The idea is to move the feedback resistor to the other end of the cable, as shown in the figure. I know this has been tried for power amplifiers, but I was wandering if it was worth a try for preamplifiers connected to power amplifiers. As preamplifier, I am intending to use a transconductance amplifier, with high output impedance. This reduces the risk of instability due to capacitive loading. So, have someone tried this? Or, is it just nothing to gain with this solution?
Attachments
the cable's parasitic capacitances cut bandwidth, and/or destabilize the loop
and make cable dielectric errors and microphonics bigger problems than the "traditional" approach
if you really want "current signaling" then you need a low Z transimpedance amp at the input of the power amp
remote "Kelvin Sensing" is used in some applications but the drive is usually close to a V source so cable C can be largely ignored, this best compensates for the series R drop of the cable resistance - not a problem for line level signals
putting a balanced receiver with high common mode rejection at the amp input is a more certain path to improvement - use shielded twisted pair, shield DC connected at preamp only
and make cable dielectric errors and microphonics bigger problems than the "traditional" approach
if you really want "current signaling" then you need a low Z transimpedance amp at the input of the power amp
remote "Kelvin Sensing" is used in some applications but the drive is usually close to a V source so cable C can be largely ignored, this best compensates for the series R drop of the cable resistance - not a problem for line level signals
putting a balanced receiver with high common mode rejection at the amp input is a more certain path to improvement - use shielded twisted pair, shield DC connected at preamp only
I intend to use an amplfier as attached, this should imply that the stability should not be worsened, although the bandwidth could be lowered (but here we are talking about very high frequencies). But how could the cable in the feedback loop worsen cable dielectric errors and microphonics? I would think the concept would be an improvement due to feedback theory.
I have to agree with you, when it come to noise suppression, a balanced connection is better.
Attachments
http://www.diyaudio.com/forums/solid-state/101321-3-stage-lin-topology-nfb-tappings.html#post1201482
includes a schematic showing feedback going to source and using source to determine gain of the receiver.
includes a schematic showing feedback going to source and using source to determine gain of the receiver.
you really need a model of the cable to put some numbers on the effects
I dislike the idea that the loop gain GBW product depends on the type and length of the attached cable (load C)
you need some clear limits on cable parameters to design this at all - and I would use those cable paramters to calculate the performance expected in "conventional" approaches to see if you're really able to improve anything
I dislike the idea that the loop gain GBW product depends on the type and length of the attached cable (load C)
you need some clear limits on cable parameters to design this at all - and I would use those cable paramters to calculate the performance expected in "conventional" approaches to see if you're really able to improve anything
Last edited:
I have tried to simulate a lumped model a coaxial cable. I have used the same cable length from the preamplifier to the power amplifier and back. When it comes to distortion, the advantage in relation to the traditional approach is only marginal. The stability is, however, much worse; the phase margin is about non-existing even at 100kHz.
The conclusion? Do not do this. A better approach would probably be to use non-feedback approach, where the transmitter is high-impedance and the receiver is low-impedance?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.