Ultra linear cascoded 'Le Pacific' JFET phono stage with MOSFET output buffer

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Yet another variant of the 'Le Pacific' phonostage. Always wanted to build a discrete FET phono stage with passive eq, while not using too many exotic components, but the 'Le Pacific' phonostage had deterrently high output impedance and looked like it could do with cascoding, so I knocked up this schematic for what should be a good phono preamp which can drive the low impedance (15K) input of my line level preamp without any trouble.

The Class A MOSFET buffer on the output was partly borrowed from ESP P83. D3 protects the gate of M1 from any excess voltage when turning off and on, should it occur. 2N7000 is an inexpensive device, I have tons lying around and wanted to use a few.

I've also a stereo collapse option (a switch between the 'stereo collapse' marker for each channel) so I can get the most out of my monoaural discs. I've put this where the signal will not have any DC bias, so as to avoid any 'thump' noises, caused by the difference in the voltage at the drain of Q1, when activated.

The +36.7 rail comes from an LM317 reg using 240R and 6K8 to set the voltage. The relay is for output muting while the circuit powers on, about 2 seconds.

What else is there to say, the schematic speaks for itself, really. But if I've done anything silly, then please let me know!
 

Attachments

  • Le Pwnage.png
    Le Pwnage.png
    44.8 KB · Views: 1,604
Check out "open Loop Follies Pt. 1" for a somewhat simpler variation on this theme. The basic circuit you presented should work, but you can get around the relay fol-de-rol for thump suppression if you design the power supply so that it comes up slowly. A cap multiplier is one option. My shunt regulators incorporate a reference driven with a current source, with a big cap across the reference for noise reduction/slow rise characteristics.
 
Schematic?

I need to get 36V or so out of a 30V 10VA transformer so need something with little voltage drop. A soft start would be useful, I could do away with the relay and D3, though and I'm thinking of simply modifying my power supply using this circuit as a slow turn on, giving R3 and C1 a time constant of about 2 seconds.
 

Attachments

  • lm317-slow-turn-on.jpg
    lm317-slow-turn-on.jpg
    28.6 KB · Views: 1,242
Administrator
Joined 2004
Paid Member
A couple of thoughts, first zeners are noisy so they should be shunted with caps or isolated with a first order LPF from the base of the cascoded NPN. (both stages) Consider replacing the zeners with a series string of LEDs set up for the equivalent voltage.

I prefer a mute relay that shunts the output rather than one that is in series with it, this removes a set of potentially distortion generating contacts from the signal path.

Not sure about your choice of 2N7000 as current source and source follower, they are characterized for saturated switch service, and while they certainly have reasonable gate capacitance Ciss, Crss, and Coss, and moderately high transconductance they are designed and characterized only for saturated switch service, some similar devices work well in linear mode, others less so.. Not sure how these would perform, and I suspect you are running them well outside of their safe dissipation limits which for the Fairchild device is 400mW at 25C.. (40mA quiescent current at 36V, combined dissipation for both devices would be almost 1.5W, and while I can't predict the power dissipated by each individual device off the cuff I am pretty confident it is way beyond the device rating..)
 
Made some mods based on what you said.

I'll be changing the zener diodes to voltage dividers with a cap across the output, I only made then zeners for simplicity in the first place.

I've just realised that I've made a mistake, R16 was meant to be 150ohms but I left out a zero when entering the values into CAD, giving around 5mA quiescent current.

Shunting the output doesn't seem like an option here as I'd need to increase output impedance (something I'm not willing to do) to make sure the current through C8 doesn't kill M1. How would the relay contacts be able to generate distortion?

What MOSFET would you recommend for this instead of 2N7000?
 

Attachments

  • Le Pwnage 2.png
    Le Pwnage 2.png
    45.5 KB · Views: 1,166
R15 and R21 are the voltage-setting divider string. The output is set for 30V at present. The LED string is the reference, and is 1.6 X 5 = 8V (you could also use 4X yellow-green GaP LEDs, but they are higher impedance/noisier.). The rest should be easy to figure.

I used the depletion mode mosfets for lower parts count /higher impedance. I originally used a p-channel mosfet + bipolar ring of two for the higher power current source (an IRF9510 would work just fine) - probably a bit higher impedance than a source with two bipolars. The source driving the LEDs is 10mA, and could be replaced with a bipolar ring of two source.
 
Made some mods on that shunt regulator to get rid of the depletion mode FETs as well as to compensate for the fact I'm using 2N5458 as the CCS for the LTP. I set the CCS at the top to 300mA and used an IRF9530 as I have loads of them at the moment.

I'm going to use a 30V supply to keep with the circuit and also because I want to increase the quiescent current running through the MOSFET follower without increasing the power of the CCS from the source to ground. To do this, I've changed the drain voltage of the JFETs to 6V.
Hopefully this won't increase distortion, would there be any disadvantages in doing this... :scratch:

I'm thinking of using a 2N5459 instead of a BC550 for the top half of the cascode for simplicity, similar to the cascode in 'open loop follies pt.1'.
 

Attachments

  • Shunt Reg.png
    Shunt Reg.png
    43.4 KB · Views: 538
  • Lolfets.png
    Lolfets.png
    41.4 KB · Views: 505
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.