AD8361: Operation outside the temperature range

Question True Power converter's (AD8361) operating temperature range from minus40C to plus85C. But storage temperature of this device range from -65C to +150C. Can this IC work at the temperature -50C? Or it will be destroyed or will work with a too large error? Answer The AD8361 is only tested and guaranteed down to –40degC. The part may operate at lower temperatures but performance is not characterized and not guaranteed. The part will not be damaged by operating at –50degC but mechanical damage to the package may result if the part is exposed to thermal shock (fast temperature changes) outside the datasheet limits.

AD8361: Using the EVAL board

Question I bought the AD8361 evaluation kit (AD8361ART-EVAL) from Silica Portugal (Avnet). When I turn it on I get a dc offset of some tens of mV. Can you tell me how to eliminate this offset? Answer The AD8361 is a wide bandwidth rms to dc converter. The bandwidth extends up to 2.5GHz. The eval board contains no pre-filtering so simply leaving the input unconnected and measuring the voltage VRMS, the AD8361 will pick up whatever spurious signals are in your lab and measure their power. Ensure that you have the board set up for Ground Referenced mode, as per Figure 2 in the datasheet. IREF -> VPOS, SREF-> GND. It is not meaningful to talk about offset as the output is a function of the power of the input signal, rather talk in terms of "intercept" and the intercept varies with the input frequency as per figure in the datasheet. To measure the intercept, use a set up (similar to TPC 21) with some accurate sine source with 75ohm source impedance through 75 ohm coaxial line. Vary the power level, plot the output level and infer the intercept.

What are the general criteria for selecting one type of RF detector over another?

Question What are the general criteria for selecting one type of RF detector over another? Answer The type of RF signal to be measured is the most important determining factor in the type of detector to use. For most general power measurement and control applications, the log type is the most useful. For pulsed RF signals, the log type is also best because of the fast response times available. In those applications where the signal has a high crest factor or a widely varying crest factor, the rms type is generally better. The crest factor is the ratio of the peak to rms value of the signal. For example, higher-order quadrature amplitude modulation (QAM) signals (e.g., 16QAM, 64QAM,and 256QAM) have high crest factors. In the case of spread-spectrum signals such as those used in CDMA and WCDMA cellular systems and orthogonal frequency-division multiplexing (OFDM) signals such as WiMAX and WiBro, the high crest factor (typically 10 to 13 dB) will change dynamically. In such applications, an RMS detector is generally more desirable.

Operating RMS Detectors at Low Frequencies

Q. Which RMS Power Detectors can be operated at low frequencies? ---------------------------------------------------------------------------------------------------- A. The term "low frequency" is relative. To someone working with microwave signals, 50 MHz can mean low frequency. In that case, all of our rms detectors, except for our current Wafer Level Chip Scale Package (WLFCSP) product line (ADL5500, ADL5502, ADL5504, and ADL5505), can readily measure frequencies as low as 50 MHz. These include the AD8361, AD8362, AD8363, AD8364, ADL5501, and ADL5902. If by low frequency you mean kHz (or lower!), then only devices that require external ac-coupling will work. Devices such as the ADL5501 have on-chip ac-coupling which disallow for lower frequency performance. The result of the internal ac coupling is a 3 dB high-pass corner frequency in the 10s of MHz which limits that power delivered to the detector. This restricts operation to 50 MHz or more. Any detector that has the ac coupling cap at the RF input externally has a controllable low frequency point. These devices include the AD8361, AD8362, AD8363, and AD8364 More information on how to operate these devices at low frequencies, including tradeoffs, can be found in the application note AN-691. http://www.analog.com/static/imported-files/application_notes/AN-691.pdf Note that the AD8363 and AD8364 (not released at the time AN-691 was publish) follow similar trends as the AD8362.

We have several AD8361s and all exhibit different levels of output voltage with no RF signal present. Is this normal?

Q. We have several AD8361s and all exhibit different levels of output voltage without any RF signal present. Is this normal? ------------------------------------------------------------------------------------------------------------------------------------------------------------------- A. In this context, offset is defined as the output voltage level that the detector exhibits when no RF input is present. The figure below shows the distribution of the output response vs. the input power for multiple devices. Below an RF input power of −21 dBm, the response of the AD8361 is no longer linear and begins to lose accuracy (see also figures 9 and 10 of the data sheet). As the input power is reduced, the spread of the output response increases along with the error. Although some devices follow an ideal linear response at very low input powers, not all devices continue the ideal linear regression to a near 0 V y-intercept. Some devices exhibit output responses that rapidly decrease and some flatten out. With no RF signal applied, the AD8361 has a typical output offset of less than 200 mV. If calibration is possible, then it is recommended that points be used in the linear region of the device, avoiding the nonlinear range at low extremes (or the high extreme where saturation is an issue). The AD8361 is accurate from about -20 dBm to 10 dBm (depending on supply, as shown in the data sheet). Fortunately, despite having the device response fan out at the low power levels (i.e. follow the ideal, rapidly decrease, or flattened out), the response curve of the AD8361 remains monotonic. The response follows a strictly increasing function and is even increasing (slowly) in the low power, nonlinear range. This behavior permits level shifting as a way to hide the offset by moving the curve down (offset being held under GND) by way of an op-amp. Derivative products follow similar offset behavior with typical output offset of less than 150 mV (ADL5500, ADL5501) or 100 mV (ADL5502, ADL5504, ADL5505).

RF Detector Recommendations

Q. We are currently using the AD8361 and are currently going through a redesign. We are looking for less temperature variation and better rms accuracy (i.e. waveform independence or the deviation from CW). What are your recommendations? -------------------------------------------------------------------------------------------------------------------------------------- A. Things have moved on quite a bit since we released the AD8361, which was our first RF rms detector. If you want to stick with a linear-in-volts transfer function and want good rms accuracy, I would make some of these suggestions: ADL5501: Similar detection range to AD8361 with better rms accuracy and temperature stability. It comes in an SC-70 package which some folks find easier to deal with than some of our newer wafer-level chipscale packages (flip chip). ADL5500: Similar to the ADL5501 in terms of dynamic range but with slightly worse (still better than AD8361) rms accuracy. This is the detector with the tightest temperature performance available. Though it has low drift across the entire 30 dB range over temperature and process variations, it boast a "sweet spot" of ±0.1 dB error from −40°C to +85°C over an 8.5 dB range with only a simple two-point calibration at room temperature. It comes in a small 1.0 mm x 1.0 mm wafer-level csp package (flip chip). ADL5504: Similar to the ADL5500 and ADL5501 but rms accuracy that is second to none. 1.2 mm x 0.8 mm wafer-level csp package (flip chip) ADL5505: A derivative of the ADL5504 with slightly worse rms accuracy but the smallest package on the market. It comes in a tiny 0.8 mm x 0.8 mm wafer-level csp package (flip chip) ADL5502: Same as ADL5504 core but with a second output that can either track the envelope of the signal or hold the most recent peak. Useful if you wanted to calculate the peak-to-average ratio of a signal, or the peak power. 1.5 mm x 1.5 mm flip chip ADL5511: Similar to ADL5502 but in a larger package. Has the separate envelope tracking output but no peak-hold. Highest detection range of all the devices listed here (around 40 dB). It comes in a 16-lead, 3.0 mm × 3.0 mm lead frame chipscale package.

AD8361 Output Offset Voltage

Question: I have an evaluation board for AD8361. With the board powered and no RF input connected the output is greater than 150 mV eventhough the specifications call out a maximum Output Intercept of 150mV. Is there something wrong with his part since its output is 157mV? Answer: Don't confuse Output Intercept with output voltage. The intercept is an extrapolation of the transfer function to the point at which it hits the y-axis. Or to put it another way, it is the output voltage that the part should provide for 0V in if the transfer function was fully linear, even at very low input levels. In practice, the transfer function becomes non-linear at low input levels (e.g. below 20 mV rms or -21 dBm). For some devices, the transfer function flattens out, yielding what looks like an output offset; for other devices, the output voltage will collapse to 0V. Take a look at Figure 50 in the ADL5501 Rev. B datasheet. This will give you a better sense for what happens with these devices. Incidentially, ADL5501 is probably a better choice than AD8361. It's not that AD8361 is not recommended for new designs. ADL5501 is just a newer better part.

AD8361 RMS Responding RF Detector - Overdrive Recovery

Question: What does the overdrive response of AD8361 look like? If I apply an RF burst which exceeds the linear input range of the device (approx +9 dBm on a 5V supply) but which is still less that the Absolute Maximum Rating (+13 dBm), will the device have an overdrive recovery time where it will stick high? Answer: Attached are four plots which demonstrate that the overdrive recovery of the AD8361 RMS detector does not show any anomalous behavior. The four plots show the output response when the device is driven by the following four signals. When the AD8361 is overdriven, the device behaves normally and does not stick high. 1. 0 dBm 900 MHz RF burst with CFLTR set to 1nF 2. +13 dBm 900 MHz RF burst with CFTLR set to 1 nF 3. 0 dBm 900 MHz RF burst with CFTLTR left open 4. +13 dBm 900 MHz RF burst with CFLTR left open