Re: Re: Regulator idea
For now it doesn't 😀
seriously, the idea is to make the first bjt from the left sense any differences between reference voltage and unregulated input. The input thru divider feeds the base, difference to be at emitter resistor (100R), this is hopefully subtracted by drop on 470R.
janneman said:
For now it doesn't 😀
seriously, the idea is to make the first bjt from the left sense any differences between reference voltage and unregulated input. The input thru divider feeds the base, difference to be at emitter resistor (100R), this is hopefully subtracted by drop on 470R.
Attachments
The second step is emitter follower (first bjt from right at original picture) with local feedback performed by middle bjt, this one also causes drop on the same 470R.
And once more sorry for quality. My computer broke some time ago and I don't have any simulator now, neither I want to run unknown circuitry without your 'approval' . That's why I drew it in MSPaint 😉
And once more sorry for quality. My computer broke some time ago and I don't have any simulator now, neither I want to run unknown circuitry without your 'approval' . That's why I drew it in MSPaint 😉
Re: Re: Re: Regulator idea
OK, the sense will work. But what does it do then? Ahh I see, you want to adjust Vout as a function of the input ripple? Kind of error feedforward? I don't think you can dimension it such that it exactly cancels any output ripple. In fact, that is what the right bjt tries to do, isn't it?
Jan Didden
darkfenriz said:
OK, the sense will work. But what does it do then? Ahh I see, you want to adjust Vout as a function of the input ripple? Kind of error feedforward? I don't think you can dimension it such that it exactly cancels any output ripple. In fact, that is what the right bjt tries to do, isn't it?
Jan Didden
Re: Re: Re: Re: Regulator idea
Probably you are right, Jan.
Something like 30dB of forward error cancellation would be good news. However I would expect this mechanism to work across relatively wide band, because it does not have the limitations of feedback, which can be very low at radio frequencies. My wishful thinking says also something about ripple transient response 😉 , but... You make great work keeping poeple down to earth 😀
Best regards
Adam
janneman said:
Probably you are right, Jan.
Something like 30dB of forward error cancellation would be good news. However I would expect this mechanism to work across relatively wide band, because it does not have the limitations of feedback, which can be very low at radio frequencies. My wishful thinking says also something about ripple transient response 😉 , but... You make great work keeping poeple down to earth 😀
Best regards
Adam
Re: Re: Re: Re: Re: Regulator idea
These feedforward schemes often depend on knowing the exact impedance or gain or whatever of the circuit you are trying to correct. That generally isn't possible, unless you make one or two parameters adjustable so you have to tune each circuit individually. It's doable but not attractive because it brings other problems.
Your idea might work in another way, like you put the output noise into an opamp with gain = 1/feedback divider and subtract it from the bottom of the feedback divider. That would work in theory, but because of uncertainties in the exact value of the feedback divider (you will need two exact resistor ratios) and non-linearities in even the best opamps, it is probable that it doesn't improve on the basic feedback regulator. These are often things that work in Spice with exact equal or ideal components but not in practise.
Jan Didden
darkfenriz said:
These feedforward schemes often depend on knowing the exact impedance or gain or whatever of the circuit you are trying to correct. That generally isn't possible, unless you make one or two parameters adjustable so you have to tune each circuit individually. It's doable but not attractive because it brings other problems.
Your idea might work in another way, like you put the output noise into an opamp with gain = 1/feedback divider and subtract it from the bottom of the feedback divider. That would work in theory, but because of uncertainties in the exact value of the feedback divider (you will need two exact resistor ratios) and non-linearities in even the best opamps, it is probable that it doesn't improve on the basic feedback regulator. These are often things that work in Spice with exact equal or ideal components but not in practise.
Jan Didden
In this kind of 'vintage' regulator, NPN pass transistor,
it is possible to let current source for V-ref be fead from the output.
I often make lowdrop ( <0.050 Volt in-out ) regulators with a PNP to pass the output.
Regulators with very low supply current consumption and low drop Voltage,
for possibility to effectively use battery for powering audio circuits.
In these PNP circuits, I have to use other methods
to be able to feed V-ref from output.
Sometimes a high value resistor from input to V-ref to give a small startup current,
sometimes I use a simple 'start up pulse generator' from input.
it is possible to let current source for V-ref be fead from the output.
I often make lowdrop ( <0.050 Volt in-out ) regulators with a PNP to pass the output.
Regulators with very low supply current consumption and low drop Voltage,
for possibility to effectively use battery for powering audio circuits.
In these PNP circuits, I have to use other methods
to be able to feed V-ref from output.
Sometimes a high value resistor from input to V-ref to give a small startup current,
sometimes I use a simple 'start up pulse generator' from input.
- Status
- Not open for further replies.