This evening i take reverse way and swap another time from VSSA to BG1-BJT and i confirm all waht i say before except about details : Vssa gives a little more in this chapter, but it really lack comparativ BG1 in lower frequency (weight, details, articulation). So the best were a mix of two.....
So next step may be to try a hexfet version. I hesitate to go for IRFP240/9240 or to go for Toshiba 2SK1530/2SJ201 with background to be limitate to one pair due the fct they are to pricy for matching session...
Under BG2 with 2 pair 240/9240 and 10uf in // 2200uf at input stage current injection point.
So next step may be to try a hexfet version. I hesitate to go for IRFP240/9240 or to go for Toshiba 2SK1530/2SJ201 with background to be limitate to one pair due the fct they are to pricy for matching session...
Under BG2 with 2 pair 240/9240 and 10uf in // 2200uf at input stage current injection point.
Attachments
If you replace in your BG1-BJT the FC caps with Nichicon Gold you will get better highs and mids (more detailed sound and cleaner), probably even better bass to.
FC is warm but somehow luck details, some places a bit muddy also shy in the bass area.
Your resisters later can be upgraded to DALE or PRP type.
Look like you are not going to use the 50V power supply, I see on your latest layout only 50V caps.
I think I'll wait until you test the HEXfet type to and based on your feedback I'll build the BJT or the HEXfet type.
According to borys the HEXfet a bit better but he didn't had the Sankens.
Let see.
If U go with 36V PS you can swap the BC550c/560c with 2SK170BL/2SJ74BL JFet without any mode.
I did it in my darlington version.
I prefer JFet sound over the BJT on the input.
Greetings Gabor
FC is warm but somehow luck details, some places a bit muddy also shy in the bass area.
Your resisters later can be upgraded to DALE or PRP type.
Look like you are not going to use the 50V power supply, I see on your latest layout only 50V caps.
I think I'll wait until you test the HEXfet type to and based on your feedback I'll build the BJT or the HEXfet type.
According to borys the HEXfet a bit better but he didn't had the Sankens.
Let see.
If U go with 36V PS you can swap the BC550c/560c with 2SK170BL/2SJ74BL JFet without any mode.
I did it in my darlington version.
I prefer JFet sound over the BJT on the input.
Greetings Gabor
Thanks for advices gabor,
I thinks i will first try hexfet at +/-36V as i have test plateform, that's why caps are 50v. I have those nichicon at house except for 100uf, but i thinks i have elna simic II for this position. So i have to decide to use whether toshiba 1530/sj201 nor irfp...I must confirm to with bjt change on BG1 that Sanken ahs is own signatur
Marc
I thinks i will first try hexfet at +/-36V as i have test plateform, that's why caps are 50v. I have those nichicon at house except for 100uf, but i thinks i have elna simic II for this position. So i have to decide to use whether toshiba 1530/sj201 nor irfp...I must confirm to with bjt change on BG1 that Sanken ahs is own signatur
Marc
Last edited:
Idefixes
The schematic for FET-hex amp is for irfp --->ugs around 3,7V each transistor ---> biasing circuit for Toshiba mosfets has to be changed !! --> I mean resistor network cause Ugs of 2sj201/2sk1530 is lower (group O and Y have different Ugs !!).
In irfp -------> around 6,7V between gates to get bias 100mA
In 2sj/sk group Y-----> prapobly around 3,5-4V to get bias 100mA
Another point is thermal behavior.
IRFP version --> hexfet in o/p and hexfet in bias (same tempco)
2sj/2sk version --> biasing should be done by the same type of tranistor
But I do not know any siutable mosfet to do proper bias compensation in 2sj/2sk version.
Maybe BJT transistor in toshiba mosfet would be better ??
The schematic for FET-hex amp is for irfp --->ugs around 3,7V each transistor ---> biasing circuit for Toshiba mosfets has to be changed !! --> I mean resistor network cause Ugs of 2sj201/2sk1530 is lower (group O and Y have different Ugs !!).
In irfp -------> around 6,7V between gates to get bias 100mA
In 2sj/sk group Y-----> prapobly around 3,5-4V to get bias 100mA
Another point is thermal behavior.
IRFP version --> hexfet in o/p and hexfet in bias (same tempco)
2sj/2sk version --> biasing should be done by the same type of tranistor
But I do not know any siutable mosfet to do proper bias compensation in 2sj/2sk version.
Maybe BJT transistor in toshiba mosfet would be better ??
Attachments
Yes i had that in mind. In all design i saw with BJT (sometime 2SC5171) involve in biassing
Marc
Hi Marc
I mentioned eaylier that your choosen Sanken model have multi emitter for higher bandwith performance witch i think lead to a aery sound compared to same model without and your changed types are probably standard single emitter. Think you can compare sound a bit like: multicore-wire=multi-emitter and solidcore-wire=single-emitter, i prefer the last which is same as traces on pcb, and then Again paralell=muticore-wire, serial=solidcore-wire.
Attach picture from old Sanken catalog.
BR Ricky
Attachments
Last edited:
Hi BYRTT,
I notice that about 2sc2922/2sa1216 i use, If theere is an effect of multi emeitter, the change to Onsemi device i have do provide a less aery sound....That won't be a good news as BG1-BJT had to win in that domain...
Marc
So i do not misunderstand you, please confirm you like the aery sound, then i can publish here ways/tricks to get aery sound.
BR Ricky
Aery sound
I hurry try to think of my experience for more aery sound, because of i hurry i may not remember all but here are few:
1: lower the bias in poweramp - very effective.
2: multicore wire in signalwire/ground - exampel solidwire against multicorewire or multicore wire where you want more aer
then get cable with even more multicore.
3: multicaps in signal trace - exampel Marc's C1 cap is 4,7uF, change it to 5x1uF, or 4,7uF and 3x470nF, just make trails.
4: multi output devices - again multiple signals that don't arive at same time.
5: multicore wire for speakers - same as no 2.
6: If using PC as player try get in to things about setting timing on memory (discount motherboards have few settings,
but midle to overclockers boards have superior settings for memory timing), here you will experience it has same effect
as bias-setting in poweramps.
In general go paralell and you get aer in sound, because we get delays between the paralells because of small
differences in hardware routes, the delays are very small nano/pico or less (A studio producer call and use this as an Effect).
I wont comment now for why i am for serial, because i am on duty at work, and it is a long explanation, just tell if you want more lines.
BR Ricky
Hi Again, to get decent lower end too, you have to eperiment.
I hurry try to think of my experience for more aery sound, because of i hurry i may not remember all but here are few:
1: lower the bias in poweramp - very effective.
2: multicore wire in signalwire/ground - exampel solidwire against multicorewire or multicore wire where you want more aer
then get cable with even more multicore.
3: multicaps in signal trace - exampel Marc's C1 cap is 4,7uF, change it to 5x1uF, or 4,7uF and 3x470nF, just make trails.
4: multi output devices - again multiple signals that don't arive at same time.
5: multicore wire for speakers - same as no 2.
6: If using PC as player try get in to things about setting timing on memory (discount motherboards have few settings,
but midle to overclockers boards have superior settings for memory timing), here you will experience it has same effect
as bias-setting in poweramps.
In general go paralell and you get aer in sound, because we get delays between the paralells because of small
differences in hardware routes, the delays are very small nano/pico or less (A studio producer call and use this as an Effect).
I wont comment now for why i am for serial, because i am on duty at work, and it is a long explanation, just tell if you want more lines.
BR Ricky
SO i have also multiwire cable and not thin, What i will try is an another coupling caps as MKS2 4u7. This as never convince me. I evcer prefer Clarity cap SA or mundorf Mcap suprem. I have Russia Pio cap to in 2u2...some more experiment to do...
Marc
I just order a couple IRFP240/9240 (to keep intact my matched sextet) and 2SB649/2SD669. By the same vendor i bought 100x BC546C and will try BC546C from mixelec.
Marc
Hi Marc
I guess you still have Onsemi output devices on BG1, then try no 1: exampel lover the the bias setting to 5mV across 0,22R emitter resistor (think you normal run it at 22mV), it should give air, and then try different settings till it suits your ear.
It is important you not think tecnical correct readings at your DMM, but use your ears (At a point you probably hit class B instead of class AB, but does'nt matter if you like air).
And finally its maybe more easy if using headphones, because of better concentration when in the silent room there, and closer to producers material.
BR Ricky
Last edited:
No no for moment i have 2sc1922/2sc1216 (see picture above) but i have in stock the three Onsemi device cathegories
Marc
This should give more thunder at Lows, and think Marc has order of 2200uF+bypass10uF instead of 1000uF sitting alone, and this gives same as removing inputcap, so result 2 x thunder (In reality the groupdelay Down there get more linear, lows arive at same time as middle to top).
But to hear it speakers most have life Down there or else we can't hear it, then try with headphones.
BR Ricky
Right BYRTT, i have 10uf under way to complete 2200uf i have in stock. The issu is that i think they don't fit in actual PCB...Will see what i can do for without damage them...For sure in Hextfet version needed place is include. I just order 50x BC556C to mixelec to complet the 100 ones BC546C coming too.Yet i am ready for a +/-50Vdc rail version. All parts i will mount on proto hexfet will in range to fit +/-50Vdc so an upgrade is quite simple.
BRYTT i just remember a AKSA advice when i told me how match my MJL => "HFE mesure is done at 55mA it's the Bias i apply in my amp"...yet when i remerber it's the bias i apply to my Nbp3 => 50mA per devices.
Marc
BRYTT i just remember a AKSA advice when i told me how match my MJL => "HFE mesure is done at 55mA it's the Bias i apply in my amp"...yet when i remerber it's the bias i apply to my Nbp3 => 50mA per devices.
Marc
Last edited:
I think that someone try input cap with LC VSSA and didnt like the sound. That capacitor has influence on signal path, has its own sonic character, so maybe is good to try without them (if u have no DC from preamp). Thunder is thunder, with or without that capa 
BTW, I got all the parts for the BG1-BJT version, but without Sankens. I will try ON and Toshiba. Over the next week, I should put it all together and listen. I will write how it sounds to me, directly against Aleph 3
BR
BTW, I got all the parts for the BG1-BJT version, but without Sankens. I will try ON and Toshiba. Over the next week, I should put it all together and listen. I will write how it sounds to me, directly against Aleph 3
BR