diyAudio (
-   Power Supplies (
-   -   [help] Full Wave rectification for VCA control voltage (

dan_thompson87 25th November 2009 04:07 PM

[help] Full Wave rectification for VCA control voltage
Hi everybody, first post here!

I'm currently designing a quad linked VCA compressor for my final year project at university, and I've hit a bit of trouble with the AC rectification for the control voltage.

The control voltage signal so far consists of a full wave precision amplifier and a low pass sallen-key filter with Fc of 20Hz & gain of 1.6. The aim of this is obviously to remove the audio component of the input signal. This setup gives a perfectly smooth RMS DC voltage at 1kHz, however as the frequency of the input increases a problem occurs with the rectifier and I have no clue what is causing it.

As the frequency increases the rectifier stops working and I am seeing very large negative pulses resulting in DC offset after my filter. However, this problem only started happening after adding the S-K filter on the end of the rectifier.

I have no idea how I am getting these large negative spikes from the rectifier (-6V for a 2Vp-p input!) so any help whatsoever would be great. My knowledge of the whole system is quite limited so its likely that I'm missing something very obvious!

Iain McNeill 25th November 2009 04:29 PM

The precision rectifier works fine until you load it with the LP filter?
Can you post the schematic?

Is something exceeding the voltage rails?

dan_thompson87 25th November 2009 05:06 PM

Ok, I've got some screenshots to help explain the problem a little better. Here is the schematic of the circuit, with reference points V1 (input) and V2 (output) for the following simulations:

Simulation for 1Vp-p at 1kHz:

Simulation for 1Vp-p at 15kHz:

The only thing changing between the 2 simulations is the frequency of the input voltage!

Thanks :)

dan_thompson87 25th November 2009 05:25 PM

I've just tried with the LP filter disconnected and I'm still getting the problem - so it must have been there all along. I've got some screenshots of the schematic and the simulations below.

Schematic (reference points V1 and V2 correspond to the simulations)

Simulation for 1Vp-p at 1kHz:

Simulation for 1Vp-p at 15kHz:

The only thing that is different between these 2 simulations is the frequency of the input. It's also worth mentioning that the negative spikes get greater in amplitude with frequency, e.g. 20kHz will have an even greater negative spike.

Thanks! :)

Iain McNeill 25th November 2009 07:11 PM

Looks like a simulator burp to me. Are your time increments appropriate?

What op-amp model are you using?

dan_thompson87 25th November 2009 08:29 PM

Sorry about the double post, I didn't realise replies were moderated!

I've had a go deleting all of the nodes in the schematic layout and rewiring it all, and the problem is solved! It seems as though I deleted a component or connection somewhere but left the net intact.

However, now I have the DC voltage a new problem has popped up! The capacitor is charging to the RMS value of the input as I would like it too, but it then drops by a few mV before levelling off. Is this just the nature of the circuit or is there a way I can keep it at constant RMS value? Here's a simulation to show what's happening:

1.747Vp-p/1.2VRMS (+4dBu line level) @ 1kHz

I have a feeling this may be something to do with the 12k feedback resistor in the filter. I was told earlier by my lecturer that a gain of 1.6 from the LP filter will improve the accuracy of the cutoff and give a flatter DC voltage. I have a feeling that this resistor is configured wrongly though as changing the value does nothing to my output voltage!

Also, is the initial charge time for the capacitor too long? 50ms is obviously a significant amount of time in a compressor circuit.

It's a TL074 op-amp chip by the way.

Elvee 25th November 2009 09:08 PM

A/ In the case of a TL074, the 12K is useless
B/ The overshoot is caused by a too high Q; try C values closer to their geometric average (~1µF)
C/ This rectifier reacts to the mean value of the waveform, not the rms
D/ Reducing the rise time can be achieved by increasing the order of the filter (for a given amount of ripple).

Steerpike 25th November 2009 09:29 PM

As LV said, your circuit is not computing the RMS value.
It computes the absolute average, which may be gain trimmed to equal the RMS value ON A SINEWAVE. But it isn't RMS; give it a non-sinusoidal signal and it will get a value different from the RMS value.

dan_thompson87 25th November 2009 09:39 PM

Thanks for the responses, they're a big help.

Everything Elvee said made sense to me - I'm going to have a go at tweaking the cap values to give a better Q, as well as trying the next order filter.

Can someone please explain to me why the feedback resistor makes no difference on the TL074 please? Would it be better to pick a different chip where it will make a difference, or exclude it altogether?

Sorry if my questions seem trivial - I only have basic knowledge of how these circuits work and I'm in the process or learning at the moment. That's the idea of a dissertation right? :)

Steerpike 25th November 2009 10:02 PM


Originally Posted by dan_thompson87 (
Can someone please explain to me why the feedback resistor makes no difference on the TL074 please? Would it be better to pick a different chip where it will make a difference, or exclude it altogether?

12k acts the same as a wire link. The input impedance of the amp is so high, and the input bias current so low that an extra 12k is irrelevant, might as well tie input directly to output.
Any resistor there will behave much like a wire link, unless you start using a value that starts approaching the amp input impedance - but that will cause other problems, like noise pickup, & offset voltage effects.

With 12k there, IF you put another resistor from the (-) input to ground, then you'd have voltage divider action going on in the feedback path, which would alter the filter gain.

All times are GMT. The time now is 12:50 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2