The LSK170/LSJ74 version
https://diyaudiostore.com/collections/jfets/products/matched-jfets
This version needs two 100pF compensation caps.
Result:
Bandwidth: 1 MHz compared to 3MHz for LSK489/LSJ689
Distortion: THD 0.0005% compared to 0.0010%
This version have almost half the dist but is slower.
https://diyaudiostore.com/collections/jfets/products/matched-jfets
This version needs two 100pF compensation caps.
Result:
Bandwidth: 1 MHz compared to 3MHz for LSK489/LSJ689
Distortion: THD 0.0005% compared to 0.0010%
This version have almost half the dist but is slower.
As far as I remember, for 170/74 the thermal stable point is 2mA with a supply voltage at the drain of no more than 12 volts. What are you modeling?
With a smaller bandwidth, and therefore a decrease in feedback depth, do distortions decrease? these are miracles - it cannot be that the worse the circuit, the better its parameters.
You have 2 amplification stages - this is already a complex amplifier design, and for the input signal it is the worst topology. There is nothing to stabilize the parameters of the jfet.
I agree. Not totally happy with the DC Offset approach. Still mulling over alternatives. Trying to find something that doesn't require adding a cap. I haven't looked at it's affect on input impedance. Maybe increasing from 100k to 1Meg would be wise.
I'll see if I can provide a provision on the PCB for different packages.
Also, I think having a provision for current limiting zeners would be a good idea.
Sorry, but first complete the principle of common-mode stabilization using Q11Q12 for the entire differential cascade and not just half of it. Then it will be interesting to everyone.
HI Vitamir - Not trying to ignore your suggestion. Just trying to stick to Lineup's design as much as possible. Your design in post 43 has considerable changes in it to the point where it no longer resembles the original design. Again, not trying to discount it, just trying to stay on topic.
The only change I made was to the VAS emitter connections; I made no value or connection changes to offset pot that was added earlier and I performed no simulations or analysis. Just throwing it out there for discussion.
But, I have built a number of amplifiers over the years that use this topology and found that it works quite well; there's a reason it's still around ...
I disagree. To say you need them simply isn't true. Look at Rod Elliott's P101. Simple approach with the VAS driving the lateral MOSFETs directly.
I agree that measured performance will improve. But the appeal of this design is its simplicity. It will be higher distortion, but the spectrum will likely be 2nd order dominate and monotonically descending. Look at the First Watt designs for inspiration. These don't measure well if low THD is your yardstick for performance. Yet they still appeal to many.
Based on a quick sim J113/J176 performs better in terms of distortion, but noise is exponentially worse... like 1000 times higher noise. I'd same this isn't worth pursing.
However, LSK170B/LSJ74B does seem to perform well. Similar noise and lower distortion. Loop gain phase and gain margin look (50db and 83 degrees). And DIYAudio sells matched quads of LSK170/LSJ74. Though at twice the price of a LSK489/LSJ689.
But might we worth allowing for either option. I'll see if this can be accommodated with the PCB.
What models are you using for the J113? My understanding is that the Linear Systems models that come in the ltspice distribution are known to be orders of magnitude off when it comes to noise predictions. Here is a Phillips model that I got from somewhere (don't recall where) that seems to give more realistic noise, but I don't know how accurate it is, either.
.MODEL J113 NJF(VTO=-1.2854 BETA=9.25964m BETATCE=-0.5 LAMBDA=3.03839E-2 RD=1.30170 RS=1.30170 CGS=1.05000E-11 CGD=1.20000E-11 PB=5.04493E-1 IS=9.86870E-16 XTI=3 AF=1 FC=0.5 N=1 NR=2 MFG=PHILIPS)
EDIT: I got the model from the standard.jft file at
https://ltwiki.org/index.php?title=Components_Library_and_Circuits
Last edited:
please don't rush to conclusions. He writes correctly, but this does not mean that you interpret correctly. Because most likely you mean that this is another emitter follower stage, in fact it is not.
Don't confuse simplicity and cheapness. The transistor structure is created for the simplicity (linearity) of the operation of one transistor. Therefore, most likely you will need a composite transistor in VAS with a low output resistance.