Making a BD139 voltage follower buffer, need a little help

[00940]

majkel.... the ground output too is buffered by a bd139

edit : you edited your post in the meantime

 

[majkel]

Quote:

Originally Posted by 00940 /img/forum/go_quote.gif majkel.... the ground output too is buffered by a bd139 edit : you edited your post in the meantime

Yep, I realised the ground buffer later. I was reading in a small IE window, you know...

 

[bearmann]

Ummmm... my I ask, why you have to use a virtual ground power supply, when using a ground channel?
So you can't use a dual power supply?

sorry for this dumb question and best regards,
bearmann

 

[tyre]

.

 

[FallenAngel]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif Why don't you build a classic symmetric PSU, instead of messing around with all that ground channel and virtual ground stuff? What you need is: - center tapped transformer or with two equal secondary windings (better), 12~17V AC output - one or two (better, only when you have double secondary winding transformer) Graetz bridges - some 'lytics - 7812S and 7912S chips Sorry for not providing the schematic ATM but I'm in work, away from my EDA software now.

Basically because I just tried that and is what I had all along. Dual secondary transformer with 2x 12V AC outputs, followed by 2 diode bridges and a 2 LM317 based power supplies (I didn't have an LM337 handy so I used a pair of these and tied the output to form a ground). The power supply is almost exactly Tangent's TREAD. It was outputting exactly +/- 12.5V. I think there was a nice and clear 60Hz buzzing noise and I can't get a very good AC ripple reading from the PSU because my DMM goes all crazy trying to measure it.

 

[majkel]

TREAD is just a positive half of your destination PSU. You have to build a complementary circuit using LM337 (with some changes) and then create the ground from the (-) voltage of the positive supply section and (+) of the negative supply section connecting them together.

 

[FallenAngel]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif TREAD is just a positive half of your destination PSU. You have to build a complementary circuit using LM337 (with some changes) and then create the ground from the (-) voltage of the positive supply section and (+) of the negative supply section connecting them together.

But if I make 2 TREADs and connect a V- from one to the V+ from another, it creates an equal ground, no?

 

[amb]

Yes, you could make two TREADs and hook it up as a split supply -- in fact this is a better solution than using a LM337 for the negative rail, because the LM317 performs better than LM337.

However, using a single TREAD and then splitting it into a dual rail supply with TLE2426 is cheaper than doing two TREADs, and offers a performance advantage in a 3-channel output application (where headphone return ground current isn't dumped into the virtual ground). It gives you the benefit of a tracking dual rail supply because the TLE2426 will self-adjust the split, without the complexity and cost of a real tracking split-rail PSU.

 

[majkel]

Using positive regulators for the negative rail is kind of weird because then you will have the negative voltage regulator referenced to the rail instead of the ground. However, it might work but is suboptimal.
Regarding the TLE2426, better replace it for a decent op-amp as a buffer + resistor divider. TLE2426 itself has got an unpleasant sonic signature when used that way. You can civilize it's influence by placing a ~100uF capacitor between it's output and either rail (try which, because the choices slightly differ - due to different phase shift at the TLE's output).

 

[amb]

Quote:

Originally Posted by majkel /img/forum/go_quote.gif Using positive regulators for the negative rail is kind of weird because then you will have the negative voltage regulator referenced to the rail instead of the ground. However, it might work but is suboptimal.

You might think it's weird, but voltages are simply a difference in potential, and when the two regulators are each referenced individually, neither one "cares" which is positive or negative, as long as it's hooked up appropriately. Using two positive regulators to make a +/- split supply is nothing new and has been done by many designers in many circuits. Even the dynahi PSU does it that way in the pre-regulator stage.

Quote:

Regarding the TLE2426, better replace it for a decent op-amp as a buffer + resistor divider. TLE2426 itself has got an unpleasant sonic signature when used that way. You can civilize it's influence by placing a ~100uF capacitor between it's output and either rail (try which, because the choices slightly differ - due to different phase shift at the TLE's output).

Sorry, but that's nonsense. In this circuit the TLE2426 does nothing more than provide a mid-way split in order to serve as the signal ground zero reference, and to provide the sub-microamp current required to bias the opamps' inputs. It swings no voltage nor is it asked to sink/source appreciable current, so what does phase have to do with it? Look at the schematic carefully and see what nodes are actually connected to the virtual ground. The TLE2426, when used in this way, imparts no sonic character to the amp at all.

Oh and btw, what do you think is inside a TLE2426 anyway? It's a resistor divider and a buffer! And a fairly good one at that.

 

[FallenAngel]

Hey guys,

I'm about to finish up with the 3-channel version, but have a quick question regarding the input resistor (latest schematic shows 4.32K). I will only be using FET opamps and I know this resistor is used to balance input impedance on FET opamps and trim DC with bipolar opamps. What's a good value, as I'm seeing different values all over the place. Mini3 v2 is 330R, PINT was 1K, majkel suggesting 910R, Pimeta/PPA are 4.32K. What would be a good value for FET-only opamps?

Thanks in advance

 

[TzeYang]

My personal favourite is 4.7k for non inverting input, and 4.7k for feedback, and the feedback ground resistor 1k~2.2k

Generally, you don't want too much or too little for the input impedance. Best of all, remind yourself to balance the input bias to minimize error.

 

[majkel]

Where TF is my post sent a couple of days ago in response to AMB's post #55?! When somebody doesn't understand my explanations this doesn't mean I was wrong.
To the mods: As a common rule I suggest sending me PM when deleting my posts and include a brief explanation.
When comparing the abilities - mine are MSc degree in Electronic Engineering + 6 years of professional experience. Tell me yours.

FallenAngel, sorry for the off-topic, that pissed me off.

 

[amb]

FallenAngel, for an FET input opamp, the value of that resistor is not very critical because the input bias current is so low, there is no need to worry about balancing the resistances on both inputs for low DC offset. A few hundred ohms to a couple of Kohms are all ok.

 

[FallenAngel]

So I built this latest 3-channel amp and I get some very strange distortion and symptoms. I can disconnect one channel of the source and I still get sound in both headphones. I have also discovered that it is not a wiring / pot problem.

Then, I started thinking - there is only 200 Ohms between ground and each channel because both are connected through V-! Wouldn't this prove a little problem?

EDIT: Sorry, bad opamp, damn OPA627 burned! Now it's fixed and sound pretty darn good! Photos soon!