audiorob said:
Because it IS positive feedback ?
Suppose you run a topless bar and it's full. To have all the seats taken, no empty ones and no standing eye-ballers, instead of controlling the in-flow at the entrance you can control the sucker-flow at the exit.
(in the restaurant business they make you wait an hour for the bill if the place isn't fully booked)
If the gain is higher than it should be, there's more current flowing through the feedback resistors, the JFET will see the one below it (R1 or R2) as a virtual higher resistance=> result is less gain.
Yeah well, i've always wanted to be an academic sleaze, or was it a sleazy academic, let's just stick to a sleaze.
Feedback of a regular non-inverting amp is also only NFB when it's fed to the inverting side of the input differential pair.
The real beauty of this thing is the split feedback, a single polarity feedback signal from the output simultaneously corrects both sides of the complementary input pair.
Feedback of a regular non-inverting amp is also only NFB when it's fed to the inverting side of the input differential pair.
The real beauty of this thing is the split feedback, a single polarity feedback signal from the output simultaneously corrects both sides of the complementary input pair.
Last time i googled for a TaTa-bar i ended up at a car shop in India.
What's puzzlin me is whether the dual-pump action is less intrusive than a single NFB loop, or still be looked at as a single gain block with error feedback. Intuitively it feels like the first, when you slice the F5 down the middle it's a symmetrical balanced system.
What's puzzlin me is whether the dual-pump action is less intrusive than a single NFB loop, or still be looked at as a single gain block with error feedback. Intuitively it feels like the first, when you slice the F5 down the middle it's a symmetrical balanced system.
Those two loops interact, they aren't independent.
BTW, in simulation when I change the feedback network from its "rhombus" or "bridge" style to an equivalent "cross" (8.2R source-R's, 8.2R to GND, 41R feedback) which represents a traditional "current feedback" single loop hookup, thus running the same DC bias points for all parts and the same overall gain, the circuit performance is practically identical for a variety of test conditions I have used, the original version being slightly better on the order of 10% or so, the harmonic distribution is overall very similar in magnitude and phase (original has slightly better even order cancelling). Maybe Nelson will enlight us what exactly made him choose the split hookup?
- Klaus
BTW, in simulation when I change the feedback network from its "rhombus" or "bridge" style to an equivalent "cross" (8.2R source-R's, 8.2R to GND, 41R feedback) which represents a traditional "current feedback" single loop hookup, thus running the same DC bias points for all parts and the same overall gain, the circuit performance is practically identical for a variety of test conditions I have used, the original version being slightly better on the order of 10% or so, the harmonic distribution is overall very similar in magnitude and phase (original has slightly better even order cancelling). Maybe Nelson will enlight us what exactly made him choose the split hookup?
- Klaus
For those in the EU interested in building the F5, our Group Buy has agreed to offer a limited number of matched FETs at "commercial" rates. All the credit received will be donated to Unicef at the end of the year.
You may contact me by PM for details. As an example, a matched pair of FQA19N20C + a matched pair of FQA12P20, matched at room temperature to 10mV with 1.6A bias, will cost 6 Euros by EU bank transfer, and 4% more by Paypal. The N devices are matched to each other, as are the P devices.
Delivery by post, at cost.
First come first serve.
Patrick
You may contact me by PM for details. As an example, a matched pair of FQA19N20C + a matched pair of FQA12P20, matched at room temperature to 10mV with 1.6A bias, will cost 6 Euros by EU bank transfer, and 4% more by Paypal. The N devices are matched to each other, as are the P devices.
Delivery by post, at cost.
First come first serve.
Patrick
I think I remember this question from many pages back. The value of the thermistor was not terribly important and since it is in series with R15/16 then their total value must also not be terribly important. Some people are not even putting a thermistor in.
see this link
Uriah
see this link
Uriah
