Don't relax if yo have no DC through resistor: excess noise modulates signal, and who knows what is worse.
Single perfect Rule of Dumbs for any place in any application?
Single perfect Rule of Dumbs for any place in any application?
I wouldn't bother with the current source, another whole can of worms in itself. If you have four of them, make a bridge and use a battery to excite the bridge. The whole affair floats so should be easy to hook up to the measurement instrument regardless of its referencing. AC-couple if absolutely necessary, if the residual d.c. offset from value mismatch is an issue (but you must already be a.c.-coupling your power supply under test as it is).
Back out the average of noise per resistor assuming that the individual resistor excess and thermal noises are uncorrelated.
Although the battery will have some noise itself, it can be measured independently as a cross-check, and its noise is of course rejected to the degree the resistors are matched.
Is Gallium ROHS-worthy? Of course, except in warm weather and with no air conditioning, one would need cable heaters 😉
hi,
Use a Lipo or SLC battery paralleled by a very large value electrolytic cap (> 0.1F) feeding a Fet CCS.
BTW, I made an open loop, cascoded J-Fet pre-pre (30dBish) for measurement systems (BNC in and out) with a few MHz bandwidth and < 50 Ohm noise.
It makes my 0.1mV 'scope have 0.003mV per division and on the EMU 1616m and AP2 it allows you to measue the noise of grass growing in Yerewan with a big enough antennae. PSU is a 9V battery (primary) with a big cap.
Ciao T
Use a Lipo or SLC battery paralleled by a very large value electrolytic cap (> 0.1F) feeding a Fet CCS.
BTW, I made an open loop, cascoded J-Fet pre-pre (30dBish) for measurement systems (BNC in and out) with a few MHz bandwidth and < 50 Ohm noise.
It makes my 0.1mV 'scope have 0.003mV per division and on the EMU 1616m and AP2 it allows you to measue the noise of grass growing in Yerewan with a big enough antennae. PSU is a 9V battery (primary) with a big cap.
Ciao T
Hi,
2SK369V at IDSS (Rs = 0) in folded cascode with 2N4403 (2 Diodes + J5XX CCD for bias, J5XX tail, all parts need selecting carefully to make it work at DC).
9V PP3 block battery with several 10V Sanyo Super Micro Size (green/silver sleeve) caps, something like 20,000uF or so IIRC, space limited the number of cap's in aprallel..
1u film cap as output coupling cap, normally into 1M.
Build into the Metal Can the expensive Mains Plugs we buy come in.
Ciao T
2SK369V at IDSS (Rs = 0) in folded cascode with 2N4403 (2 Diodes + J5XX CCD for bias, J5XX tail, all parts need selecting carefully to make it work at DC).
9V PP3 block battery with several 10V Sanyo Super Micro Size (green/silver sleeve) caps, something like 20,000uF or so IIRC, space limited the number of cap's in aprallel..
1u film cap as output coupling cap, normally into 1M.
Build into the Metal Can the expensive Mains Plugs we buy come in.
Ciao T
That's indeed the practical way to do it, Wheatstone, sorry I forgot about you. Actually, don't need a battery; my prototype supply is by now quiet enough as it is for that kind of thing, and even faster and quieter parts will be coming in next week. The apparent paradox is that to create perfect DC, you need unlimited bandwith.
vac
Hi T
I think KBK (Ken) is using Galistan. (see his patent app: US 2010 0193243)
There has been done experiments using Galistan instead of Mercury on wetted relays.
One problem is oxidation (see: JOURNAL OF MICROELECTROMECHANICAL SYSTEMS, VOL. 21, NO. 2, APRIL 2012 )
Cheers
Stein
There is a very interesting article about regulators in the April 2012 issue of Power Electronics Technology, "An Unconditionally Stable Linear Voltage Regulator", by Sandler and Hymowitz of Picotest. Although they give some performance results and claim an approximate regulator bandwidth of 300MHz, they don't show a schematic. However the circuit is apparently on their Picotest Voltage Regulator Test Standard kit.
The article appears to be available here:
An Unconditionally Stable Linear Voltage Regulator
This is a concrete spec that can be tested, the scientific community knows of numerous brands of resistors with very low excess noise. In fact there are folded cacode input op-amps with corner frequencies below 3Hz using (yuch!) 400 atom thick thin film resistors.
BTW by bad do mean to say everyone here using "ordinary" MF resistors is just going from bad to worse with their designs.
In the absense of DC current there is no excess noise as long as some care is taken to keep the two ends of the resistor at the same temperature (not that hard actually).
I repeat this exhaustive survey of excess noise, plenty of choices (if you can sort out the colors 🙁).
https://dcc.ligo.org/public/0002/T0900200/001/current_noise.pdf
https://dcc.ligo.org/public/0002/T0900200/001/current_noise.pdf
They might be but there are too many similar colors for me to sort by eye, heaven forbid you're color blind (this is a government program they probably violate some rule 😀).
EDIT - IIRC the Vishay S102K are the crappers for sound. 🙄
Last edited:
Very nice link.
So a new pseudo-theory explaining "resistor sound" should be created? 😉
I happened to measure hundreds and hundreds of the Susumu resistors (0.1%, 25PPM tempco) to hand select parts out of two resistor kits for very precise pair of stepped attenuators. It surprised me to find the part to part variation so tight. Many of the values would have met tighter accuracy tolerances (0.01%). Not knowing how to listen to resistors myself, I go with larger wattage and lower tempco when available.
.
.
Last edited:
Wow - very nice paper.
What is striking is the difference in noise peformance - orders of magnitude between resistors of the same value but from different manufacturers
Would be interesting to have a similar set of results for uV/V linearity and inductance.
- Status
- Not open for further replies.