So I was the lucky one who got to do the listening here and I'd like to post some comments on it.
In this house, the reference output stage against which others are judged top out with a 3a5 output stage powered by Ale Moglia's gyrator and Rod Colemans filament supplies. I have built this 3 times now for myself and friends who all agreed to put their money into the parts to make one. Others used for testing include the Pedja Rogic diamond buffer, an AC coupled cathode follower, the older Sen also from Patrick, also his current mirror, the new complementary SEN as well, a SRPP using 6n2 as well as the opamps of course.
Patricks new design meets and is at least the equal of the 3a5 output stage in terms of just sitting back and listening, it really is that good but at a fraction of the floor space needed and I would imagine the measurements are also orders of magnitude better than a valve stage. Really tight strong bass, that liquid midrange that we all love and clear clean highs that go long beyond my hearing. It is the kind of output stage that you will listen to for a very long period and not have any listening fatigue. I think a special mention is needed for the soundstage, which was very good indeed. There is precise positioning within the sound stage (deep and wide) and it easily resolves things like two singers placed beside each other. I think its quite the triumph to be honest.
And then I broke it, badly. 🙁😡🙁
I hope to listen to it again soon, and I plan on replacing the AC coupled valve stage in one of the two Miro dac builds that are in near daily use here with it instead.
"so what did you use as I/V resistor and decoupling cap"
AKM is voltage out. No IV required, just 3rd order filter.
For PCM56, Riv is S012K.
I don't have caps in the signal path.
Or you mean something else ?
Patrick
AKM is voltage out. No IV required, just 3rd order filter.
For PCM56, Riv is S012K.
I don't have caps in the signal path.
Or you mean something else ?
Patrick
And some expectation management for the new IV connverter.
The simplified schematics of the new Folded Cascode CEN IV has been posted here :
https://www.diyaudio.com/community/threads/zen-i-v-converter.173291/post-6927415
What is not shown is the new thermal tracking current source, as well as the Sallen Key filter between Riv and output buffer.
The tricky part of the design has been described here :
https://www.diyaudio.com/community/threads/zen-i-v-converter.173291/post-6994628
It took a few iterations before getting it right, which you do not see, of course.
But at least, you can see that on the BoM of one IV, there is one matched pair of 2SK170/2SJ74, and another pair of matched 2SK170.
So that gives you a good indication of what is involved.
There are 11 active devices in total, including JFETs, MOSFETs and BJTs, all requiring matching.
Making PCBs available for the through hole version is the easiest part.
The SMD version needs SMD JFETs that even I have limited stock of just 30 pairs.
I shall not buy from unknown sources. So in real life terms, it is truely unobtanium.
All parts on the PCB are SOT23, 0603 and one 0402.
So not easy to solder either.
And you might now be just happy that Ripster made the CM IV GB available for many of you.
Patrick
The simplified schematics of the new Folded Cascode CEN IV has been posted here :
https://www.diyaudio.com/community/threads/zen-i-v-converter.173291/post-6927415
What is not shown is the new thermal tracking current source, as well as the Sallen Key filter between Riv and output buffer.
The tricky part of the design has been described here :
https://www.diyaudio.com/community/threads/zen-i-v-converter.173291/post-6994628
It took a few iterations before getting it right, which you do not see, of course.
But at least, you can see that on the BoM of one IV, there is one matched pair of 2SK170/2SJ74, and another pair of matched 2SK170.
So that gives you a good indication of what is involved.
There are 11 active devices in total, including JFETs, MOSFETs and BJTs, all requiring matching.
Making PCBs available for the through hole version is the easiest part.
The SMD version needs SMD JFETs that even I have limited stock of just 30 pairs.
I shall not buy from unknown sources. So in real life terms, it is truely unobtanium.
All parts on the PCB are SOT23, 0603 and one 0402.
So not easy to solder either.
And you might now be just happy that Ripster made the CM IV GB available for many of you.
Patrick
Last edited:
Thanks for the feedback.
Yup...I was talking about curent output dac chip and what to use in a smd narrow design. I find the differences between resistors make those curent DAC outputs from non optimal to pretty good. But no experience with smd hence my question.
For the caps I talked about the power supply near decoupling...not on the tiny board but what you liked when testing it in your device...but I thougth you tested the board on the Miro1360 pcb. So forget please the question if not.
Good job ...as usual 🙂
Thanks. But I do not want try Ali there anymore as in the last pair one had a broken lead that can not be fixed and this is why btw I need a standalone chip. Nothing hurry though...
yeah but sadly paddy cant get them anymore from Rochester or so I read on other thread, so sad who else is going to buy them...
They are not inexpensive, but your solution apparently works above woodturner-fran reference, so I think paying the price for the matched quads being offered at the diayadio store might be worth it. I believe the other active devices involved are more easily available,
You need, for 2 channel single-ended output :
1 matched quad of 2Sk170/2SJ74
1 matched quad of 2SK170 (LSK170)
1 matched quad of ZVN3306/ZVP4424
You need to group match the Zetex to get good yield.
The rest, as you said, are easy.
Are there any replacements possible for the Toshiba JFETs ?
Yes, but you need to include an input servo (JFET opamp).
Then you could use, if you can find genuine Toshiba or Onsemi :
1 matched quad of 2SK246/2SJ103BL, or 2N5458/5461
1 matched quad of 2SK117BL or 2 matched quads of 2SK209GR
Patrick
1 matched quad of 2Sk170/2SJ74
1 matched quad of 2SK170 (LSK170)
1 matched quad of ZVN3306/ZVP4424
You need to group match the Zetex to get good yield.
The rest, as you said, are easy.
Are there any replacements possible for the Toshiba JFETs ?
Yes, but you need to include an input servo (JFET opamp).
Then you could use, if you can find genuine Toshiba or Onsemi :
1 matched quad of 2SK246/2SJ103BL, or 2N5458/5461
1 matched quad of 2SK117BL or 2 matched quads of 2SK209GR
Patrick
"1 matched quad of 2SK246/2SJ103BL, or 2N5458/5461
1 matched quad of 2SK117BL or 2 matched quads of 2SK209GR"
Don't blame me if you only get fakes.
Plenty around.
DIY store is the safest, and best performance.
Patrick
1 matched quad of 2SK117BL or 2 matched quads of 2SK209GR"
Don't blame me if you only get fakes.
Plenty around.
DIY store is the safest, and best performance.
Patrick
I know you do not liks TO90 case anymore but do you think we can come close with such transistors today...what is lost with longer traces is winned with a better thermal behavior ? Are there still good enough Tosh and Zetex in this size for papys not wanting to mess with0405 smd size or soic ?
I still have a simple 6 Bjt bc550/560 Rodgic diamond I like, but not matched (a little more H2?). Is it hearable to reduce more the noise with nowadays ref ?
I think i/v resistor is still animportant part of this stage and I will report soon a funny happy test to allow people to tweak their Miro board there. I surmise you use a smd thin film resistor in your tiny setup. Advantage of a little bulkier design is a little more choice with this resistor imho. YMMV, hearing test could be fun...sad world since cv19 to meet for such cool testingABX wine parties...
I still have a simple 6 Bjt bc550/560 Rodgic diamond I like, but not matched (a little more H2?). Is it hearable to reduce more the noise with nowadays ref ?
I think i/v resistor is still animportant part of this stage and I will report soon a funny happy test to allow people to tweak their Miro board there. I surmise you use a smd thin film resistor in your tiny setup. Advantage of a little bulkier design is a little more choice with this resistor imho. YMMV, hearing test could be fun...sad world since cv19 to meet for such cool testingABX wine parties...
Last edited:
