Are Jfets drain and source interchangable?
actually, i beleive the drain and source are interchangable without difference primarily for lower gain devices. i'm not sure the 2SK170 is such a device, but it won't be damaged and will likely still work ok.
mlloyd1
quote:
... in theory Drain & Source are interchangable for jfets... (One of the unique features of jfets) ... [/B]
Theory may say that this is the case...it all depends how the device was architectured or developed.
If you refer the following spice model you can see that CGD (capacitance from gate to drain) is about haf the value than CGS (cap from gate to source)....25pF vs 56pF. This may not mean very much on this application about reversing the drain and source. However it will put more cap into the signal path which will increase the Miller cap of the diff pair.
So be aware of simple observations. My 2cents worth.
.MODEL 2sk170 NJF
+ VTO=-5.211e-001 BETA=3.683e-002 LAMBDA=4.829e-003
+ IS=1.000e-009, RD=0.000e+000 RS=0.000e+000
+ CGS=5.647e-011 CGD=2.562e-011
+ PB=4.860e+000 FC=0.5
actually, i beleive the drain and source are interchangable without difference primarily for lower gain devices. i'm not sure the 2SK170 is such a device, but it won't be damaged and will likely still work ok.
mlloyd1
quote:
... in theory Drain & Source are interchangable for jfets... (One of the unique features of jfets) ... [/B]
Theory may say that this is the case...it all depends how the device was architectured or developed.
If you refer the following spice model you can see that CGD (capacitance from gate to drain) is about haf the value than CGS (cap from gate to source)....25pF vs 56pF. This may not mean very much on this application about reversing the drain and source. However it will put more cap into the signal path which will increase the Miller cap of the diff pair.
So be aware of simple observations. My 2cents worth.
.MODEL 2sk170 NJF
+ VTO=-5.211e-001 BETA=3.683e-002 LAMBDA=4.829e-003
+ IS=1.000e-009, RD=0.000e+000 RS=0.000e+000
+ CGS=5.647e-011 CGD=2.562e-011
+ PB=4.860e+000 FC=0.5
HI
Today I finished one channel of the Symasym amp . I fired up at first with out no problem .No offset . I set up the amplifier 350mA , I have large heat sink . I made some listening test , the sound is very promising . Very detailed ,good bass , clean sound .
I would like say a big thank to Michael for such a good amplifier .
Thank you very much .
Michael I would like to give a try for jets . I have at home some 2SK170 jets .
Which type you use BL or GR ?? I have both .
Do I have to replace some resistors with different value .
One more question , please let me know the 2SK170 it goes in the PC board the same position how the MPSA18.
Thanks one more time , tomorrow I will put together the another channel and more listening tests .
Regards
Today I finished one channel of the Symasym amp . I fired up at first with out no problem .No offset . I set up the amplifier 350mA , I have large heat sink . I made some listening test , the sound is very promising . Very detailed ,good bass , clean sound .
I would like say a big thank to Michael for such a good amplifier .
Thank you very much .
Michael I would like to give a try for jets . I have at home some 2SK170 jets .
Which type you use BL or GR ?? I have both .
Do I have to replace some resistors with different value .
One more question , please let me know the 2SK170 it goes in the PC board the same position how the MPSA18.
Thanks one more time , tomorrow I will put together the another channel and more listening tests .
Regards
Hi Mike,
DARN! I wanted to try that first. You beat me to it buddy. 😎
The sound you heard is what I was trying to describe to you. What Idss figure were your pairs? Now you have to test with your soundcard again.
-Chris
DARN! I wanted to try that first. You beat me to it buddy. 😎
The sound you heard is what I was trying to describe to you. What Idss figure were your pairs? Now you have to test with your soundcard again.
-Chris
MikeB said:
Hi Mike,
could you tell what Idss were they?
Once again I beg you to change the input pair to a constant power triplet (someone's earlier post) a'la pfeufeu! I expect further improvment.
Terry Demol said:
This is a good point! Once you do that (the degeneration), you get to the same gain but less linearity with the jfet version. Need to compare apples with apples!
Jan Didden
Of course there is no need to exchange source and drain, the sk170 datasheet also does not mention that they are interchangeable.
Well, when replacing mpsa18 with sk170, the sk170 is rotated 180° compared to mpsa18, just like BCxxx.
I tried degeneration to the bjts, it did not improve the sonics. Its quite sure that it is not the reduced nfb that increased sonics here.
Padamiecki, i did not measure the idss itself, i matched them by measuring Vgs at 1ma. Values were -268mv to -270mv.
Yes, cascoding/cfp will be interesting, but needs new prototype, i can't do these changes on the pcb.
Gaborbela, i have BL-type sk170. Nothing else to change. (If supplyvoltage is not >36v)
Chris, i suspect that the THD rised, having lowered nfb. I got too curious to wait for your results... 😀
Yes, time to hook up the amp to the soundcard again !
I have no idea which jfet feature improves/changes the sound, but this is an observation i already made with earlier amps. Its good possible that the jfets simply have the better euphonic distortion.
They have a transfer function that is more similar to tubes than to bjts, maybe simply giving less high order harmonics.
What sims show is that the jfets improve low freq behaviour (especially psrr, +40db at ~1hz) because of their ~infinite current gain.
Mike
Well, when replacing mpsa18 with sk170, the sk170 is rotated 180° compared to mpsa18, just like BCxxx.
I tried degeneration to the bjts, it did not improve the sonics. Its quite sure that it is not the reduced nfb that increased sonics here.
Padamiecki, i did not measure the idss itself, i matched them by measuring Vgs at 1ma. Values were -268mv to -270mv.
Yes, cascoding/cfp will be interesting, but needs new prototype, i can't do these changes on the pcb.
Gaborbela, i have BL-type sk170. Nothing else to change. (If supplyvoltage is not >36v)
Chris, i suspect that the THD rised, having lowered nfb. I got too curious to wait for your results... 😀
Yes, time to hook up the amp to the soundcard again !
I have no idea which jfet feature improves/changes the sound, but this is an observation i already made with earlier amps. Its good possible that the jfets simply have the better euphonic distortion.
They have a transfer function that is more similar to tubes than to bjts, maybe simply giving less high order harmonics.
What sims show is that the jfets improve low freq behaviour (especially psrr, +40db at ~1hz) because of their ~infinite current gain.
Mike
Hi mike designer of symasym
just want to ask
why you have used this type of vas topology
where is the improvement with it
thanks
john
just want to ask
why you have used this type of vas topology
where is the improvement with it
thanks
john
MikeB said:
I wish it was true, probably it is a key for better sound: less high order harmonics!
MikeB said:
WRT to jfets vs tubes vs bjt's, it depends on how much current
you run through them and how much the bjt is degenerated.
Specifically when talking about current it is AC signal current versus
quiescent.
Jfets are much better behaved when run nearer their class A limit,
IOW when the AC signal current is a large proportion of the
quiescent they still produce fairly low order harmonics.
If bjt's are run with a high ratio of quiescent to signal current and
equiv degeneration they exibit extreme linearity and benign
harmonic structure.
Having said all the above we have found front end devices to
make a big difference to sound more than one would think
due to just the transfer curve.
I think there is more to it. Capacitances, drive impedance versus
base current, thermal modulation of the junction are some
things that come to mind.
cheers
Terry
One important thing about JFETs: they are normally used at tail currents signifficantly higher than BJTs, because increasing the tail current increases gm with practically no other penalty save power dissipation. There are minor changes to the noise figure but this should be insignifficant to the application we are discussing.
One more thing you might want to try is a pair of small TO92 MOSFETs. Their advantages are similar to those if JFETs, but since they are enhancement devices, you are not restricted by Idss, and gm is higher to start with. You will find that Cgd and Cgs are on the order, or even smaller than with high gm JFETs. Vds is also not a problem - look at BS107, BS170, 2N7000 - 70V or higher is common. The down side: Cgd is quite nonlinear (stil, so it is with JFETs) so keep Vds high (normally not a problem), and there is more noise.
One more thing you might want to try is a pair of small TO92 MOSFETs. Their advantages are similar to those if JFETs, but since they are enhancement devices, you are not restricted by Idss, and gm is higher to start with. You will find that Cgd and Cgs are on the order, or even smaller than with high gm JFETs. Vds is also not a problem - look at BS107, BS170, 2N7000 - 70V or higher is common. The down side: Cgd is quite nonlinear (stil, so it is with JFETs) so keep Vds high (normally not a problem), and there is more noise.
PMA said:
Sounds like a worthwhile venture, mos generally slightly higher gm
than jfets
What do u recommend?
T
PMA said:
Hehehe Pavel, am I right in thinking you have some experience with this too? 🙂
So far, this has worked very well for me, but i do tend to run them at tail currents which would be completely impractical for BJTs.
Terry Demol said:
I have tried the aforementioned 2N7000, BS107, BS170.
Also, VN1720, IRFD110, 120, 210, 220, and of course a bit of an overkill with IRF610. All in different applications so it's not easy to compare.
I would look at 2N7000 and BS107 in this application. I also have loads of these which I bought to be able to pair them up. I have found that, compared to bigger MOSFETs, the Vgs tends to be in a far narrower spread. It's not too difficult to find a few pairs with only mV differences between Vgs out of a bunch of 20, which is actually rather good considering the Vgs treshold is high 🙂
I've used these at tail currents up to 30mA (when necessary with small heatsinks originally used for TO style BC109 BJTs back in the dawn of recorded history 🙂 ). Subjectively, the results confirm what every Pass Zen user will tell you - more current (just on a smaller scale), better sound. I tend to explain this with increased and flatter gm vs Id 🙂
Increased tail current of course requires changing of some resistors on Symasym...
One more thing: you mkight want to include gate protection on high impedance inputs.
Hi Mike,
I was going to check the harmonic structure of the distortion products. I can live with higher overall THD as long as the orders are lower. This is not to say the BJT units are bad.
Since we are now looking at an input voltage waveform rather than converting it to a resistance, the behaviour of the input stage will be completely different. This is from a signal point of view. So higher Idss figures may be bettter than low Idss units to accomodate music peaks. Input protection using diodes may be a sticky issue. It's possible they may change the sound in some way. You never know until you try. I might prefer Jfets rather than mosfets for this reason.
Is it really important to go for high transconductance here? What advantages do you see aside from higher open loop gain?
-Chris
I was going to check the harmonic structure of the distortion products. I can live with higher overall THD as long as the orders are lower. This is not to say the BJT units are bad.
Since we are now looking at an input voltage waveform rather than converting it to a resistance, the behaviour of the input stage will be completely different. This is from a signal point of view. So higher Idss figures may be bettter than low Idss units to accomodate music peaks. Input protection using diodes may be a sticky issue. It's possible they may change the sound in some way. You never know until you try. I might prefer Jfets rather than mosfets for this reason.
Is it really important to go for high transconductance here? What advantages do you see aside from higher open loop gain?
-Chris
johndiy said:
Several reasons:
- Super balanced operation, no other topology i know of give that good balance to the input diffamp (includig current mirrored)
- Ultra low DC-offset, without the thermal problem in the 2nd stage current mirror, DC-offset would stay far below 1mv. In fact, Pavel tried openloop operation, that gave ~8v offset.
- Very high gain with high bandwidth, giving less work to 1st stage
- No complementary matching necessary (except outputstage), but still having a push-pull vas output. Means combining advantages of asymetrical and symetrical topologies. (-> SymAsym)
- good sounding
Mike
anatech said:
I would need to avoid higher open loop gain, symasym is quite close to the edge here.
Mike
Hi Mike,
Your amplifier will have lower open loop gain with the J Fet anyway. Now we are looking at that lower magnitude as a reference. You did degenerate your BJTs some also. Mine is running "wide open" so to speak. I was going to add degeneration once my measurement stuff is running.
-Chris
Your amplifier will have lower open loop gain with the J Fet anyway. Now we are looking at that lower magnitude as a reference. You did degenerate your BJTs some also. Mine is running "wide open" so to speak. I was going to add degeneration once my measurement stuff is running.
-Chris
ilimzn said:
Ilimzn,
why is the BS107 preferred, not BS170 ?
You mean those folded springmetal high Napoleon hats ?
anatech said:
I am very curious as to what a FFT plot will show!
You mean, converting it to a current? Actually, it all boils down to a 'gm' but the difference is of course, the input impedance.
Of course, you need to avoid coming anywhere near Idss in normal operation, but this is usually not a problem as distortion rises massively if you don't. The thing is, low Idss units usually have lower gm - I am saying this relevant to construction, so across different JFET types, not relevant to selection within the same type (as in GR, BL etc subtypes).
I agree that adding any sort of device in parallel to the input signal source of finite impedance is to be avoided if it's absolutely not necessary - and 'absoultely not' tends to be difficult to prove. JFETs have a 'diode' inherent in the structure, and in fact it can sometiems be the reason for latch-up. In the grand majority of cases, the parallel resistance at the input (47-50k usually, or lower) and input RF filter cap will be more than enough to save the MOSFET gate. Certainly I never had a problem with prototypes that never had protection, and I connected all sorts of god awful things to them on fly. If you want to make the thing truly reliable, it's worth thinking about protection diodes - like 15V zeners or something like that, but as you say, the proof will be in the measurements and listening tests. IMHO the added capacitance of the diode will be negligible compared to the MOS gate and RF supression cap.
Ah, although this has a ring of 'what was fiorst, chicken or egg' to it, one reason for higher gm is lower delta Vgs, which means that any parallel protection element is less likely to make a difference 🙂
Still, I am nitpicking here, delta Vgs should be in the mV range if even that.
Higher gm is only useful as long as stability is not impaired as it alowes you more of the 'good' feedback to be applied. Whatever harmonic spread you get will be further pushed down into noise, which is always a good thing.
MikeB said:
Give that the gm of a small MOSFET, even one at rather high Id, is a lot smaller than that of a BJT (especially undegenerated), I don't think you'll run into problems with instability. Still, because MOSFETs are not limited by Idss, and can win the gm vs Cgs/Cds race with most JFETs, especially by increasing Id, they are at the very least worth a try. Some actually have sufficient gm for some degeneration to be used, although at highish Id. It's interesting to see a small VMOS have a gm on the same order of magnitude as a 7A lateral MOSFET 🙂