DEF Amp

Founder of XSA-Labs
Joined 2012
Paid Member
If you cascode it you can adjust the Vgs for a given current by setting the
Vds of the JFET via the cascode. In the F3, this is only a couple of volts,
but if you raise it to, say, 4 or 5 volts you can have a higher Vgs, something
to match your P channel part perhaps.

I use IRFP9240 so Vgs is around 3.8v to 4.0v. Can you give an example of a cascode that will allow the LU1014D to work with the 9240? What about the resistor divider network between the output and the P gate? Could that be set to allow the LU1014D to balance with the IRFp9240?
 
Check R085 for gate leakage.

I have two R085s. One does not have a gate leakage current. The second R085 has a gate leakage current of ~15 mA in a properly biased circuit. I either messed it up in my goofy experiments or it came this way. I measured the resistance between its gate and drain and found it to be is ~ 500 Ohms versus infinite for the other one.

This leaky R085 gave me heartburn.
 
The attached schematic shows a DEF which uses the leaky R085 I have. A constant current source [CCS] of ~15 mA bypasses its gate current to the negative rail. The CCS also maintains a high input impedance for the follower. If I wish to use a resistor instead of a CCS, its value is 1.28K; calculated from: [23-3.8] divided by 15 mA. The value of the joint Vgs = -3.8V is arbitrary.
 

Attachments

  • Draft8.png
    Draft8.png
    25.4 KB · Views: 2,242
If you cascode it you can adjust the Vgs for a given current by setting the
Vds of the JFET via the cascode. In the F3, this is only a couple of volts,
but if you raise it to, say, 4 or 5 volts you can have a higher Vgs, something
to match your P channel part perhaps.

I didn't think about that. Of course, we would have to watch the dissipation on the little LU1014's. Has anyone tested the part at higher Vds and Vgs? The datasheet is useless beyond -1Vgs.
 
Founder of XSA-Labs
Joined 2012
Paid Member
First try at DEF

Veroboard P2P of DEF using LU1014D and IRFP9240. I am using a 500R pot between gate of P chan and output and 1k pot to set negative bias from -20v rail with 4.7k from bias source to input gates. 220R gate stoppers on both gates. 0.33R degen resistors on both sides. It doesn't blow up but I am running out of range to set bias to high current and DC offset to zero. I got it to 2v offset and only 72mA before hitting the stops.

Too bad I have to leave lab now for more travel so cannot debug until I get back in a few days. I think it will work though. Maybe I need that cascode but am not smart enough to design it.

Here is what I have so far:

583844d1480881949-def-amp-def-amp-lu1014d-irf9240-proto-01.jpg


Here is schematic as built:

583845d1480882304-def-amp-def-amp-lu1014d-irf9240-proto-sch-02.png
 

Attachments

  • DEF-amp-LU1014D-IRF9240-proto-01.jpg
    DEF-amp-LU1014D-IRF9240-proto-01.jpg
    122.7 KB · Views: 3,030
  • DEF-amp-LU1014D-IRF9240-proto-sch-02.png
    DEF-amp-LU1014D-IRF9240-proto-sch-02.png
    10.5 KB · Views: 2,991
Last edited:
Veroboard P2P of DEF using LU1014D and IRFP9240... Maybe I need that cascode...

Perhaps try something like in the attached schematic, which uses the ZV9/F3 cascode. R9 should be a trimpot. I have the bias at 1A and the LU1014 has about 10.4V across it. Of course, note that this is a model only, so actual required values may vary dramatically. I won't have a chance to test the real thing for some time.

According to the LU1014 model, having -4.5Vgs or so (which one would need for the IRFP9240) would require Vds in excess of 20V, so you'd be dissipating 20W or more through the tiny LU1014. Possibly not safe long-term. So, perhaps the low-Vgs 2SJ201 is a better match than the IRFP9240.
 

Attachments

  • def-lu1014.jpg
    def-lu1014.jpg
    188.9 KB · Views: 1,904
Last edited:
The one and only
Joined 2001
Paid Member
A 2Sj201 with a 2.x Vgs would probably be a decent fit against a cascoded
LU1014, and might a 2SJ618.

But then I also count about 7 Ixsys depletion mode parts in TO-247's with a
depletion range of about -.7 V to -2.7V.

And we haven't begun to explore the SIT possibilities, including the Sony VFETs.
 
A 2Sj201 with a 2.x Vgs would probably be a decent fit against a cascoded
LU1014, and might a 2SJ618.

But then I also count about 7 Ixsys depletion mode parts in TO-247's with a
depletion range of about -.7 V to -2.7V.

And we haven't begun to explore the SIT possibilities, including the Sony VFETs.

Hello soundhappy. You made an interesting point. A follower using an N device which has a pentode-like characterisic [R085], and P- device which has a triode like characteristic; be it a Schaded FET or a SIT; say 2SJ28. What will the resultant subjective sound be due to mixing a pentode and a triode characteristics?

Sit's or Vfet's could work together with & TO-247 IXYS puck's in push-pull stage that realy possible ?
Very interesting thread i stay tuned
Thanks Gentleman's have a good time :umbrella::rain:
 
Founder of XSA-Labs
Joined 2012
Paid Member
Perhaps try something like in the attached schematic, which uses the ZV9/F3 cascode. R9 should be a trimpot. I have the bias at 1A and the LU1014 has about 10.4V across it. Of course, note that this is a model only, so actual required values may vary dramatically. I won't have a chance to test the real thing for some time.

According to the LU1014 model, having -4.5Vgs or so (which one would need for the IRFP9240) would require Vds in excess of 20V, so you'd be dissipating 20W or more through the tiny LU1014. Possibly not safe long-term. So, perhaps the low-Vgs 2SJ201 is a better match than the IRFP9240.

Thanks for the circuit and sims. My Vgs on 9240 was measured at 3.33v so not as bad.

I was wondering why don't we just make it a symmetric circuit and use IRFP240 and LU1014 on top and bottom and in that way, the bias is exactly matched automaticallly. You would end up with something similar to what I am doing with a headphone amp.

513590d1447102908-mosfet-follower-headphone-amplifier-euvls_hpamp.jpg


Just use IRFP240 and adjust Zener to get appropriate bias current.
 
Founder of XSA-Labs
Joined 2012
Paid Member
Yes, I only ask because since I will be adding an IRF240 cascode and the concern that Needtubes said about too much dissipation when matched to single IRFP9240 on bottom, it occurred to me that if the bottom half was same as top half, we wouldn't have this Vgs disparity. This then reminded me of the other thread. So is there a benefit to using a n channel JFET on top and P channel MOSFET on bottom vs n channel JFET and N channel MOSFET on top and on bottom. Would the lack of symmetry of the DEF have a different sonic character? More H2 vs H3 whereas other way around for symmetric top and bottom?