TL071 low voltage biasing question

[Gus]

http://twin-x.com/groupdiy/displayimage.php?pos=-2349

This is the idea I was posting about

Use low side switches to switch in bias change resistor and also the CC or a resistor.

Bias 100K and 220K at 9VDC and with the mosfet on 100K and 110k (220K || 220K)

You can adjust the bias string values etc for what you want.

24VDC turns on the mosfets at about 12VDC at the gate. You could use a LED in the resistor string to show you are running at 24VDC. If you use a tip and sleeve the sleeve will short the mosfet bias resistors to ground making sure they are off. You can find dual mosfets in little dip packages.

EDIT to make things clear
the three S in the circle all connect together
The other two circles has supply written inside, they are connected together

Do you like/understand the auto switching of bias and the CC?

Also I would make the 24V current limited to make sure it does not break when you insert a Tip Ring Sleeve (TRS) plug the 24VDC to ground

Should also have a 100 ohm in series with the output.

 

[tv]

You know, yours and PRR's ideas are _too_slick_ for an onboard preamp. They both have that rockman-ish vibe - i.e. "smart switching" etc. They are clearly suitable for outboard, boxed devices (and smart marketing buzzwords ). OTOH, the onboard must fit onto a 2,5 x 3,5 cm pcb ... which means it must be similar to russian wristwatches - as simplistic as possible.

What you and PRR suggested calls for 4066's, like rockmans imo.

In supply I'm using 5819 schottkys for diodes, and the mid-ring (phantom) diode gets a 220ohm 1W resistor in series for the onboard. The Psu also has a 470ohm 1W series resistor - when plugging-unplugging there can be shorts. The config I use is dead simple and just works. In audio path I have slightly more elaborated circuit which also just works.


One thing started to intrigue me, though. Namely, I did some extensive tests on my prototype and the circuit (with a 071) consistently sounds more focused and tighter when biased @2/3 V++ ( !!! ). I mean this is NOT how it should be, right? Even with chips from different manufacturers!

Now I tested the circuit with a dying 9V battery, with a fresh 9V battery and two fresh batteries (18V), and in each case it sounds better mis-biased ... (I'm thinking of permanently biasing it @ 2/3V++)


Is somebody else interested in putting together a simple buffer with a 071 and maybe confirming what I'm hearing (i.e. testing the sound when biased @V/2 versus being biased @ 2/3 V++ ... on whatever audio source you prefer)


I mean - this _shouldn't_ be happening, right?

:shock: :shock: :shock:

 

[PRR]

> would be a ZENER stable at so low current?

Zener data page 6 figure 7 shows that a Zener sold as "3V at 5mA" may tend to be 2V at 0.1mA, 1.5V at 0.01mA, which is near your proposed 6V/440K= 0.014mA operating point. Low. But a "4V" or "4.7V" may (or may not) be consistently ~~3V at 0.01mA. -Maybe- a "4.7V" will reliably be 3V at 20uA, or maybe it will leak to lower voltage. Have to try and see. Do put a heat-lamp on it to learn if it will change on a hot day.

A JFET may be happier; however you don't get to pick the voltage. N-type JFET Drain to 9V-24V rail, Gate to ground, Source gets a 22K-47K resistor to gound. The Source voltage will rise to roughly the "pinchoff voltage" Vp or Vth of the JFET. Bring your 220K+220K divider to the Source pin.

Obviously the JFET will pull a supply current near 0.1mA, enough higher than bias-divider current to be stable, significantly lower than TL071 supply current, though not neligible.

Vp is not precisely specified, usually a range like 2V-4V. It cannot be controlled in the foundry; they brew wafers with Vp ranging from 1V to 10V, and sort them into two or three ranges. If you find a JFET which gives over 3V, you could reduce the Source resistor to pull it down, or wire two resistors as a voltage divider.

If you can find current regulator diodes with low current ratings, put a resistor in series to give the desired voltage.

> PRR suggested calls for 4066's

I dismissed CMOS right away. They can be handy, but not here.

> sound when biased @V/2 versus being biased @ 2/3 V++ ... I mean - this _shouldn't_ be happening, right?

I doubt I would hear it. I have no doubt it might make a difference. These are 30V parts with a 7V reference and JFETs with ~~2V-4V pinchoff. Under 9V power, sagging below 7V as the battery fades, the JFETs and reference are "starving". And since these voltages appear "below" the bias/signal voltage, it does seem that the bias might like to be a bit high of "center".

 

[Gus]

CMOS?no cmos in my schematic. Look at the IRF mosfets SMD or dips.
Why are you using a zener in the input bias string? resistors are easier. And if you are happy with 2/3 at both voltage one less switch is needed. So all that is added is two 100K resistors and a mosfet for the CC autoswitching

The 5.6K is from the old Jung White pat4 rebuild in The Audio Amateur (TAA) around 1980 ( I will look for the issues to make sure I am remembering correctly). It is not a true CC however if the signal swing is small and you are running at 9VDC(2/3 about 6VDC to ground) to 24VDC(1/2 about 12VDC to ground) it might be a "good enough" and SMALLER solution. And like I posted I am not sure it made any difference with the phono preamp I built compared to the output stage without them

 

[tv]

Update:

I have now run tests with 9V, 18V and 27V supply (batteries) and biasing to 2/3 V++ does wonders to the sound in each and every case. This means I'll permanently bias to 2/3V++ and leave the zener out.

I have recorded the thing - TL071 - @V/2 and @2/3V++ (27V), and listened later with fresh ears. The difference in tone/definition is comparable to pickup upgrade ( !!! ). :shock: :shock:

As for the Aclass biasing, I'll have to reconsider. Maybe I'll leave the zener "sensor" :grin: OUT and borrow/mod the Gus' idea but using a BJT: "listen" to the +Voltage on the mid-ring ...

Or alternatively take on PRR's JFET ccs sugestion (only 5 holes to drill :green: ) ... IF I "hear" a remarkable imrpovement in tone when biasing to Aclass ... (if I do, Gus has to wash his ears :green: )


If the only thing needed is to rebias to 2/3V++ this means I can use some of my older preamp pcbs. To be honest, I was hoping to get me a new batch - figure out where to get these sexy blue pcbs. I'm using neat blue metaloxide resistors and neat orangeish tantalums (for more grind, like bartolini)

 

[tv]

Another update:

I tested the circuit @18V supply, 2/3V++ biasing and Aclass pulldown (a resitor only, no CCS).

When at cca. 1mA pulldown (11K), there was a barely recognizable "improvement" (slightly more "oily" feel), but not worth the hassle.

At cca. 2,5mA / 4K7 (similar to what Gus built) there was (IMO) a "downgrade" in sound, the preamp felt "sloppy".


Conclusion: in the final unit, I will only use asymmetrical biasing (reuse my old pcb's as well) and will in time being concentrate on the divider ratio - which one works best for the given circuit. Props to all who gave input/feedback (I'll borrow some of your ideas for a "breakout box"). :guinness: :guinness: :guinness: :guinness: :guinness:

 

[Gus]

TV

How does the circuit "sound" with

stock opamp no CC?

a true CC pull down, FET or BJT type?

If I understand the post you don't like the resistor pull down.

 

[tv]

Hi Gus. I only did the initial tests with a _resistor_ pulldown. Results were either neglibile or slightly worse (sloppier) than what I got without it. So I discarded further experiments (i.e. haven't actually built the CCS pulldowns) _and_ won't use any pulldowns in the final circuit. Not worth it - just biasing it asymmetrically makes it sound defined and expensive. (If _anybody_ told me such a cheap trick, I honestly wouldn't believe it, period.)

Do you still have that circuit you mentioned before? You said it used 5K6 pulldowns (pulling cca 2,7mA iirc). Imo you could make this experiment: assymetrically bias it (more towards V+ rail, I used 2/3V++) and use circa 1mA pulldown resistor (according to your voltages). Maybe this could make it sound better?