TL071 low voltage biasing question

[tv]

I'm designing a recording buffer for a bass guitar that _must_ run off a 9V battery, it'll probably use TL071 (alternatively TL061 or AD820). In addition to battery supply, it will have a feature to run off a "phantom" in range 24-28V which is going to come into the guitar via the mid-ring (stereo jack - i.e. if standard guitar cable is inserted it will run off battery, if a "phantomized" stereo cable is inserted, it will be powered via phantom). The "logic" for power supply is going to be handled with 1N5819's and a 5W 24V zener (plastic). In 9V mode it needs as low current draw as possible with the 071 device (== large V/2 biasing resistors, cca 220K each), but in "phantom mode" it will get A-class biasing via a CCS (it'll have a zener to check for voltage higher than 12volt to load down the opamp only then and not in 9V mode in order to mantain as low power draw as possible).

Now to questions:
According to Self, the 071 wants asymetrical supply at low voltages to mantain low distortion (fig 4)
http://www.dself.dsl.pipex.com/ampins/webbop/072.htm

How should I pull this off: when in 9V mode, the 071 should be biased closer to 2/3ds of V+, not V/2, no? But in phantom (24V) mode it should be close to V/2. I was thinking a ~3V zener to ground in series with biasing resistor voltage divider, but arent resistances a bit high for a zener (2 x 220k) - I need a very low added current for 9V mode. Any alternatives/suggestions? A led? ... ? It needs to be very stable, no drifts etc.

Further, I need a hint on Aclass biasing for 071's. I said I will use a "smart" CCS that will load 071's output to ground _when_ in phantom mode (or when larger than 12V supply is present). Namely _what_current_ (!!!) is best for Aclass biasing of 071 devices. Any suggestions - schematics, from working devices etc? I'd like to mantain as high quality as possible.

TIA

p.s. would similar be valid also for tle2071's or are they free of behaviour as described by Self? Additionaly, are 061 devices similar enough to be treated like Self described? What about lf441's?

 

[Samuel Groner]

I do really advise you to just use an OPA134 and to forget the silly idea of asymmetric rail voltages or the I-bias-my-opamps-in-class-A. Distortion will be miles better than what you'll ever be able to squeez out of a TL071, and noise lower as well.

Samuel

 

[clintrubber]

Twice 220k will result in say 110k input-'impedance'.

I'd go a bit higher than that, unless it's for an active bass (in which case I wouldn't bother to record it at all :twisted: )

It doesn't make sense to go much beyond 1MOhm resulting input-Z because of various parasitics already dominating then (as also pointed out on various spots here).

Regards,

Peter


BTW, if your not going for 0.000...% distortion figures, a simple JFET+PNP circuit could do the trick as well, can easily be phantom powered and could have a valid 'all-discrete class A'-label on the box :wink: .
It may throw a bit of 'nice bend' in as well.

 

[Gus]

I would not worry about it to much it is for a bass and who cares about AB/A output just don't load the TL071 output stage to much. I am guessing non inverting.

Do the "standard" two resistor divider with a cap at the node. Use a 220K to 1meg as the input resistor and an input cap of .01uf or less to the + input(taste).

Now is it a buffer gain of one or > one gain stage? If a buffer you should be fine if a gain stage more than X3 at 9VDC, a passive bass can clip the opamp gain stage. Do you want that distortion? it might be cool.

 

[JohnRoberts]

I'm not sure if there are better low current opamps available these days, but I recall in the deep dark recesses of memory that some bifets reverse their outputs if the inputs pull too close or below the - rail. This is not just a little more distortion, but loud pops in the middle of tones as the opamp pegs from rail to rail. So with a hot bass pickup you may want to clamp the input to prevent overdriving some bifets.


There is new stuff that claims rail to rail inputs and outputs but I have no real experience with them.

JR

 

[tv]

Just to clarify: the biasing resistors are just for that - biasing. The input "loading" resistor needs to be 355k (pickups/harness want precisely that, measured). All the impendances etc. have been measured/recorded etc. sofar and "set in stone" for that particular config (AND gain, but it's only sligtly over 1).

This bass sounds best with a jfet opamp (adds just a _hint_ of "active" tone "centeredness"), jfet/bjt discrete stage didn't sound as convincing as an opamp in this case. (f.e. in one of my other basses, a single bjt follower sounds best, but it was specially designed for _just_ that bass.)



What I'm asking now is those pesky DC conditions as described in OP - how to take care of them? And Aclass biasing _current_ ???

This is going to be a recording bass, I need the best _linear_ tone (i.e. NO equalizers, a twin passive tone controls a-la early jazzbasses - or a lespaul - instead) and transient response out of it within the given limits.


I need answers _precisely_ on what I asked, please. I.e:

a) how to make a biasing network that would do what described in OP and

b) what _current_ must be the CCS designed for in order to _best_ Aclass bias the 071 device at cca 24volts.

 

[clintrubber]

[quote author="tv"]
I need answers _precisely_ on what I asked, please.[/quote]
Sorry, I can't be of any further help here, but I'll be following this thread with great interest.

I mean, you're using 355k-exactly for just any song; no adjustments to that when songs in other keys show up? :shock: :roll:

Just kidding though, you obviously know what you're doing (or with all respect, the other option...) and that's for this topic way more than I thought there could ever be going on.

Regards,

Peter

 

[Gus]

The bias part is easy use a switch to switch in another resistor in parallel to adjust the bias all is is is a simple voltage divider. Two divider resistors bypassed with a electro cap and the input R at this node to the + to set the input resistance. To find what you want temp use a pot in the middle of the two to find the ratio you want to divide at.
If you want make it sense the power supply voltage change and switch it with a semiconductor switch you like to use.

Next part you need to think about the power rating of the chip and the thermal changes you will cause with the PNP part having a current shunt around it forcing the NPN at the "top" to work harder in the chip. For a starting place look at the source and sink current that the 71 is speced for.

FWIW 79,80 I built a passive phono preamp using TL07x gain stages with output "class a" resistors to the -15 volt supply (5.6k IIRC). I did not note any change with the resistors vers the output stage as stock

 

[tv]

Yep, 355k for _just_any_song_. It's golden. Atm I'm having a metal one in the prototype, watch out when I put the carbon mojo mofo in it. Itll reap stockings and panties apart.

@Gus:
so you did a 2,7mA bias

the funny part of the biasing I described in OP: it has to stay @2/3 V+ @9V but close to V/2 @24V. I thought a ~3V zener in series to the divider (to ground) BUT the current trough dividor has to stay as small as possible for operation @9V (071 with 1,4mA is a bit much already). Would a zener be stable at such small currents? The blocking cap is going to be a 4,7uF tant.

 

[clintrubber]

[quote author="tv"]Yep, 355k for _just_any_song_. It's golden. Atm I'm having a metal one in the prototype, watch out when I put the carbon mojo mofo in it. Itll reap stockings and panties apart.[/quote]
Just curious, not in jest (this time): What bass ? Playing style ? String-type & string-age ? Pick/fingers ? Amp ? Speakers ?

As in: I could certainly understand arriving at a certain 'feels best' range, but can't imagine that slight variations of such a value (I almost get the impression it's to be with +/- 1% accuracy) can even be noticable. At least a few orders of magnitude away from the influence of the factors I mentioned above.

Regards,

Peter

 

[Gus]

For 9VDC you have the 2/3 ratio make the 2 part say a 22K and the 3 part a 33K. Scale them if you want but don't make them two big

+9VDC then 22K then 33K then ground.
At the middle of the divider connect an elecro + (47uf what ever) and ground the -

Also at the divider node connect one end of the 335K and connect the other side to the + input.

Now when you switch to 24VDC calculate the resistor value that will give you 22K when added to the 33K in parallel. connect one end to the divider node and switch the other end to ground. This will allow low side switching with say a mosfet with a low Rdson or just use a slide switch Toggles often get broken on effects that go on the road. If you use a DPST you could switch the bias and pulldown CC at the same time.

The resistor pulldown value used was from the Jung, White PAS preamp mod in TAA IIRC 79 80 sometime around then.

 

[tv]

This time it's a sort of a "golden" peavey grind (momentarily with stock pu's). Quite possibly she be upped to barts m45cbc and bourns "plastic" pots (500k). The 355k on stock pu's is in similar ballpark to what experience I have with (soapbars) barts, a slight "retuning" is expected.

For some reason "j"-"pj" basses I have really want ~1meg input resistor, even with 500k pots - IME.

The difference is mainly in playing feel plus how does linearly recorded bass stand in mix, attack/length of sustain and how annoying are the highs (I want them to be "inside" the tone, so they can be sweetly EQed in a mix).

"Compactness-definition-3Dness" I'd say. Mostly I use fingers and "down" slap. 40-100 and 45-125. lately picatos/lowriders.

I ve tinkered with "passive + buffer" for some 20yrs now, I ackquired the "feel" for it but not much theory (apart from what's already written). I can't express what I know in technical terms adequately. Well for me it's important to get the "feel" anyway. I'm done with oversmart recording engineers. Maybe I'm geting an active bass to punish them



@Gus:
The 22/33K are way too small, they'll help to suck battery's life :grin: I KNOW :wink:
I need the "switch" to happen automagically (see OP) when phantom-powered. I'm afraid that 220k resistors won't pull the zener sufficiently (that's my primary concern).

Any alternatives?


now .... to the logoff

 

[Gus]

Tip ring sleeve using the ring For 24VDC?

How are you going to protect from shorting the 24VDC to ground when inserting the plug?

What is the current draw limit for 9VDC you want?

Are you using two diodes for auto power supply switching?

If 24VDC from the ring. Power the switching section from that. What I would do is use two low Rdson mosfets as low side switches to add a resistor to the bias and low side switch the CC in.

make a resitor divider network connected to the ring connection that has say 5VDC to 15VDC at the gates to turn the mosfets on and this shorts the ring to to ground shutting the mosfets off with just a tip ring plug.

 

[mikep]

somebody else already hinted at this. instead of biasing the output stage you really want to put a buffer after the 072. If you like the sound of the 072 on a particular bass then OK, I can believe that. but believe me, that opamp will sound beter unloaded regardless of biasing! try the jfet/pnp buffer AFTER a TL072, parhaps inside the feedback loop. load the buffer with an NPN CCS, the bias for the CCS coming from a cap-bypassed divider, the top of which is connected to the positive supply... higher voltage input (your phantom option) equals more bias current. doubt it will sound any different but wy not try it. you could use a similar scheme if you are dead set on your original idea.

mike p

 

[PRR]

> some bifets reverse their outputs if the inputs pull too close

The TL0 will do this. Violently. I usually can mind-design such that the first-draft PCB needs just one oversight-fix, but my TL074 cascade needed a major chop so the 2nd stage didn't slam the 3rd stage to the opposite wall.

And yet, TL071s are commonly used as pickup buffers. Even on 9V. We may be worrying about things that don't really matter to the music.

> it has to stay @2/3 V+ @9V but close to V/2 @24V.

Be sensible. If it "works" at 9V with 3V headroom on the plus side, it would work fine with 24V split 2/3 and 8V headroom on the plus side. So just bias it 2.2M+1Meg, or 1.065Meg+532.5K if you prefer that 355 sound.

> In 9V mode it needs as low current draw as possible

I did a buncha outdoor sound with 324 and 072 chips running on 9V battery. When the field generator wanders from 105V to 130V, it is nice if some of your EQ gear stays stable. And I wasn't getting overpaid. But I don't remember battery costs being at all a problem. A '072 will run a very long time on a fresh alkaline 9V batt. Is your music worth less than mine? (I won't argue... I hosted some very fine acts but also some real disappointments.)

> I need answers _precisely_ on what I asked

You don't always get precise answers unless you -force- them. Whips. Chains. Candy. Beer. Gold. Or something interesting and easy. This does not look interesting (at first), your "precise" demands don't make it more interesting (at first). There "ought to be" a 10-minute solution. If there isn't, the amount of beer needed to chain a mind to this problem is probably worth more than a year's supply of batteries.

-
-
-
There "ought to be" a 10-minute solution.
-
-
-
This kinda silly problem is my meat.
-
-
-
There really "ought to be" a 10-minute solution.
-
-
-

If there is a 10 minute solution, I'm blind. I've whacked at it for 2 hours, even cheated with a computer. You don't use isolation diodes in 9V systems if you need max headroom and battery life. As you say, a 3V Zener bias offset wants near as much idle current as your TL071. You don't use thrifty CMOS logic in 24V systems. All problems are solvable, but some don't come easy.

This will work. It is NOT finished, not optimized... already I see that R6 Q3 are moot since I had to jam D2 in there to stop a sneak-leak path. And it has the bug/feature that the battery may get a slight charge if left on while Phantom is applied... no harm to disposable batts. No pilot light, no tubes nor (functional) FETs, no sex in this city. And the drawing is a frikkin' maze.

Lot of simulation-only parts. All the switches, R8, and 0V are sim-setup. R17 could be '071 gate leakage, but really nulls a useless sim-"error". J1 is equivalent to '071 power supply current. C2 makes the graph prettier (and is not a bad value to clean the bias). D1 is 24Vz, transistors are 9-cent general-purpose NPN and PNP.

The sim-switchery sets four different combinations of power: off, phantom, both, battery. While "both" is not precisely what you asked, it -will- happen, and I won't give you something that smokes, and preferably not drain the battery. The fact that "both" might actually "charge" the battery is an accident. "Fixing" that would cost you more.

Oh... if the '071 really benefits from "Class A" (I don't think '071 has the guts to be a good "A" amp), and if that CCS can go to the negative rail, then the CCS can flow to Q1 collector.

 

[tv]

for clarity: here's the schem of "main" part without the rest of the clutter.



Guys, you are over-reacting :guinness: I outlined clearly what's bothering me in the OP. This is a small TH onboard design, not a spaceshultle computer :thumb: , i.e. size, elegance, stability, durability and simplicity matters.

(power switching is done similarly clean&simple with 2 schottkys and a power zener)

As of info gathered by now, I'll design the CCS to load the chip with cca. 2 - 2,5mA when under phantom. Are there any advantages of using f.e. 1,5V zener versus the 2,7V zener (that's what I drew in the schem).



NOW TO PROBLEMATIC PART:
heres a zener in biasing dividor marked as "~3V". Ideally, this should DO what I want (or what Self says TL071 wants), namely bias it close to 2/3V+ @9V and closer to V/2 @24V.

BUT would be a ZENER stable at so low current? Should I replace it with something else? A led?

(that's precisely what I asked in the OP)


P.S.: another thing is bothering me - when battery dies and voltage goes lower than 9V minus schottky drop, that sort of biasing as I drew it would start to become more and more problematic as the voltage drops further.

Wouldn't it be simpler to make a dividor just of say 120K and 220K resistors (similar to PRR's suggestion) and be done with it? The question remains: HOW would that affect the 071's / 2071's operation/performance when powered by full 24volts? In a BAD or in a GOOD way?


@gus: there is going to be a series resistor in the phantom power supply (7824 lifted with a 6v8 zener) as well as in the preamp. There's a zener built in preamp, as I mentioned. This is a time-tested config, simple and idiotproof. The only bad thing is the 7824 can't be directly mounted to the metal casing.

 

[Samuel Groner]

Size, elegance, stability, durability and simplicity matters.

Exactely. That's why I wrote my first post in this thread. Half the parts count, quarter of the troubles. Your approach is neither elegant nor simple if you ask me...

But would a zener be stable at so low current?

Very likely yes. It will be rather noisy though, at least you got a heavy low-pass filter there. But look at the opamp current draw: whether the zener draws 20 uA or 200 uA doesn't really matter, right?

Samuel

 

[tv]

half the parts but 4,5mA current. If only powered by a phantom, yes, otherwise no. If only powered by battery I'd take the AD820 as I said in OP. (well not really half the parts; only ONE zener more - OPA's benefit from Aclass biasing aswell afaik)

I'd rather have tha 20uA on zener because the opamp already goes @ 1,2-1,5mA @9V. TLE2071 a bit more, but tolerable.

Knowing the inside of several instrument onboard preamps I think what I have is on the very elegant side of things in existance, but that's IMO.

The drawing _is_ ugly, though.



P.S. I forgot to add that in my 071 deadbug prototype there is a noticealble difference in mids when biased to 2/3V++ @9v. Not sure _what_ this is, but there's definitely more "mid definition" and "wateryness" of TL071 goes away. :?: :!:

... afaik Fender also mis-biases 071's in their preamps (39k / 47k dividor). I used more radical 270K/470K in my test (w/o the zener atm).

 

[Gus]

re read my posts Drawn out the circuit

define phantom in your use with this circuit. A 48vdc with two 6.82 resistors in the -+ legs balanced?
OR
24VDC on one of the lines and unbalanced?

Don't get stuck on using the zener resistors are cheaper and you can set the ratio for bias

Also how much is the input bias node going to wiggle with the 220Ks divider.

Is that schematic correct 15K output to inverting input and then a 100K to AC "ground" bias node?

Do you understand what I posted about low Rdson mosfet low side switching?

If the gain is about one I would just use a resistor it should be close enough for the CC with the output centered at 12 volts with smaller swings.

 

[tv]

I understood what you said perfectly. The main circuit is as-is, tailored to that particular bass. The phantom is custom (read above). "Just works" and time-tested.

I am tinkering on making a "box" that would enable use of the standard 48V phantom, though.

Don't get stuck on using the zener resistors are cheaper and you can set the ratio for bias

The main idea is that the biasing network would mis-bias (@2/3V++) only when powered by 9V battery, but stay close to V/2 when on 24V phantom. A network with a zener added does precisely that, but has a drawback that I described earlier.



I am wondering if my questions weren't written well? (Not native english speaker)