Flash ADC Back-End Digital Encoder

[neoflash]

bubble error flash adc

Hi, folks:

I want to survey experienced guys on flash ADC back-end encoders.

Let us assume the flash ADC is for data communication. What is the standard, or most commonly used, back end digital encoder, are used to solve bubble issue and metastability.

I've seen "NAND3" + "Gray ROM" + "Completion detection" combined as total solution. Just want experience designers to comment its performance and other popular ways in nowadays.

Regards,
Neoflash

 

[rfsystem]

Yes bubble error detection and correction is standard and helps to reduce static and dynamic accuracy requirements for the comparator stages.

I did not have a number but some sources say that the area could be reduced by a factor 2 or more for the same yield spec by correcting one bubble errors.

By the way the coding is straight binary and the complementary lines are used for error detection if also in the presence of 1 bubble correction there are further errors.

Both techniques are standard for more than 15 years now!

 

[neoflash]

rfsystem: You mean the coding rom is just binary? Then how about the metastability issue? If one word line metastable incur more than one LSB error, that will be a big problem.

 

[rfsystem]

Yes metastability is an issue everywhere but these are very rare situations. The gray encoder helps only if there is a latch architecture after the decoder. So if one code gets wrong that the numerical error is low. But the metastability is much worse at the flash comparators. I guess it is orders of magnitude more occurrence there than in later latches. Therefore the bubble correction is implemented.

The binary ROM encoder could be built with the error detection. This signal would also be very helpful for testing and online system diagnosis.