Comment on my pre-amp schematic?

[AudioGeek]

Attached is my buffer with step-up transformer pre-amp...

could you please comment on it?

problems?

The 82K resistor is input resistor of following stage.. i simulated it with a 20K (SS amp etc) and seemed to work good.

Thanks!

[darkfenriz]

how about using current source to make a mosfet a better follower?

[maylar]

That design will clip the entire negative half of the input signal and most of the positive too. The transistor won't start conducting at all until the input exceeds the gate threshold voltage. You need to add some gate bias to bring the transistor into its linear range. About 20 ma of drain current would be right.

[AudioGeek]

well this is a P channel follower and connected this way I do not need to bias... the mosfet will be at about 3-4v bias.

will the volume control work at the input of the mosfet this way?

Thanks!

[AudioGeek]

to me a current source is essentially an amplification stage.

my goal is:

To develop a pre-amp that uses a transformers ability to amplify voltage without any impedance matching issues associated with transformers... i.e. use buffers.

hence I don't want to use a SS or tube amplification stage.

[PRR]

> my buffer with step-up transformer pre-amp...

Sure, it will work. No major problem.

> could you please comment on it? problems?

Some comments as you requested:

N-type FETs are a penny cheaper than P-type, and come in a wider assortment.

The Source-Drain voltgage is equal to Vgs. A MOSFET will live that way, but you are close to the Triode Region where an FET works different. That may not be an issue: you only want 2 or 3V out so 0.2-0.3V at the Source, a very small swing compared to the 2V-4V Vgs.

What is the THD at 5KHz when VR1 is at 50%? FET input capacitance is nonlinear with voltage and THD will rise with high source impedance (as with a pot set half-way). If I recall, it is worst at very low voltages such as this. Might be several percent. Might be less because you have no Miller multiplication. Might be musical. Isn't accurate.

On paper, performance would be "better" with a little gate bias. It's cheap: no significant DC current. I do like your simple elegance.

I don't think a current source would add anything much to this design. You are asking for 0.3V peaks out of a 12V supply. In effect you are using 75% of total supply voltage and power just to feed a "passive current source" (resistor). The MOSFET's Gm will swamp the DC resistance. The plan is inefficient but has few side-effects other than heat.

> will the volume control work at the input of the mosfet this way?

Look at some tube amps. This is a very common affair. The only issue with doing it on a MOSFET is: you can't kill a tube with static electricity, but you can kill a MOSFET. It is unlikely here, because you will normally never have the volume pot all the way up when plugging-in, so there will be some series resistance to work with the MOSFET's protection diodes. If you sold 100/day, you'd have to get more serious about input protection because 1% of the buying public are total klutzes, but in DIY you will use common sense and also you can fix it quicker than a klutz can write a nasty letter to the factory.

> i simulated it with a 20K (SS amp etc) and seemed to work good.

IHF standard used to be 10K load. And I have seen lower input impedances.

The real "problem" is that transformer. Assuming you allow for 10K loads, it looks like 100Ω:10KΩ. That's an unusual part. Do you know where to buy one? (150Ω:10KΩ is reasonably available as a pro-audio part.) Transformers are NOT "free gain": aside from being costly they distort in ways very different from tubes/transistors. Unless it is low-ratio or low-impedance (10K is a fairly high impedance for a hi-fi transformer) it will limit bandwidth to not much more than 20Hz-20KHz, and can be prone to ringy highs (and with coupling caps: bumpy bass). Selection of a good hi-fi transformer is really harder than refining a transformerless circuit.

In SPICE, set the transformer parameters to: Lpri=1H, Lsec=100H, K=0.99, and load the output with 10K and 1,000pFd. Sweep from 1Hz to 100KHz. The bass will roll-off; perhaps at an acceptable point but perhaps humped due to C1. The treble will roll-off badly, and may even ring. Actually K=0.99 is a pretty bad transformer. A power transformer (12V:120V) will do about that good, a hi-fi transformer will have K up around 0.999 or better, and you can probably get more than 1H without excess copper loss. But still it is a significant bandwidth limit.

And transformers are "lively" things. There are magnetic forces inside which make the windings and laminations shake and even buzz microscopically. Such effects are minimized in manufacture, but impossible to eliminate, and can make or break the sound of a transformer. One guru talks about the type of rust on the laminations.

Having said the rational objections: a good transformer is a very good audio device, and some less-good transformers can be very "musical" in the right situation. Recording engineers often turn to transformer equipment to get "a sound". (And also to solve thorny ground problems.)

The 100Ω transformer input will need more than 39uFd coupling cap, easy to change.

> To develop a pre-amp that uses a transformers ability to amplify voltage

You understand a concept, but you may be looking at only one side of the horse.

In hi-fi, we really want current gain as much as voltage gain. Consider a "unity voltage gain" device (say an equalizer that happens to be set "flat"). We conventionally demand that it have input Z over 10K, output Z under 1K. Even with unity voltage gain, we need available current gain of 10. If we also have voltage gain of 10, we need available current gain of 100. You have shifted the voltage-gain problem out of the tube/transistor, but increased the need for current gain in that active device. The MOSFET is pretty good current gain; it would be far harder to do this in one BJT or tube.

> I don't want to use a SS or tube amplification stage.

It IS an "amplification stage". It isn't our ordinary "volt-amp", but a strained current-amp has problems too. It is different, not certainly better. It may be better, but I note that few designers go this way, and they are not all stick-in-the-mud guys.