!

Here are details about the com:
- 4 parallel streams (no TX, just RX)
- 64 contiguous bits at 8MHz

Is it possible with SPI and EasyDMA?
Thanks again!

(let me know if there's any other info I could add)

[haakonsh]

No, I'm afraid that wont be possible. SPI is not a contiguous bitstream, but separated into individual bytes controlled by an SPI Master's clock and slave select signals. F.ex:

An option is using four external 64-bit shift registers, they usually have SPI and/or TWI/I2C interface. 

[Cedric Honnet]

That's what I was worried about ;(
...is there a way to record the states of at least 1 GPIO at about 8MHz using EasyDMA then? (for about 64 bits)
Thanks again!

[haakonsh]

No, I'm afraid not. You might be able to read a pin at 8MHz, but I doubt it. The GPIO does not have access to a shift register and will therefore have to be manually sampled at 8MHz intervals (8 CPU cycles). 

External shift registers are probably the only option you have. 

Do you have a link to the datasheet of the device your are interfacing with?

[Cedric Honnet]

I wished...

It's to decode signals of the HTC Vive LightHouse V2.0 (pre-cleaned by ts4231), a reverse engineer session was started but it's still a work in progress:

• https://youtu.be/Ij1oqvcllUs?t=8844
• https://github.com/cnlohr/esptracker/issues/1#issuecomment-410549534
• https://github.com/cnlohr/esptracker/issues/4#issuecomment-412240505

The simplest is with an FPGA but they do it with an ESP8266, so hopefully the nRF52 has enough power!

Thanks again,
Cedric ;)

[haakonsh]

Do you have a scope of the bitstream? I've read the ts4231's datasheet but it does not specify the final bitstream. It seems the nature of the bitstream it generates is dependent on the IR source bitstream. 

The ESP8266 has a 160MHz CPU so it might just be able to capture the stream.
Do you have the actual ESP8266 implementation? It would be useful to know whether they've got some HW support for capturing bitstreams or if their just bit-banging it with the CPU.

If they're bit-banging it and decoding it with a LFSR where both is implemented in the CPU then the nRF52 is just not fast enough. 

You might be able to make an LFSR with descrete components, but that is kinda where FPGAs start to make a lot more sense.