Post Go back to editing

Hi,

I am not an expert in RF; I am trying to design a signal power meter for satellite application using AD8307 log Amp.

Below is the schematic that uses BPF (BPF-C70+) to filter unwanted signals and to allow only 70MHz signals. I am able to see the 70Mhz RF signal in spectrum but the voltage measured at output of AD8307 is not as per calculated value 25mv/db

Please let me know where I am doing wrong.

Below are the different cases we observed while testing:

Case 1: (20dB input/fc:70MHz) consider filter Insertion Loss at 70MHz is 6.4dB the input to AD8307 is 13.6dB

• Input Signal to BPF: Signal Absolute Level = -82.3dbm / Noise Floor = -102.3
• Output: Voltage at Pin 4 of AD8307 is 0.51V without signal 0.48V (just at Noise floor level)

Case 2: (22dB input/fc:70MHz) consider filter Insertion Loss at 70MHz is 6.4dB the input to AD8307 is 15.6dB

• Input Signal to BPF:  Signal Absolute Level = -36.3dbm / Noise Floor = -58.3
• Output: Voltage at Pin 4 of AD8307 is 1.45V without signal 1.41V (just at Noise floor level)

Case 3: (25dB input/fc:70MHz) consider filter Insertion Loss at 70MHz is 6.4dB the input to AD8307 is 18.6dB

• Input Signal to BPF: Signal Absolute Level = -61.3dbm / Noise Floor = -86.3
• Output: Voltage at Pin 4 of AD8307 is 1.05V / without signal 1.04V (just at Noise floor level)

I also tested the same circuit by just replacing BPF (F455K000S011) with 455Khz one. below are the results observed.

BPF specification:  Bandwidth (+/- kHz) 3 kHz, Insertion Loss 6 dB, Input/Output Impedance (Ohms) 2000 Ohms.

• Input Signal to BPF: Signal Absolute Level = -27.3dbm / Noise Floor = -56.3
• Output Observed on Pin-4 of AD8307 is ranging from  1.834V to 2.136V and difference is 302mV. Ideally, I am expecting 25dB/V x 29dB = 725mV as per the datasheet.

Added input signal details to the 455khz filter configuration.
[edited by: ANURAG at 7:50 PM (GMT -4) on 25 May 2021]
• Hi ANURAG,

The setting of intercept adjustment R2 will affect the results. Consider taking data with INT pin left open at first, to make sure your measured data can be similar to the datasheet response of Figure 7, Otherwise the response will be more like Figure 9,

In all cases below, if input signal has complex modulation present, the measured results will be a little higher or lower, compared to CW (no modulation). To eliminate this error, an RMS-responding detector is necessary.

Case 1: RF signal in= -82dBm -6dB = -88dBm. Looking at Figure 7, -88dBm is well below the plateau of response curve. Results seem as expected.

Case 2: RF signal in= -36dBm -6dB = -42dBm. Looking at Figure 9, results could be anywhere from 1.1 to 1.4V, so signal level results seem close enough to expected. Noise level measures too high. Suggest using standalone RF Signal Generator for the test.

Case 3: RF signal in= -61dBm -6dB = -67dBm. Looking at Figure 9, results should be approx. 0.4 to 0.75V. Because measured results disagree, consider hooking up RF Signal Generator to test the detector independently.

Going from 70MHz to 455kHz, some capacitors may need to be increased in value, to support the lower input frequency. External capacitor on COFS pin is needed (ref. datasheet Figure 30 and accompanying text).

Simple log detectors such as AD8307 cannot separate signal from noise. The device will respond to the aggregate of both signal+noise.

For best measurement of the noise level (no signal), it may be necessary to also include a small RF 50 ohm attenuator between filter and detector (i.e. -3dB). This may help AD8307 provide lowest output voltage for the RF filter bandstop region where filter terminal impedance is a mismatch, either very low or very high Z.

Note: many AD8307 customers convert to AD8310, which has buffered output. In this case, it should eliminate the need for U3.   -Bruce H.

• Hi Bruce,
Thank you very much for the reply. As you suggested, I will float the INT pin and observe the results. I will add 1uF capacitor to COFS pin.
Can I change my design from Figure 44 in datasheet where intercept is shifted from -84dbm to -64dbm as noise floor in my design is -56dbm? And also could you please let me know how can I calculate the values of  C1,C2, R1, R2 and C2 for the desired intercept and  input RF(in my case -56dbm intercept and input RF: 455khz).

If I use AD8361, would it detect true rms power and ignore the noise floor?

• Hi ANURAG,

1. You mention Figure 44, but this figure in AD8307 datasheet pertains to LF operation. Did you mean some other figure?

2. Normal intercept is approx. -84dBm equiv. sine wave power into 50 Ohms. Shifting intercept down 20dB to -64dBm would be quite a large shift. I'm not sure that it is possible with the provided INT pin 5 provision. A preferred workaround would be to do this shift with external op-amp circuitry, for example the U3 circuitry on your schematic.

3. AD8361 is a lot different. Not to say it will not work, its just different. The measurement of noise is different when comparing log detector against rms detector (rms is more accurate in the case of noise). But a true rms detector will not ignore noise. Conversely, it will measure and respond to noise in a more accurate way compared to log detectors like AD8307 or AD8310.

Hope that helps.  -Bruce H.