I am making a Direct conversion receiver from this circuit at this website.

direct conversion receiver schematic

I made it for the 80m band and I changed:

  • The frequency of VFO to about 1.75 MHz
  • L1 and L2 to a bandpass filter for 80m
  • and the VFO coil is not like the one in this circuit.

Everything works very well, but on some frequencies I hear loud AM broadcast stations. I think this is caused by harmonics from the VFO, but I am not sure.

How can I get rid of this AM broadcast interference? Find out what frequency the AM broadcast is on, and try to filter it?

  • 1
    $\begingroup$ Hello Mathias, and welcome to ham.stackexchange.com! We look forward to seeing more of you here. I have edited your question slightly to make it a little easier for others to understand and answer it. Please consider reading the tour and help pages to get the most from this site. $\endgroup$ Commented Jan 26, 2020 at 18:14

3 Answers 3


It might be harmonics of the oscillator, as you suppose, but it also might be that any mixer doesn't only produce the difference frequency of its two inputs ($|f_\text{RF}-f_\text{local oscillator}|$, here), but also higher-order intermodulation products (like $|2f_\text{RF}-f_\text{LO}|$ and $|f_\text{RF}-2f_\text{LO}|$).
I think I've posted multiple introductions to intermodulation product/mixing here, but the one I can find right now is this.

Either way, since you can't change that your mixer somehow mixes a station outside your band of interest into your baseband or onto your IF, the only way to stop it is, as you proposed, to filter it out before it reaches the mixer stage.

Luckily, that's a very sensible thing to do: Since you're designing your radio to use the second order intermodulation in your mixer, the lowest frequency that gets mixed down is also the one that is of interest to you. So, low pass filter your signal!

That could happen at the output of your RF preamplifier¹, so, you could just add a series resistor followed by a capacitor to ground after the "100" capacitor (hope there's some document stating what the unit of "100" is). That'd form an RC low pass. However, that would limit the mixing performance of the two 1N4148 diodes², making a second preamplification state desirable – ugh, complexity.

You could also experiment with a capacitor in parallel to the 470 kΩ resistor between T1's base and collector (though that can lead to stability problems), or you very simply add a capacitor from base to ground. That's probably the stablest method of getting higher frequencies out of your input, but requires a bit of experimentation when it comes to figuring out the optimal capacitor value – it's a bit hard for me to estimate the source impedance for the signal coming out of L2.

Generally, reduce that 470 kΩ to maybe 100 kΩ from Base of T1 to ground, anyway – the 470 kΩ is there for the purpose of negative feedback making this thing more stable, but at these resistances and without a ground path, that's negligible. Use the 470 kΩ as second half of the voltage divider from base to ground.

Now, looking at that circuit: you can definitely improve the rejection of EMI by adding more stabilizing ceramic capacitors here and there; placing a 10 nF – 100 nF with short leads close to the top of the 4.7 kΩ of your preamplifier stage would swallow quite a bit of HF noise.

¹: Sadly, the schematic doesn't label any of its passives, or most of its active components, that would make it much easier to talk about it, and also make it easier to understand what one is doing while building it – sigh, let's explain this nonetheless. Notice that "label your components" is really what EE students learn in their first lecture that actually deals with circuits in their first week of university. There's no good excuse not to label your components!

²: generally, interesting choice in this application: these are typically used as switching diodes, so you get a lot of your higher-order intermodulation products, because they are significantly nonlinear beyond a quadratic term. However, this is not subject of your post. If in doubt, try using a diode whose "knee" in the I/V curve is further out.

  • $\begingroup$ Did you mean HP filter? And even attenuation on the input is not going to help this, right? Some of the schematics on that site use front-end AM BCB filters on other direct conversion receivers, which are prone to this issue. $\endgroup$ Commented Jan 26, 2020 at 22:15
  • $\begingroup$ Thanks allot for that answer, and allot of advices for me try out. I am not educated myself into radio or electronics (self learned, with a little help from my father) but i love it, and studying it as much as i can so i can one day become a radio amateur :). The circuit is not my own, just something i found on internet, and i agree that it is a shame to not label the components (i guess 100 capacitor means 100 pf, though he label other capacitors with "nf"). $\endgroup$
    – mathias
    Commented Jan 27, 2020 at 12:21
  • $\begingroup$ I will try out your advices, and see if i at least can get the Am broadcasts not so loud. They don disturb much, as soon a strong signal from any radio amateur is on the frequency. Only made regenerative receivers before, so this was surprisingly much better result for ssb and cw stations. Thanks again for spending time answering. 73!/Mathias $\endgroup$
    – mathias
    Commented Jan 27, 2020 at 12:21

Everything works very well, but on some frequencies I hear loud AM broadcast stations. I think this is caused by harmonics from the VFO, but I am not sure.

What you're possibly hearing is intermodulation distortions created by your receiver from the mixer and amplifier. A local AM broadcasting station has high radiated power, making the problem even worse.

The first trouble is when you apply a pure sine wave and a pure oscillator to a mixer. In an ideal mixer, when an RF input (e.g. an RF signal from the antenna), and a LO input (e.g. a VFO) are applied to a mixer, the sum and difference frequencies appears at its output.

  • $ f_\text{RF} \pm f_\text{LO} $

But in a real mixer, due to its nonlinearity, a wide range of frequency products appear at its output,

  • $ m f_\text{RF} \pm n f_\text{LO} $

Two RF signals, $ f_\text{RF1} $ and $ f_\text{RF2} $ (e.g. signals from two radio stations) at different frequencies can create similar troubles. Due to the nonlinearity of amplifiers and mixers, they also create unwanted modulation between themselves.

  • $ m f_\text{RF1} \pm n f_\text{RF2} $

In a receiver, the intermodulation from two RF input signals created by an amplifier can be modulated further by the local oscillator, creating even more unwanted frequencies.

  • $ c(a f_\text{RF1} \pm b f_\text{RF2}) \pm d f_\text{LO}$

These higher-order intermodulation products always exist, but normally as the order goes up, their power decreases rapidly. The worst situation occurs when an amplifier or a mixer is overwhelmed by a powerful signal, these otherwise hard-to-see intermodulation products suddenly appear all over the spectrum.

The following example is a two-tone test of a RF amplifier, showing how it will create significant intermodulation when the input signal power is excessive.

An example of a two-tone test

Source: Wikimedia Commons, by Ice Ardor, License: CC BY-SA 4.0

In amateur radio, powerful radio stations at different bands often create a lot of troubles. The three common culprits are:

  • Local AM, Mediumwave radio stations

  • Local FM, VHF radio stations

  • Local/Remote AM, Shortwave Radio stations

In general, all radio receivers are vulnerable to this problem, especially when a bandpass filter or preselector is not used. A ultra-wideband direct-conversion software defined radio, like a RTL-SDR, is the most vulnerable radio receiver. Severe interference from a "blocker" radio station it can make a radio receiver become utterly useless, you'll see radio stations everywhere, and nothing else.

The first step to solve the problem is making a bandpass filter for your band-of-interest to protect your receiver from being overloaded by unwanted out-of-band signals. You can make a bandpass filter for the 80-meter band, for example.

But in this case, only the AM stations are creating the troubles, you can as well making a high-pass filter to remove all frequencies below 2 MHz, it's easier to make than a bandpass filter, as lower-Q inductors are acceptable.

  • $\begingroup$ Thanks allot for that answer!. I use a bandpass filter then allready (even though first time i tried winding the coils for it, so not sure it works perfectly), i now got some advices what could cause this, and will experiment further on it. Thanks allot! $\endgroup$
    – mathias
    Commented Jan 27, 2020 at 12:30

That's an oddball receiver, employing a so-called Polyakov mixer. Its local oscillator is supposed to run at half the received frequency. For 80M, the local oscillator runs at about 1.8 MHz.
The RF amplifier ahead of the mixer should be very linear...this one is biased weakly, and could easily distort when presented with strong input signals.

This type mixer is supposed to work well in this application. One can only guess at possible causes:

  • Local oscillator doesn't drive mixer with 50%/50% duty cycle

  • Mixer input signals not balanced on +ve swing versus -ve swing.

  • Local oscillator feeds into audio amplifier, making it into a conventional mixer.

  • RF amplifier is distorting on strong signals.

  • wrong local oscillator amplitude: you should not over-drive nor under-drive this mixer.

Balance for this mixer is key. Diodes should be matched, but more important: both local oscillator and RF amplifier input signals to this mixer must be symmetric on +ve swing and -ve swing.

At these low RF frequencies, it is also important that the audio amplifier sees audio input signals. A more aggressive filter that attenuates RF might be used between mixer and audio input.

  • $\begingroup$ Thanks for that answer. Yeah i know it is critical with this 50/50 input. I made change i the circuit, where i used a transformer from collector to the resistor (12 turns on primary side and 6 turns on secondary, to the mixer, then i use the "original frequency on 3.5 mhz" and changed the diodes to germanium diodes. Now it works perfect without any strong am signals from broadcasts..and even more sensetive..I am learning this stuff so i try understand all you say..i kind of mange it :) I can try post my circuit if any one is interested. But i am very happy with the result. $\endgroup$
    – mathias
    Commented Feb 1, 2020 at 16:55
  • $\begingroup$ the vfo might be a bit unstable then, but not much, only when i make a fire to heat my house :) $\endgroup$
    – mathias
    Commented Feb 1, 2020 at 17:00
  • $\begingroup$ i allso changed the silicon diodes to germanium...will post the circuit $\endgroup$
    – mathias
    Commented Feb 1, 2020 at 21:04
  • $\begingroup$ This mixer should be fine with silicon diodes, with a larger local oscillator injection compared to Ge diodes. I'd guess that the Colpitts oscillator won't supply a symmetric sinusoidal signal...which will cause imbalance. $\endgroup$
    – glen_geek
    Commented Feb 1, 2020 at 21:45
  • $\begingroup$ It works fine with silicon in this circuit, and not fine at all with germanium, but when i changed the output, things changed, germanium works, well. Inot ttried with silicopn in the new construction. I will post the new circuit, and i would love to know what you people think about it. $\endgroup$
    – mathias
    Commented Feb 1, 2020 at 22:05

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .