Simplistic NJFET RIAA

[Salas]

Would you test some k170BL with 8mA and 7.5mA ?

Would like to confirm my findings.

Later on when I can find the exact IDSS you require, of course.

 

[Joachim Gerhard]

Fine, you have the traces now. It already gave me a lot of information.
Ricardo, the bigger brown one, i think that is a German made wire wound.
I used it in the old times in speakers.
What are the smaller green ones ?

 

[Salas]

After some more hours listening I am glad to report the tone is really good. No stridency no bass blurring, sweet mids and that magical super wide soundstage.

I read on the papers of Allen Right that matching riaa components between channels is really important to preserve a very tight gain/phase relationship so as not to corrupt the stereo image.... totally agree.

In this case I was not able to follow the prescribed cap / resistors values due to stock shortage but am getting really good at simulating riaa components.... This time it was a total hit from the start... no need for further trimming.

Please bear in mind that 15.15nF are meant to be 15.15nF any deviation will blurr everything. If you can no get the exact values, trimming must be done in "all" riaa components.

I am willing to help of course.

I have stressed, building the folded simplistic with precision as given by tested circuits is key, its more strict than previous. That is why I linked it as "experts version MC only" in post #1. I am glad you came to same conclusions.

 

[Joachim Gerhard]

OK, this are Toshiba drivers.
Nice specs, they are fast and even have fairly high Hfe.
I found that many medium power transistors match well.
It seems that controlling early voltage in PNP BJTs is easier when the chip inside is bigger.

 

[RCruz]

The big brown one is a NOS Allen Bradley (In this case a 1M26 that I placed on top of the 46600 ohm to get 44938 ohm I needed)
The smaller green ones are kiwames (koa speer) quite nice 2w carbons... noiseless and with very good tonality.

The smaller brown one is a 209000 r Dale that I used to reduce the partnering 6760 ohm kiwame to it's calculated 6548 riaa resistor.

 

[Joachim Gerhard]

That Allen Bradly looks exactly like the wire wounds i used.

 

[Salas]

Fine, you have the traces now. It already gave me a lot of information.

Quickly confirmed that close HFE or IDSS genuine Toshiba semis give excellent matching in curves also usually. Expected, but nice to know there is no renegade semi when building open loop especially. In simple designs of course, because I would not have the energy to trace 88 BJTs after matching HFE out of 200pcs each sex in "some" designs.

 

[RCruz]

Merlin

Did you get my answer ?

 

[Joachim Gerhard]

Here the are :
VITROHM | Produkte | Drahtwiderstände, umpresst, vollisoliert
OHH, they are even made in Portugal.

 

[RCruz]

Quickly confirmed that close HFE or IDSS genuine Toshiba semis give excellent matching in curves also usually. Expected, but nice to know there is no renegade semi when building open loop especially. In simple designs of course, because I would not have the energy to trace 88 BJTs after matching HFE out of 200pcs each sex in "some" designs.

Hehe... it took me more than three days to match those 88bgt's ... and the first time was even worst because there where some more mirrors.

 

[Joachim Gerhard]

Yes, matching each and every device in the Paradise is too much work.
I do not think that it is necessary. Maybe in the input stage and the buffer.
Some of the cascode J-Fets had too low Vgs.
The mirrors are uncritical.
Matching for 10% variety of Hfe works just fine.

 

[RCruz]

Here the are :
VITROHM | Produkte | Drahtwiderstände, umpresst, vollisoliert
OHH, they are even made in Portugal.

Thank you so much for the link... but mine are 1.26M... those seem to stop at 2k4

 

[RCruz]

Some of the cascode J-Fets had too low Vgs.

Would you please explain ?

 

[Salas]

Hehe... it took me more than three days to match those 88bgt's ... and the first time was even worst because there where some more mirrors.

We pick on Joachim a bit about PITA building but we hope he won't mind. This thread always had humorous to hilarious trends anyway.

 

[RCruz]

Joachim, I hope you do not matter.. we tend to laugh a lot around here
Anyway I allready built your design so I know it is really something special.

 

[Salas]

Yes, matching each and every device in the Paradise is too much work.
I do not think that it is necessary. Maybe in the input stage and the buffer.
Some of the cascode J-Fets had too low Vgs.
The mirrors are uncritical.
Matching for 10% variety of Hfe works just fine.

I kept my channels at under 5% but the input ones even closer because I was afraid of gain differences between channels. They should be no more than 0.3dB. Did not match the J310s but they ended up OK luckily. It was a 100pcs Fairchild old Ebay stock in nice price from USA. But I had done sample checks they are high Vgs before from this bag.

 

[merlin el mago]

Merlin

Did you get my answer ?

Yes thanks.

 

[RCruz]

I kept my channels at under 5% but the input ones even closer because I was afraid of gain differences between channels. They should be no more than 0.3dB. Did not match the J310s but they ended up OK luckily. It was a 100pcs Fairchild old Ebay stock in nice price from USA. But I had done sample checks they are high Vgs before from this bag.

How do you determine the fet vgs out off the circuit ? Are you talking about Vp ?

 

[Salas]

I made the real cascodes with the real to be used IDSS below fets. And I measured the Vgs of the above fet i.e. the VDS of the lower, same thing.

 

[RCruz]

So it is a matter of luck ?

Must one breadborad the thing to know the vgs ? or can we estimate by using a higher Vp ?