I just replaced the KSA992 in the supplemental current source (the one that actually carries the current) for the input stage with 2SA965. This is a TO-92L vs. a straight TO-92, so there should be lower temperature rise. I don't know if that will affect the drift in VDS, as I think that will be more influenced by the sensing transistor of the ring-of-two current source.
I also added small heat sinks to the TO-220 depletion mode mosfets I'm using to provide supplemental bias for the PN4391 2nd stage gain cell. I'll see what effect that has on the drift in bias voltage at the output.
I finally got to scan the RIAA preamp sections of this preamp after a humiliating experience last weekend (locked myself out of the lab and had to slink home in disgrace...). Match between channels is not quite as precise as earlier versions of this amp, but still OK.... I think wat happened in this case was I was only able to match the PN4341-based 2nd section of this preamp at the max current capability of my sorting jig, which is ~15 mA. Actual bias current was more, so the second stage gain elements likely had a less precise matching at the higher bias current vs. the max 15 mA of the sorting jig. I think that I will go back into the jig and mess with the bias resistors on my reference so I can get at least 30 mA capability. 0.58 dB of difference between channels at 1 kHz is still not all that shabby.
Attachments
Last edited:
Well, I think I fount the reason why the 1kHz gain was different between the two channels, so it's back to the bench for a bias re-tweak and some gain-phase scans.
Here are the results from this afternoon's G-P scan. I found that I had used 4.53k instead of 4.32k as the gain resistor on the 2nd stage of channel 2. Using the proper value of gain resistor (and a bit of rebiasing) reduced gain mismatch between channels to 0.25 dB, down from 0.58 dB.
Attachments
Yes, my approach was very different, as I was looking at not only using more prosaic jfets for input and output, but I was also looking to solve the distortion issue with the classic Pacific preamp second stage. I solved the problem using much higher bias current in the 2nd stage using PN4391 jfets instead of the classic 2SK170. Both stages of my approach are cascoded, which should also help with amplified Miller capacitance, which could impact use of the classic Pacific preamp (esp the 1st stage) with cartridges having fiddly input capacitance limitations (the whole lineup of AT MM cartridges comes to mind). I have another approach using feedback (horrors!) that also gets around the fiddly capacitance issues, such that I can use a classic AT MM cartridge (AT440MLB) without it sounding thin and shrieky.
Last edited: