acquisitionZONE Products for the week of September 20, 2010
Texas Instruments Says…
THS770006: Op Amp Delivers Low Distortion for Driving High-Speed 16-bit ADCs
Op amp maximizes performance across the signal chain for wireless broadband communications, high-speed data acquisition
Texas Instruments Incorporated (TI) introduced a high-linearity, low-distortion, fully-differential operational amplifier (op amp) that delivers 16-bit full-scale precision up to 200 MHz IF to maximize signal chain performance for wireless base stations, high-speed data acquisition, test and measurement, medical imaging and defense applications. The THS770006 op amp has an output third-order intercept (OIP3) of 48 dBm and the lowest distortion in its class, with a third-order intermodulation distortion (IMD3) of -107 dBc at 100 MHz, at least 14 dB better than competitive amplifiers.
"The advancements we've made in the performance of the THS770006 op amp will enable greater levels of receiver dynamic range in wireless systems – a requirement for base station manufacturers to meet the adjacent channel rejection and blocker requirements of next-generation networks," said Steve Anderson, senior vice president of TI's High-Performance Analog business unit. "Our customers can have confidence knowing the THS770006 op amp will help them realize the full potential of their high-speed signal chain."
The THS770006 differential ADC driver amplifier is being presented very much as an RF channel device by TI. I have only once ever seen what amounts to a wide-band op amp having noise figure specifications called up for it (see below). And I have never seen noise figures quoted for high-speed ADCs themselves. On querying this, Jim Karki, Manager, High-Speed Amplifier Strategic Marketing at TI commented, "Adequacy of noise performance of the THS770006 for driving 16-bit ADCs can be seen by comparing the noise figure of the THS770006 to the noise figure of a 16-bit ADC. Typical noise figure of a high performance 16-bit ADC NF (sic) is 30 dB to 33 dB, while that of the THS770006 is 10.5 dB to 13 dB. The 20 dB (or so) lower noise figure shows the THS770006 will be able to drive a 16-bit ADC will little impact on its noise performance."
These mathematics are indisputable with the THS770006 offering the quoted noise figures at 50 MHz and 200 MHz and throwing overall noise performance of the chain back to the source, which is probably going to be an LNA and downconverter. In quoting suitability for 16 bit conversions, however, we are more likely to be keener at looking at the input noise of the driver and the effect of that level of noise on resolution. In this case the input referred voltage noise is quoted at 1.7 nV/rtHz (>100 kHz) which is just about at the limit for effective 16 bit conversions to follow.
The THS770006 is probably partially targeted at the ADI AD8351 market (there is also an AD8352 which costs more, has poorer quiescent and slew rate but offers better distortion numbers) which makes no claim to 16-bit driving performance with an input referred voltage noise of 2.7 nV/rtHz. Comparing the two devices – as much as one can with data from two different vendors – the THS770006 is hungrier on power (with a single +5 V rail) at a typical 100 mA (800 nA power-down) compared to 28 mA while the second harmonic distortion is comparable in the upper range of -70 dBc above 100 MHz and the third harmonic distortion is much better at -86 dBc at 100 MHz. The slew rate of the TI part is considerably higher at about 3000 V/µs (compared to 1300 V/µs) and the small-signal bandwidth is slightly higher at 2.4 GHz (compared to 2.2 GHz).
Close comparison should also be made with the three-year-old Linear LTC6401, which is also a 2.2 GHz part with a 3400 V/µs slew rate and a quiescent of just 45 mA. The input referred noise of the LTC6401 is also low at 2.2 nV/rtHz, not, in my opinion, low enough for full advantage of 16 bit resolution conversions to follow (but these levels are low enough to be getting very subject to test conditions), noise figure is quoted at 12.1 dB, and IMD3 is a poorer -80.5 dBc at 140 MHz.
The 0.1 dB flatness of the TI part is typically 475 MHz while S11 is -20 dB at 200 MHz (compared to -30 dB for the LTC6401). The part operates with a nominal 6 dB gain and offers an output offset of ±1 mV with a CMRR of 60 dB. The differential input resistance is a typical, and very useful, 100 Ω.
The differential output voltage is 4.85 V with a 400 Ω load and it can drive 80 mA into 10 Ω (why not offer a comparison at 400 Ω?)
The THS770006 is approaching ideal for 16-bit IF analog-to-digital conversions. It is, inexplicably in a package where a full ten pins are either not connected or unused. The pricing is over a dollar higher than the LTC6401 – which might be tough to justify – but it is, nevertheless, going to do extremely well in the market. TI might do well to keep ahead of the game by offering a dual version of the amplifier to be able to get precise channel timing for wide-band I/Q modulation techniques: this is a problem area that analog vendors are only now going to have to face, are still largely unaware of, and one that is going to be huge within a few years. TI is suggesting the THS770006 could be mated with a new ADC, the ADS5493 (16 bit, 130 Msample/s), which does not appear to have had a news release yet.
The THS770006 differential op amp is in production in thermally-enhanced QFN-24 and is priced at $4.10 in 1000-piece lots. An evaluation module is available. The ADS5493 is sampling in thermally-enhanced VQFN-48 with production scheduled for Q1 2011 when it will be priced at $65, also in 1000-piece lots.