Question: I'm looking for an RF RMS detector with high range (greater than 40 dB) which will work at very low input frequencies. Which device would you recommend?
Answer: There are a number of options. HMC909, HMC1010, HMC1020 and HMC1120 are dc-coupled and can all operate at arbitrarily low input frequencies but not at dc. All of these devices have an internal offset compensation loop. It detects and removes internal offset voltages and low frequency signals so that the detector core can detect very small signals. You can think of it as being analogous to an input high-pass filter. The corner frequency of this offset compensation loop can be externally set by attaching a capacitor to a particular pin on the device. The individual datasheets explain this and provide equations for choosing the value of this capacitor. The high-pass corner frequency of this offset compensation loop should be set to be lower than the expected minimum input frequency. A good rule of thumb is to set the corner frequency lower by approximately a decade than the minimum expected input frequency.
When applying relatively low input frequencies, there is a second consideration. With every rms computation, there must be averaging. This is the "m" portion of r(oot) m(ean) s(quare). To set the averaging on these devices, you use the SC1-SC4 control pins. SCI 0000 corresponds to the least amount of averaging and SCI 1111 corresponds to the highest amount of averaging. Each step increase in the SCI setting corresponds to a reduction of the corner frequency of the averaging by half, or one octave.
Because low frequency input signals will also have a relatively low bandwidth (e.g. the maximum bandwidth of a 10 KHz carrier is 20 KHz), these signal will require a high SCI setting (I.e. low averaging corner frequency) to ensure enough averaging for a valid rms computation to take place.
On HMC909, HMC1010 and HMC1020, the max allowable SCI setting is 1100. Exceeding this setting these devices will not cause any damage but will result in incorrect operation.
On HMC1120, the max SCI setting is 1111 which facilitates heavier averaging of the input signal. As a result, this device is probably the best choice for low frequency operation.
There is a good plot in the HMC1120 datasheet that relates rise time, fall time and residual ripple to SCI setting for an 4 Carrier WCDMA
signal and for an 8-tone signal with 100 KHz carrier separation. Notice how the wider band signal (4C WCDMA which has a bw of approx. 20 MHz) has less output ripple because it is getting more heavily filtered than the other carrier which has a bw of 800 KHz. This is a bit counter-intuitive but makes sense if you think about it for a while. So at a particular SCI setting the narrower band signal has relatively more energy inside the pass-band of the averaging filter compared to the wider band signal which has relatively more energy outside of the pass band. More energy inside the pass band equates to more residual output ripple/noise.