Maxim Integrated Products introduced the MAX109, the industry's highest performance, 8-bit, 2.2Gsps analog-to-digital converter (ADC). This device offers excellent wideband dynamic performance that has been optimized for capturing input frequencies in the second Nyquist zone. Fabricated using an advanced SiGe process, the MAX109 integrates a high-performance track/hold (T/H) amplifier, a quantizer, and a 1:4 demultiplexer on a single monolithic die. At a sample rate of 2.2Gsps and an input frequency of 300MHz, the ADC achieves a spurious-free dynamic range (SFDR) of 62dBc and a signal-to-noise ratio (SNR) of 45dB. The SNR remains flat (within 1.6dB) for input frequencies all the way up to 2GHz. The MAX109 can also digitize/synthesize signals directly at RF, thus reducing or eliminating the need for RF downconverters. This excellent wideband dynamic performance and integrated functionality make the MAX109 ideal for applications such as broadband communication receivers, radar systems, and high-end instrumentation systems.

  

 

This has been a fascinating 8-bit ADC product family to follow: from the 250 Msample/s MAX100 through the 1000 Msample/s MAX104 to the 1500 Msample/s MAX108 and now up to 2200 Msample/s with the MAX109. The latest offering has also resulted in an increase in package pin count because of the move to a 1:4 demultiplexed LVDS output system. It can also interface directly with external FPGAs with both dual (DDR) and quad (QDR) modes.

 

National Semiconductor produces much lower power high-speed ADCs but Maxim seems to just eclipse them in the ac specifications they extract from their parts. But, as always, this is an applications area where what you get in the finished product does not always tally with what you get on the bench -- or in the data sheet. There are also delivery questions about some of the products that have been announced elsewhere. The SiGe process being used allows for decent power rail voltages avoiding the noise hazards, and headaches, of sub-2-V rails.

 

The MAX109 uses split 5 V rails with the output circuits operating at 3.3 V. Total power is 6.8 W (typical) and both negative and positive PSSRs are 50 dB. I did not see any shutdown, or sleep mode, information.

 

With 2.2 Gsample/s clocking SINAD ranges from 44.1 dB at 300 MHz input to 40.1 dB at 2500 MHz input. Over the same range SFDR numbers are 61.7 dBc to 45.0 dBc, and THD numbers are -55.6 dBc to -43.7 dBc. Clocking at 2.5 Gsample/s worsens all the numbers, a little, with the biggest changes of about 3 dB on the THD returns.

 

The input small-signal bandwith is flat to 200 MHz and then falls off close to -2 dB at 1 GHz. It is then peaked to achieve the same at 2 GHz and falls to -4 dB at 3 GHz. The input can be single-ended, making it much easier to deploy in RF systems. The voltage range at the input is 0.5 V.

 

The most negative thing about the MAX109 is that, because of the increased pin count, it is not back-compatible with its earlier family members so updating existing designs is made much more complicated. The power numbers are also a little frightening. But it will succeed in spite of those things and will be extremely profitable. Pricing is a very sensitive topic for high-speed parts like this and Maxim has not quoted one; but it is probably worth $350 - $400 in 100+ piece lots and the part will find its way into all sorts of high-speed RF systems most of us will never know about, plus the more prosaic communications systems and digital oscilloscopes. None of these arenas have customers with really tight budgets.

 

The material from Maxim never exactly says the MAX109 is shipping -- but it is certainly implied. It will be in SBGA-256 and, as noted, pricing has not been disclosed.