You could try something like this.
It will short one output to ground in AC, but in your case this shouldn't be a problem.
The common mode range of the input is something like |Vs|-4V
Attachments
A discrete amp with balanced input! Although its other unusual features, this will be a great significant feature for me anyway. Maybe it is not opamp style but my aim is not related with this as you know.
Thank you for great support.
Last edited:
View attachment 225174
View attachment 225180
View attachment 225181
w
If you see anything else don't hesitate to point it out.
If there isn't any termal issue between Q1 - Q2 (Q4-Q3 in original schematic) PCB layout is very good but if they should positioned face to face, is it possible in this layout?
I think I tend to go with 2sc2240/2sa970's and populate parts on vero board. This is not related with current PCB layout.. just not being in a mood of making a pcb.
Regards.
Last edited:
Putting those transistors face to face is rather cosmetic: they themselves dissipate a moderate (~25mW) and strictly identical power, and on the board, there are no big sources of heat, therefore placing them side by side is quite sufficient.
You could wrap them in a common heatshrinking sleeve, but that's pure luxury.
Even if there is a one or two °C gradient between them, this will translate into 2 to 5mV DC offset at the output, and nothing else: the Circlophone circuit forces a balanced operation.
That's an excellent choice, and veroboard is fine, I made all my prototypes that way
TO5 heatsink is sometimes right size to clamp a pair of TO92, not always.
I see size variations in TO92, usually deviant on the small side of normal.
I got some matched pairs that just fall out of TO5 heatsink.
This pair wouldn't fall out of a heatsink...
No time for sceenshot, my clearwire connection has been failing.
Uptime in minutes, when it connects at all... you might need a
library, search Duncan Amps. I'll bundle in zipfile later.
I want to change floating sense to Taylor, and tail to current mirror,
so I can set a threshold lower than 1 emitter drop, and determinism.
But serf circle becomes unstable whenever I try that arrangement.
I see size variations in TO92, usually deviant on the small side of normal.
I got some matched pairs that just fall out of TO5 heatsink.
This pair wouldn't fall out of a heatsink...
No time for sceenshot, my clearwire connection has been failing.
Uptime in minutes, when it connects at all... you might need a
library, search Duncan Amps. I'll bundle in zipfile later.
I want to change floating sense to Taylor, and tail to current mirror,
so I can set a threshold lower than 1 emitter drop, and determinism.
But serf circle becomes unstable whenever I try that arrangement.
That was like the thirteenth attempt to post above. My connection keeps
getting dropped mid-post, while typing. I kept it short, though I shoulda
just typed offline, then quickly pasted. Anyways, thoughts I wanted to add...
The tOObage was merely plausible excuse to invert the input topology at
expense of CFP driver stage. There is no benefit Triode over JFET or BJT
when input impedance is only 10K, and Mu here has absolutely no function.
Plate held constant 30V doesn't bother 6DJ8 made for cascode operation,
but isn't showing us anything a solid state part couldn't do well or better.
Open loop gain is rather low, and drive current is too low for 4 ohm loads.
Both can be fixed by adding drivers back in. Would need to be Darlington
rather than CFP, if N type input stage is wanted with N type output...
I am interested if it is possible to sense one transistor in float, and the
other at the rail? That would enable possibility to cross real transistor
currents, rather than perceived currents at load. Which crossing would
be superior anyways? These crossings would never be same with N BJT
outputs, as bottom one wastes base current into the rail. The top one is
getting an assist from base to emitter and base bypass currents into the
load. Right now, I am crossing at center of Schottky currents, which is
not center of BJT currents...
Elvee's circuit is crossing emitter vs collector current, perhaps not in
exactly equal proportions. Is there some fudge factor built into the
Taylor voltage references at each rail to find the actual center? Note
the resistor at the bottom is only 1/2 that of the top, what was the
reasoning behind that?
getting dropped mid-post, while typing. I kept it short, though I shoulda
just typed offline, then quickly pasted. Anyways, thoughts I wanted to add...
The tOObage was merely plausible excuse to invert the input topology at
expense of CFP driver stage. There is no benefit Triode over JFET or BJT
when input impedance is only 10K, and Mu here has absolutely no function.
Plate held constant 30V doesn't bother 6DJ8 made for cascode operation,
but isn't showing us anything a solid state part couldn't do well or better.
Open loop gain is rather low, and drive current is too low for 4 ohm loads.
Both can be fixed by adding drivers back in. Would need to be Darlington
rather than CFP, if N type input stage is wanted with N type output...
I am interested if it is possible to sense one transistor in float, and the
other at the rail? That would enable possibility to cross real transistor
currents, rather than perceived currents at load. Which crossing would
be superior anyways? These crossings would never be same with N BJT
outputs, as bottom one wastes base current into the rail. The top one is
getting an assist from base to emitter and base bypass currents into the
load. Right now, I am crossing at center of Schottky currents, which is
not center of BJT currents...
Elvee's circuit is crossing emitter vs collector current, perhaps not in
exactly equal proportions. Is there some fudge factor built into the
Taylor voltage references at each rail to find the actual center? Note
the resistor at the bottom is only 1/2 that of the top, what was the
reasoning behind that?
Last edited:
Hi Kenpeter
the asc file is impossible to simulate due to absence *.inc file called in directives
Generic triode
Generic pspice model dmtriodep.inc you need.
Or rip them out and replace with JFET, U309 seems to work...
Or rip them out and replace with BJT, 2N3904 seems to work...
Or rip them out and replace with MOSFET, 2N7002 seems to work...
Higher loop gain with substitutes may aggrivate ringing after clip.
Generic pspice model dmtriodep.inc you need.
Or rip them out and replace with JFET, U309 seems to work...
Or rip them out and replace with BJT, 2N3904 seems to work...
Or rip them out and replace with MOSFET, 2N7002 seems to work...
Higher loop gain with substitutes may aggrivate ringing after clip.
Last edited:
Hi Elvee, Kenpeter!
back to xquad idea, you can use 4 IRF610 and input voltage 2V will not saturate irfs...
You realize what happens to drain capacitance at 2V?
Transition toward symmetry with source capacitance.
Deceptively low Miller in spec sheet is rated at 10VGS
and 0.8xBVDSS. Thats 160Volts, my friend, Not two...
Going by Figure 11 of Fairchild spec sheet: I see about
90pF Gate to Drain at 2V... Climbing to 120pF at 1V...
I would assume at 0V that maxes to 135pF, same as
gate to source? Anyways, its much bigger than 16pF
suggested value...
Crazy varactors do seem the ideal way to aggrivate
an already unstable xquad into complete madness...
Tentacles, flutes, the whole nine yards...
Last edited:
I am reminded of Blumlein's garter bias, which has a half loop gain of 1/2 and can't latch.
Maybe something of that sort, but push the half loop gain to 0.9 or therabouts. Yet if
the xquad was not a barely tamed superregenerative monster, it might not prove any
more interesting than a Darlington LTP...
Maybe something of that sort, but push the half loop gain to 0.9 or therabouts. Yet if
the xquad was not a barely tamed superregenerative monster, it might not prove any
more interesting than a Darlington LTP...
Last edited:
You are right.
That's very strange, in principle the netlist should be common to the schematic and PCB.
Anyway, you just have to put a big enough blob of solder to unite the two pins 1 of C3
As Ken said, using MOSFETs there could be troublesome.
It could be made to work by tying a cap large enough between the inputs, to combat the effect of capacitances.
The Tringlotron is a sort of asymetric Xquad, and it is made to work with MOS transistors.
But as I said above, if the Xquad is part of an amplifier, saturation is not a problem (except for latch-up).
It is only when you want to use it as a stand alone amplifying stage that you could be confronted with the problem.
And there is a very simple workaround with bjt's: you just have to increase the Vbe of the transistors by using a zener, or stacking more junctions in series with the base. (shunted by a big cap, of course)
Last edited:
Now, which resistor(s) would you be talking about? I've been using my Circlophone with +-39V and 1/4W 1% MF for a while now, no problems.
The only resistor that could benefit from an upgrade is R17, the main feedback resistor.
http://www.diyaudio.com/forums/atta...y-little-cheap-circlophone-thecirclophone.gif
With relatively high supply rails, it could dissipate up to ~50mW, and to eliminate any possibility of thermal distortion, a 10x factor could be applied, leading to 0.5W.
But this amplifier doesn't target the ppm, and a normal 0.25W metal film should not degrade significantly the performances under normal circumstances.
I know about thermal distortion, but I am so satisfied with the sound that I am not going to change anything. Maybe I'll put it in a solid block of epoxy
