AD8676BRMZ-REEL
Product Specifications
AD8676BRMZ-REEL
Analog Devices — Analog & Mixed-Signal ICs
Product Overview
The AD8676BRMZ-REEL from Analog Devices is a high-performance analog / mixed-signal IC providing precision signal acquisition, conversion, and conditioning. Analog Devices (ADI) is a world leader in high-performance analog, mixed-signal, and DSP ICs for instrumentation, healthcare, automotive, and communications.
Superb Automation partners with leading brands including Qorvo, TI, ADI, XILINX, NVIDIA, Intel, ST, MACOM, Renesas, and Coilcraft to provide complete BOM and supply-chain solutions.
Key Features
- High-resolution data conversion with excellent SNR and THD
- Wide dynamic range and high linearity across operating conditions
- Integrated voltage reference with low temperature drift
- Flexible digital interface — SPI, I2C, or parallel options
- Industrial/automotive temperature range with guaranteed specs
- Low power consumption with power-down modes
Key Specifications
| Resolution | bits |
| Channels | — |
| Sample Rate | kSPS |
| Interface | — |
| Supply Voltage | V |
| Operating Temp | °C |
| Package | — |
| INL | LSB |
| SNR | dB |
| THD | dB |
Applications
AD8676BRMZ-REEL: 12 microvolts of offset and the 99 microvolts of drift behind it
The AD8676BRMZ-REEL is the dual member of the AD8676 family from Analog Devices, an ultra precision 36 V amplifier with 2.8 nanovolts per root hertz of voltage noise, a 12 microvolt typical offset and a rail to rail output. It sits between two familiar poles: the bandwidth and noise of a wideband op amp, and the dc precision and low bias current of an OP07 class part. The MSOP version is supplied in tape and reel only.
What the vendor publishes
| Parameter | Value |
|---|---|
| Channels | two, the AD8675 is the single |
| Noise | 2.8 nV per root hertz at 1 kHz, 0.10 microvolts peak to peak from 0.1 to 10 Hz |
| Offset | 12 microvolts typical |
| Drift | 0.2 microvolts per degree C typical, 0.6 maximum |
| Bias current | 2 nA maximum |
| Open loop gain | 120 dB |
| Bandwidth | 10 MHz typical |
| Output | rail to rail |
| Supply | plus or minus 5 V to plus or minus 18 V, specified from 5 to 15 |
| Temperature | -40 to +125 C |
| Package | 8 lead MSOP, also 8 lead narrow SOIC |
Drift is eight times the offset
A 12 microvolt typical offset sounds like the whole error budget. Over the full 165 C span from -40 to +125 C, the 0.6 microvolt per degree maximum drift is 99 microvolts, which is 8.25 times the offset the part starts with. That reverses the usual instinct. Nulling the offset at room temperature with a trim does nothing about the term that dominates at the ends of the range, and a circuit that was characterised at 25 C and shipped to a cabinet at 70 C sees the error triple before the trim is looked at again. The honest way to use the number is to take the drift times the span and compare that with the signal.
Noise against a 16 bit converter
At 2.8 nanovolts per root hertz the amplifier integrates to 280 nV over a 10 kHz bandwidth and 2.8 microvolts over 1 MHz. One LSB of a 16 bit converter at a 10 V full scale is 152.6 microvolts, so the wideband noise is about one five hundred and fortieth of an LSB. The conclusion is not that the amplifier is free, it is that the resistors and the reference in front of the converter usually contribute more. Gain is 120 dB, a ratio of one million, and with 10 MHz of bandwidth the closed loop bandwidth at a gain of 100 is still 100 kHz.
Bias current against real source impedance
Two nanoamps of bias is nothing until it meets a source impedance. Into 10 kohm it is 20 microvolts, which is already larger than the 12 microvolt offset. Into 100 kohm it is 200 microvolts, which dwarfs both the offset and the drift over a short span. For a bridge or a thermocouple the source impedance is the first number to write down, before the offset column is even read.
Two questions buyers ask
Is this the part for a thermocouple? For the amplifier stage yes, if the cold junction and the source resistance are handled. The 0.10 microvolt peak to peak low frequency noise is the figure that matters at thermocouple bandwidths, and it is far below the 40 microvolt per degree C of the junction itself.
Dual or single? The AD8675 is the single channel version in the same family, which matters when four channels have to fit on a board without four separate packages.
SUPERB quotes the AD8676BRMZ-REEL against this row and confirms tape and reel quantity before shipment. Amplifier alternatives are compared under amplifier cross reference, a same footprint check starts at pin to pin cross reference, and the category index is analog and mixed signal ICs.
AD8676BRMZ-REEL carries the 60 uV limit, not the 50 uV one
The 12 uV offset in the summary is a typical figure. The ordering table splits the maximum across the package: the B grade in the SOIC body is 50 uV, the B grade in the MSOP body is 60 uV, and the A grade is 100 uV in either body. This order code is the B grade in the MSOP, so the limit that applies is 60 uV at 25 C and 160 uV over -40 C to +125 C. A board qualified on the SOIC line and then moved to the smaller body for space keeps the grade letter and loses 10 uV of guaranteed offset.
Drift is larger than the offset it is measured against
Offset drift is 0.2 uV per degree C typical and 0.6 uV per degree C maximum. Over the 165 degree industrial span the maximum figure is 99 uV, which is 1.65 times the 60 uV offset limit of this package. Trimming at 25 C therefore buys very little unless the drift term is already inside the budget, and the distribution curves in the data sheet show most units well below the limit, which is the reason a typical based budget appears to work.
Rail-to-rail output with an input that stops 2 V short
The output swings to within 100 mV of either rail into 10 kOhm, 4.90 V minimum on a 5 V supply, and to within 200 mV into 2 kOhm. The input is a different story: the common mode range is plus or minus 3.0 V on plus or minus 5 V supplies and plus or minus 12.5 V on plus or minus 15 V, so it stops 2 V and 2.5 V from the rails. A unity gain follower of a signal that reaches the rail will clip before the output does, and the 105 dB minimum common mode rejection is quoted over the usable range only.
10 MHz small signal against 2.5 V per microsecond
The gain bandwidth product is 10 MHz and the closed loop response into a 2 kOhm load with 35 pF holds 55 degrees of phase margin. Slew rate is 2.5 V per microsecond, which at a 20 V peak to peak output gives 39.8 kHz of full power bandwidth and at 2 V peak to peak gives 398 kHz. A stage that needs 10 MHz of large signal bandwidth is 250 times outside what this slew rate can deliver, and a stage that only needs small signal bandwidth is inside it.
Bias current against the source, and the 0.7 V clamp
Input bias current is plus or minus 2 nA maximum at 25 C and plus or minus 5.5 nA over temperature on plus or minus 5 V. Into 100 kOhm the over temperature figure is 550 uV, nine times the 60 uV offset of this grade, and the two terms cross at about 30 kOhm. The absolute maximum differential input voltage is plus or minus 0.7 V because of the input protection diodes, so the part is not a comparator and a fault that puts more than 0.7 V across the inputs needs an external clamp.
Thermal and sequencing
Each amplifier draws 3.4 mA maximum at 25 C on plus or minus 15 V, so the dual is 6.8 mA and 0.204 W. In the MSOP body with 210 C per W that is a 43 C rise, which puts the junction near 128 C in an 85 C ambient; the SOIC body at 158 C per W gives 32 C instead. The data sheet also states the sequencing rule: the supplies have to be established with or before the input signal, otherwise the input current must be limited to 10 mA.
What to check before the order
- Package and grade together, 60 uV for the B grade MSOP and 50 uV for the B grade SOIC.
- Input signal swing against plus or minus 3.0 V on a plus or minus 5 V supply.
- Load, since the rated output current is plus or minus 20 mA with a 40 mA short circuit limit.
- Large signal bandwidth from the 2.5 V per microsecond slew rate rather than from the 10 MHz gain bandwidth.
SUPERB supplies the AD8676BRMZ-REEL as available through distribution, with the grade letter and the body confirmed on the quote. The wider amplifier range is listed under amplifiers and comparators, the surrounding analog devices sit under analog and mixed-signal ICs, and replacing an obsolete precision amplifier inside a mixed bill of material is covered in BOM shortage recovery.
AD8676BRMZ-REEL in front of a 16 bit converter: the gain sets the settling time, and the feedback resistors set the noise
The two figures that decide whether this amplifier can drive a 250 kSPS channel are 10 MHz of gain bandwidth and 2.8 nV per root hertz. At a gain of 100 the closed loop bandwidth is 100 kHz, one time constant is 1.59 us, and 13.8 of them, which is what settling to 0.001 percent takes, is 22 us. A 250 kSPS channel allows 4 us. The same part at a gain of 10 gives 1 MHz, a 159 ns constant and 2.2 us of settling, which fits a 4 us channel with about 1.8 us left for acquisition. The gain is therefore a conversion rate decision before it is an accuracy decision.
What the gain does to the offset and the drift
A non-inverting stage multiplies the input offset by the gain. The 60 uV limit of the B grade in the MSOP becomes 6.0 mV at the output at a gain of 100, which is 39 counts of a 16 bit converter on a 10 V range at 152.6 uV per count. The 160 uV over temperature limit becomes 16 mV, 105 counts. Trimming at 25 C removes the first term and leaves the second, and the second is the larger of the two at the ends of the range, which is the reason a gain of 100 stage is usually trimmed in system rather than at the amplifier.
The feedback network, not the amplifier, is the noise source above a gain of 10
A gain of 100 needs a 99 kohm feedback resistor against a 1 kohm ground resistor, and 99 kohm at 300 K produces 40.5 nV per root hertz of thermal noise. That is 14 times the 2.8 nV of the amplifier, so the resistor dominates the input referred noise of the stage and no amount of amplifier selection recovers it. Dropping to 9.9 kohm and 100 ohm brings the resistor noise to 12.8 nV and still leaves the amplifier in second place; the cost is 10 times the load current in the feedback network and a heavier output drive for the same signal level.
| Gain | Closed loop bandwidth | One time constant | Settling to 0.001 percent | Channel time at 250 kSPS | Feedback resistor noise |
|---|---|---|---|---|---|
| 1 | 10 MHz | 15.9 ns | 0.22 us | 4 us | not applicable |
| 10 | 1 MHz | 159 ns | 2.20 us | 4 us | 12.8 nV at 9.9 kohm |
| 100 | 100 kHz | 1.59 us | 22.0 us | 4 us | 40.5 nV at 99 kohm |
| 100 with a 1 kSPS channel | 100 kHz | 1.59 us | 22.0 us | 1 ms | 40.5 nV at 99 kohm |
Where the output swing and the load meet the settling number
The output reaches within 100 mV of either rail into 10 kohm and within 200 mV into 2 kohm on a 5 V supply, and the rated output current is plus or minus 20 mA with a 40 mA short circuit limit. A 99 kohm feedback resistor at a 4.9 V output carries 49 uA, so it does not load the stage, while a 9.9 kohm network carries 490 uA and is still an order of magnitude below the rating. Settling to 0.001 percent into a 2 kohm load is slower than into 10 kohm because the output stage has to charge the same capacitance with less available drive, so the 22 us figure belongs to the light load case and a 2 kohm load should be treated as a longer channel rather than the same one.
Comparisons against other precision amplifiers are on the amplifier cross reference page, the grade and body checks before a swap are on how to verify a drop in replacement, and the Analog Devices second source route is on ADI part second source.
Request Quote — In Stock, 100% New & Original, Full Warranty