Radio telescope filters, low noise amplifier, downsampling.

Go To Last Post
41 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello folks!

I am contemplating the idea of building myself a radio telescope from an old c-band big dish, with a computer-controlled azimuth/elevation control. The mechanical aspect of it will come later, for the moment I am more concerned about the main electronics...

First, the antenna will have a bandpass filter. I am interested in a 2.5MHz range centered at 1420MHz. I would prefer staying with fully analog design, because of the cost of the parts needed for an active filter... From there split into 8x 312KHz bands, then into 8 LNA inputs. Ideally I would like to then use these outputs to drive a 8-bits per band 8 bands VFD display.

Now I am not used at all with working at these frequencies with analog filters, are there any pitfalls I must avoid, or any suggestions on the design? I am still investigating this and putting ideas together, but I figure I will need a crapload of precision resistors and caps. Is there a way to design this filter digitally without spending a whole lot of money?

The 1420MHz center frequency is the frequency of radio emissions from superheated hydrogen in space BTW... ;)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

At 2 Ghz the dish has to be really big to get a reasonable narrow beam. Getting the input satge right is probably also not easy, allthough I am not an expert on this. There are probably enough pitfalls, because at 2 Ghz every milimeter counts. For fine filtering you will most likely have to go digital anyway, its just a question at which bandwidth.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Oh and I will be using "Analog and Digital Filter Design (2nd Edition)" by Steve Winder to help me along with the maths, any other suggested reading is appreciated... ;)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Forget "active filters" at that frequency.

I would bet that you want to down convert from 1420MHz before splitting into the narrower bands. a 312KHz bandwidth at 1420MHz implies minimum resonant Qs at least as large as 1420/0.312 = 4500 and THAT is a challenge at any frequency. If you do a first IF in the 100-250MHz area, lets say, that reduces the needed Qs by a factor of 10. Then, you can use nice helical resonators with achievable Q of 500 or so, and that should make you some nice filters. Also, you can achieve good noise levels at that lower frequency.

I strongly recommend the books by Wes Hayward on "RF Circuit Design". There is information about mixers, low noise amplifiers, filters of all kinds (but not one digital filter - forget that at RF).

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Sounds like you are planning on 8 separate LNAs @~1400MHz. Why not go the classical route and have one LNA(or a couple) and downconvert to a more reasonable frequency for further processing?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks for the comments guys...

Kleinstein: Dish will be an old C-band 5.2m diameter, shouldn't be too bad in terms of FoV. I will have to design a custom feed horn and the whole input stage from scratch though... I am used to designing tuned vented speaker enclosures though, so the feed horn adjustment itself shouldnt be so hard... ;)

Jim: Thanks for your book suggestion, I have this one on a shelf already, I will give it a good whiff... Yes indeed down sampling somewhere along the way is necessary, probably via a PLL driven filter, I do however need time resolution down to about 5ns (it's not going to drive just a VFD forever... ;) so I guess down conversion to 200MHz should be ok. What will I lose along the way?

CountZero: You know, I thought about this for a while and what I feared was the need for higher power precision filter components if I filter after the LNA (which will have a gain of about +90dB), as opposed to filtering and separating, then amplifying each band separately. I will probably go back and forth on this. Amplifier will be done from GaAs FETs, and they are pretty pricey, which will probably dictate my final choice...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

"Designing digitally" and "implementing digitally" are somewhat different things.

Implementing filters at that frequency, or even at 20MHz, just isn't there at the current state of the technology. But, designing digitally (that is, using software to design an analog filter) is the norm, nowadays. I can provide some assistance in this area off-list. At 1420MHz, you won't be using capacitors and inductors, but lots of copper tubing and circuit board with, maybe, a few select variable capacitors for fine frequency adjustment. This will all change if you down-convert; if you have an IF in the 100-250MHz area, then lots of inductors and capacitors and stuff.

Your initial title included the term "down sampling". I really doubt that sampling is what you want to do. I good diode mixer (say, from MiniCircuits) and a local oscillator is far more preferable (with the current state of the technology) and a whole lot easier to do. To sample, you would need a sampling system with sample-pulse bandwidths approaching 10GHz and I suspect that is a rather long reach.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Down sampling was a misnomer you are right, I was meaning down conversion, but you get my drift... Hehehe.. I realize I have some reading to do, and I will probably fail miserably at my first attempts, but hey, isn't that what's fun? ;)

On the plus side, I am a quick learner, and I have access to the necessary plotting software. Do you have any suggestions of specialized software for designs in these frequency ranges? (I am used to something like Electronics workbench for standard simulation, but I also have the latest Mathematica to play with, still shrink wrapped... ;)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I recommend getting a (free) copy of Sonnet Lite. That will teach you a LOT about transmission lines on circuit boards, resonators, impedance matching, etc. Basically just go through the tutorials.

Another excellent source is the set of books from the American Radio Relay League (ARRL) such as "UHF/VHF Handbook" and others:

http://www.arrl.org/catalog/inde...

These have far more about construction of real hardware than all the other books combined.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

OT

If you (or anyone else interested in Radio Astronomy) is interested there is an electronic engineering position open here in Charlottesville, VA (ALMA project) and a software engineering position open in Socorro NM (ALMA also I believe). Neither position has made it to the website ( http://www.nrao.edu/administrati... ) yet but has been posted internally. I believe these positions have to be posted internally for 10 business days and then they go on the website. So, if interested check out the above website in a week or so. I am in no way involved in the interview or hiring process. I was just mentioning it here for the benefit of fellow freaks.

www.nrao.edu

www.alma.nrao.edu

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Just to be fair, sampling CAN be used as a down converter. That is how the newer radios that go "direct" from RF to digital do it, especially the ones that are called "SDR" (software defined radio).

You start with a numerically controlled oscillator (either PLL or DDS). From this, you create a very fast rise and fall, and very narrow (small relative to one period of the RF being sampled) pulse. You use this to drive a very fast sample gate. You do not have to sample every RF cycle, only fast relative to the modulation bandwidth. The result is a reconstructed carrier (at +/-Frf +/- N*Fsample) with the modulation preserved. For example, if your sample pulse rate is about 500MHz and the RF is 1420, then reconstructed carrier would be at 3*500MHz - 1420MHz or 60MHz.

This process is pretty amazing BUT the technology that is running now requires custom GaAs ICs or bleeding edge custom CMOS and is not for the casual hobbiest. There is a little on the market but it is for specific uses (cell phone, WiMax, etc).

So, that state of affairs still directs you to multiple conversion superheterodyne.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


(Click for full size image)

Did You mean something like this one ?

I got the disc as a donation. This specimen had been serving as data link over 100km span. The disc itself is light but the support structures - those are heavy iron.

I made some experiments with it and then my wife told me to get rid of it...

So I donated it to one of my collagues and now he is watching Astra2 with it. The Astra2 has it's main beam pointing to UK and we only get some really weak side beams here at finland.

It turned out to be a very good disk but because of that the alignment was really critical. That I knew in advance. So I welded a good support for my dear friend and now he can watch his BBC without any disturbances.

I also designed a rotary support for that disk but that plan did not walk out of it's closet. Pictures >>here<<

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
You start with a numerically controlled oscillator (either PLL or DDS). From this, you create a very fast rise and fall, and very narrow (small relative to one period of the RF being sampled) pulse. You use this to drive a very fast sample gate. You do not have to sample every RF cycle, only fast relative to the modulation bandwidth. The result is a reconstructed carrier (at +/-Frf +/- N*Fsample) with the modulation preserved. For example, if your sample pulse rate is about 500MHz and the RF is 1420, then reconstructed carrier would be at 3*500MHz - 1420MHz or 60MHz.

...

So, that state of affairs still directs you to multiple conversion superheterodyne.

This is very interesting stuff, I was just reading about superheterodyne receivers, it is a very simple and elegant design, I like it... So if use a PLL (or a couple) setup running off a reference crystal frequency and outputting 1420MHz sine, then mix it with the incoming signal, I have just added a carrier signal to the incoming analog signal. Now I take the crystal frequency and divide it to achieve a phase correct 5MHz, mix this with the 1420MHz reference, filter the 1415MHZ sine, mix with 1420MHz carrier with signal, I have just shifted my 2.5MHz band down to a 5MHz carrier, which once filtered I can feed to an ADC or a off the shelf spectrum analyser.

I'm sure this has missing pieces and gotchas, but does that oversimplistic explanation pretty much sums the objective?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Nice eskoila! ;) I will be using something more similar to this.

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Cool,
here is my back year with my two dishes:

JC

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

OK,

Just kidding. :D

That shot was taken at Disney a few years ago.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

OK, a little tutorial here....

A mixer functions, mathematically as a multiplier. That is, if there are two input signals, sin(F1*T) and sin(F2*T), you get sin(F1*T)*sin(F2*T). If you do the basic trig identity stuff, you get {sin[(F1+F2)*T] + sin(F1-F2)*T]}/2 (assuming F1 > F2). This is true whether you use a diode mixer or an active mixer. F1 + F2 and F1 - F2 are often called "sidebands" or "mixer products"; the latter is a little more appropriate in this case.

For the practicalities, lets assume that you are interested in a 2.5MHz wide band centered at 1420MHz (as you said in the original post). That means that the band extends from 1418.75 to 1421.25MHz. Suppose that you have a local oscillator at 1318.75MHz (100MHz below the lower edge of the band). Any signal at 1418.75MHz will be converted into 100MHz (the lower mixer product) or 2737.5MHz (the upper one).Likewise, a signal at 1421.25 will be converted to 102.5 or 2740MHz. From this, you can see that the band of interest has been shifted, intact (all the same modulation and frequency relationships) to these two new frequency bands.

If the mixer is followed by a filter to remove one of the two mixer products, then you have the signal translated to a single frequency. In this arrangement (local oscillator below the signal frequency), the band is translated downward with no inversion. This remaining frequency is called an "intermediate frequency" or IF.

The conventional arrangement is to put a filter and maybe an amplifier, depending on noise requirements, ahead of the mixer, and a second filter at the output of the mixer. In recent years, there has been a big improvement in understanding of what these filters need to do.

The input filter is GENERALLY not a very sharp filter. Whether or not a sharp filter is needed depends on adjacent signals. But, not for the reason you might think. A nearby signal, say 5MHz above the center frequency will get removed by the IF filter (which CAN be sharp because it is at a much lower frequency). But, a strong adjacent signal can overload the low noise amplifier (LNA) if you have one or the mixer. This can, in turn, cause non-ideal mixer behavior so that it no longer works the way it should. Sensitivity drops and noise level rises. However, the sharper you make the input filter, the more loss it has, and that wrecks your noise figure. You basically can't win; you can only try to make things least bad!!!

As described, above, the mixer is followed by a filter. This is called the IF filter. If there is more than one mixer and intermediate frequency, then they are called "First IF" and "Second IF" and so forth. In very recent years, it has been understood that best performance (in this case, often meaning noise figure and sensitivity) is achieved if you pay attention to the apparent impedance seen by the high mixer product. More about this later.

This is probably enough to digest for now. I'll be happy to add to it as needed.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I would like to add a further recommendation.

On this first try, don't attempt to design your own LNA and mixer. You can get modules from places like MiniCircuits. Put your effort into filters and circuit boards and power supplies and local oscillators (for the front end, anyway). With your probable lack of good test equipment (such as spectrum analyzers and signal sources), it would be an exercise in frustration to try to do the active parts.

If you make this as a series of interconnected circuit boards, then you can come back at a later time and do your "best in class" front end. But, for starters, use modules. DO make sure that signal paths between boards use coax.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Unixwhore,

I'm tinkering with my first real circuit doing analog at 10 MHz and struggling some, (OK, struggling lots). Venturing into the GHz range will, I imagine, be quite an undertaking.

I found these two books helpful:

Basic Communications Electronics
by Jack Hudson and Jerry Luecke
Master Publishing, Inc
copywrite 2005, (ie pretty recent)

This is a large, 8.5" x 11" format, large text, big diagrams, back to basics with easy to understand explanations and examples.

Practical RF Design Manual
by Doug DeMaw W1FB
MFJ Enterprises, Inc
Second Printing, 1997
ISBN: # 1-891237-00-4
MFJ-3507

This is another RF reference, aimed at Hams, with circuits. The circuits are generally transistor based with a few op-amps, it does not include recent RF chips and chips with integrated mixers, if strips, etc.

Hope these help.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

On this first try, don't attempt to design your own LNA and mixer. You can get modules from places like MiniCircuits. Put your effort into filters and circuit boards and power supplies and local oscillators (for the front end, anyway). With your probable lack of good test equipment (such as spectrum analyzers and signal sources), it would be an exercise in frustration to try to do the active parts.

I do have signal generators and DSO with Fourier transform functions available for up to 60mhz bandwidth, which should do fine after down conversion... I can also borrow some equipment from a friend that works at Marconi... I do get your point though, but the expected frustration is the whole point for my dabbling in there ;) This is my first high frequency venture (>1GHz) and i expect to learn lots.

Quote:

If you make this as a series of interconnected circuit boards, then you can come back at a later time and do your "best in class" front end. But, for starters, use modules. DO make sure that signal paths between boards use coax.

That was the plan, I have a bunch of small cast aluminium enclosures and a lot of panel mount BNC connectors as well as a roll of RG6 lying in a corner... ;)

DocJC: Thanks!

Fortunately I have some experience figuring out audio crossover networks and the likes, and also have worked plenty with high frequency digital signals, but as far as RF goes closest I got is a short range wireless UART over a 915MHz link... But then again, this should be technically simpler, as there is no demodulation or decoding to do, I am looking for NOISE.

Anyways, still reading and slowly putting the pieces together, I'll figure something out soon enough...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Here's a question: since the signal I am looking for doesn't have a carrier frequency as it is noise, I MUST add one in before I can down convert yes? This is not entirely clear to me, although the principle seems to suggest it is. Or could I simply add in a 5MHz carrier right away (since RF signal has no carrier, 0MHz RF + 5MHz IF = 5MHz output carrier)?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No carrier gets added.

That "noise" is just a very broad signal. It will convert quite nicely. But, be aware that any other noise will also. This means LNA noise or noise "created" by lossy filters or a mixer. You will be even less able to distinguish one from the other than in a "normal" communications receiver.

Further, while your antenna filter will reduce susceptibility to out of band signal coming in through the antenna, it will do nothing for noise generated by the preamp. This gets us into the problem of images. There is another signal that will make it through the mixer and IF filter. It is 1x IF frequency on the other side of the local oscillator. So, if your LO is 1318.75MHz, as I suggested earlier (by the way, there is NOTHING magic about this, I simply chose it for simple arithmetic), then a signal in the range of 1218.75 to 1216.5 will also make it through the IF filter. Guess what, the LNA, being a broad band device, also makes noise there. You mixer cannot tell this noise from other noise, and it converts. This is referred to as "image noise". For this reason, alone, you need a filter between the LNA and the mixer to nock down this noise. Clearly, this filter does not have to be very sharp, but, without it, the noise from the LNA will be doubled.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
For this reason, alone, you need a filter between the LNA and the mixer to nock down this noise. Clearly, this filter does not have to be very sharp, but, without it, the noise from the LNA will be doubled.

I would suggest putting most of the 1420mhz selectivity *before* the lna. Keeping the off-channel noise, etc. from reaching the lna will increase the system gain and give the best image rejection. A multi-cavity filter is easy to make and you should be able to obtain a nicely defined 5mhz passband without too much trouble.

edit: goof

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Eski, That is very nice!! Over my head...perhaps as far as the dish is capable of but, nice nonetheless!!

John

Just some guy

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I agree with Tom. At that frequency, filters will need to be cavities and such in order to get sufficient Q.

The filter between the LNA and the mixer does not need to be sharp. If it is down by 10db or so at the image frequency, things should be good. The challenge will be to get reasonable Q (here, to reduce losses rather than to make a sharp filter) in a space that does not require lots of cables (that just add more loss). All that loss just kills the noise figure.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ok cool.. As for the noise issue, let me first explain what I *think* I set my mind on... (feel free to change it... ;))

First, I will have a +10dB@1420MHz LNA, feeding a balanced mixers setup, itself fed clock from a PLL setup outputting a 1413.25MHz. The clock generator will most likely be in a separate shielded enclosure, and I plan to make all interconnections balanced via coax to minimize further noise injection. The output of the mixers will go through a second order inverse Chebyshev (6dB/octave, smooth passband, gentle roll off, steep skirt, 0.01dB ripple on passband, 3dB ripple on stopband), good bandpass filter centered on 6.75MHz with a -3dB bandwidth of 2.5MHz (If I am not mistaken this means a needed filter Q of 2?) @ 6.75MHz, and then into a 3dB attenuator to eliminate the stopband ripple. From there into balanced opamp setup to bias it to DC then into an ADC... ;)

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
From there into balanced opamp setup to bias it to DC then into an ADC...

I'm following along for my own learning, not because I have experience in this, but just to be clear, doesn't this (single/multiple) op amp stage have to incorporate a precision AM detector? This is not the same as just a bias offset.

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This came out wrong... lol.. Sorry this is the consequence of doing 4 things at the same time..

"Balanced" is in reference with cabling impedance between modules. Noise substraction via differential amplifier, additional 3dB gain...

And I now realize I completely omitted my narrow band filters for the 8 bands... Hehehe...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
The output of the mixers will go through a second order inverse Chebyshev (6dB/octave, smooth passband, gentle roll off, steep skirt, 0.01dB ripple on passband, 3dB ripple on stopband), good bandpass filter centered on 6.75MHz with a -3dB bandwidth of 2.5MHz (If I am not mistaken this means a needed filter Q of 2?) @ 6.75MHz, and then into a 3dB attenuator to eliminate the stopband ripple.

This seems too low of an IF frequency, your image rejection will be very poor. Also, remember "gain bandwidth product"! Your IF bandpass filter has little selectivity as described, and your IF gain will be low as a result. The IF filter should have steep skirts for good system performance. Same thing for the 8 sub band filters.

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Would getting down to this IF via 3 stages be better?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Would getting down to this IF via 3 stages be better?

I don't think you have to go *that* far, but your first IF conversion should be perhaps at least 50 mhz for good image and LO rejection. I have a gps that uses single conversion to a 70mhz IF, for example, which then goes straight into the a/d.

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Mind you I could go a bit higher, but I would like to stay below 60MHz, so I can test the output around with the DSO I have on hand...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ah ok well 50MHz it will be... Let's see thats a needed filter Q of around 20 yes?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It was mentioned using a "3db pad to reduce filter ripple". Sorry, it does not work that way! The ripple will be exactly the same whether terminated in a 50 ohm load or a 50 ohm load preceded by an attenuator. No diff, ripple-wise.

Yes, an IF below 10MHz is way too low. I'd choose something in the 40-50MHJz area. That way, the image will be around 80MHz away and you can still do a reasonable filter between the LNA and the mixer to reduce the LNA's image noise. And, you have hope for a reasonable cavity antenna filter. You should be able to get reasonable IF gain at that frequency, also, and the components are not very big, physically.

I would also consider dual conversion. Use a 1st IF that is 2.5MHz wide and get some gain out of it. Then, follow that with your sub-band mixers. Those 2nd LOs are not nearly as critical (in this application) and the filters with a few hundred KHz BW are easy to do.

You won;t be able to use an AVR ADC. The required sample rate is at least 2X the highest frequency component and that will have to be at least a few hundred KHz. You CAN, however, use active filters at these frequencies.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

But the big question is - wouldn't it be simpler just to get a data packet from Seti?

Seeing as you're slap in the middle of the hydrogen band, I'm assuming that you're hoping for beacons?

Or you could just put your ear at the focus and listen very very hard? :mrgreen:

Neil

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Simpler, sure. As fun and rewarding? NOT! ;)

Quote:

It was mentioned using a "3db pad to reduce filter ripple". Sorry, it does not work that way! The ripple will be exactly the same whether terminated in a 50 ohm load or a 50 ohm load preceded by an attenuator. No diff, ripple-wise.

An Inverse Chebyshev response curve has a clean bandpass and a 1-3db ripple with zeroes on the bandstop. Stop band attenuation at 50MHz should be around 30dB. You are telling me that using a 3dB attenuator will not "lose" the injected 1-3db ripple on the bandstop? Isn't an attenuator the exact opposite of an amp, adding "negative gain", or loss to the signal?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

What kind of measurement stuff have you available ?
I believe designs in that area require at least
a network-analyser and a spectrum-analyser.
If it comes to low-noise designs I think this is
very ambitious.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The attenuator would just subtract 3db from everything. Same transfer function, just shifted downward by 3db. An amplifier would shift it upward. Poles and zeros stay in the same location and and is what defines the ripple.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ossi wrote:
What kind of measurement stuff have you available ?
I believe designs in that area require at least
a network-analyser and a spectrum-analyser.
If it comes to low-noise designs I think this is
very ambitious.

For the moment my tools are a good calculator, a bunch of books, and a pen and paper... ;) When the time comes I will have what I need...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Good low noise transistors are not so expensive: http://www.mouser.com/Search/Ref...

But as others have said start with something easier first. Hunting for oscillations in your LNA is perhaps not the first thing you want to do. These have very high gain at these low frequencies.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0