Doug Self Line Input

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

atavacron

Well-known member
Joined
Jan 28, 2009
Messages
344
69AB2AB0-248C-461A-A26E-00358AAF6591.jpeg
Has anyone experimented with Self’s original or updated line in? Fig C is in Small Signal Audio Design and dates back quite a ways, while Fig D is applied to a Balance control in the updated hifi preamp design from Linear Audio a few years back.

I’ve run a zillion numbers based on these and it’s a stretch to create a +/-10dB line trim. +/-6dB is easy. Significant gain is within reason, still low noise up to about +20dB if your input impedance is a conservative 2K. Just curious about real world implementations, and how people like the gain curve. The main challenge for me is getting 0dB to sit at center of travel — I can’t quite grok that calc.
 
FWIW this multiple feedback concept does apply to differential topologies….you can use it both with the double balanced input approach and an inverting FDA. You could even use it with a non-inverting diff gain stage if your Rg had to be a fixed value (OPA1622 or INA1620 with their 1200nA bias current for instance).
 
Last edited:
Hey thanks @abbey road d enfer . I'm guessing that an equal value + and – range with a linear would put unity somewhere up above the halfway mark, and to keep unity dead center one has to stick with the ratios as drawn?

I apparently posted this in the wrong forum, should have gone into Drawing Board. Unsure if that's changeable now.
 
I think the +5.8dB / -3.6dB ratio can come pretty close to +10dB / -6dB, which might make for a good line trim that keeps 0dB at center.
 
Last edited:
Hey thanks @abbey road d enfer . I'm guessing that an equal value + and – range with a linear would put unity somewhere up above the halfway mark, and to keep unity dead center one has to stick with the ratios as drawn?
Correct. Apparently, you can't have symmetrical boost/cut and 0dB in the center, unless you use law-steering resistor on the pot.
For a symmetrical range of +/-4.6dB and 0dB in the center position, you need to increase the input resistors to 930 ohms, and add a 6.8k resistor between the wiper and bottom (ground) of the pot.
I'm not a genius, I haven't spent hours calculating that. I just used LTspice, which is free and can learn in a few hours, or any other decent simulator.
I apparently posted this in the wrong forum, should have gone into Drawing Board. Unsure if that's changeable now.
Done!
 
I’ll admit to having tried LTSpice a couple of times and having been unable to get it going for me.

Here is a better phrasing of the question at hand…with an example load of driving a cable. Just pretend I got the input polarity right, and the bottom 2K || 665 is a 499R.

409901DD-B16E-420E-AB4B-10CDE935D882.jpeg
 
Last edited:
Shorthand for an OPA2211 with an OPA210 for CM null. .025% match on the CM feed resistors. Waffling on whether the 332R Rs would be necessary at the 2211 NI inputs.
 
I’m only seeing a gain concern in the potential pot travel value mismatch. Given that significant CMR degradation is essentially off the table with this form, the concern then becomes the (admittedly minor) balance issue between the phases. I’m expecting that while an individual phase might be off after the trim calibration, the differential signal could totally cancel.

E3E350CE-3512-42CC-8426-E09E8C0379DE.jpeg
 
A calibration rig for the center trim, even…probably done more efficiently with an AD8271 though. And less offset.

9FE710F5-DD2B-4E9A-99E7-2FE5113864EC.jpeg
 
Back
Top