Symmetrical input stage

[hjelm]

Hi,

Trying to design a low gain bipolar line amplifier stage and the first problem i find is that i get too much harmonic distorsion in the inputstage when i try to go from voltage to current.

The idea i had for the gain stage are dual current mirrors and then terminating the current in a resistor which determines the output impedance.

Any suggestions on a bipolar inputstage that has low distorsion.

Input impedance is not too critical 10k is fine.



Thanks
Mikael

Edit: The Subject was confusing so i changed it.

[traderbam]

What output Z do you need?

[peranders]

If you have some "what-if" questions, why don't you try LTSpice (freeware) and fool around with this program. LTSpice has also FFT.

I have simulated four projects and the result has been rather good.

The most interesting simulation was for my QRV-06 headphone amp
Current feedback high speed headphone amp with diamond output buffer
You can have the simulation files. Send me a message.

If you want low gain and not the lowest noise possible, try to insert emitter resistors (at the input transistors), 47-220 ohms or something like that. You could also test cascodes. This is good for distortion also. Check the result with LTSpice.

[hjelm]

Traderbam:
For the input stage or the whole line amp?
For the input stage since i would like to have it current out i suppose as high as possible but small enough to not cause distorsion due to the following (currentmirror) stages is fine.

Peranders:
Already running winspice simulations. The problem is that i cannot seem to get the THD down below 2% with my current input stage which is a stage based on as you suggested emitter resistors. Where i take the collector currents to the current mirrors. Tha problem i have is that almost all the distorsion seem to stem from the input stage.
I should note though that i have used 1k-10k emitter resistors but the distorsion didn't seem to go down if i went for the smaller values.

Another factor that might be limiting in my application is the low supply rail +-10 and the input swing of 1.5v AC.

Thanks for the replies.

[traderbam]

hjelm,
Could you explain a little more about what you are trying to achieve?

It sounds to me like you are trying to make a line amp (voltage gain?) with 10k input Z (output Z?, max output V?).

You seem to be having trouble with the first stage linearity - but what is the second stage? Are you intending to use feedback or are you trying to make an open-loop design? Are you talking about having dual diff pairs feeding current mirrors then feeding a common resistor to ground?

For a diff pair stage the best linearity occurs when the differential current is very small compared with the bias current and the bias current with no signal is identical in each transistor.

In any case your best sound will be if you use a simple asymmetrical design, diff pair feeding CE with feedback.

[hjelm]

Traderbam

The thing I am trying to achieve is a line amp with gain 2.

This is simple enough and I could probably find something on the net or in this forum, or use an opamp.

I would however like to create something by myself if possible, I would probably reinvent the wheel or at least an octagonal version of it in the process but it would be a learning experience.

My plan was as you say to use a dual transistor input with emitter resistors, not differential, and two current mirrors and feed the current through a common resistor to ground. The idea was that the symmetry would remove some distorsion.

The design criteria were bipolar devices BC550C BC560C open loop class A.

This doesn’t seem to work and it seems that the emitter resistor idea is the one to blame.

Your last statement is interesting but I would like to have an open loop amplifier if possible.

I am attaching the first version of the line amp i thought could work.

[peranders]

Hjelm, you could try a current feedback topology, like my amp. Your circiut is starting to like a current feedback amp.

Download LTSpice and test because this software (or any other SPICE software) is excellent if you just want to test. If you want to know the gain the software is pretty accurate.

[hjelm]

P-A
Is it not possible to do a gain of 2 with distorsion down at 0.01% without feedback?

[peranders]

It is certainly possible but I'll suspect that you must use cascodes in the current mirrors.

You can check my diamond output stage, both in the QRV-06 and in the separate buffer QRV-05. I have tested this buffer with excellent freeware Rightmark Audio Analyzer. With my soundcard I could draw the conclusion that the distortion was below 0.008%

[hjelm]

Finally i found the information on bipolar cascodes i wanted, passlab where else? Wouldn't it be good to have cascodes on the input as well? It is not the current that is unlinear rather the Vbe but that might improve as well with cascodes?