# Frequency mixing at low frequencies (with AD835)

I am using the AD835 to downconvert signals from around 1 kHz to around 10 Hz. While testing I found that the mixer behaves differently when using for example an LO at 1 MHz and RF at 999,99 kHz opposed to an LO at 1 kHz and RF at 1.01 kHz - both yield the same IF at 10 Hz, but have different amplitudes. Calculating the amplitudes using the given transfer function $$W=\frac{(X_1-X_2)\cdot(Y_1-Y_2)}{U}+Z,$$ where $X_1,X_2$ and $Y_1,Y_2$ are either of the differential RF or LO inputs, $U=1$ and $Z$ an additive term set to zero in my case, I can not match the results (which of course should be equal regardless of the frequency range of the RF and LO as long as IF is the same)?! Just for convenience the equation I get out of this assuming two sinusoidal signals: $$\frac{A_1\cdot A_2}{2\cdot U}\cdot \{[cos(\omega_1-\omega_2)t-\varphi]-[cos(\omega_1+\omega_2)t+\varphi]\}$$ with RF amplitude $A_1$ and frequency $\omega_1$, LO Amplitude $A_2$ and frequency $\omega_2$ and phase shift $\varphi$, if there is one (at first I generated both LO and RF with function generators with unknown phase shift, later I used a NI 6361 to generate signals with known phase shift).

While reading the datasheet again very thoroughly I noticed all plots being for frequencies starting in the MHz range. In general the mixers I find are designed for high frequency applications. In my opinion at least in theory there should not be a problem with low frequencies though.

So my question is: Is there any reason why this multiplier (and also others that specify the 3 dB maximum frequency and are wideband) should not work for low frequencies? If there are, are there recommendations for mixing devices or other approaches that fit my frequency range (from some kHz down to around 0-100 Hz)?

• How much of a difference are you seeing? – mng Jan 16 '15 at 19:06
• Don't neglect that $\omega_1 + \omega_2$ term. When your signal and LO are at 1 MHz, that unwanted output term is at 2 MHz, but when they are at 1 kHz, the unwanted term is at 2 kHz, three orders of magnitude closer to your desired signal. How exactly are you filtering the output? – Dave Tweed Jan 17 '15 at 2:12
• @ mng: I get differences ranging from -2 dB even up to -12 dB, depening on the frequency region. The maximum value in this cases is not at the expected IF frequency but slightly shifted. – Irenaius Jan 19 '15 at 7:37
• @ Dave Tweed: I do not filter the output just yet, but since I evaluate using a AD converter card or a spectrum analyzer I do not think it would make a difference. – Irenaius Jan 19 '15 at 7:39