balanced diff. input stage


2002-06-02 11:47 pm
I am wondering what the best way to design a balanced input stage for a differential amp is. I need to reinvert the inverted line and combine the noninverted signal with it to cancel out mutual line noise. So, do I slap a transistor on the inverted line such that the base goes to the input, the emmitter goes to negative, and the collector goes to a big resistor and the now reinverted signal out? Then the inverted signal out would connect to the noninverted in. Or is there a better way of doing it, or any special conciderations?

Also, I need both balanced input as well as output. I am guessing it is better to have a single feedback loop, so I'd convert to unbalanced for the input stage, then back to balanced for the output stage by the same method. Does that make sense?



2002-06-02 11:47 pm
inverting inputs..

Thanks much for the info, guys. The articles were very interesting.

But I am new to this and a little confused.. Balanced inputs consist of non-inverted, inverted, and ground. Both of the signal inputs are AC.. But in the articles you two gave me there was a positive in, and a negative in, both DC. Or are they supposed to be offset a few volts, so each doesn't cross zero?


2002-06-02 11:47 pm
Cool, thanks for all the help. It's a fun learning experience.

Hmm... I'm having trouble determining the cap values for the inverted input. For instance, to make the leach amp accept balanced inputs, one would use the grounded part of the feedback network (below C6 and C7) as the input. Not sure about the ground below C8...

It uses a strange scheme which takes feedback from different areas of the amp (driver and output), though, so I'm not sure how to calculate the correct values on the caps. Why are those caps necessary, anyway?

Do the resistors have to be changed, too? The current gain setting (R19/(R17 + R18 + R19)) should work with balanced inputs, right?

Or should I just leave all as is, and replace the ground below C6 and C7 (leave C8 alone?) with the inverted input?

I hope I'm not asking too many questions. Thanks again, guys.


  • leachfdbknet.gif
    3.3 KB · Views: 451
Okay, the simulation is now working. I had to cheat as I don't have a very good selection of power BJT models at my disposal. Instead, I've used IR power MOSFETs in place of the driver and output devices (Q16-Q21) in the Leach 4.5 schematic.

This works pretty well in simulation land, so maybe I can make a preliminary comment or two. First, the dual feedback comes into play above the audio range (about 150 kHz, I believe), and even then, C8 is isolated by the 11 k ohms of R18 so I don't think it will matter to yout input signal.

For a tip on how to wire in your negative input phase without it shorting to ground through C6/C7 and without removing this 100% DC feedback, I suggest checking out the schematic for the Aleph 30 or 60 available from (I think you go to Products, then click the Aleph link at the bottom of that page.)

That said, and with all due respect for Prof. Leach, I'm not convinced of the need for this fancy feedback arrangement. If I leave the lag compensation caps (C10/11) in place, I can stabilize the amp easily enough by shorting out R20 and removing C8.

While my simulations can generally identify a tractable design, they don't always (or even often) nail the optimum compensation cap value. Be prepared to experiment with C9 in the working amp if you decide to simplify the NFB. It's nice to have an assortment of mica caps on hand for just this purpose, say from 10-100 pF.

I'll report back if any of this needs updating in light of further progress with the model.
One thing I overlooked in questioning the complexity of the v4.5 feedback network is that it would offer greater rejection of RFI pickup via the speaker leads. This point in its favor could argue against simplifying the network if RFI is a problem in your area. It's conceivable that Dr. Leach encountered it in his GA Tech lab, which I know to be close to several broadcast facilities, both on and off campus.