C3g Aikido Phono

[simon]

Will do.

Tripling the fets the common resistor beneath the source resistors needs decreasing to 12R, which gives a gain of 28.7dB. And knocks another 5 or 6 units off the noise.

4 paralleled fets needs the common resistor decreasing to 10R, giving a gain of 28.5dB. And knocks another 5 or 6 units off the noise.

5 paralleled fets needs the common resistor decreasing to 8R, giving a gain of 28.5dB. And knocks a couple more units off the noise.

All very interesting. And a lot less effort than actually building something!

[simon]

Thinking about the power supply for this MC preamp stage I can't help thinking there's good potential to create ground loops if I'm not careful.

Perhaps the easiest way to get a 12V PS is to use 3no. 4V or 5V zener on the end of the 230V shunt reg. The audio ground could then just loop from the input phonos to the MC preamp stage to the C3g.

But I wonder if the zeners will be good enough.

Maybe better to have a separate reg with its own trafo and raw DC, which if left floating should get its ground from the audio circuit, which ultimately gets its ground reference from the resistor which is connected to the output phono at one end and mains AC earth at the other.

Does this seem reasonable or have I missed something?

[Nick]

I would strongly consider batteries of some form. Floating, quiet, whats not to like.

[simon]

2 x 9V in series is easy enough. Even without a switch they'd last a long time I guess.

[simon]

Whilst thinking about drafting a pcb for this it occurred to me that the drain resistor I'd drawn for the second LSK170 is bypassed by the paralleled output from the second LSK170. The more I think about this the more I'm confusing myself.

I found a Linear Systems application note which touches on paralleling fets
https://www.digikey.com/Site/Global/Lay ... y%203%20dB.

This is Fig 3b on page 8:
.

Paralleled FETs.png (30.39 KiB) Viewed 1365 times

.
So I think I need a common drain resistor rather an individual one for each fet. Sized for the increased current. Am I on the right lines?

[Nick]

Whilst thinking about drafting a pcb for this it occurred to me that the drain resistor I'd drawn for the second LSK170 is bypassed by the paralleled output from the second LSK170. The more I think about this the more I'm confusing myself.

So I think I need a common drain resistor rather an individual one for each fet. Sized for the increased current. Am I on the right lines?

Err, think I mentioned this some time ago

I think there is a error though. The 40r source resistor is shorted out

I would suggest a combination of a individual drain resistor for each fet combining to a shared one under them all. The individual ones will help balance the individual fet current out.

[simon]

Okay, you're suggesting the opposite to what I was thinking. But what I can't understand is with separate drain resistors, the outputs from each drain are in effect bypassing one of the drain resistors. Which is why I thought I might need a common drain resistor. What am I missing? (I sorted out the bypassed to ground 40R under the source.)

[Nick]

Not sure I understand the problem. Something like this:

[Nick]

I think I confused source and drain. Sorry about that.

[simon]

Ah, got it. Yes that makes sense now. I still learnt something as I sussed something wasn't right

[simon]

And a padded envelope has just dropped through my letterbox, thanks muchly .

I'll rerun the sims for the change in D and S resistors and see whether the noise drop continues per each additional fet. I was thinking of using a pair, possibly 4, but is 8 per channel recommended? Or do I need to match them for best performance?

[Nick]

Don't know, the ones I sent are not matched in any way other than they are outside (on one side) the range of values that I want to use in my phono stage, but that's just to allow simple "build and it just works", So they will be closer matched than just buying a bag from the makers. 2 or 4 sounds like a good compromise, I just included ore to give you the option. TBH, one of them is quiet enough for 300uv cartridges, so I would just play and see what happens.

[simon]

Just measured them. Oh the evenings of fun you must have had measuring big bags of these things

[Nick]

I have a phono stage PCB with a transistor socket for the jfet. Just sort by the anode voltage on the ECC88 on top of the fet. I have found that that single measurement is good enough to sort by op point in the phono stage and gain is directly correlated to that voltage.

[simon]

A bit of an update for record purposes having remodelled the parallel fets.

With the single fet model I tried bypassing the source resistor with 470uF and this added 5dB of gain, nearly 33dB of gain now. But the noise increased significantly from ~53nV/Hz1/2 to ~84nV/Hz1/2.

With two paralleled fets, common 1500R drain resistor, 10R source resistor under each fet with a 18R beneath the two 10R resistors, the gain is 28.5dB. Adding a 470uF bypass cap to the 18R resistor increased gain to 32dB, +3.5dB.

Adding a 470uF bypass cap around the two 10R source resistors increases gain again, to 33.3dB.

Noise with the unbypassed paralleled fets is around 40nV/Hz1/2. With the single bypass cap for the 18R resistor the noise increases to 55nV/Hz1/2. Also adding the bypass caps around the 10R resistors increases noise to 62nV/Hz1/2.

So noise is better without the bypass caps if the lesser gain is adequate. Paralleled fets with a lower bypass cap has noise similar to an unbypassed single fet.