Post Go back to editing

AD8332 only 40dB gain in VGA


I am using AD8332 for low frequency NMR. RF part of my board closely resembles the evaluation board for AD8333. The difference is that i use onboard signal splitter for LNA from minicircuits SCP-2-1-6+, and connect LNA to VGA. Post-detection amplification is different, but it is of no concern.

Frequency of interest is 800kHz-10MHz. Signals as low as my generator can handle (in this case it is -127dBm). My problem is that i cannot seem to get full gain out of the VGA inside AD8332 chip.

LNA in this chip is said to give 19dB per channel in differential mode, and 13dB when measured single ended. LNA is working fine, and gives promised gain.

LNA is coupled to VGA via 100nF capacitor, as in figure 81, page 30 of the AD8332 datasheet.

Output has series termination of 100 Ohms per each of differential pair line, then is being fed to AD8333 through 100nF capacitor and a bias network as in Figure 60, page 26 of the AD8333 datasheet.

Gain pin gets 1V from external voltage reference and potentiometer, i can see the output signal react on potentiometer setting, reach full power at 1V, and doesnt go beyond, even though i have headroom of additional 0.25V.

Datasheet for AD8332 mentions that VGA is not intended to be used in single-ended mode (per channel), but i dont have a differential probe, so single ended measurements have to do. I expect 6dB loss in measured voltage because of this.

Enable for VGA and LNA are set to ON, HILO is set to HIGH, Mode is set to posive gain slope. AD8332 has his own LDO with 800mA of output current, and is well supplied.

Signal used for testing is 1MHz, -40dB. The signal splitter gives expected loss of 3.5dBm. -43dBm (1.6mV) is fed to LNA (50Ohm matched, but found that matching doesnt change my situation), and LNA outputs 6.8mV single ended which is roughly 12.5dBm gain. This is acceptable.

So this -30dBm signal is fed to VGA via 100nF, and VGA output i see is 246mV (+0.5dBm). Accounting for the single ended measurement, i add 6dB, and that gives me +6.5dBm per differential pair per channel at output, and total gain of 36.5dBm. With HILO setting set to HI, and Gain pin at 1V i would expect over 50dB of gain per channel on the VGA.

What am i doing wrong?

I checked this for 3 different pieces of AD8332, 2 samples and 1 bought. I get the same result when i decouple AD8332 from AD8333 by tombstoning the 100Ohm termination resistors.

The whole board is able to detect -127dBm signals on FFT (demodulated to 10kHz band, with post-detect amplification), and -122dB is visible in time domain on a basic scope. The missing 14dB would mean a world to me, as expected signals in real experiment are expected to be less that -127dB.

Best regards,