While experimenting on the practical build of the analogue multiplier, I ran into a problem. The clone ATX power supply that I typically use for my projects is pretty much adequate when testing circuits that need voltages of 3.3 V and upwards. However, when it comes to small signal circuits, such as the analogue multiplier, power supply ripple becomes an issue worth taking into consideration. Let me demonstrate.
From the analogue multiplier post, we saw that the expected output is as below
However, if I were to use the clone ATX directly without any filtering, this would be the outcome
Looking at the differential output, it kind of follows the normal trend that is expected, but the individual outputs are extremely overridden by noise.
Of course one of the best way to avoid the whole noisy output is to use a well designed power supply, but sometimes the right equipment isn’t at hand and you have to use what you have at hand and a little bit of hacking here and there.
Eliminating power supply ripple is an easy enough problem to solve, you simply use a capacitor multiplier, there are a dime a dozen tutorials out there. The problem that I ran into was that most capacitor multiplier circuits were mostly for filtering the positive rail of the power supply, but I needed to filter also the negative rail of the power supply. On top of that, I had to make sure that in the multiplier circuits that I built I had access to ground. So let me explain each of these problems and why they are important.
Why Filter at All?
Let’s take a look at the three signals below
a) is when input A and input B are grounded, this can be considered the ideal case absolutely no noise. b) is AC coupled +15 V DC and -15 V DC. Looking at this we have AC ripple of up to 100 mV riding on the 15 signals. This is bad because the input signals of your small signal circuit could be 100 mV, thus this ripple acts as an unwanted signal compromising the integrity of your output. c) is the +15 V DC and -15 V DC after a bit of filtering. The ripple has been reduced to about 20 mV and it isn’t as pronounced. You can actually trace the straight DC line.
So that’s one of the main reasons why you need to filter.
The Capacitor Multiplier
When it comes to filtering power supplies, the capacitor multiplier is one of the most prominently documented go to solution. The configuration is as below:
It is essentially an RC filter with an Emitter follower circuit added on to “multiply” the value of the capacitance. You can watch EEVBlog’s video on the same for an elaborate treatise of the theory behind a capacitor multiplier.
The next problem I ran into was, most of these circuits talk about filtering the +V terminal, what about the -V? Well the following circuit could be a solution:
Ok, this could work, but then comes the problem about ground.
Let’s Talk About Ground
Typically ground is defined as a point of Voltage reference in your circuit and that’s why you are able to get positive and negative voltage, depending on what your point of reference is. However, when building and testing your circuit, depending on the equipment you use, the ground could be very much well defined.
The clone ATX supply that I have and the oscilloscope are mains earth referenced, meaning their ground potentials are referenced to the mains earth. Why does this matter?
The circuit below is a very ugly approximation of what it would look like when the circuit is hooked up and you are probing it:
From the above circuit, what would be happening is that you’d effectively be short circuiting the output from the emitter follower with the ground lead of your power supply
Essentially beating the purpose of the circuit. Furthermore this could probably lead to your oscilloscope blowing up or you equipment or circuit getting damaged, Dave of the EEVBlog explains very succinctly in his How NOT To Blow Up Your Oscilloscope!
The Negative Rail
Okay, so that would not be the best way to solve this particular problem. So what needs changing? The solution is simple (but I will admit, took me quite a bit more thinking than I care to admit, hence the reason I am writing this post), replace the NPN transistor with a PNP transistor and flip your polarised capacitor to match the new polarities:
From this configuration there is still a direct connection to ground between the unfiltered input and the filtered output, meaning our probe won’t be introducing any dangerous short circuits.
PNP transistors are not as common as NPN transistors in practical use, but I seemed to have forgotten that the analysis is more or less exactly the same between the two with only a change in polarity.
So there, a very simple hack to filter your knock off ATX power supply as you proceed in your quest to build your own home workshop on a budget, well until you can save up and get a proper power supply.
Till the next time, keep building stuff. Oh, and below is a little simulation of the two configurations I have talked about.
PS:I did not put any keen attention in the selection of the resistor and capacitor value, you may want to in your design.