Kind request of help _ basic but low THD discrete buffers

[ginetto61]

Hi thanks for the valuable info Speaking of buffers i would focus more the output stages of circuits A buffer by definition has the duty to provide current and lower the output impedance without adding any gain
I am also looking at diamond buffer but i cannot reach low THD ... the problem is as always to have a little idea of what i am doing That is the problem

 

[tomchr]

Meanwell are very good but all SMPS has some noise on the output. You can attempt to simulate this by adding a spiky HF content to the supply.

A better approach would be to simulate the ripple voltage directly. Most SMPSes that operate in continuous mode will have a square wave ripple waveform on the output. For a good SMPS it's typically a few mV peak-to-peak in amplitude.

Also note that the switching frequency of an SMPS is always placed above the audio band. Some even switch above the AM radio band (so > 1.8 MHz). This means that any noise on the output of the SMPS won't be audible. It could present an EMC issue, but that can be avoided by good PCB layout.

I understand that chip regulators have usually poor PSRR above certain frequencies

That's true to a point. The older regulators (such as the LM317 from 1976) tend to have poor regulation above the audio band as they were optimized to deal with mains hum. Do keep in mind that rectification hash can extend well into the audio band, so the regulators aren't necessarily all that effective at dealing with that either.

Modern voltage regulators have good ripple rejection well above the audio band. The TPS7A4700 by TI, for example, has >55 dB of ripple rejection up to 10 MHz.

I followed also a very interesting thread on capacitance multiplier I think that active filtering should be more efficient than passive filtering

For the most part, yes. That's often especially true once you account for the parasitic components in the passives.

Capacitance multipliers do well if the input voltage is stable. They can have issues in scenarios where the input voltage droops below the voltage on the capacitor in the cap multiplier.

Nevertheless i hope to be able in the near future to sim also some passive LC or RC filters

I've definitely found an LC filter or CLC filter at the output of an SMPS to be an effective way of removing switching hash. If you use a shielded inductor you (likely) won't have EMC (electromagnetic compatibility) issues either.

If i had discovered this SW long ago my approach to audio would have been more rational it's never too late

Heh. I'm so glad to hear you say that. Circuit simulation is a powerful tool indeed.

Getting back on the original topic: A plain vanilla emitter follower should have pretty low distortion, especially if you crank the bias current. There are better options (the diamond buffer springs to mind) but I thought I'd mention it.

Tom

 

[ginetto61]

Hi thank you very much indeed for your kind and very valuable advice I am getting great information from great experts here in this wonderful site
I am very beginner without the needed background in electronics I tried to understand the theory but failed I am a little retarded
Yes i am captured by this LTSpice a lot I am changing parts without knowing completely what i am doing I started simulating some very well reviewed commercial products and found that at least from a simulation point of view they are not completely optimized I can only imagine how hard must have been in those days when such sim programs were not available
In design i like the approach to try to extract the most from very basic circuitries Nelson Pass is a Master at this
But i understand that very basic circuits have intrinsic limits expecially when feedback is not used
This means passing from -80dB to -100db of 2nd order harmonic distortion
And also it seems easier to make simple but low THD buffers than preamps No pain no gain seems to apply also to preamps
Thanks a lot indeed for your helpful advice

 

[Elvee]

Another 2-Bjt option is the repeater or duplicator circuit.
It relies on cancellation to reduce the THD.
This is the simple follower, for reference:

The THD is 0.018%.
If a complementary transistor is added in front, cancellation occurs and the THD is reduced by more than 20dB:

Of course, the resistance ratio's have be carefully chosen for a good cancellation to occur.
It is also possible to connect the collector of Q2 to emitter of Q1, but the resistors values need to be adapted

 

[tomchr]

I've done a lot of follower-up + follower-down in IC design. It's useful for keeping the signal between the supply rails.

The double-follower is also useful as its input impedance is pretty high.

Tom

 

[Ken Newton]

it is the old lh0033 first stage schematic. both fets are operated in the idss mode, the lower one acting as current source load for the follower. indeed drain and source can be swopped. with matched fets and symmetrical voltages bothinput fets operate under the same conditions and offset should be minimal, operation at zero Vgs.

Sure, but what seem usefully unexpected are the 600 Ohm performance figures.