Derivator Dienoiser
What about this?
With RC prefilter. R = 1 ohm and C = 4700 uF:
With RCRC prefilter. R = 0,5 ohm and C = 4700 uF:
The 10K trimpot setting allows you to find the most optimal rejection point.
Is it not surprising?
Best regards
Using an opamp in a similar (but not that extreme) role has already been done, by Apex among others, and he probably lifted it from another source.
Without compensations, I predict huge stability problems, and the scheme will only work for PSRR and impedance, not noise: an opamp, even a good one is noisier than a transistor, even a humble BC337
You could start with a cap in // with the 560K, and a resistor in series with the input cap.
Note that even if it looks stable in .tran, that doesn't guarantee the stability of the actual circuit.
It has to be tested for real.
Yes, the gain of an opamp is huge compared to one or two transistors, the question is: will all this gain be usable?
Hello Elvee,
I have a question about the your denoiser...the old C945 is a good (better) substitute for the BC337? As for the datasheet it is low noise and operating in a linear Hfe range.
I don't think so, at least for usable devices: you might find that a large power transistor using the same technology as the ZTX like the the 2SC5200 has an even lower Rbb, but using one would not be practical: because of the bulk, the parasitic capacitances, and the behavior at low currents
Depends on the source impedance. For a 10k source impedance almost any transistor will be queiter!!
For voltage noise its up there with the best, for current noise its not. For best performance at a particular source impedance you want the ratio of voltage noise to current noise to roughly match the impedance so that the combined noise is minimal. A ZTX851 at 10mA is matched to 5 ohms or so.
If you allow yourself to use transformers to transform impedances then the lowest noise device has lowest noise _power_, measured in zeptowatts per √Hz
I am very happy to know that now your DAC has found its voice. Enjoy it and thanks for sharing your experience.
The greatest of my respects to you
Note that using a non-linear component as R2 will seriously blunt the performance of the Die- or Denoiser: it will shunt its output with a low dynamic impedance of forward-biased diodes.
Time to remind that the denoiser (and the Dienoiser) should only be applied to plain-vanilla regulators (as per datasheet examples)
Time to remind that the denoiser (and the Dienoiser) should only be applied to plain-vanilla regulators (as per datasheet examples)
I definitely doubt the LT3042 measurements, starting with the missing absolute scale.
My measurements look this way:
(ed. that does not seem to load in finite time)
Circuit is from the data sheet and results are like the data sheet. (red trace)
The strong rise at low frequencies goes on the undersized input capacitor
of the preamplifier. That has been changed in the meantime.
Increasing the output cap of the 3042 beyond the recomendend value creates
a noise peak. (green line). If you insist in a larger cap, you need some damping R.
The other picture is the LT3042 with an external DH44 BJT for large output current.
Also 1:1 from the data sheet.
< spectrum_lt3042_DH44 | Noise spectrum of the LT3042+ D44VH10… | Flickr >
The other pics to the left/right show the noise of some Zeners, LEDs and other references.
All plots are scaled that 0 dB is 1nV/rtHz, which is abt. the equiv. input noise density of an AD797 or LT1028.
Gerhard
Attachments
Last edited:
PL802,
I took the liberty to simulate your regulator. First two use the green LED you tried, though only one in my case to get the output you need. Your LED must be different.
Then I replaced the LED with a 680 ohm resistor, also to get to the voltage you need.
In this case the comparison was between PSRR, and the curves speak by themselves. Elvee seems to be right about the LED worsening the results, and impedance or noise should be worst than using a simple resistor. Why don't you try that option, instead of the LED?
Also have a look at the improvement you should have by just adding a 1 ohm resistor before your input filter cap, as suggested by Diego. On both versions.
I took the liberty to simulate your regulator. First two use the green LED you tried, though only one in my case to get the output you need. Your LED must be different.
Then I replaced the LED with a 680 ohm resistor, also to get to the voltage you need.
In this case the comparison was between PSRR, and the curves speak by themselves. Elvee seems to be right about the LED worsening the results, and impedance or noise should be worst than using a simple resistor. Why don't you try that option, instead of the LED?
Also have a look at the improvement you should have by just adding a 1 ohm resistor before your input filter cap, as suggested by Diego. On both versions.
Attachments
> Your LED must be different.
For starters, in the direction. LEDs are no Zeners.
As drawn, the ADJ pin floats somewhere near the output without
any connectivity to GND. The LED carries just some nA of backward
current.
If you turn the LED around, you may be able to set the output voltage
but most of the amplified noise voltage is shorted to GND through the
diode's small differential resistance. So, almost no correction.
For starters, in the direction. LEDs are no Zeners.
As drawn, the ADJ pin floats somewhere near the output without
any connectivity to GND. The LED carries just some nA of backward
current.
If you turn the LED around, you may be able to set the output voltage
but most of the amplified noise voltage is shorted to GND through the
diode's small differential resistance. So, almost no correction.