I have 4 2SJ28/2SK82 pairs and would like to participate in the "mix and match" exercise suggested earlier in this thread (if it ever starts 😉).
Best Regards
Sven
Best Regards
Sven
Hi Sven!
With Jama we see the problem long time ago and start to mix our goodies, trying to individualize with trace curves, lookly He has the tools for make a good and valuable evaluation. The problem could be: which reproducible method could we make with the objet of see in a common pool of data?
I am to interested.
With Jama we see the problem long time ago and start to mix our goodies, trying to individualize with trace curves, lookly He has the tools for make a good and valuable evaluation. The problem could be: which reproducible method could we make with the objet of see in a common pool of data?
I am to interested.
Question to those who have heard both the source follower output stages with 2SJ28/2SK82 and output stages with IRFP240/9240 heavily biased: Is there a big difference in the sound or is it just a fine nuance? What is your experience?
Nelson,
I noticed on startup a 2.5+ amp surge through the Vfets until the bias supply comes up to place. If I want to slow down the RC time constant on the regulator do you recommend increasing the size of C1 and C2 and/or increasing R11 and R12, or should I not worry about it?
Thanks,
I noticed on startup a 2.5+ amp surge through the Vfets until the bias supply comes up to place. If I want to slow down the RC time constant on the regulator do you recommend increasing the size of C1 and C2 and/or increasing R11 and R12, or should I not worry about it?
Thanks,
Hi,
It is mandatory to have the bias source stabilized first, and only then apply the main supply. Because the vfets are depletion devices, they conduct while unbiased.
Either vfets or the power supply can be damaged.
It is mandatory to have the bias source stabilized first, and only then apply the main supply. Because the vfets are depletion devices, they conduct while unbiased.
Either vfets or the power supply can be damaged.
@Metallicus, first bias and then VCC...
Without Bias they are wires.
Remember when Walter burnt many Vfets with CSX1?
Without Bias they are wires.
Remember when Walter burnt many Vfets with CSX1?
It was me who burned too many VFETs in the CSX1 😡.
In my defense that was not a bias problem per se, but an electrical short between the primary and secondary of the Cinemag transformers that I was using. On power off this was destroying the gates of the VFETs completely turning them into very expensive 1 ohm resistors. I suspect a flyback voltage spike when the power was cut.
Please do be careful with them, they are a dwindling resource!
In my defense that was not a bias problem per se, but an electrical short between the primary and secondary of the Cinemag transformers that I was using. On power off this was destroying the gates of the VFETs completely turning them into very expensive 1 ohm resistors. I suspect a flyback voltage spike when the power was cut.
Please do be careful with them, they are a dwindling resource!
Easy enough to increase C1 and C2, or decrease C3, C4. You can also
reduce R37 and R38. Any these by a factor up to 2 probably.
2.5 Amps is not likely to break anything, but no reason not to take it down
at bit.
😎
Nice! You are the first!
Did you build the CSX1 also?
Let us know when you finished the second channel, then hopefully we can hear some listening impressions 😉
Looking at your picture, did you use ZTX450 an 550 transistors for Q13 and Q14?
I guess the K2013 and J313 are the ones with the heatsinks? Or did you use other transistors.
I have been listening to the first channel mixed with the CSX1 on the second while building the second mono block. Sounds good so far. first impression, more detail on top.
I'm using BC550 and 560 for the cascode.
you are correct the K2013 and J313 have the heatsinks.
I increased C1 and C2 to 440uF, this helped with the turn on surge.
Not easy. Everybody could write (or engrave 😀) a unique number on each vfet, then send it do Jama. Jama could do the measurements and publish the results in a Google spreadsheet. All "owners" would then see the results and start a endless and completely chaotic discussion with owners of other pairs to get a better match. 😀
Regards
Sven