...it might not be the most simple, but the diamond buffer with signal depending bias is an elegant beauty.
The diodes are essentially free and the 12 bipolars are 5-10 cents each. The JFET's can vary but there are some < 25 cent options. This exercise was an attempt to also "make the numbers".
The noise is just under 2nV, if you really wanted 1 nV I would entertain the inductor trick and some paralleled FET's on the input. If you wanted 30 Ohm drive I would just use a good closed-loop buffer or high current op-amp as an output stage. Those MJE devices have big Cje's to drive and two of them are bigger than the op-amp would be anyway. Notice the JE990 drops to -75dB THD at 75 Ohms.
Note on those step responses, there was no input BW limiting or small cap in the feedback circuit to smooth the response.
This is sophisicated and refined. Much needed and appreciated by all. Now its in the ether of the Internet for all time. The SW-OPA. You're a class act, Scott. Thx-Dick
Thanks, as said before all and in part in the creative commons. Pete B. had enough info that I think I could easily cobble together a FET JE990 with cheap available parts, but it will have to wait a few days.
I couldnt agree more, i've been waiting in the wings letting those who actually know what they are talking about speak, waiting for the inevitable mastery with this dreamteam and we havent been let down.
i'm in need of a slightly lower gain circuit though for my custom in ear headphones (under 10ohms at some frequencies, they are multidriver 28ohms nominal with 3 way WW-MM-TT 6 driver). previously actual opamps are some of the few things that can drive them properly.
Scott, I dont really like the idea of strapping an IC buffer to such a great discrete circuit; you call the last circuit a couple of pages ago a buffer, is that the buffer you are referring to in this last circuit for under 30ohms load impedance, or something else more standard?
is there any noise penalty for using actual diodes vs diode connected transistors? I was thinking of trying 2sk246/j103 or perhaps even some confused bisexual 2sk170 for both positions in the front end differential; any opinions?
Last edited:
You can't divide anymore this amp on "input stage" and "buffer", the amp now is more like a tight nice knot.
was this in reply to me at all? the section of the circuit in post 1380 sure looks like a buffer to me, but I know what you mean, very neat and integrated. besides what do I know, I feel pretty humbled in this company
but its OK, dont worry Scott, I read that post more carefully and you were only referring to the current state of the circuit as you havent done enough tweaking to provide reliable operation and predicted performance at low gains. still not sure what the recommended route for driving low impedance might be with either circuit (preferably the second one), but i'm sure that will be covered in time. it will hopefully drive my 300ohms HD600 OK as is though
Last edited:
Very elegant Scott! Are we to assume that these diode connected transistors are 1n4148 or 1n6263 or is there some new transistor when connected as a diode outperforms both the ones I mentioned. There is no transistor nomenclature. You mentioned the 2n5551 and the 2n5401. Did you model with these or are these 2n4403 and 2n4401s as in your first posted circuit. Also, forgive my ignorance in these matters but what is the quiescent current in the output devices and what procedure are we to use in setting bias. I'm an architect not a EE. Thanks a lot. Ray
1N4148 should be fine, I used the 2N4401/4403 in those plots. The 2N5550/5401 seems a little less good but the 2SA970/SC2240 look much better and very encouraging for lower closed loop gain versions.
The .1u caps in the input might not do anything anymore, a holdover from some experiments.
The .1u caps in the input might not do anything anymore, a holdover from some experiments.
Guru Wurcer, we grovel at your feet.
But could you give the unwashed masses (us) an explanation of perhaps one of the simpler circuits you have deigned to show us? eg the one in your #1380 post.
In particular the reasons for
-all the cross-coupling
- certain diodes like Q12.
- what does Q15 do and how does it do it
But could you give the unwashed masses (us) an explanation of perhaps one of the simpler circuits you have deigned to show us? eg the one in your #1380 post.
In particular the reasons for
-all the cross-coupling
- certain diodes like Q12.
- what does Q15 do and how does it do it
I don't get it with the 2SK170.
Could you elaborate?😱
while k170 are not truly symmetrical, they are reversible to a degree, so you can use them as a P channel by reversing the pins. some say performance is fine like this and they are pretty much as close as many complimentaries, others disagree. they work and work quite well like this depending on parameters of the circuit, but whether the change will mean it functions here or not I cant say.
Last edited:
I was thinking myself also at that property, and I wonder why nobody tried yet (AFAIK), leaving the search for complementaries in the dust. LSK389 is in this list? Because if it is, our life is much easier.
No way. JFETs are symmetrical, mostly, for an interchange of the leads labeled "source" and "drain", but that doesn't make them function like P channel parts. Life would be nicer if they did, although very confusing for solid state physics.
You may be able to use them as n channel
with source and drain switched, but not as
p channel.
Where do you get the idea that there is asymmetry in the Toshiba 2SK170, 2SJ74 jfets? Are you confusing these with American jfets?
aahhaa the distinction, sorry yes badly described and conceived. use them as reversed N channel in a differential then. completely right guys, thanks for the correction. 'acting as P channel', but not actually complimentary. so what does this mean for harmonics cancellation?
army of responders haha, OK so I havent bothered to actually do this in an amplifier due to lack of info, i've only read about it a few times. last time I read about it, there was mention of a small difference in.... Ciss perhaps, something, not completely the same one way or the other (for these jfets in particular). maybe John Curl can clarify, it was discussed in his blowtorch thread.
obviously the polarities of the supplies will be switched if you are using it in place of a Pchannel, will it actually work, or not? my google-fu was unable to get much info on it last time or this time, but I have seen it discussed several times on the forum.
doh morning head, now i'm not sure at all... was a throwaway line to throw a cat amongst the pigeons to maybe get some into from people who know, its so rarely mentioned and I have a crapload of k170, not many j74
army of responders haha, OK so I havent bothered to actually do this in an amplifier due to lack of info, i've only read about it a few times. last time I read about it, there was mention of a small difference in.... Ciss perhaps, something, not completely the same one way or the other (for these jfets in particular). maybe John Curl can clarify, it was discussed in his blowtorch thread.
obviously the polarities of the supplies will be switched if you are using it in place of a Pchannel, will it actually work, or not? my google-fu was unable to get much info on it last time or this time, but I have seen it discussed several times on the forum.
doh morning head, now i'm not sure at all... was a throwaway line to throw a cat amongst the pigeons to maybe get some into from people who know, its so rarely mentioned and I have a crapload of k170, not many j74
Last edited: