DC reduction mod II
Sure TA2024 DC Offset Mod 2
- Category: Sure TA2024
- Published: Sunday, 12 June 2011 20:59
- Written by Justblair
In in another article, I showed how the Sure Tripath Boards unacceptable DC could be reduced significantly by removing a pair of resistors from the PCB effectively disabling Sure's attempt at a DC offset reducing circuit. This article takes the modification a step further to eliminate DC offset completely.
Now a few mv of offset on an amplifier is not especially harmful to either your amplifier or your speakers. Indeed in most speakers the most sensitive part to the speakers, the tweeter will have a capacitor in the crossover that will block any DC anyway. In my project I will be using an active crossover arrangement where the filter is before the amplifiers. I wanted to reduce the DC offset to zero to avoid damage.
Two of my amplifiers were giving DC offsets in the 100's of mv, even after my first DC offset modification. This is a little higher than I feel comfortable with, even if it is probably paranoia on my part.
Sure's circuit which attempts to reduce DC offset at first glance appears to be a shot in the foot for the company. In a standard circuit, without any DC offset reducing circuit, the Tripath 2024 chip produces moderately small DC offsets. As can be seen in the following table, with the circuit removed the DC offset varied between 12mv and 128mv on the boards in my sample.
|Amp 1|| |
|Amp 2|| |
|Amp 3|| |
|Amp 4|| |
|Amp 5|| |
Looking at the DC offset figures that the boards were producing before the disabling of Sure's DC offset circuit, you can see how far off the mark Sure is with its circuit. There are reasons for this:
- The DC offset is dependant on the Supply voltage to the board, as the board can accept between 9v and 14v safely (and higher voltages have been run without mishap). Therefore in the Sure board, the DC offset circuit could at best be tuned for 1 value for input voltage.
- Variances in the TA2024 Chip mean variances in DC offset, even between channels of the same chip. For Sure's circuit to work successfully, each board would have to have different resistor values for R1, R2, R15 and R18 chosen.
- Tolerances in resistors are typically between 1% and 5% (and sometimes higher) so again, using static chips in the circuit would require some trial and error in Sure's manufacturing...
So we can see that whilst Sure's DC circuit is a valiant attempt at curing DC offset, its implementation of the circuit was unlikely to work and when it does, it is more by chance than anything else.
Looking at the schematic that Sure have provided, we can see how their DC reducing circuit is configured. R1, R2 and R3 form a voltage divider. R1 and R2 divide the 5v output from the Tripath chip R3 then ties the DC voltage of the Tripath 2024's input to the divided +voltage from R2 and R3. By varying the Dc offset at the input, the DC offset at the output can be adjusted. The equivalent circuit for the TA2024 from Tripath's Data-sheet shows more.
At the input (In1) of the Tripath TA2024 chip is an op-amp. This allows us to set the gain for the chip, and also gives the chip some impedance. Traditionally and op-amp is supplied with a +/- voltage, but of course the Tripath 2024 chip is designed to be run with +12v only. In order to allow the chip to function with a positive supply, the op-amp is biased into DC. The input capacitor C is of course now essential, because the op-amp is biased this way.
What we need to do then to make Sure's circuit work correctly, is adjust the R1,R2 resistors until we get the correct DC offset at the input to supply the correct (i.e. zero) offset at the output. We could play all day with resistor values until we get the correct DC offset, however a quicker way is to use an variable potentiometer (or trimmer to give it a friendlier title) to set the correct DC offset.
A variable trimmer has three inputs, You can see in the image of the trimmer that I have used they are numbered 1, 2 and 3. Inside the trimmer, is a wiper that makes contact with a resistive trace. Where the wiper sits, there is then a resistance between pins 1 and 2 and between pins 2 and 3 Effectively we get the same effect as we get in Sure's board with Resistors R1 and R2. Resistor R3 should be connected to Pin 2 on my trimmer and the Sure DC reduction circuit is recreated.
This time however the resistances within the voltage divider are now variable, allowing adjustment of the DC offset at the input of the Tripath TA2024 chip.
Looking at the Sure Tripath TA2024 based board, R1, R2 and R3 are located close to the inputs to the board, The R3 resistor has been removed already on my board as a result of my first DC mod.
R1 and R2 have 20K resistors present, though on two of my other Sure Tripath TA2024 based boards, 22K values were used.
First job then was to remove the R1 and R2 resistors so that i can rebuild the Sure DC offset reduction circuit with my trimmer.
Poking around with my multi-meter, set to its continuity checking setting, revealed how the tracks (hidden under the black silkscreen) are arranged.
- The pads on the left of R1, R2 and R3 are all connected and represent the point where the 5V is divided.
- The right hand pad of R1 is tied to the 5v
- The right hand pad of R2 is tied to Analogue Ground
- The right hand pad of R3 is tied to In1 (The Tripath TA2024 input)
To rebuild the Sure circuit, pins 1 and 3 need to be attached to the right hand pads of R1 and R2. Pin number 2 will be connected to a resistor and then to the right hand pad of R3
To make it fit, I started by bending the pins of the trimmer. The trimmer is a bourns 3296 copy that was in my parts bin. This is a precision 25 turn trimmer, an ideal design for my board. Having 25 turns allowed for very precise adjustment of the resistances within the trimmer. A single turn trimmer could be used, but adjustment would be more difficult.
I bent pin number 2 upwards, and pins 1 and 3 were manipulated to make them fit the pads of the PCB. Although this trimmer is designed to be through hole mounted, it is possible to solder the pins to the surface mount pads.
I bent Pin 2 sideward as well, ready to accept a 1 Meg resistor
Here you can see the 1 meg resistor soldered to Pin 2. Its other leg is bent to the side and though the picture doe not show it clearly, pins 1, 3 and the loose resistor leg are in fact in line and proportioned to allow them to solder to the Sure Tripath TA2024 PCB. The assembly is ready to mount on the board.
I attached the trimmer to the slanting side of the input terminal block with some double sided sticky foam. This will take the strain, so that the trimmer is not relying purely on the solder connections to the pads of the PCB. This offers some strain relief to avoid accidentally lifting the pads through accident.
The slight tilt the terminal block offers makes it easier to get the fine tip of a soldering iron under the trimmer to solder the pins to the pads.
Finally I soldered the trimmer in place.
Testing the new DC offset reduction circuit is as simple as attaching the probes of a multi-meter to the outputs of the amplifier and selecting the 2000m setting of the voltage meter section. The trimmer was set at dead centre when it came out of the parts bin, the DC voltage measured +49mv. The trimmer I adjusted while the circuit was live, watching the readout of the voltage meter to make my adjustment.
A few turns of the trimmer and the DC offset of the Sure Tripath TA2024 based amplifier board was reduced to 0.0mv.
At this point I tried varying the DC voltage supply to the board, which as expected threw out the DC offset.
I know that some people at this point measure the resistances set on the trimmer and substitute resistors to the same values, but I see little point. The variable resistors (trimmers) are cheap items, and having the ability to readjust means that if I wish to play around with the boards input voltages, I can re-trim the DC offset back to zero
For the second channel, the resistors that we are interested in are R15, R16 and R18.
The circuit works in exactly the same way as it does for the first channel, R15 and R18 divide the 5v voltage. R16 ties the input (IN2) to the divided voltage to create the voltage.
The components for the modification are the same, A bourns 3296 (or a copy like mine) and a 1 meg resistor.
The variable resistor this time will connect to the right hand side of pads R15 and R18 however, R16 is this time not in line.
The variable resistors pins this time need to be manipulated differently. Pins 1 and 3 are pent in one direction, and a slight "step" is put in place. Pin 2 is bent in the opposite direction for the 1 meg resistor to be placed.
In order to create a little strain relief for the trimmer and the pads it solders to, and also to insulate against shorting out pads, I attached a double sided, sticky foam pad over the exposed pads of R15 and R18 on the left hand side.
The assembly was then soldered in place.
Again the adjustments to the trimmer are done with the circuit powered up, my multi-meter plugged in to the speaker outputs of the board. The offset could be trimmed right down to zero with the circuit in place.
For most people this modification will not be required, however if you are wishing to directly drive a tweeter without a passive crossover in place, or if the first modification did not bring the DC offset as far down as you would like it to, this modification will be worthwhile doing. Sound quality wise, I did not here any difference in the modified circuit, but then a few mv of offset is unlikely to make any deterioration of the sound quality.
It is on the other hand a simple modification to carry out, and for the very low price in parts, another way to look at it is why would you not have zero offset in your amplifier?