Mpp

[Netlist]

Topic, please. Last I looked it said MPP, not each other and not objective/subjective.
We have other threads open for the latter, the former can be done on the myriad of social networks out there.

http://en.wikipedia.org/wiki/List_of...rking_websites

/Hugo

[Joachim Gerhard]

Thank´s for the reminder. New circuit ideas will follow soon. Maybe i can put subjective and objective in one glass and shake up a new cocktail.

[Netlist]

[Joachim Gerhard]

I am a fool to love you. Billy Holliday

First a quote from an aplication note of Analog Devices concerning intrumentation amps:

"is single-ended with respect to a reference terminal.
Most commonly, the impedances of the two input
terminals are balanced and have high values, typically
109 , or greater. The input bias currents should also
be low, typically 1 nA to 50 nA. As with op amps, output
impedance is very low, nominally only a few milliohms,
at low frequencies.
In contrast, if a standard op amp amplifier circuit were
used in this application, it would simply amplify both the
signal voltage and any dc, noise, or other common-mode
voltages. As a result, the signal would remain buried under
the dc offset and noise. Because of this, even the best
op amps are far less effective in extracting weak signals."

Intrumentation amps intrigued me for a long time.
In my commercial stages i use both the transimpedance and the transresistance version. One of the lowest noise Inamps is the INA163 with 1.0nV/qHz. Considering the feesback resistor it is hard to get any better then 1.2nV/qHz. To circumvent this i use massive paralleling. For example 8 AD797 in the Spiral Groove Phono Stage for a result of 0.5nV/qHz balanced for example. This is good for series production because no sorting and no servo is needed.
In DIY we can efford to use a discreet input. Hfe and Ube matching is time consuming but why not if we do it in our spare time. i looked at posibilities of input biasing and see two major posibilities. One is the classical shunt feedback version that lends itself to an unbalanced input and one is the "direct bias" version that could be a good candidate for balanced operation. The input pair can be cascoded for better PSU rejection but may require some dominant pole compensation. See my 3 concept circuits with default values that do not represent the "real" values. I see no reason why to go to unbalanced immediately after the input stage so fully balnced RIAAs in various disguises are a posibility. I will alaborate on that later.

[Joachim Gerhard]

Errata. there was a rooting mistake in the cascode version so here comes the real thing.

Further work on the DC coupled Fet MPP:
I studied the instability problem further and also took advice from Jürgen Ultee to give the servo less "authority" as D.Self likes to call it. The situation was the following: the servo supllied ca.1V into 470 Ohm for DC of under 1mV at the output. I changed that for 3.5V into 1.8kOhm. The low level moovement of the cones at insane volume now is minimal. At "normal" level, that means 100dB peaks in my case ( around 25 W), anything more hurts my ears, the naket eye can not detect anything. Surprisingly the sound got better still. I am now totally happy and can imagine that some people prefer the result over the Low Z versions because it is very smooth and human. It definetely sounds like music. I can hear no loss of detail any more still it does not have the dramatic slamm of the Low Z versions. I did not find the time to try the 1kOhm Caddock in comparison to the 3Kohm YAM2. That is on the argenda tomorrow.

[scott wurcer]

Originally Posted by Joachim Gerhard

Errata. there was a rooting mistake in the cascode version so here comes the real thing.

Some serious schematic problems in the second one?

[Joachim Gerhard]

I have looked at the schematic you send me and can not understand how the input transistors are biased. Everything seems to be the oposite way around.
When the cascode in my circuit does not invert phase then it is nonsense.

[Joachim Gerhard]

Could that work!
I have no experience with Rush cascodes.
Anyway, i first will try the uncascoded versions and then work my way up.

[chrissugar]

Hi Joachim

The balanced version to work correctly as an instrumentation amplifier, you should leave out R12. You only need R11, R15, R13, R5 around OP3.

Isn't R14 a bit large? Is there any reason for this?

Probably you should choose some suitable opamps because most of them are not happy to drive loads as low as 1k or lower (distortion?). Feedback resistors should be considered as part of the output load.

chrissugar

(my first post at diyaudio)

[Joachim Gerhard]

Hi Chrissuger ! I got a bit confused and have drrawn the resistor after the lower inverterstage double. Sorry! The first version of the cascode should work after removing that resistor in principle. I posted after a long day of work and was a bit tired. When you read my text carefully i put in only default values, so i made all resistors 1kOhm and all caps 1uF and so on. So this circuits are only concepts and i am still working on the "real" values. Sorry again for any confusion. I will first build the balanced, uncascoded version and i have also put more thought into the cascoded version that will eventually include a constant current source. When i have the circuit ready i will post it.
On another issue i have again followed advice from Scott Wurzer to make the input stage of the High Z Fet Headamp trimable. When the cascode transistors are selected for Hfe and Ube using the trimmers can reduce second harmonic and DC offset the same time and not much selecting of the input devices are necerarry.
I think this is very convenient. Thanks for that advice.