Create project Need help?

A innovative White Noise Generator

Status: Proposal
July 7, 2013

The project describes and explaines a really new kind of a Noise-Generator-Technology, on the base of a prebiased and on its own potential floating Audio-Source, respectively for the best noise possible - as a standard transistor. This gets in a arrangement together with a symmetrical amplifier and gets then enhanced with some filters, two cmos-pulse-generators and also some other wellknown parts from the classic "Elektor-Formant Noise-Module". On the end, there is then performed a stand-alone and widely experimental Synthesizer-Module, wich enables the usage of a high quality, but also variable Noise-Source, in conjunction with the ability of being full experimental, expandable and coming hereby with unknown and countless possibilties, if one changes the said transistor to any desired impedance as choosen and adjusting a preselectable bias.

With some of the here published Circuit-Variations, I got my Oscilloscope to a complete and almost unexplainable malfunction (except the High Voltage Section), but without damaging anything. The event was not really reproduceable, neither periodically, nor predictable, but - so to say - not seldom and somekind of crazy!
About 2-5 times per hour, while measuring on the circuit, the whole electronic (inside of that osci) made a smoothed, but hearable "zoomph" and then nothing more, and with the beam standing as a point in the middle of the tube.
On the first time I worried about that and I thought on a defect, but after switching a new power-up, the device was ok again and working like known at anytime. It is still ok until today and almost every tried explanation to this happening failed, but it seems to me, like somehow the produced noise signal of the measured circuit, knocked out on some charge carriers at the working semiconductors inside of the oscilloscope (maybe back to the nuclear plant??)...
The shown noise-signal on that osci was also somehow of irregular (??), or maybe better told as somekind of astonishing!
I have only very few pictures left from this special experiments, except some smd´s on a verostripe, also this was in early 2010 with some circuits on a breadboard and on a time, when I didn´t know about the possibilities of the elektor-labs. Making this public, was of coarse an idea, but I didn´t see any possibilty at that time.

For about two or three decades I`ve been doing some electronic developments on circuits for Analog-Synthesizers and hereby almost with the backgrounds around a suitable usage of the circuits in modular Systems.
By the years on doing that, I´ve been of coarse also coming on a lot of garbage and therefore not to tell, but there were also some nice things and where I want to report now about a very useful unit, for my first project-description and introduction to the elektor-labs-community.
Here I have prepared a circuit, that I`m calling for being "floating", as told in the headline and this is directly named by the working principle to the core-function of the device.

The main features and goals of the Floating-Noise-Project:

1.)    A White Noise Audio-Generator on a "best choice noise quality performance", with Bias Control and interchangeable Noise Element (possible external adaptor recommended). Therefore the unit gives the background-abilities also to a possible Zener-Diode-Noise, or maybe a biasing of a Germanium-Transistor for generating the noise, for experimental studies, etc. and so on.
2.)    Symmetrical and adjustable Gain Amplifier, for several possible Noise-Levels and Plug-ins on a Transistor-Socket - Six Potentiometer-Controls in one row (4/5 HE) - Smart White-Noise-Module variation, for a stand-alone usage.
3.)    Six separate outputs for noisy Audio-Products (White-, Pink-, Red-, Blue-Noise, one fixed and one variable Formant-Random-Voltage-Circuit) and two digital Pulse Generators (CD4538 - One with fixed Pulse-Length and one variable)
4.)    Layouts for handmade Printed Circuit Boards, wich are fenced by a 0.4 mm line on the top and bottom, for a optimized fixing of the PCB to the exposuring unit. This is enabling very accurately exposured and reproduceable prints, double-sided and on possible home production, with the vias as rivets.

How the circuit works:

The circuit works by the well known usage, of an especially for this purpose selected noisy transistor, but almost in a innovative and unusual way! The published main schematic for the project, describes in the end not only a possible, interesting Synthesizer-Module, but this could also be a very good product unit, for a professional usage and maybe with then some further enhancements around the thing, what is also predestinated to get somekind of a Laboratory-Standard.
As a daily, universal noise as working tool, this may also get into a little box, with a 9V-battery and a DC/DC-Converter for a negative supply, then a small PCB with connectors and two Controls on Print-Potentiometers, where the unit can get useful in almost every Audio-Service-Environment and at very good data.

The basic circuit to the project was fully tested some years ago, with several variations on a experimental breadboard and additionally in a second step-up and this until today also in a Tina-Simulation. These Files are here prepared for the Attachment, what you can download for trying it, if you´re using that software.
For the practical useage in a simulation, the noisy-transistor got replaced by an AC-Generator (50mV) to drive all kind of possible audio and wich is coupled with a capacitor onto a input. This helpful adddition I have set upon the circuit, just only because of the fact, that I could not find any useful spice-models for implementing noisy transistors, respectively for testing them in that kind of arrangement.

The here told data and facts to the project I am giving for promise (only if the projects gets built like it is recommended), because the circuit is really a new kind of a concept and as I think so, almost interesting to all persons, who are used to some practice on audio-electronics.

All published files, that I have uploaded here, are copyright signed and commercially protected for the core-circuit since the very first tests, around from 2010, but this gets not produced anywhere as available unit. I do not know about that in any configuration, like this here presented publication, so maybe a similar professional electronic equipment is working on some parts of the principle, but I have searched on a lot of websites around the internet and other possibilities to get almost for shure, there will be nobody being working on the same circuit arrangements, or really right on that special principle of a floating noise source, as you will see, when you are reading the descriptions to this publication, and reading it to the end.  

The accurately routed PCB´s (!! as I think so, smallest parts = 0603/0805, R11/R21) to this project are coming in Full-SMD´s as Eagle-Files and over all this is suited for a handemade product, hereby comes a first unit, with only the White-Noise-Generator, using two Control-Potis for an adjustable Bias-Current and a Gain for the Amp (12 available divisions for units from 1 Euro-Multiple-Card - this layout is prepared with a standard of 1mm-spaces for easy uploading to PCB-Services and this is available on a request).

The core-function to the project is based on the performance of the TLE2037A Excalibur Opamps by Texas Instruments in SO-08 Packages (= Burr Brown OPA637 Equivalent), but of coarse this is variable.
(for the said opamps I could help to support interested readers on more than 100 possible units/modules, also the recommended opamps aren´t completely obsolete, so they are still available in SO-08 (C-Versions) on element14 or by digikey, and they are really excellent! A parallel version with a possible used INA114-Amp by Burr Brown, is just prepared to get routed.)

The main circuit comes also as a 2´nd Version, with a complete extension to a very useful Synthesizer-Module, with 6 Controls arranged in a row, where several Noise-Colors are performed by Filters (parts of Paia Design and on their commercially copyright - very fine, and on the layouts prepared for DIL-14 vintage quad opamps), then a variable Random-Voltage, as known from the Formant-Noise-Module (with the copyrights on Elektor-Formant - also excellent) and two variable Pulse-Generators/Monoflops based on CD4538 in DIL16 (there are 3 available Pieces for Units/Modules, when made from 1 Euro-Card. An additional Sample/Hold-Module is just in development/sim and will be coming soon).
For full details on all possible versions, please refer to the Attachment-Files, some coming also maybe later uploaded, as Eagle.PDF, Eagle.brd, Eagle.sch!

Enhancements to the Circuits for a practical use in modular Synthys

All versions of the Modules give the advantage of an enhancement on a externally used Transistor-Socket, for changing the Noise-Source-Transistor to an external adaptor (TO-39 Socket on the PCB, maybe patchable onto banana-jacks), or perhaps for using some other possible impedances, to get extended functions out of the thing.
This option offers very interesting features for experimenting, also for Owners of Modular-Synthies, by connecting variable and selected transistors, with different noise-products, or phototransistors, LDR´s, or maybe any other practical sensor to the circuit, etc., so also a "pre-biased Guitar-Pickup" would be possible, to connect on that input and filtering therefore the audio, etc.
Each of the above said versions comes with the mechanical dimensions for an economic partitioning of multiple-devices for Euro-Card ordering, maybe on the basis of a collected joint venture to the project (as possible), so there could be saved much costs on the needed PCBs, coming with very high quality from Europrint of coarse and maybe sometime by the offer of the Elektor Print Service, as I hope so.

The published Noise-Project in Steps

The first background to the project was the following: A couple of years ago, I decided to develope a circuit, for a more effective and easier selecting of noisy transistors out of a bunch.
Then, by that on the way and of coarse to get more parts out of my charge, I just got (optionally) on that action also a new kind of a noise-generator at all and by the implementation of a few more parts in the configuration, I got a much better signal amplitude of the produced noise.  

Figure 1:    A Standard-Circuit, to produce a "White Noise" with a Transistor

So here, for a start on a useful description to the new circuit, let´s first have a look onto the method, wich is therefore known as a standard and wich is used quite a lot in industrial circuits for music-synthesizers.
The basic concept to this kind of noise-generating is shown in Figure 1 (as also performed for the very first Formant-Noise-Modules in the 70´ties),
where a Supply-Voltage around 9-15 Volts gets over two Resistors to an especially selected Noise-Transistor, wich is connected in a Zener-Diode-Configuration, with an unused collector and with a fixed reverse-current thru the Base-Emitter-Junction, on about 100µA. The produced noise is then amplified by an Opamp, with a fixed amplification of Av = 100-300 (1000), and hereby set upon a preferred output level of the device, but after all, this is generally very high and where the known standard opamps often get onto their usable limits.
With this circuit on a breadboard, I started on a test for a first charge, with about 25-30 pcs. of new BC548B transistors and I got here 2 pieces with an acceptable noise-level of about 1,5 Volt resulting AC, measured with an oscilloscope. The rest of these transistors was also some noisy, but almost not with acceptable values and with this kind of selection-method.
A first conclusion was hereon, that maybe the noise-circuit couldn´t be very useful. On this result, I thought about shurely not being pleased with that, so maybe the basic circuit could be modified, to get more noisy transistors out of that charge.

Figure 2:    With a Current-Mirror the Noise gets transduced onto an Opamp

And this can be done very well, by the additional use of a current-mirror, where in the end almost every new transistor can get "animated" to start on a noise and acting like somekind of a "Resonator" in conjunction with another addition, a therefore used ceramic capacitor, suitable with about 10pF to 120pF and wich stimulates the noise, by the low resistance for all kind of AC-Signals.
Right here on that point comes the funny question, on how to get into a "Resonance on a Noise (?)", and this can get pretty difficult, because a white-Noise couldn´t origin anyhow, to get then told or explained as a resulting oscillation!
But this circuit works, and so the audio signal gets coupled out as a product on the output of the current-mirror, as shown in Figure 2. For this circuit, the two formerly used electrolytic Caps in Figure 1, got replaced by the ceramics, and so to say not saving only money, because there has to be seen, that now comes a changed working principle to the circuit.
Hereby the current-mirror means an acting current-source, and this is a much better solution, to get a transistor to work as a noise-generator, because the part represents here a noise-function, wich is performed by a voltage-source (the one in the PN-Junction on the base-emitter current-path) and this is on a much better way driven with that current-source, for getting the needed energy to the part, where the noise origins.
With this circuit, one can get almost every new transistor, wich is connected like this, to generate a noise, by also selecting the right value for C1 and together with the right wiperposition of the used potentiometer, on a controlled biasing current for the part (at a limited maximum of probably 5-10mA).
If this could become a real "White-Noise", is almost impossible, by it´s physically told definition (White Noise = a infinite number of different frequencies, with an energy amount by 1), so this would be in the end a good question, but I haven´t got a spectrum analyser for testing that and almost I think, this spectrum is of coarse different, because the capacitor represents a defined and fixed frequency parameter, and therefore of coarse this is also a resonator, but in a really unusual kind of arrangement!
Especially there is to say, that the preselected high-noisy transistors did not need that cap for getting in that "generator mode" at all and also therefore this is then maybe more resulting in form of a really White Noise, what on other parts could be told as a somekind of an ignition on a selected frequency, just for starting on a kind-as-noisy action, but with a restricted frequency spectrum. As I told above, I had no spectrum analyser to measure that.
Each transistor, that gets tested with this circuit, has then it´s individual and unique point, for the best value on a biasing-level for the part, to start a desired and generated noise, and to be seen on the osci.
Then as a next step, to prepare on the circuit, was another enhancement, by increasing the inner resistance of the used current-mirror, with a doubled configuration (as in detail shown in Figure 3) and this has guided directly to a first product of the circuit, with a useful "MM5837-Replacement" and where this circuit comes with a realizable noise-level of up to 4-5 Volts rms. This circuit needs then a proper opamp, with a good gain/frequency product, where the used LM321 (by National Semiconductor) has a value of 1 MHz and comes in a small and useful SOT-23-5 SMD-Package!

Figure 3:  A doubled Current-Mirror (Q2-Q3) with higher Resistance enables a               MM5837-Replacement on a small SMD-PCB in DIL08 or SIL04-Size.

The MM5837-DIL08 NMOS-IC was a digital noise-generator from around the end of the 70´ties, manufactured also by National Semiconductor and is obsolete since years. The most advantage of that IC was a high amplitude on the noise-level, well prepared for a usage without any other amplifying parts needed and because of that it was implemented in quite a lot of synthesizers and other applications. Another advantage to this NMOS-Device was the fact of a possible supply voltage between 12V and 15V, so it could get easy being used in the most non-computerized synthies at that time.
For the MM5837, there are known several good working replacements, but some of them need a modification to the destination device, for a restricted 5 Volts usage in 12 or 15 volt systems, where the described and shown circuit, with the LM321 for the output, doesn´t need any, so this means the user can plug this thing in, and ok.

Picture 1:    The PNP-Current-Mirrors BCV62C in SOT-143 can get easily fixed on a                         piece of Vero-Stripe, for a plug-in-usage on a Expo-Breadboard.

The shown circuit uses for the current mirrors BCM857BS SMD-Transistors in SOT-363-packages. These parts are not very easy to handle, mostly not on a breadboard, or also not on a tentative strip for SMD´s, so the first tests were made with BCV62B in SOT143-packs, wich got soldered on a piece of vero-stripe as shown in Picture 1 (the only photos that I just keep left from these tests).
To arrange the current mirrors with BC557B/C, or similar parts, is of coarse also possible, but that doesn´t give in the end such good devices, as with the in-package-thermocoupled BCV62B/C mirrors. For these also closely matched parts, the manufacturers NXP and Infineon are giving some information about the thermocoupling for mirror usages, where the performed mirrors are not manufactured on the same die, but for a good coupling inside of the SOT-143 SMD-package, they guarantee.

The finally presented Schematics and Layouts for the Project

To get a real "Floating Noise" out of that thing, as the name is told for, the audio-source needs again to be configured in another way, where the shown circuit in Figure 3, gives almost a good noise, but wich is still electrically oriented on the Ground-Potential as a part of the noise generating element. Of coarse this is useful, but this cannot be told as a real professional thing, or let´s say on a Lab-Performance.
Therefore the noise-source should get in a real state of being "floating" on a independent potential, wich is no more (directly) on the GND-Potential oriented anyway and what is here done by an another, additional Current Mirror, where the first one sits on the positve supply side and now the second Mirror sits on the negative supply, as shown for the basic principle in Figure 4. The upper, positive supplied mirror is hereby realized on PNP-Transistors and the one on the negative supply side, is then arranged with NPN-Transistors.
Another extention to the circuit is made also for the used Amplifier, where now comes for this job an Instrumental Amplifier (shown as Symmetrical), wich is arranged in the published circuit with the High-Speed Excalibur-Opamps TLE2037A. For the finally circuit you can refer to the attached File named "Elektor-The Art of Noise.sch/brd - or the PDF´s". Additionally there is to say, that the circuit got once again extended, with a CMOS Analog Switch ADG211 by Analog Devices, if the user prefers a versatile selection of the used smd-capacitors, then possible by individual values as recommended (refer to the schematics - the useage of these additional parts was not tested on breadboard, but for getting a real experimental product and for the said versatility, this is shurely very useful!):

- Some switches on a frontplane and displ. by LED (R1-R4 = 680 Ohms).
- Only selected with Switches/jumpers (8 Jumpers, R1-R4 = 47 kOhms)

Figure 4: The basic Principle to the Floating Noise, as a symmetrical                configuration and by the usage of an Instrumental Amplifier

For the finally, very smart construction on a PCB, as it comes with the Eagle-Files named as above, this small unit is thought for getting mechanically fixed only by the Potentiometers and there are no possible holes anywhere, for some screws on the surface available. For this I recommend a lightly PCB on 1mm material and all the rest-data for standard.
The next and full version (Elektor - Floating Noise.pdf, .brd, .sch), with all filters, monoflops, random, etc., has three positions for possible screws on the print and then to get fixed hereby additionally.
Conclusive there ist to say, that the shown partlist for the schematics, is with mixed SMD´s and conventional Parts, routed upon my storage-parts for availble numbers and this can get changed on request, if you´d send me an email therefore.
Important Notice:    If you want to use these Board-Files (Eagle Vers. 5.11), please be aware, that all resistors indexed as "Minimelf-LL" are not recommended for automatic positioning (less tolerances!) and need then to be interchanged with metallfilm-resistors in 1%-1206-Packages, to prevent crashes. The routed, shown and used p

Read the full post
Show less
BCV62 PNP-Mirrors on Verostripe
Dr, Boehm Transistors without any Noise
Figure 1 - Standard Noise - Text
Figure 2 - Noise w. Current Mirror - Text
Figure 3 - MM5837-DIL08 - Text
Figure 4 - Floating Noise - Text

Loading comments...