Search

JESD204C: A New Fast Interface Standard for Critical Applications - Electronic Design

kembaliui.blogspot.com

Download this article in PDF format.

As data-conversion circuits have sped up over the years, the current serial interfaces haven’t been able to keep pace. But a new interface has arrived, called JESD204C. It’s the next iteration of the JESD interface standards. This article introduces JESD204C, explains its features and benefits, and highlights the differences with its predecessor JESD204B.

 Sponsored Resources: 

What is JESD?

JESD204C is a standard of the Joint Electron Devices Engineering Council (JEDEC). It’s a high-speed interface designed to interconnect fast analog-to-digital converters (ADCs) and digital-to-analog converters (DACs) to high-speed processors, FPGAs, and ASICs.

The standard first appeared in 2006, offering a 3.125-Gb/s data rate. Subsequent releases implemented improvements in function and speed. The JESD204A version in 2008 added lane synchronization that kept multiple serial lanes aligned with multiple converters. The JESD204B upgrade in 2011 boosted speed to 12.5 Gb/s and incorporated deterministic latency. With the new JESD204C version, the interface data rate jumps to 32.5 Gb/s, along with other improvements in the mix. By the way, the newer versions of the standard maintain some backward compatibility with previous versions.

Year after year, ADCs and DACs continue to push the envelope when it comes to speed. Gigabit rates are common. And those circuits transmitting or receiving this high-speed data, such as FPGAs or fast MPUs, have been able to match these perpetual speed increases.

However, the limiting factor has revolved around interconnecting these devices.

The previous JESD204B standard became widely used to allow these fast circuits to talk with one another. Data exchanges take place between chips on a PCB, on backplanes, and via short cables between modules or subassemblies in an instrument or other equipment. But as some have found, the JESD204B specification is no longer adequate to interface the fastest data-conversion circuits. Now, JESD204C brings the interfaces up to speed.

Applications

The JESD204C interface can benefit almost any design where data converters run at high megabit and gigabit data rates. A sampling of the applications includes:

  • Test instruments
  • 5G cellular equipment
  • Other cellular base-station gear
  • Medical apparatus
  • Military/electronic warfare (EW)
  • Satellite
  • Almost any direct-conversion software-defined radio (SDR)

JESD204B and JESD204C Compared

The JESD204B standard was a real breakthrough and has been widely adopted. It offered a 12.5-Gb/s data rate, multiple lane support and synchronization, 8B/10B encoding, and deterministic latency. The JESD204C standard has all of the features of its predecessor plus some added new benefits such as the 32.5-Gb/s data rate, 64B/66B encoding, and deterministic latency.

The key addition to the standard is the 64B/66B encoding. It also supports a 64B/80B encoding option. The encoding method used in the earlier B version of the standard is 8B/10B, which means that each 8 bits to be transmitted is first converted into a unique 10-bit number. There are a couple of good reasons for such encoding:

  • The encoded character now contains more bits and bit transitions that are helpful in clock recovery and the alignment of the data streams.
  • The encoding ensures that the average dc on the line is constant. Random numbers of bits cause a varying or wandering dc average on the transmission path that can introduce bit errors and cause other annoying operational characteristics.

While 8B/10B encoding is beneficial, it also slows down the transmission. Ten bits are transmitted instead of eight, so it takes longer. While the data rate (R) is still based on the bit time (t), the net data rate is 20% less (R = 1/t).

The JESD204C standard uses 64B/66B encoding. It not only improves dc balance, clock recovery, and data alignment, but also has a much smaller bit overhead of 3.125%, considerably less than the 8B/10B method.

The signal processing includes a scrambler prior to the encoding. A scrambler is a digital circuit that processes the serial data to create a highly randomized version. Like a cyclic redundancy check (CRC), it uses an algorithmic process based on a polynomial, in this case 1 + x14 + x15. The end effect of scrambling is a serial digital signal with a spectrum that greatly reduces the generation of EMI.

Deterministic Latency

Determinism is the characteristic of being able to predict or determine when certain actions will occur. The opposite of determinism is randomness. Something that’s deterministic can predict when operations will happen with precision and repeatability.

Latency is the time delay that occurs between the start of an operation and its end. Latency in general is the total time that it takes for an electronic operation to take place or a signal to travel from one part of the system to another. Latency is measured in seconds or clock cycles.

The combination of the two is critical to the actual performance of this device or equipment. Examples include equipment that have two or more data converters for things like I & Q conversion. Interleaved converters are another example. When multiple inputs or outputs must be phase-coherent, the JESD204C standard can provide that condition. The resulting deterministic latency must be constant and repeatable for proper operation in most systems.

The latency of an ADC is the time between the capturing of the analog sample, the subsequent processing by the ADC, and the appearance of the digital output. The latency of a DAC is similar—the time between the digital input to the DAC circuits and when the corresponding analog signal first appears. This latency is measured in seconds or by the number of clock cycles.

Deterministic latency is generally defined as the time between the occurrence of the data input to the transmitter and output at the receiver. It doesn’t include the ADC and DAC latencies (see figure). The ADC develops the signal that’s processed into the frame of data to be transmitted. The transmitter (TX) circuit processing includes scrambling, 8B/10B or 64B/66B encoding, and some alignment of data. The data is transmitted over the serial link, possibly in parallel with other serial links associated with multiple ADCs.

At the receiver (RX), the signals are decoded and synchronized, then descrambled. A buffer memory is used to provide some timing adjustment if needed. Note the total link latency that must be consistent to achieve deterministic latency.

In most high-speed applications, it’s essential that data from multiple streams pass through the system with a consistent known delay. The standard defines some signals to achieve that goal. Different signals and clocks help deliver lane synchronization, lane alignment, and error detection.

The primary clocks are the data frame clock and the Local Multi-Frame Clock (LMFC). The SYSREF and SYNC signals are used to provide time reference alignment. SYNC is a receiver-to-transmitter signal that’s used for synchronization requests and error reporting. A SYSREF signal is a phase timing reference that influences the LMFC.

 Sponsored Resources: 

Download this article in PDF format.

Traditionally, resistance-temperature-detector (RTD) sensor resistance is measured by applying a precision current source and measuring the developed voltage. This approach usually requires a precision voltage reference to create then current source, followed by a high-quality analog-to-digital converter (ADC) to measure the voltage.

This isn’t difficult to achieve at room temperature, but when you consider that the temperature of your measuring system can be in the range of −40 to +55 C°, the task becomes more daunting.

A brute-force approach to this problem would be to use an expensive temperature-stable voltage reference, ADC, and other components combined with software calibration to compensate for the temperature drift of parameters. This approach is complex and will fail to achieve the high precision that borderlines the sensor accuracy.

A better approach was discovered using 5 ppm/°C ultra-stable resistors with 0.1% accuracy as a reference for RTD measurements. This approach requires two onboard ultra-stable resistors for calibration (1k and 2k) to achieve high RTD precision. Those resistors are used to calibrate the RTD reading and compensate for temperature-drift errors.

The design uses the Q1-Q3 transistors in combination with the R1 resistor to form a constant-current source that sources about 1 mA using a 2.5-V ADC voltage reference (see figure). Calibration resistors R4 and R5 along with the RT1 and RT2 RTD sensors can sink this current when the corresponding GPIO pin is driven low. When not used, the GPIO pins are tri-stated. The voltage is measured at the ADC output.

The Q1-Q3 transistors and R1 resistor form a constant-current source that sources about 1 mA using a 2.5-V ADC voltage reference.The Q1-Q3 transistors and R1 resistor form a constant-current source that sources about 1 mA using a 2.5-V ADC voltage reference.

For calibration, we will need to read two resistors and calculate the constant-current source value and combined errors, which we call VOffset. The calibrated Icc and VOffset values are used to convert RTD temperature readings.

Vardan Table

The calibration results (see table above) are applied using following formula:

Eq1

where:

  • RRTD is the measured RTD resistance.
  • VADC is the ADC voltage reading.
  • ICC is a somewhat constant-current source.
  • VOffset is the voltage offset of cumulative errors. Note that this variable is a combination of multiple error voltage sources. Therefore, it might be beneficial (though not necessary) to split into its components to get better precision.

To calculate VOffset and ICC, we need to make a few assumptions to derive the formula below:

  • First assumption: Calibration resistors are ideal and have 1,000- and 2,000-Ω values, respectively.
  • Second assumption: The ICC current source is stable for the duration of measurements.
  • Third assumption: ADC conversion results are perfect.

Following those assumptions, we can write that:

Eq2

In Equation 2, VCAL1K and VCAL2K represent voltages developed on calibration resistors when ICC current is applied.

By solving those equations for ICC and VOffset, we get the following equations:

Eq3

Eq4

Experimental Setup Measurements

The experimental setup has two calibration resistors and two RTDs mounted on different locations. We used an ADC with 10-bit resolution and surface-mount RTDs rated 1k at room temperature. Notice how the calibration values change when board temperature changes between sample 2 and 3 in the above table.

To gather the data, the software followed these steps:

1.   Read ADC voltage levels on calibration resistors and RTDs.

2.   Calculate Voffset using Equation 3. 

3.   Calculate ICC with Equation 4.

4.   Determine RTD resistance utilizing Equation 1.

5.   Convert RTD resistance values to temperature with the table lookup and piecewise interpolation.

Vardan Antonyan is Senior Electrical Engineer at Glenair Inc.

Let's block ads! (Why?)



"interface" - Google News
August 19, 2020 at 12:19AM
https://ift.tt/315OY2U

JESD204C: A New Fast Interface Standard for Critical Applications - Electronic Design
"interface" - Google News
https://ift.tt/2z6joXy
https://ift.tt/2KUD1V2

Bagikan Berita Ini

0 Response to "JESD204C: A New Fast Interface Standard for Critical Applications - Electronic Design"

Post a Comment

Powered by Blogger.