2SJ74 subs

[rafafredd]

Thanks Tamas!

How much Idss do you think is necessary on the output pair to drive a 600Z transformer?

The BL series goes up to 12mA according to datasheets, but I don´t know in real life...

I don´t think it would be a nice idea to use a matched P-pair (2sj109) for Q6 and Q7, because Q6 can be in BL range and Q7 should be more than 10mA, so if I´m going to select the higher Idss ones for the outputs, that would be a waste, no?

I think it would be better to use two pair for two opamp and as those are matched pairs, I would only need to to select once for a pair of opamps.

So, I´ve found 2sj109, but they don´t specify if it´s GR, BL, or V, but I hope they are high Idss... Will get some for testing.

 

[rafafredd]

So, got my 170s on the mail today.

Will start sorting out.

Do you think that 170 and 174 BLs with high Idss, like 11 or 12 mA will be enought for a 600ohms load? Or will I really need the V version for 600Z out?

 

[tk@halmi]

Rafa,

I would not worry that much. If you read Fred's paper carefully on this first generation opamp you will note that it makes no claims that it will drive a 600 ohm transformer, does it? It says for loads above 600 ohm it will do 24dB. Watch the careful wording... How much above?

JFETs are terrible power devices in the first place. Reading Doug Self and Sloan's books you will get the impression that JFETs should be banned to Siberia. Of course, their analysis tends to be very objective and focused on numbers. Something can sound good to you (subjective) even though it has high distortion.

If you can manage matching 10ma devices for the output then try it with a transformer. What can you loose? If you do not match for Vgs and you have large DC offset put a capacitor between the output and the transformer. The worst case scenario is that low frequencies will not be driven as well as you would like. Just listen and decide for yourself if you like it.

Erno Borbely connects JFETs in parallel to get more current drive. You could try that as well. In that case you will want to match both Idss and Vgs within 10% or less for best results. When Borbely needs more power he uses MOSFETs.

Currently, I am waiting for a bunch of ZVP3310A and ZVN3310A MOSFETs to try on the output. They are in TO-92 cases so they do not take much space, but they can dissipate 50% more power and handle currents hundred times over the 2SK170/2SJ74 devices. With MOSFETs there is an additional BJT required to help bias the MOSFETs into conduction.

Cheers,
Tamas

 

[Bauman]

Tamas is our official DOA tweaker!

Lots of good info going on.!

cheers,
Fabio

 

[rafafredd]

Thanks again Tamas. really nice info.

So, to use the JFETs in parallel I just need to wire two of them or there are any other components needed, like low value resistors from the +-V rails?

I think that maybe the best think would be a k389BL (both parts in parallel) and a J109BL (same thing) to get nice low end. Those I can get...

Are you gonna try those mosfets on the output of this opamp or onother tottally different design?

 

[chrissugar]

If you read Fred's paper he says that at the output you will have +24dBu into loads above 600 ohms.
Lets make some calculus, +24dBu is aproximately 12volts pp, that is 6 volts per alternance, that means 6volts for the 2SJ74 and 6volts for the 2SK170. Now if you divide 6volts with 600 ohms you have 10mA.
The conclusion is that if you use fets with Idss higher than 10mA it is OK even at 600 ohms.

chrissugar

 

[tk@halmi]

So, to use the JFETs in parallel I just need to wire two of them...

I have seen Borbely do that. I suspect that doing so requires excellent matching of both Idss, Vgs and gm parameters. My suggestion would be to try using the 10 ohm source resistors (R5 and R6) for EACH device. Basically you double up the output section. You should not need anything between the drain and the power rails. Still need to match as well as the pool of devices you have allows.

Are you gonna try those mosfets on the output of this opamp or onother tottally different design?

My intent is to change the output only and leave the rest of the opamp the same. I do not know if that is feasible as I have not progressed beyond simulations. A prototype might prove me wrong and probably I shouldn't have spoken until I tested it.

 

[tk@halmi]

[quote author="chrissugar"]The conclusion is that if you use fets with Idss higher than 10mA it is OK even at 600 ohms.
chrissugar[/quote]

Thanks Chris. You are right, the math does check out.
Even though any modern DIP8 chip opamp can do 10mA I have never seen a design that hooks them up to a 600 ohm output transformer directly. As PRR explained in the past you need far more power to drive a transformer at low frequencies than elsewhere. (And the trouble with the JFET output is that once you hit Idss there is no more juice)

 

[tk@halmi]

Hello,

I am finding that the J271 works in place of the 2SJ74. Its short coming is the higher input voltage noise, but the Vishay part is pretty good. Stay away from the Fairchild parts.
The PN4393 kind of works in place of the 2SK170 with about twice the input voltage noise figure.

Finally, I have a working prototype of the Forssell Gen1 opamp using small MOSFETs on the output (ZVN3310A and ZVP3310A). These little buggers are extremely durable flat profile parts that can be cooled well by pinching them between two metal plates like Chrissugar has done before.

Here is what I have done:
1. Left out the 10 ohm source resistors (R5 & R6) on the output
2. Changed R3 to a 1K resistor to bias the MOSFETS to ~15ma standing current. This is best replaced with a 1.5k trimmer pot to adjust as needed. The value also depends on the current supplied by Q4. It was about 6ma in my test case.
3. Added 100 ohm gate stopper resistors to the MOSFETs. These are absolutely a must otherwise the output pair will start oscillating and it will destroy itself within a minute.

It runs like a champ. Just needs a PCB.

Cheers,
Tamas

 

[rafafredd]

GREAT NEWS. JUST GREAT NEWS!

Maybe it will drive low impedance transformers easily with those mosfets at the output, no?

Where did you buy those mosfets from?

Any other recommended mosfet part numbers?

 

[tk@halmi]

Rafa,

Last I checked Farnell carried the ZVN3310A and the ZVP3310A. I buy them at Digikey or Mouser by the truckload.

They will drive your 600 ohm transformers without blinking. The MOSFETs can source over 100ma, but they cannot dissipate much power due to the small size so you should not set the standing current to more than 20ma.

Tamas