this circuit, using VGA's appears in the current issue of EDN -- claiming 3nV Hz^1/2 noise
http://www.reed-electronics.com/ednmag/contents/images/10401di.pdf
interesting and less expensive than a bank of relays:
http://www.reed-electronics.com/ednmag/contents/images/10401di.pdf
interesting and less expensive than a bank of relays:
Attachments
ensen.
If you take a look at http://www.diyaudio.com/forums/showthread.php?postid=246385#post246385 you will find a similar approach for gain/attenuation control in binary dB steps, but now even without (opamp) buffers in between the stages. The trick is to keep the characteristic impedance of each attenuation stage constant, irrespective of that stage is attenuating or not.
Steven
Steven
> It looks like it only gets a maximum attenuation of -45dB and in fairly low resolution steps of 3dB. Is that right?
> take a look at... you will find a similar approach for gain/attenuation control in binary dB steps, but now even without (opamp) buffers
A VGA is not the same as a VCA!
The EDN Idea is all-AMPlifier: 0dB to +45dB. It might be the start of a microphone booster amp. It is not a hi-fi volume control. (It is very likely intended toward a sonar, radar, or ultrasound image system preamp.)
x-pro's posted plan is just an ATTenuator. Sure you can put a booster behind it to make an AMPlifier, or even to get the +20dB/-60dB Amplify/Attenuate function we want in hi-fi system volume controls. A attenuator followed by an amplifier is normally not lowest-noise (nor does it need to be for hi-fi volume function).
Anyway: the EDN Idea is a university term-project. It demonstrates a concept with enough range to show it works. This one is built with two quad-chips, which is why it has 2^4 (16) steps and 0 to +45dB in 3dB steps. Obviously you could add two more switches of 1.5dB and 0.75dB to get 0.75dB resolution, at the cost of more switches and more control-wires. You could even add another +48dB stage for 0dB to +93dB range in 0.75dB steps and 7 control wires. Or 0 to +64dB in 1dB steps with 6 wires. But the kid just wanted to make the idea work on breadboard, get his grade, get his publication credit, get his degree, and get OUT in the Real World to go design bigger stuff. The quad-chip version meets the goal: working model not bogged-down in complexity and parts-cost.
Read the whole article. He says a better op-amp than the quad he used would give even lower noise.
That is an interesting switch-chip. It could be used for a pure-ATTenuator, which is probably what we want in hi-fi. It can't drop-in on x-pro's plan, because that uses a double-throw switch. You can get similar switch-chips in DT, or use two ST switches for each DT switch, or revert to potentiometer non-constant-Z attenuators.
BTW: the wiring of IC2D is wrong. And it will have slight distortion in the low-gain position.
An objection to the EDN Idea is that the front amplifier runs at max-gain all the time, and will be clipping on strong signals. The signal path bypasses that stage when set for lower gains, but we audio-heads would object to a stage sitting there clipping and spewing garbage throughout the box and supply rails. The EDN Idea way is simple, and works fine for many chores, but there are other ways to build a variable amplifier and we'd want something else even though it will cost more.
> take a look at... you will find a similar approach for gain/attenuation control in binary dB steps, but now even without (opamp) buffers
A VGA is not the same as a VCA!
The EDN Idea is all-AMPlifier: 0dB to +45dB. It might be the start of a microphone booster amp. It is not a hi-fi volume control. (It is very likely intended toward a sonar, radar, or ultrasound image system preamp.)
x-pro's posted plan is just an ATTenuator. Sure you can put a booster behind it to make an AMPlifier, or even to get the +20dB/-60dB Amplify/Attenuate function we want in hi-fi system volume controls. A attenuator followed by an amplifier is normally not lowest-noise (nor does it need to be for hi-fi volume function).
Anyway: the EDN Idea is a university term-project. It demonstrates a concept with enough range to show it works. This one is built with two quad-chips, which is why it has 2^4 (16) steps and 0 to +45dB in 3dB steps. Obviously you could add two more switches of 1.5dB and 0.75dB to get 0.75dB resolution, at the cost of more switches and more control-wires. You could even add another +48dB stage for 0dB to +93dB range in 0.75dB steps and 7 control wires. Or 0 to +64dB in 1dB steps with 6 wires. But the kid just wanted to make the idea work on breadboard, get his grade, get his publication credit, get his degree, and get OUT in the Real World to go design bigger stuff. The quad-chip version meets the goal: working model not bogged-down in complexity and parts-cost.
Read the whole article. He says a better op-amp than the quad he used would give even lower noise.
That is an interesting switch-chip. It could be used for a pure-ATTenuator, which is probably what we want in hi-fi. It can't drop-in on x-pro's plan, because that uses a double-throw switch. You can get similar switch-chips in DT, or use two ST switches for each DT switch, or revert to potentiometer non-constant-Z attenuators.
BTW: the wiring of IC2D is wrong. And it will have slight distortion in the low-gain position.
An objection to the EDN Idea is that the front amplifier runs at max-gain all the time, and will be clipping on strong signals. The signal path bypasses that stage when set for lower gains, but we audio-heads would object to a stage sitting there clipping and spewing garbage throughout the box and supply rails. The EDN Idea way is simple, and works fine for many chores, but there are other ways to build a variable amplifier and we'd want something else even though it will cost more.
PRR, you're right. I was too quick posting and didn't even realize it was about programmable gain instead of attenuation. I didn't read the original article. I just wanted to indicate that by proper resistance dimensioning you can do without buffers in between the stages.
Well, in this case it is not about buffers, but about amplifiers. An alternative approach would now be to put the x-pro scheme in the feedback loop of a single amplifier to get programmable gain with one amplifier instead of one per bit. This also improves the dynamic range of the circuit (like you said: no 24dB gain in the first stage, whether being used or not).
Steven
Well, in this case it is not about buffers, but about amplifiers. An alternative approach would now be to put the x-pro scheme in the feedback loop of a single amplifier to get programmable gain with one amplifier instead of one per bit. This also improves the dynamic range of the circuit (like you said: no 24dB gain in the first stage, whether being used or not).
Steven
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.