Mmmmm . . .
Absolutely no idea. Now, if there were capacitors connected from the transistor bases to ground, that would constitute a ripple eater and there I could see a possible reason for your findings.
Absolutely no idea. Now, if there were capacitors connected from the transistor bases to ground, that would constitute a ripple eater and there I could see a possible reason for your findings.
Mmmmm. . .
Similar drawing appeared here, post #1904
http://www.diyaudio.com/forums/chip...g-audio-integrated-opamps-96.html#post2805970
Similar drawing appeared here, post #1904
http://www.diyaudio.com/forums/chip...g-audio-integrated-opamps-96.html#post2805970
I'm missing to as Bonsai said couple of capacitors each one connected from base of transistor to GND. Then you get the basic capacitance multiplier, which is defined by the capacitor capacitance and transistor gain. R supplies the charging current as well as the transistor's (denoted by Q) base current. I'm olmost shure, that capacitors are missing from the circuit. IMO this is for stabilizing the pover to the IC.
Nope no caps missing, this is what I though a cap multiplier, here they are on Ebay.
eBay Australia: Buy new & used fashion, electronics & home d?r
Cheers George
eBay Australia: Buy new & used fashion, electronics & home d?r
Cheers George
It seems to me Chris Daly is the author or designer of this, I think he should be asked what the benefits are, if any.
Q1 and Q2 act as resistance dividers, so the main effect is to insert a resistance in the supply rail of R/beta. I can't see what benefit this will bring; why would anyone want to deliberately degrade their DC supply?
Hi
Thanks for your interest, the benefit with an emitter follower on each rail in op amp circuits is in a number of areas firstly if used in a stereo installation left right information is correctly represented. Direct comparison with the same op amp in this case AD825 without pass transistors stereo is a muddle appearing from the centre soundstage.
Secondly turn on and off thump is very much reduced.
I started this exploration examining dreadful implementation of op amp power supply in a Quad 405 which led to thoughts of locating transistors directly on the DIP adapter. Because there are many different locations of ground in op amp circuits. It dawned on me that leaving the capacitor out is still a valid circuit, obtaining the benefits of negative feedback arising from collector to emitter. It was then a matter of tailoring the base current to suit. Much experimentation found 6ma to be ideal.
The circuit in practice on a DIP Brown Dog adapter is extremely consistent delivering the same improvement time after time.
In October after much work I perfected a Dual version using a single adapter, Construction details are at my blog site opamp - opamp construction
Cheers / Chris
article here I believe validly explains an emitter follower:
Emitter Follower
"Due to this deep negative feedback, the voltage gain of the emitter follower is smaller than unity. However, the circuit is drastically improved in terms of its input and output resistances. In fact the emitter follower acts as an impedance transformer
Comparing this with the input resistance and output resistances of the common-emitter transistor circuit, we see that the emitter follower circuit has very favorable input/output resistances.
Although the emitter follower does not amplify voltage, due to its high input resistance drawing little current from the source, and its low output resistance capable of driving heavy load, it is widely used as both the input and output stages for a multi-stage voltage amplification circuit."
Thanks for your interest, the benefit with an emitter follower on each rail in op amp circuits is in a number of areas firstly if used in a stereo installation left right information is correctly represented. Direct comparison with the same op amp in this case AD825 without pass transistors stereo is a muddle appearing from the centre soundstage.
Secondly turn on and off thump is very much reduced.
I started this exploration examining dreadful implementation of op amp power supply in a Quad 405 which led to thoughts of locating transistors directly on the DIP adapter. Because there are many different locations of ground in op amp circuits. It dawned on me that leaving the capacitor out is still a valid circuit, obtaining the benefits of negative feedback arising from collector to emitter. It was then a matter of tailoring the base current to suit. Much experimentation found 6ma to be ideal.
The circuit in practice on a DIP Brown Dog adapter is extremely consistent delivering the same improvement time after time.
In October after much work I perfected a Dual version using a single adapter, Construction details are at my blog site opamp - opamp construction
Cheers / Chris
article here I believe validly explains an emitter follower:
Emitter Follower
"Due to this deep negative feedback, the voltage gain of the emitter follower is smaller than unity. However, the circuit is drastically improved in terms of its input and output resistances. In fact the emitter follower acts as an impedance transformer
Comparing this with the input resistance and output resistances of the common-emitter transistor circuit, we see that the emitter follower circuit has very favorable input/output resistances.
Although the emitter follower does not amplify voltage, due to its high input resistance drawing little current from the source, and its low output resistance capable of driving heavy load, it is widely used as both the input and output stages for a multi-stage voltage amplification circuit."
Emitter followers may be fine, in the right circumstances. That circuit is not an emitter follower. An emitter follower has a voltage supply to the base; that circuit has a current supply. As I said, it is a resistance divider. It adds resistance to the supply. Given the low supply current of most small-signal opamps even an emitter follower would not work because the emitter impedance would be relatively high. I suggest you go and do some more reading about emitter followers.
In the attached article an Emitter follower diagram is provided: note Rb is a resistor delivering current to the base from Vcc, An emitter follower as provided has a resistor from Vcc to the base. In the article a AC coupled signal is shown. Ignoring complexity of
AC coupling to process audio. The voltage source in my implementation is Vcc which also delivers current to the base exactly as the article provides. The load is provided on the emitter - in the article lifted from ground. In my implementation the load is also fed by the emitter to the relative op amp pin.
Cheers / Chris
AC coupling to process audio. The voltage source in my implementation is Vcc which also delivers current to the base exactly as the article provides. The load is provided on the emitter - in the article lifted from ground. In my implementation the load is also fed by the emitter to the relative op amp pin.
Cheers / Chris
No, you have misunderstood how it works. In the AC-coupled emitter follower given in those lecture notes the coupling capacitor provides an AC voltage source (e.g. low impedance, assuming the signal source has low impedance). Your implementation lacks this - you have removed the thing which makes it work. As I said, you need to do some more reading.
To a first approximation, the emitter impedance of an emitter follower is 25/Ie + Rb/beta. The first term is strongly non-linear; the second term is often zero (but not in your case). Note that the follower given in the linked article will have a DC output impedance which is higher than the AC output impedance, because the input coupling cap has no effect at DC.
To a first approximation, the emitter impedance of an emitter follower is 25/Ie + Rb/beta. The first term is strongly non-linear; the second term is often zero (but not in your case). Note that the follower given in the linked article will have a DC output impedance which is higher than the AC output impedance, because the input coupling cap has no effect at DC.
As I stated ignore the complexity of AC coupling. This is a DC circuit using a pass transistor to improve op amp DC rails. Think DC and compare this to a transistor buffered zener circuit which allows current to bias the zener tied to ground. Now without the zener such a circuit is still a pass transistor circuit, It is not pretending to be anything else. Same here, the transistor in this simple implementation follows path from collector to base via a resistor and the load is fed by the transistors emitter. It exhibits the same characteristics as a pass transistor DC circuit. However the control element is absent. In its absence the circuit drops the typical .7v of a transistor junction.
In practice it works exceptionally well, Try it and see, compare it to a standard implementation where DC is applied to an op amp, and notice how it instantly improves stereo separation.
In practice it works exceptionally well, Try it and see, compare it to a standard implementation where DC is applied to an op amp, and notice how it instantly improves stereo separation.
Last edited:
As I stated, you are simply adding a non-linear resistance to your supply rails. You can't think just DC when your opamp supply pins carry AC too. In fact it could be worse than that as for larger signals the supply pins carry half-wave rectified AC (if the output stage moves into Class B).
I suppose if the opamp has very good PSRR then your circuit won't do much harm.
I suppose if the opamp has very good PSRR then your circuit won't do much harm.
Totally agree with DF96.
I suppose for decoupling it has the beneficial property of decreasing impedance with increasing load current but as DF96 points out this is at the cost of non-linear variation of the op-amp bias.
Hope this helps
-Antonio
I suppose for decoupling it has the beneficial property of decreasing impedance with increasing load current but as DF96 points out this is at the cost of non-linear variation of the op-amp bias.
Hope this helps
-Antonio
I have tried with OPA627, OPA1611 and AD825 . The AD825 is exceptional. My source is a Marantz CD80. listening through headphones I have also exhaustively copied sound to a Yamaha Hard Disk Recorder, for each op amp enabling comparison. In every case this little circuit proves itself as a fine addition to improve audio performance. I have yet to hear it not improve.
In every case stereo performance improves as well as sound stage.
Cheers / Chris
In every case stereo performance improves as well as sound stage.
Cheers / Chris
Dear George
These transistors are mounted as gyrator circuit.
This improve PSRR of the power supply and decrease impedance source "seen" by OP amp. So sound would be better !!!
Cheers / Dan
These transistors are mounted as gyrator circuit.
This improve PSRR of the power supply and decrease impedance source "seen" by OP amp. So sound would be better !!!
Cheers / Dan
In your circuit, the transistors are converted to simple, two terminal devices. No emitter followers or gyrators here. Essentially, they work as if you would insert a diode into each supply rail. It is a nonlinear device, which provides a small voltage drop, nothing more.
Stereo separation is given by the recording and can only be ruined by bad layout (at high frequencies) or power supply coupling (at low frequencies). The latter happens if you use a common power supply for both stereo channels with relatively high impedance; i.e. without proper individual regulators. You can of course artificially expand the stereo base, but I don´t think that´s what you mean.
Can the insertion of diodes as shown reduce the crosstalk between stereo channels in the case of a common channel power supply with too high impedance? I have never investigated this, but it seems unlikely, because the diodes are dc-biased and thus essentially ac-transparent. The diodes will introduce some nonlinearity, which could result in distortion. How much? Probably not much, with high PSRR opamps.
Stereo separation is given by the recording and can only be ruined by bad layout (at high frequencies) or power supply coupling (at low frequencies). The latter happens if you use a common power supply for both stereo channels with relatively high impedance; i.e. without proper individual regulators. You can of course artificially expand the stereo base, but I don´t think that´s what you mean.
Can the insertion of diodes as shown reduce the crosstalk between stereo channels in the case of a common channel power supply with too high impedance? I have never investigated this, but it seems unlikely, because the diodes are dc-biased and thus essentially ac-transparent. The diodes will introduce some nonlinearity, which could result in distortion. How much? Probably not much, with high PSRR opamps.
The drawing is incomplete. I was using diodes series on rails (instead of resitors) since 10-20 years ago - and the drawing basically this is what it is.
But you really need to follow them with storage capacitors (to ground), otherwise is degrading the signal.
It is helpfull on pre-driver stage of a power amplifier, not so much on OpAmps rails (where currents are very small).
But you really need to follow them with storage capacitors (to ground), otherwise is degrading the signal.
It is helpfull on pre-driver stage of a power amplifier, not so much on OpAmps rails (where currents are very small).
Makes no sense to me. A cap, placed on the pins, would do more good. Even a cap from Base to earth makes sense, as a capacitance multiplier. It sure looks like no gyrator I've ever seen.
It is in effect a low value, non linear resistor. I would think level 2 of this
upgrade is a 'linear' resistor, sans bjt. Cheaper too. 🙂
Reminds me of a mixing console upgrade I did years ago.
Replaced a bunch of TL072's with newer faster OPA's. Since the console was
old and had archaic opamp bypassing, I added bypassing to each OPA for
fear of the faster OPA's oscillating.
Long story short, the console sounded a lot better, in almost every way, but
it had an added HF edge to the sound which was a bit unnatural. The
engineer perceived this as 'added HF'.
I had to think real fast as we were doing this on site, late at night and I
didn't want to have to undo all my work. I had an educated guess that due
to the old thin ground trace that all the bypass caps were connected to, and
the fairly low Z nature of the modern caps, they were 'ringing' or interacting
with each other at some VHF causing the pereceived 'edge'.
I inserted a low value (around 10 ohms) R feeding every opamp PS pin (PS side
of the bypass cap) and voila HF edge gone but all other + attributes still
there.
T
- Status
- Not open for further replies.