Updating Leach FET preamp: New transistors?????

[grhughes]

HI! I posted this same question over at A asylum (Prop Head Plaza) but I would like to get other's opinion here if possible. In about 77 or 78 I built Marshall Leach's low distortion preamp that was published in Audio Magazine. It used 2n5210/2n5087/mps8099/mps8599 transistors and 2n5457 n channel FETs. I carefully matched all transistors as to beta/hfe/d-s current. It would seem to me that transistors should have improved since that time. Am I wrong? I realize that most interest has turned away from discreet devices to large scale integrated circuits but I would also assume that the process should have improved the same process whereby discreet devices are made. My question is: Are the newer Toshiba/Hitachi discreet devices better, have fewer impurities and are easier to match than the older Motorolla, National, and Fairchild transistors I used in 1978? It is time to re-capp the preamps as their electrolytics have leaked and while I was there I was thinking about making the output stage more robust by subbing Hitachi 2sc3421s for the all Npn output stage mounted on heat sinks. I was pondering driving a 1:1 transformer (600 ohm:600 ohm balanced line) Or MJE181s?? Bad idea, Good Idea???? Thanks for your opinions. Ray

[Jonathan Bright]

I don't know about being easier to match but there are a few low noise Toshiba Fets around that should be an improvement on the 2N5457's.
Cheers, Jonathan

[rayfutrell]

The 2N5457 jfets have a rather modest transconductance of 1 mS. You can't get lots of voltage gain. The low gain and the 3 pF gate drain capacity results in a very small Miller effect. Dr Leach advertised the input capacity of his jfet preamp as 20 pF. That is astonishingly low and very useful if you are using a high inductance phono cartridge which needs a small terminating capacity.

The famous 2sk147 jfet had a gm of 25 mS which was achieved by a large interdigitated gate structure. The large gate geometry resulted in a gate drain capacity of 18 pF and a potentially big Miller effect. LUX had a phonepreamp where they used these jfets in a cascode circuit to reduce the Miller effect.

[grhughes]

Thanks: I was under the impression that the first FET stage didn't have much gain and most of the gain was in the second stage 2n5087 pnp diff amp stage. I have added zener referenced current source to that tail. What about the 2sk170BL as a replacement for the 2n5457? Ray

[mlloyd1]

i built the leach preamps years ago with the 2n5457 jfets. it sounded nice and nowadays, might be considered noisy compared to some recent diy designs. if you redesign and choose the higher gain, quieter toshiba jfets, you may want to cascode them due to the significant extra capacitance, especially if you're using mm cartridges.
i'd also tweak the riaa eq network and add an RC filter at the back to get the rolloff more correct.

however, having said all that, give it a shot, the results should be promising!

mlloyd1

[grhughes]

Thanks! I have a scheme of cascoding from a book on FET written by Siliconix. I have some 2n4416A and some 2n3821 FOR THE SECOND jfet transistor in series of the cascode or should I use all 2sk170BL? Motchenbacher and Fitzen wrote a book and said the 2n3821 was the lowest noise n channel jfet they had tested. Thanks again! Ray