Simple I/V for TDA1541

[Pedja]

Originally posted by Kuei Yang Wang
I think AD also has a transconductance Op-Amp (743? not sure) where the transconductance node is available externally and that can be used open loop thusly....

AD743 is quite conventional opamp, but AD844 is most likely what you are thinking about.

Pedja

[Elso Kwak]

Originally posted by Kuei Yang Wang
Hi,



Not off-hand, but it really should be obvious from the Datasheet. Here the basics for the OPA660.


L8er T

Hi Thorsten, congratulations, you just constructed a current feedback amplifier but without the usual feedback resistor.
This whole thread seems to explore and reverse-engineer and possibly improve the circuit by Jocko.
I strongly disagree with you Thorsten that a discreet circuit would be worse than integrated. Why on earth would Jocko use four dozens of transistors on his IV-board if he was satisfied with a simple AD844?. BTW that is real current feedback amplifier.
I constructed as a first attempt to a discrete design a circuit as in the OPA627 datasheet and it sounded a lot better than the original.
The whole idea of the IV-converter seems to circle around the input impedance of the IV-converter as the output of the DAC chip has limited compliance. What does that mean?
Well to qoute from Horowitz:
"<B>Ideal current-to-voltage converter</B>
Remember that the humble resitor is the simplest I-to-V converter. However it has the disadvantage of presenting a nonzero impedance to the source of input current; this can be fatal if the device providing the input current has very little compliance or does not produce a constant current as the output changes" [due the current]

The transresistance configuration can be used that is a opamp with non-inverting input grounded and the current going to the inverting input. A feedback resistor is connected between the output and the inverting input. The opamp does not need to be IC-type. It works very well with a discrete opamp. This is the configuration I am using with cascoded JFET's.
Now for Jocko and many others feedback seems to be a ugly F-word so I asume Jocko is using a current feedback configuration without the feedback resistor just as you do Thorsten.
As for the offset there are ways around Jocko wrote. I see some IC's on Jocko's board and suspect these are servoamps for offsetcountermeasure. A full complementary circuit would also be a asset.
In the datasheet of the AD844 it is explained that the current feedback amplifier has some advantages over a normal voltage feedback amplifier as a IV-converter. This is too technical to repeat or explain here but has to do with phase margin and stability. Also in fig. 28 of the datasheet is a simplified schematic in which the first part is almost excactly the scheme by Thijs. (the other part is the double diamond outputbuffer, LH002 style) I have been using the AD811 and OPA603 current feedback amplifier during some years in my DAC. After reading the Pooge Chronicles by Walter Jung and Hampton Childress. Also tried the AD844. Thereafter I switched to my discrete circuit as it sounded superior to the IC-currentfeedback amps. See Sonny Andersons thread.
Jfet/bipolar PCB and Schematic (finaly)
(SCM.zip)
His circuit is similar to mines.

Of the three transistor configurations possible the common base amplifier (input goes to the emittor) has the lowest inputimpedance. This happens to be the inverting input of the current feedback amplifier.
With the transresistance configuration a virtual ground at the inverting input is created.

[Pedja]

Hi Elso,

Sorry, but I want to ask you, what is the (current) feedback without the feedback? No feedback, no current feedback. Simply. But signal goes to emitter. That is not feedback at all. That is common base. The result is buffered DAC’s current output, and its voltage compliance is not critical afterward. That is all. And not every current feedback monolithic opamp can serve for this purpose. But OPA660 and AD844 can.

However, it is little dubious if cfb opamp like AD844 is really useful this way as it’s inverting input impedance is 50 Ohms (non-feedbacked, typicaly, according to datasheet, SPICE model actually shows 65 Ohms), generally low, but higher than we would like. Btw, OTA’s emitter of OPA660 has some 7 Ohms input impedance.

Pedja

[ThorstenL]

Hi,

Hi Thorsten, congratulations, you just constructed a current feedback amplifier but without the usual feedback resistor.

Actually, Burr Brown constructed (and manufacturerd) the device, so congratulation for that to them.

And congratulations to you for my newest favourite oxymoron, namely:

"a current feedback amplifier but without the usual feedback resistor"

BTW, I did not construct ANYTHING LIKE what you say, but it sounds good. Congratulations.

This whole thread seems to explore and reverse-engineer and possibly improve the circuit by Jocko.

I don't know about "the circuit by Jocko", but I do have in my little black book copies from Audio in the earlier part of the 80's of the last century which show the same principle as discussed here and if I may be so bold, much better implemented than anything I have seen in this thread.

I strongly disagree with you Thorsten that a discreet circuit would be worse than integrated.

I know you do.

I also wish you a good time selecting dynamically matched complementary transistors (and matched at least a few MHz so this whole shebang really works) and to minimise parasitics in the layout. I for one have long given up discrete solid state circuits, in the basic area where I need them to work I find the right IC's applied competently outperfom OBJECTIVELY AND SUBJECTIVELY ANY discrete circuit presented to me.

But that's just me and I have only been playing this game for two decades, so WTFDIK.

Why on earth would Jocko use four dozens of transistors on his IV-board if he was satisfied with a simple AD844?.

Beats me. But IIRC he prefers oversampling too. Maybe an innate prefernce for the complex over the simple.

It takes all sorts.

I constructed as a first attempt to a discrete design a circuit as in the OPA627 datasheet and it sounded a lot better than the original.

Please qualify your remarks by saying:

"To me, in my system and based on my personal preferences and prejudices it sounded better."

Or if that is not correct state the "rahmenbedingungen" (cornerstones may be the best contextual translation of rahmenbedingungen). unqualified remarks assert usally superiorty that does not exist even in subjective terms, never mind the type of subjectivism usally falsely peddled to us as objectivism (see Hume).

Now for Jocko and many others feedback seems to be a ugly F-word so I asume Jocko is using a current feedback configuration without the feedback resistor just as you do Thorsten.

Feedback is feedback.

What I propose does not include ANY form of feedback, not even local degeneration. Not knowing "Jocko's Circuit" I shall not speculate about what it is and what it not is. I do know what the circuit I propose is however and it does not include feedback in the I/V conversion..

As for the offset there are ways around Jocko wrote. I see some IC's on Jocko's board and suspect these are servoamps for offsetcountermeasure.

Well, I prefer to do without servos. I have in purely subjective sonic terms (though my sample size in these investigations considerably exceeds the number one it is not large enough to allow statistical significance at a .05 level) always found servos on the liabilities side of the balance sheet.

A full complementary circuit would also be a asset.

I am proposing to use one. And indeed one that due to the special process used is truely complementary, unlike ANY discrete circuit based on so called "complementary" types, selected for equal Ft and Beta (or transconductance, treshold voltage and Ft in FET's) or not. So what is your problem. Apart from the fact that the suggestion came from me, that is?

After reading the Pooge Chronicles by Walter Jung and Hampton Childress. Also tried the AD844. Thereafter I switched to my discrete circuit as it sounded superior to the IC-currentfeedback amps.

Again, you fail to sufficiently qualify your remarks. I personally prefer to use a combination of transformers and valves for I/V conversion (incorrigable reactionary is me). But what holds true for me is not the universal truth.

Sayonara

[Elso Kwak]

Hi Pedja and Thorsten,
I meant the <B>topology</B> of the current feedback amplifier of the course. The AD811 has 14 Ohm inputresistance on the inverting input and the OPA603 30 Ohm according to the datasheets.
I tried to make clear the similarity between Jocko's circuit, Thijs's & Rudolf's schematic, OPA660 and common current feedback topology. Apparently it is not appreciated by all
Thorsten I am surprised by your reaction. Just before you left the AudioAsylum you wrote that "most" solid state was fatally flawed.
http://db.audioasylum.com/cgi/m.pl?f...lawed&session=
Now the right IC's are OK?

[Elso Kwak]

Originally posted by Elso Kwak

of the course.

Please read instead "of course".
And it is LH0002 style buffer.

[ThorstenL]

Hi,

Thorsten I am surprised by your reaction. Just before you left the AudioAsylum you wrote that "most" solid state was fatally flawed.

Yup, I made clear MOST. Certain modern IC's are indeed "alright", few if any discrete circuits are, for the simple reason of severe dynamic mismatches between active components.

Sayonara

[Pedja]

Hello Elso,

Opamp’s open-loop input impedance is important if you could use it in open loop. AD8009 also has low inverting input impedance (8 Ohms, available only in SO package though), but the question is how to use it (and AD811, AD8001, OPA603) without the feedback. You only have their final (voltage) output available for connection. The only way to use them as I/V is to use them like classic opamp. Or I am wrong?

Again, that no_feedback approach can be meaningless if feedback doesn’t shorten bandwidth, and cfb doesn’t (… and if bandwidth is what we need here ).

Pedja

[Elso Kwak]

Hello Pedja,
I only used the AD811, AD844, OPA603 and my own discrete opamp in the transresistance configuration. I.e with feedback applied. {Non-inverting input grounded and the current going to the inverting input. A feedback resistor is connected between the output and the inverting input. }
With the TDA1543 you can connect Vref to the non-inverting input of the opamp just as in the datasheet. You can even null the offset at the output by adjusting the resistor from Vref to ground but I found the sound better without the resistor.
I am not sure it will work without feedback. Maybe it works with the OPA660. Kuei Yang Wang, is your schematic a <B><I>working</B></I> circuit?
As said earlier I am not against feedback.

[rbroer]

Hi Elso,

I had been using opamps for I/V like most others, and much preferred the LM6172 as least evil above others like OPA2604, AD811, LM6181, AD797, OPA627, AD828, AD826, AD825, etc. etc.

About 1.5 years ago I made a real big step forwards using circuit shown; applied it to TDA1541 and TDA1543 cd players.
Didn't use opamps as I/V anymore since then.

The common base circuit is another step forwards and to quote Droopy "I'm happy".

If the Thorsten proposed circuit SOUNDS better, I'll use it, it is minimalistic enough for my DIY activities.
So, if anybody could send me a few OPA660's, I'll try it, compare it to my current common base circuit and post the results.

In the mean time, I started this thread to let other DIY know they can get good sound from a very simple discrete circuit, using generic parts and sounding so much better than the opamp I/V circuit used in many cd players and dacs.

As always, I'll try to pick up as much hints and suggestions as possibe, to complete my quest to audio nirvana...