Decibel Dungeon


Nothing in the world of technology stands still for very long and even something as simple as the Gainclone develops at an amazing pace!
We have progressed from the very basic and simple inverted circuit, through buffered variations, added Zobel networks, regulated power supplies and now we have the T-network feedback.
Fortunately, this time, the modification to a basic IGC is fairly easily achieved and with little cost. It involves replacing the single feedback resistor (or two if you parallel them like I have in my first two GC's) with a network of three resistors. Hence the name! The T part of the name comes from the circuit diagram of the T-network where two resistors are used in series between the output and input of the chip amp; and another runs from the junction of these two resistors to signal ground.

Inverted Gainclone circuit with T-network feedback
The T shape of the resistor network is quite apparent!

Site menu Page menu

Resistor values

You can see in the above circuit diagram that I used the following resistor values (with the LM3875 chip):
  • Input - 22K
  • Output to inverting input - 10K and 10K (in series)
  • Junction of series feedback resistors to signal ground - 100R
  • The value of the resistor from non-inverting ground to signal ground will vary according to the chip that you use and will even vary between chips of the same make and model!
The best way to determine the value of the resistor from non-inverting (NI) input to signal ground is to start of with a 10K trim pot. Use a multi turn type as this will make fine adjustment much easier!
  1. Adjust the trim pot so that it is about 10K and then solder it between NI and signal ground. Power up that channel and measure the DC offset across the speaker terminals.
  2. Adjust the trimpot and remeasure the offset.
  3. Continue like that until you get close to 0 mV offset.
  4. Measure the resistance across the trim pot and find a resistor as close in value as you can.
  5. Solder the resistor in place and recheck the offset.
You may be unlucky to find that you cannot achieve 0 mV offset even with the trim pot adjusted as far as it will go (0R). If this is the case you have a couple of options. First, you could accept the lowest DC offset that you can get. Anything under 100 mV is fairly safe. Second, you could try a different chip.

Why use these values?

The values that I have used in the above example were suggested by Joe Rasmussen after a lot of calculations and were chosen because they should work with most of the chips bearing in mind their slightly different parameters. You can use other values if you wish but you should only do so if you are confident that you know how to work out the gain using the values that you choose.
Franz Gysi came up with a T-Network resistor calculator but I have never used it myself. Click here to download it.

Inverted Gainclone circuit with T-network feedback
The T shape of the resistor network is quite apparent!

Site menu Page menu



Why use a T-network feedback instead of the single resistor? One advantage with the Inverted Gainclone is that it allows you to increase the input impedance. This could negate the requirement of a buffer between source and amp.
The second reason is that it is said to sound better. I am not totally convinced that it sounds better yet. More detailed, certainly but that doesn't necessarily mean better! But don't take my word for it. We are only talking about a handful of resistors here so do try this for yourself (and let me know what you think!

Can I use the T-network with the NIGC?

Yes you can! The advantage of raising the input impedance is not an issue with the 'standard' NIGC though. And of course, you will need to connect the T-network between the output and inverting input with the 'variable' resistor between inverting input and signal ground.

Does it matter what type of resistors that I use?

That depends on how much you think that the type of resistor affects the sound quality! I like to use carbon films and have used them for the input and feedback positions. I have used metal films for the grounded resistors. BTW - I measured the 5% carbon fi,m resistors to get exact matches for each channel. I may try something different like Welwyn RC55's at a later date.

Does it matter that this necessitates a longer signal path?

I would say that if you keep the resistors soldered directly to the chip pins, then probably not. I have been told that there is nearly an inch (25 mm) of pins inside the chip body so I do not subscribe to the idea that a few millimeters more or less makes much (if any) difference.

Does the T-network affect the working of the GC in any other way?

Not as far as I can tell. Certainly, in my two experiments it did not affect the temperature that the chips worked at.
If you have any more questions please use the contact form to ask me!
Site menu Page menu