Forums

Automatic RF noise cancellation and audio noise measurement

Started by Jason Hsu August 5, 2003

Tweetldee wrote:

> > Allow me to infuse a different idea. Think about noise as a fairly > broadband phenomenon. The same atmospheric noise that you hear on 21.500 > Mhz will also be heard on 21.510 Mhz. Consider the possibility of biulding > two identical receivers, the main receiver tuned to the desired reception > frequency, the other tuned just a few Khz away, but out of the received > signal's bandwidth. Then, you detect the noise on the second receiver, > invert it, and sum it into the first receiver's signal chain. This concept > is used in ultra-low noise, high sensitivity deep space receivers to almost > totally eliminate noise in the received signal.
It's also the same scheme widely used for years in VHF FM mobile receivers. There is a completly separate receiver that receives the noise which is inverted and applied out of phase to the signal channel to cancel the noise pulses received through the receiver. The scheme obviously works much better with pulse type noise. Dick
In article <3F31B264.CAA78C3E@ieee.org>, Jerry Avins <jya@ieee.org> wrote:
>Tweetldee wrote: >> > ... >> >> Now, as to your concern to keep 100W of RF from blowing up your noise >> canceller. That's what antenna changeover relays are for. They normally >> connect the antenna to the receiver, but when the transmitter is keyed up, >> the relay connects the antenna to the transmitter output, and disconnects it >> from the receiver. Simple, but effective, and has been used for many many >> years in amateur as well as commercial radio equipment. > >The line from the change-over relay to the receiver should be shorted >when the receiver is disconnected and either short or a half wave, or >else open when the receiver is disconnected and a quarter wave long. > >Jerry
It's important to protect the both the noise antenna input and the main antenna inputs on the noise canceller from RF - but trying to figure out if you should short or open the receiver input based on the wavelength is a little difficult when the frequency range covers a continuum of multiple octaves. Not to mention the problem with putting a quarter or half wave chunk of feedline between the changeover relay and the receiver input when the wavelength is 160 meters. That technique works well for any single frequency (or very narrow frequency band as a percentage of the center frequency) and is commonly used at VHF frequencies and above with diode and relay switching circuits for the additional isolation it affords. An important consideration is the voltage on the changeover relay and the components connected to it. The obvious part is the 100 W and 50 Ohm situation, but consider what happens to the voltage when you have a high (SWR for the hams / VSWR for the engineers). Jason, since this is your project, I'll leave you with the research to figure out the potential range of reactive loads with various antennas as well as the calculation of the resultant voltages. Remember a good simulation models the real world, not just a convenient part of it. I will tell you the result if you don' t give this part of your design its due. Think about PCB traces vaporized, capacitors shorted, transistor "fuses", relay housings melted and what we used to call "essence of Allen Bradley" from the burned resistors. It's about the same as a lightning hit or running 1000 watts through a 100 watt unit! This can be a safety issue as well as causing unncessary delay and cost. One of the other people noted that it was important to do a lot of reading, research and analysis of your own - I'll second that! If you haven't run across it yet, start with the term "Null Steering". 'nuff said for now, Randy Gawtry Timewave Technology Inc.
Randy,

Speaking of the horse's mouth! I am concerned that Jason might have too high
an expectation of what the noise canceler can do. See if the below is right.

1. It can cancel out the noise from a single stationary noise source that is
received via groundwave, and will in general do nothing for atmospheric
noise as you would have on 75m.

2. For an automated adjustment method to work, and keep it within his means
and time frame, he will have to tune to a frequency where there is no signal
present. He should then be able to retune to a differenr frequency within a
few KHz, and still have the nulling.

3. Consistent with 2, the receiver's AGC voltage is as good an indication of
noise amplitude as any. In fact, it can't be measured AFTER the AGC, as for
instance the audio output.

Tam/WB2TT
***********************************************
"Randall R. Gawtry" <rgawtry@timewave.com> wrote in message
news:vj3q1bhol1tm9d@corp.supernews.com...
> One of the other people noted that it was important to do a lot of
reading,
> research and analysis of your own - I'll second that! > > If you haven't run across it yet, start with the term "Null Steering". > > 'nuff said for now, > > Randy Gawtry > Timewave Technology Inc.

"Dick Carroll;" wrote:
> > Tweetldee wrote: > > > > > Allow me to infuse a different idea. Think about noise as a fairly > > broadband phenomenon. The same atmospheric noise that you hear on 21.500 > > Mhz will also be heard on 21.510 Mhz. Consider the possibility of biulding > > two identical receivers, the main receiver tuned to the desired reception > > frequency, the other tuned just a few Khz away, but out of the received > > signal's bandwidth. Then, you detect the noise on the second receiver, > > invert it, and sum it into the first receiver's signal chain. This concept > > is used in ultra-low noise, high sensitivity deep space receivers to almost > > totally eliminate noise in the received signal. > > It's also the same scheme widely used for years in VHF FM mobile receivers. > There is a completly separate receiver that receives the noise which is inverted and > applied out of phase to the signal channel to cancel the noise pulses received through the > receiver. The scheme obviously works much better with pulse type noise.
It even works for _on-channel_ jamming TX'ers off in another direction from the desired signal (cable co's use the technique). Separate directional antennas are used for the desired and the undesired signals. The undesired signal (jammer) is gain adjusted and phase shifted and then summed into the "main" receiver signal path at 180 degrees (and equal amplitude) from the same jammer entering into the main antenna/rx'er path. The obvious null for the undesired signal occurs. Nulling of the desired signal is highly improbable given the directional aspects of the antennas. The sum can occur in the rx'er front end. Yes, this is signal processing down at the microvolt level. One problem is tuning drift.

Tarmo Tammaru wrote:

> Randy, > > Speaking of the horse's mouth! I am concerned that Jason might have too high > an expectation of what the noise canceler can do. See if the below is right. > > 1. It can cancel out the noise from a single stationary noise source that is > received via groundwave, and will in general do nothing for atmospheric > noise as you would have on 75m.
Correct!
> > 2. For an automated adjustment method to work, and keep it within his means > and time frame, he will have to tune to a frequency where there is no signal > present. He should then be able to retune to a differenr frequency within a > few KHz, and still have the nulling.
Correct!
> > 3. Consistent with 2, the receiver's AGC voltage is as good an indication of > noise amplitude as any. In fact, it can't be measured AFTER the AGC, as for > instance the audio output.
Correct!
> > Tam/WB2TT > *********************************************** > "Randall R. Gawtry" <rgawtry@timewave.com> wrote in message > news:vj3q1bhol1tm9d@corp.supernews.com... > >>One of the other people noted that it was important to do a lot of > > reading, > >>research and analysis of your own - I'll second that! >> >> If you haven't run across it yet, start with the term "Null Steering". >> >>'nuff said for now, >> >>Randy Gawtry >>Timewave Technology Inc. > > >
Tam,

Dave Schrader gave you the right short answers to your questions in a parallel 
post to this one; however, question number 2 is a bit constraining.  Consider 
the fact that there are some substantial variations of the ANC-4 and MFJ noise 
cancellers that others have built as well as few that are floating around 
in my head and, I am certain, a lot of other peoples' heads.  Nobody has 
mentioned these on this thread yet.  Also think about the fact that armed with 
the same information and skill sets, engineers tend to simultaneously invent 
the same thing about the same time.  Jason seems to have a pretty good tool 
box and a good understanding of how to add tools to it - witness this thead 
and his last project.  So he may figure out a way to completely avoid a 
situation like the one you describe, maybe even costing less and being quicker 
to develop!  If he cannot for some reason, then the challenge is to figure out 
a quick and easy way to do something like what you describe.  The only problem 
is finding a different frequency a few kHz away with no signal present - 
especially on the ham bands! So maybe a little more reaching question is how 
does he find a better way to null a noise canceller?  Of course, having said 
all that, sometimes the best engineering approach is brute force - do the 
obvious the way you already understand and get it shipped (or turned in to the 
instructor)!

Randy Gawtry, K0CBH
Timewave Technolgy Inc.



In article <PqKcnd9A-LwlDa-iXTWJhg@comcast.com>, "Tarmo Tammaru" 
<t-tammaru@home.com> wrote:
>Randy, > >Speaking of the horse's mouth! I am concerned that Jason might have too high >an expectation of what the noise canceler can do. See if the below is right. > >1. It can cancel out the noise from a single stationary noise source that is >received via groundwave, and will in general do nothing for atmospheric >noise as you would have on 75m. > >2. For an automated adjustment method to work, and keep it within his means >and time frame, he will have to tune to a frequency where there is no signal >present. He should then be able to retune to a differenr frequency within a >few KHz, and still have the nulling. > >3. Consistent with 2, the receiver's AGC voltage is as good an indication of >noise amplitude as any. In fact, it can't be measured AFTER the AGC, as for >instance the audio output. > >Tam/WB2TT >*********************************************** >"Randall R. Gawtry" <rgawtry@timewave.com> wrote in message >news:vj3q1bhol1tm9d@corp.supernews.com... >> One of the other people noted that it was important to do a lot of >reading, >> research and analysis of your own - I'll second that! >> >> If you haven't run across it yet, start with the term "Null Steering". >> >> 'nuff said for now, >> >> Randy Gawtry >> Timewave Technology Inc. > >

Tweetldee wrote:

> "Jason Hsu" <jason_hsu@my-deja.com> wrote in message > news:f7d9a152.0308051412.54d4c913@posting.google.com... > > THE BIG QUESTION: How do you measure the audio noise level in a > > receiver? Does anyone here know of any circuits that show the overall > > audio noise level? > > > > Let me tell you what this is all about. > > > > SIDE NOTE: Thanks again to those of you who helped me with the > > SWR/wattmeter project I worked on last semester. This project taught > > me more about ferrite cores, op amps, and diodes than ANY class > > possibly could. I learned about stray capacitance and what rails > > mean. I also learned that not all 1N34A diodes are alike. > > > > In this coming semester, I will be working on an even more ambitious > > project - an automatic noise canceller for 160m to 10m. It will > > definitely be a control system, and it may even use DSP. The device > > will also require a motor for automatically adjusting the controls. > > The idea is to design and build an automatic version of the Timewave > > ANC-4 or the MFJ-1025/1026 RF noise cancellers. > > > > The ANC-4 and MFJ-1025/1026 are manual devices. There are automatic > > noise cancellation devices, but their adjustments are at the audio > > level rather than the RF level. > > > > The back panel of my device will contain: > > 1. Connector for the signal antenna > > 2. Connector for the transceiver > > 3. Connector for the noise antenna > > 4. DC power connector > > > > The front panel of my device will contain: > > 1. Power switch and the obligatory idiot light to show that the power > > is on > > 2. 2 audio connectors: one connects to the transceiver, and the other > > connects to an external speaker > > 3. Frequency range control for choosing the inductance values, as the > > proper inductance values in the RF noise cancelling circuit varies > > with frequency > > 4. Phase range knob so that the phase shifter can cover all 360 > > degrees > > 5. Manual noise phase knob > > 6. Manual noise gain knob > > 7. LED display to show the noise level > > 8. Noise phase adjustment push-button: Pressing this button activates > > the control system to adjust the phase shift of the noise. > > 9. Noise gain adjustment button: Pressing this button activates the > > control system to adjust the gain of the noise. > > > > Other features: > > 1. I need to design my automatic noise canceller so that parts won't > > blow up if I transmit 100W through it. This probably requires some > > type of automatic bypass circuitry. > > 2. Transmitted power must not reach the noise antenna (at least not > > without a great deal of attenuation) > > 3. It also needs a low insertion loss. > > 4. A preamplifier would be desirable. > > > > How it would work: > > 1. You manually turn the phase and gain controls to the minimum > > settings. (Or I could put in a reset button to do that. No, that > > would add too much complexity to the circuit.) > > 2. You select the band using the band inductance switch. > > 3. You move the phase polarity switch to the negative setting. If > > the noise cancellation procedure does not work, it probably means the > > opposite setting is required. > > 4. Press the noise gain adjustment button. The device adjusts the > > noise gain until a change in the signal level (or the S meter) is just > > noted. (Some type of derivative function would come into play.) > > 5. Press the noise phase adjustment button. The device adjusts the > > noise phase shift so as to produce a null in the signal level (or the > > S meter). > > 6. Steps 4 and 5 could be repeated to make further adjustments. > > There might be away to activate the control system with just one > > button. > > > > Does anyone know of circuits that measure audio noise? Making the RF > > noise canceller automatic requires an audio circuit to monitor the > > noise level. This noise level needs to be converted into a large DC > > voltage, which feeds the motors that turn the variable capacitor and > > potentiometer. > > > > Jason Hsu, AG4DG > > usenet@@@@jasonhsu.com > > Allow me to infuse a different idea. Think about noise as a fairly > broadband phenomenon. The same atmospheric noise that you hear on 21.500 > Mhz will also be heard on 21.510 Mhz. Consider the possibility of biulding > two identical receivers, the main receiver tuned to the desired reception > frequency, the other tuned just a few Khz away, but out of the received > signal's bandwidth. Then, you detect the noise on the second receiver, > invert it, and sum it into the first receiver's signal chain.
Hmmm.... surely the noise will be phase shifted and altered in amplitude by a time-varying transfer function?You will need to estimate this first. Tom
Randy,

Thanks for your response. The obvious thing you would like to do is to have
the software be smart enough to distinguish between signal and noise. But
then the hardware becomes moot, and you have invented a better DSP noise
reduction circuit.

BTW, I am considering getting an ANC4 to get rid of power line noise on 6
meters. Fortunately, the noise is coming from the North, a very seldom used
beam heading.

Tam/WB2TT


My ANC-4 is sitting on top of the ICOM 746. I have a short piece of 
coax, 7 to 8 feet, to a doublet with a total length of 8 feet on the 
floor against an outside wall.

Works Great!

I've used the ANC-4 for almost two years. While in Massachusetts the QTH 
was 100 meters from the New England Power Grid with three sets of high 
tension [> 67,000 volts] three phase power lines  [nine wires total]. 
The power lines were generally quite quiet but during summer humidity 
and/or thunderstorms they would get 'noisy'. The ANC-4 reduced the noise 
to S1 typical levels.

Here in New Hampshire, my local noise source is the local LAN where the 
router is four feet from my receiver. Similar results with the ANC-4.

BTW, it was given to me as a gift by W1KSZ who picked it up on e-bay for 
about $60.00. Thanks many times over, Dick.

DD, W1MCE
+ + +

Tarmo Tammaru wrote:

> Randy, > > Thanks for your response. The obvious thing you would like to do is to have > the software be smart enough to distinguish between signal and noise. But > then the hardware becomes moot, and you have invented a better DSP noise > reduction circuit. > > BTW, I am considering getting an ANC4 to get rid of power line noise on 6 > meters. Fortunately, the noise is coming from the North, a very seldom used > beam heading. > > Tam/WB2TT > >
Dave,

Thanks for the info. In my case, the source seems to be a bank of 4
transformers on a pole about 300 feet away. Weirdest thing I had was S9
white noise from about 40 to 55 MHz with about 50% amplitude modulation at
around 50 KHz. Turned out to be a 60 amp cartridge fuse in my fuse box about
to go bad. I think the 50 KHz was a mechanical resonance in the fuse.

I don't know about the 746, but my 756 has dual watch capability, which lets
you listen to two frequencies at the same time. It would be interesting if
they would let you vary the phase of the secondary channel..

Tam/WB2TT
"Dave Shrader" <david.shrader@comcast.net> wrote in message
news:EcOYa.65732$It4.35429@rwcrnsc51.ops.asp.att.net...
> My ANC-4 is sitting on top of the ICOM 746. I have a short piece of > coax, 7 to 8 feet, to a doublet with a total length of 8 feet on the > floor against an outside wall. > > Works Great! > > I've used the ANC-4 for almost two years. While in Massachusetts the QTH > was 100 meters from the New England Power Grid with three sets of high > tension [> 67,000 volts] three phase power lines [nine wires total]. > The power lines were generally quite quiet but during summer humidity > and/or thunderstorms they would get 'noisy'. The ANC-4 reduced the noise > to S1 typical levels. > > Here in New Hampshire, my local noise source is the local LAN where the > router is four feet from my receiver. Similar results with the ANC-4. > > BTW, it was given to me as a gift by W1KSZ who picked it up on e-bay for > about $60.00. Thanks many times over, Dick. > > DD, W1MCE