How can an effect audio processor add echo to something that comes after it?

I wired my audio processors like this. You can see that there is an fx processor that processes microphone data and adds effects, and a sampler that plays sample sounds. After all the processed sound, I add a recorder:

void WiringProcessor::processBlock (juce::AudioBuffer<float>& buffer, juce::MidiBuffer& midiBuffer) {
        fx->processBlock(buffer, midiBuffer);

        sampler->processBlock(buffer, midiBuffer);

        float **inputChannelData = buffer.getArrayOfWritePointers();
        simpleRecorder->processInput(const_cast<const float **>(inputChannelData),buffer.getNumChannels(), buffer.getNumSamples());

The problem is that the sampler sounds are coming out on the speaker with some echo. Even though fx processes before sampler, the sample sounds come out with an echo. I used headphones to rule out the speaker being captured by the microphone. How is it possible?

When I mute fx, that is, set its gain to minimum, there’s no echo on the sampler sounds. How can the fx alter the sample sounds if it processes things before them?

Notice that I make fx and sampler share the same processBlock because I want to record both into simpleRecorder. I could make them both in different processBlocks but then it’d be hard to join them together again.

Is there a better way to do what I’m doing?