hogstedt@ae.chalmers.se (PER HOGSTEDT) (01/03/91)
I need to produce a voltage that is going to be used to bias a high sensitivity magnetometer (intrinsic noise < 1 pT/sqrt(Hz)) exposed to the earth's magnetic field. The long term stability demands on this voltage source are moderate, but the voltage noise has to be extremly low. I have tried some "reference voltage" IC:s and they are indeed very stable, but the noise level is too high. The noise level would have to be below what corresponds to <1/2LSB of a 22 bit A/D converter to utilize this beast to it's full. Can you send me a dozen :-) As a kind of last resort, I would like to know what the noise is of a regular 1.5V (e.g. alkaline) cell, assuming the cell is kept at room temperature and loaded only by some microAmps? Yes, if you have that super duper low noise circuitry I'm looking for at hand I'm certainly interested in any pointers. The bandwidth of primary interest is DC to maximum 50 Hz. ------- Per Hogstedt Internet: hogstedt@ae.chalmers.se Lindholmen R&D or: hogstedt@plab.se P.O. Box 8714 Phone...: +46 3 50 70 50 S-402 75 Gothenburg Fax.....: +46 3 51 53 13 Sweden MiniCall: +46 74 35 68 49 -------
larry@kitty.UUCP (Larry Lippman) (01/03/91)
In article <1991Jan3.001303.483@ae.chalmers.se>, hogstedt@ae.chalmers.se (PER HOGSTEDT) writes: > I need to produce a voltage that is going to be used to bias > a high sensitivity magnetometer (intrinsic noise < 1 pT/sqrt(Hz)) exposed > to the earth's magnetic field. The long term stability demands on this > voltage source are moderate, but the voltage noise has to be extremly low. > ... > As a kind of last resort, I would like to know what the noise is of a > regular 1.5V (e.g. alkaline) cell, assuming the cell is kept at > room temperature and loaded only by some microAmps? > The bandwidth of primary interest is DC to maximum 50 Hz. The essential noise component of a simple electrochemical primary cell is thermal noise (Johnson Noise, for Johnson fans). According to the good Dr. Nyquist, only three factors need concern you for this application: (1) absolute temperature, (2) noise bandwidth, and (3) internal resistance of the cell over the noise bandwidth. Since your bandwidth is only 50 Hz, we can ignore any frequency effects upon cell resistance. Since you want to run at room temperature, we can figure 290 deg K for absolute temperature. Since you are asking about a 1.5 volt alkaline cell, I would guess internal resistance to be around .25 ohm. My off-the-wall guess is that thermal noise will be less than 1 nanovolt. You should be able to readily obtain the thermal noise equation, plug in your values along with Boltzmann's constant, and get a more exact figure. If you were drawing some serious power from the alkaline cell, I would consider some other factors in evaluating noise (like temperature shifts, gas generation, etc.), but offhand I can't see anything other than thermal noise for a uA load. Larry Lippman @ Recognition Research Corp. "Have you hugged your cat today?" VOICE: 716/688-1231 {boulder, rutgers, watmath}!ub!kitty!larry FAX: 716/741-9635 {utzoo, uunet}!/ \aerion!larry
kludge@grissom.larc.nasa.gov ( Scott Dorsey) (01/03/91)
In article <1991Jan3.001303.483@ae.chalmers.se> hogstedt@ae.chalmers.se (PER HOGSTEDT) writes: >As a kind of last resort, I would like to know what the noise is of a >regular 1.5V (e.g. alkaline) cell, assuming the cell is kept at >room temperature and loaded only by some microAmps? I don't know what the value is offhand, but it's low-but-measurable, and fairly well distributed. A reference oscillator I use has a reactance tube which uses a battery as the bias source for low noise. Alkaline cells are better than dry batteries, and mercury cells are much lower noise than the alkaline ones. I haven't tried lithium or magnesium cells, but they might be worth fooling around with. It's so easy to get batteries, so give it a try. -- Scott Dorsey/ Kaptain Kludge NASA Langley Research Center, Aircraft Guidance and Control Branch Disclaimer: Neither NASA nor Lockheed really know anything about what
whit@milton.u.washington.edu (John Whitmore) (01/04/91)
In article <4285@kitty.UUCP> larry@kitty.UUCP (Larry Lippman) writes: (replying to query on low-frequency noise from dry cell) > > The essential noise component of a simple electrochemical primary >cell is thermal noise (Johnson Noise, for Johnson fans). > > If you were drawing some serious power from the alkaline cell, >I would consider some other factors in evaluating noise (like temperature >shifts, gas generation, etc.) Some researchers here who needed a steady voltage tried out dry cells and found that some gave 10 mV sawtooth output voltages (about 1Hz frequency). Gas generation was their explanation; a small pressure buildup in the cell eventually pushed past the seal on the battery. The cell voltage, of course, is depressed when the gaseous byproduct has significant backpressure. Even at very low currents, the leakage current in a dry cell is guaranteed to generate some gas. The folks found that their best bet (Zeners and IC references were too noisy) was a stack of standard cells (which have no gaseous byproduct). They bought a few dozen for their 90V bias battery. It was expensive. Then they put the cells in a temperature-regulated box (chilled, I believe). Different battery systems have different stabilities; get a few different types and test 'em. The only ones I KNOW won't work are carbon/zinc. John Whitmore
larry@kitty.UUCP (Larry Lippman) (01/04/91)
In article <13749@milton.u.washington.edu>, whit@milton.u.washington.edu (John Whitmore) writes: > > The essential noise component of a simple electrochemical primary > >cell is thermal noise (Johnson Noise, for Johnson fans). > > > > If you were drawing some serious power from the alkaline cell, > >I would consider some other factors in evaluating noise (like temperature > >shifts, gas generation, etc.) > > The folks found that their best bet (Zeners and IC references > were too noisy) was a stack of standard cells (which have no gaseous > byproduct). They bought a few dozen for their 90V bias battery. The standard cell (Weston cell) is certainly a stable voltage source, but from an electrical noise standpoint is undesirable to bias a device subject to great amplification due to the high internal resistance of the cell. Typical unsaturated (the most common secondary standard variety) standard cells have internal resistances between 500 and 1,200 ohms. When used as a voltage standard, the noise component is not significant when compared to the accuracy and precision of the cell, and when one considers that standard cells are usually compared potentiometrically in a system having virtually zero frequency response bandwith. > They bought a few dozen for their 90V bias battery. > It was expensive. I bet it was expensive, considering that one would require at least 88 unsaturated Weston cells to get 90 volts! Even the miniature unmounted Weston cells made by Eppley sell for at least $ 50.00 apiece in quantities. > Then they put the cells in a temperature-regulated > box (chilled, I believe). The most common standard cell temperature is 28 deg C. Current drain from standard cells must be kept to a minimum. Eppley recommends an absolute maximum of 100 uA for 5 minutes. A definite breach of etiquette is to measure the voltage of a standard cell using a conventional VOM. It's really a silly thing to do, when one considers that the precision of a standard cell is many orders of magnitude greater than a VOM meter movement! However, put a standard cell and a VOM in a lab full of students, and once can rest assured that at least one student will use the VOM to measure the voltage! :-) It seems to be an irresistable temptation, and I freely admit that I have committed this fau paux in my younger days. > Different battery systems have different stabilities; get > a few different types and test 'em. The only ones I KNOW won't work > are carbon/zinc. I certainly agree with that! Bear in mind that lower internal resistance will result in lower thermal noise. Mercury cells make great stable voltage sources, but they have fallen into disfavor for obvious reasons. A favorite "casual" voltage standard during the 1960's and 1970's was a multi-tapped mercury cell made by Mallory. It had taps every 1.34 volts, had 8 cell sections, and was sealed in a robust metal case with binding posts on the top. Larry Lippman @ Recognition Research Corp. "Have you hugged your cat today?" VOICE: 716/688-1231 {boulder, rutgers, watmath}!ub!kitty!larry FAX: 716/741-9635 {utzoo, uunet}!/ \aerion!larry
strong@tc.fluke.COM (Norm Strong) (01/08/91)
In article <1991Jan3.001303.483@ae.chalmers.se> hogstedt@ae.chalmers.se (PER HOGSTEDT) writes: }I need to produce a voltage that is going to be used to bias }a high sensitivity magnetometer (intrinsic noise < 1 pT/sqrt(Hz)) exposed }to the earth's magnetic field. The long term stability demands on this }voltage source are moderate, but the voltage noise has to be extremly low. } }I have tried some "reference voltage" IC:s and they are indeed very stable, }but the noise level is too high. The noise level would have to be below }what corresponds to <1/2LSB of a 22 bit A/D converter to utilize this beast }to it's full. Can you send me a dozen :-) } }As a kind of last resort, I would like to know what the noise is of a }regular 1.5V (e.g. alkaline) cell, assuming the cell is kept at }room temperature and loaded only by some microAmps? } }Yes, if you have that super duper low noise circuitry I'm looking for at hand }I'm certainly interested in any pointers. } }The bandwidth of primary interest is DC to maximum 50 Hz. I happen to know the answer to that question. The very best battery for that use is the Duracell RM-42R mercury cell, a 1.35 volt "D" sized cell made by a variety of manufacturers. It's rated at 14 ampere/hours, and if the current consumption is kept in the microamp region, it has a noise resistance of about 2 ohms. -- Norm Strong (strong@tc.fluke.com) 2528 31st S. Seattle WA 98144
apn@Apple.COM (Alex Novickis) (01/24/91)
In article <1991Jan3.001303.483@ae.chalmers.se> hogstedt@ae.chalmers.se (PER HOGSTEDT) writes: >I need to produce a voltage that is going to be used to bias >a high sensitivity magnetometer (intrinsic noise < 1 pT/sqrt(Hz)) exposed >to the earth's magnetic field. The long term stability demands on this >voltage source are moderate, but the voltage noise has to be extremly low. > >I have tried some "reference voltage" IC:s and they are indeed very stable, >but the noise level is too high. The noise level would have to be below >what corresponds to <1/2LSB of a 22 bit A/D converter to utilize this beast >to it's full. Can you send me a dozen :-) > >As a kind of last resort, I would like to know what the noise is of a >regular 1.5V (e.g. alkaline) cell, assuming the cell is kept at >room temperature and loaded only by some microAmps? > >Yes, if you have that super duper low noise circuitry I'm looking for at hand >I'm certainly interested in any pointers. > >The bandwidth of primary interest is DC to maximum 50 Hz. > >------- > Per Hogstedt Internet: hogstedt@ae.chalmers.se > Lindholmen R&D or: hogstedt@plab.se > P.O. Box 8714 Phone...: +46 3 50 70 50 > S-402 75 Gothenburg Fax.....: +46 3 51 53 13 > Sweden MiniCall: +46 74 35 68 49 >------- I don't recall as to waht purpose it was for, but I remember using a mercury cell for a reference of this nature. Degradation of the cell over long-term time can be _computed_ and therefore compensated out. Short term noise was in the 22nd bit range. -- Alex P. Novickis, Real Time systems demi-guru. (W) 408-370-4541 ALINK:alex.n (PAGE) 989-6678 UUCP:{amdahl,claris,pyramid,sun,decwrl,well,ubvax,ames}!apn@apple.com,apn@nonvon "I think... I think it's in my basement. Let me go upstairs and check"-Escher.