>
> Even in a perfectly balance mixer, you will have a strong third order
product. > A mixer is a multiplier, period. Check the mathematics. Using a
rectangular > waveform results in additional products because this waveform consists of
a > chain of harmonic components. In a perfect square wave, you will have only
the > odd order harmonics and with the imperfect waveform you will have
increasing > amounts of even order products.
>
It's the exact same characteristic (squarewave LO harmonics) that causes the excess NF penalty. It not for this, the NF for a diode ring mixer would be somewhere around 3dB (equal power split to f1+f2 and f1-f2). I managed to get the NF of my active mixers down to this level with much better IMD performance, but the squarewave LO signal still incurred a NF penalty. Chris ,----------------------. High Performance Mixers and / What's all this \ Amplifiers for RF Communications / extinct stuff, anyhow? / \ _______,--------------' Chris Trask / N7ZWY _
| 3020
| 2009-05-12 07:30:17
| Chris Trask
| Re: Mixer Purity
|
>
> It's the exact same characteristic (squarewave LO harmonics) that
causes > the excess NF penalty. It not for this, the NF for a diode ring mixer
would > be somewhere around 3dB (equal power split to f1+f2 and f1-f2). I managed
> to get the NF of my active mixers down to this level with much better IMD
> performance, but the squarewave LO signal still incurred a NF penalty.
>
I didn't word that properly. I should have said that I got the NF of my active mixers down to that of a diode ring mixer (6dB). I managed to do this with just under 0dBm of LO power, but the IMD performance was comparable to that of a diode ring mixer using +17dBm of LO power. Chris ,----------------------. High Performance Mixers and / What's all this \ Amplifiers for RF Communications / extinct stuff, anyhow? / \ _______,--------------' Chris Trask / N7ZWY _
| 3021
| 2009-05-12 15:15:43
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Chris Trask schrieb: > I didn't word that properly. I should have said that I got the NF of my
> active mixers down to that of a diode ring mixer (6dB). I managed to do
> this with just under 0dBm of LO power, but the IMD performance was
> comparable to that of a diode ring mixer using +17dBm of LO power.
>
Hm. I just read that image and harmonics downconverted will increase NF considerably especially in Gilberts. But doing this will reduce the need for a good input filter. And read reducing LO drive will increase NF. So one must balance both for performance. As an intermezzo: What do you think about the combination of JFET+SA602+JFET as mixer+IF -combo? I extracted the circuit diagram detail here: http://www.ehydra.dyndns.info/tech/emrfd/JFET_SA612_JFET.gif The origin doc is here: http://www.mydarc.de/df1fo/DOC/fjrx84.doc In the end I ask myself when I replace for example a SA602 with one of your designs, will I find a noticable performance improvement? regards - Henry -- www.ehydra.dyndns.info |
| 3022
| 2009-05-12 17:18:22
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
ehydra wrote: >
> Chris Trask schrieb:
>> I didn't word that properly. I should have said that I got the NF of my
>> active mixers down to that of a diode ring mixer (6dB). I managed to do
>> this with just under 0dBm of LO power, but the IMD performance was
>> comparable to that of a diode ring mixer using +17dBm of LO power.
>>
>
> Hm. I just read that image and harmonics downconverted will increase NF
> considerably especially in Gilberts.
> But doing this will reduce the need for a good input filter.
>
> And read reducing LO drive will increase NF.
>
> So one must balance both for performance.
There's nothing inherently "noise adding" with an ideal +/- 1 gating mixer. That is little more than a mixer made from zero-ohm DPDT switches, and is what a diode ring mixer approaches with high drive level. The higher noise figure happens only because of its below 0dB gain at the fundamental LO conversion. The rest of the signal power is lost at harmonic outputs. No extra "noise" is converted down to the IF output due to harmonic input mixing, unless the signal source contains wideband noise. With active mixers such as Gilbert Cells, a source of wideband noise is already available from the transistors within the mixer, so noise at LO harmonics can also appear at the IF output. > As an intermezzo: What do you think about the combination of
> JFET+SA602+JFET as mixer+IF -combo?
> I extracted the circuit diagram detail here:
> http://www.ehydra.dyndns.info/tech/emrfd/JFET_SA612_JFET.gif
>
> The origin doc is here:
> http://www.mydarc.de/df1fo/DOC/fjrx84.doc
>
>
>
> In the end I ask myself when I replace for example a SA602 with one of
> your designs, will I find a noticable performance improvement? |
| 3023
| 2009-05-12 17:55:15
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Hi Russell - Maybe I misunderstood the AppNote from Agilent. I don't know. http://origin-eesof.tm.agilent.com/pdf/mixer_noise_figure.pdf Pages 17 and 20 So Chris would certainly say, we need Gilberts with inside filtering. - Henry -- www.ehydra.dyndns.info Russell Shaw schrieb: > ehydra wrote:
>> Chris Trask schrieb:
>>> I didn't word that properly. I should have said that I got the NF of my
>>> active mixers down to that of a diode ring mixer (6dB). I managed to do
>>> this with just under 0dBm of LO power, but the IMD performance was
>>> comparable to that of a diode ring mixer using +17dBm of LO power.
>>>
>> Hm. I just read that image and harmonics downconverted will increase NF
>> considerably especially in Gilberts.
>> But doing this will reduce the need for a good input filter.
>>
>> And read reducing LO drive will increase NF.
>>
>> So one must balance both for performance.
>
> There's nothing inherently "noise adding" with an ideal +/- 1 gating
> mixer. That is little more than a mixer made from zero-ohm DPDT switches,
> and is what a diode ring mixer approaches with high drive level.
>
> The higher noise figure happens only because of its below 0dB gain at
> the fundamental LO conversion. The rest of the signal power is lost at
> harmonic outputs. No extra "noise" is converted down to the IF output
> due to harmonic input mixing, unless the signal source contains wideband
> noise.
>
> With active mixers such as Gilbert Cells, a source of wideband noise
> is already available from the transistors within the mixer, so noise
> at LO harmonics can also appear at the IF output.
> |
| 3024
| 2009-05-12 18:08:57
| Chris Trask
| Re: Mixer Purity, JFET+SA602+JFET
|
>
> Maybe I misunderstood the AppNote from Agilent. I don't know.
> http://origin-eesof.tm.agilent.com/pdf/mixer_noise_figure.pdf
> Pages 17 and 20
>
> So Chris would certainly say, we need Gilberts with inside filtering.
>
No, I wouldn't. The load impedance is far too low and the source impedance much too high to accomodate any filtering. Someone tried that a few years ago and the results were not worth repeating. Chris ,----------------------. High Performance Mixers and / What's all this \ Amplifiers for RF Communications / extinct stuff, anyhow? / \ _______,--------------' Chris Trask / N7ZWY _
| 3026
| 2009-05-13 11:56:09
| Allison Parent
| Re: Mixer Purity, JFET+SA602+JFET
|
| 3029
| 2009-05-13 13:26:05
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Allison Parent schrieb: >> And read reducing LO drive will increase NF.
>>
>> So one must balance both for performance.
>
> Most gilbert cell style mixers have a rather limited range of performance and
> for overload and noise. most of the integrated ones are optimized for low power.
> At HF external environmental noise is less a factor than overload performance.
>
Yes. Maybe I pronounced to less that I'm primary interested in mobile/low-current apps. If we define a figure-of-merit for mixers as FOM=IMD/Icc : Defined as such, the SA602 makes a good value!? I would like to keep the discussion on-going up to the point where we find a useful overview as such: With parameters a,b,c then use mixer-type X. >> As an intermezzo: What do you think about the combination of
>> JFET+SA602+JFET as mixer+IF -combo?
>> I extracted the circuit diagram detail here:
>> http://www.ehydra.dyndns.info/tech/emrfd/JFET_SA612_JFET.gif
>>
>> The origin doc is here:
>> http://www.mydarc.de/df1fo/DOC/fjrx84.doc
>>
>
> My pass at it is there is opportunity for overload there. The 612/602 is
> a low dynamic range device (at 8ma it's to be expected) designed for
> 2nd converter use or less demanding applications. So between it's
> moderate noise figure, good conversion gain but limited large signal
> handling it's excellent for potable use where battery life is criteria. It's
> a great experimenters part.
>
Chris noticed that Gilberts have poor NF. As I read it depends on wether the Gilbert Mixer has pre-distortion (=linearizing) interals. Whereas the DBM is a attentuator only and the Gilbert has gain, can we compare both if the gain is substracted for IMD? > I use the part when power use and simplicity are more important than
> dynamic range. Te only caveat is applications where the input bandwidths
> and other more tightly defined parameters such as signal levels are controllable
> such as a product detector after a gain controlled IF.
>
>> In the end I ask myself when I replace for example a SA602 with one of
>> your designs, will I find a noticeable performance improvement?
>
> Depends.. If your building a RX that beats commercial designs you will notice.
> If your experimenting or working toward low power portable then unknown.
> Depends on your goals for performance.
>
Q: You recognized that the first two JFETs and the JFET at the output are all controlled by VCC ("Abschwächer" means RSSI)? - Henry -- www.ehydra.dyndns.info |
| 3030
| 2009-05-13 23:01:44
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Chris Trask schrieb: >> So Chris would certainly say, we need Gilberts with inside filtering.
>>
>
> No, I wouldn't. The load impedance is far too low and the source
> impedance much too high to accomodate any filtering. Someone tried that a
> few years ago and the results were not worth repeating.
>
Chris, what do you think about this patent. As I read it first, I scanned it for 'Trask', really! ;-) http://www.ehydra.dyndns.info/tech/emrfd/pat5379457.pdf regards - Henry -- www.ehydra.dyndns.info |
| 3032
| 2009-05-14 06:44:30
| Chris Trask
| Re: Mixer Purity, JFET+SA602+JFET
|
>
>
>>> So Chris would certainly say, we need Gilberts with inside filtering.
>>>
>>
>> No, I wouldn't. The load impedance is far too low and the source
>> impedance much too high to accomodate any filtering. Someone tried that a
>> few years ago and the results were not worth repeating.
>>
>
>Chris, what do you think about this patent. As I read it first, I
>scanned it for 'Trask', really! ;-)
>
>http://www.ehydra.dyndns.info/tech/emrfd/pat5379457.pdf
>
It does not solve the NF problem of the basic mixer core, which is due to the wideband noise of the driver transistor and the subsequent spectral folding. Chris ,----------------------. High Performance Mixers and / What's all this \ Amplifiers for RF Communications / extinct stuff, anyhow? / \ _______,--------------' Chris Trask / N7ZWY _
| 3035
| 2009-05-16 08:07:20
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Hi Chris - Meanwhile I googled around and found that message: http://qrp.kd4ab.org/2003/030316/0019.html So why not test it against SA602 ? Is the augmented version the latest and the diagram ready? I like to solder one together. What is the relationship with phase noise of LO ? For testing it, do I need a very clean LO? I have a Si571 but I'm not sure if it is good enough. regards - Henry -- www.ehydra.dyndns.info Chris Trask schrieb: >>
>>>> So Chris would certainly say, we need Gilberts with inside filtering.
>>>>
>>> No, I wouldn't. The load impedance is far too low and the source
>>> impedance much too high to accomodate any filtering. Someone tried that a
>>> few years ago and the results were not worth repeating.
>>>
>> Chris, what do you think about this patent. As I read it first, I
>> scanned it for 'Trask', really! ;-)
>>
>> http://www.ehydra.dyndns.info/tech/emrfd/pat5379457.pdf
>>
>
> It does not solve the NF problem of the basic mixer core, which is due to the wideband noise of the driver transistor and the subsequent spectral folding.
> |
| 3036
| 2009-05-16 08:59:00
| Chris Trask
| Re: Mixer Purity, JFET+SA602+JFET
|
>
>Meanwhile I googled around and found that message:
>http://qrp.kd4ab.org/2003/030316/0019.html
>
>So why not test it against SA602 ?
>
It would be a waste of time, like comparing a tank to a skateboard. Chris ,----------------------. High Performance Mixers and / What's all this \ Amplifiers for RF Communications / extinct stuff, anyhow? / \ _______,--------------' Chris Trask / N7ZWY _
| 3037
| 2009-05-16 09:29:24
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Hm. I don't understand how do you mean it! What is the tank, what is the skateboard? Or is it just personally? - Henry -- www.ehydra.dyndns.info Chris Trask schrieb: >> Meanwhile I googled around and found that message:
>> http://qrp.kd4ab.org/2003/030316/0019.html
>>
>> So why not test it against SA602 ?
>>
>
> It would be a waste of time, like comparing a tank to a skateboard. |
| 3038
| 2009-05-16 09:42:20
| Allison Parent
| Re: Mixer Purity, JFET+SA602+JFET
|
| 3039
| 2009-05-16 09:43:14
| Allison Parent
| Re: Mixer Purity, JFET+SA602+JFET
|
| 3040
| 2009-05-16 10:12:32
| bobtbobbo
| Re: Mixer Purity, JFET+SA602+JFET
|
I agree with Allison suggesting the AD831. I have been very satisfied with them. One key point is to keep the VFO level not more than -25 dBM. Over that and you will get birdies. Certainly its performance may not be as good as a switch-mode mixer like the one in the Elecraft K3, however the surrounding circuitry is much simpler. A problem, for some, is that the AD831 comes in a 20 pin PLCC package, slightly more difficult to work with physically than a DIP part. That aside, the results are rewarding. If you decide to use a single power supply I would recommend 9 volts. On 10 volts I have had overheating problems. Because of its limitations I have only used the SA602 for product detectors. Good luck with your experiments!
Bob, K1AO
|
| 3041
| 2009-05-16 10:45:24
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Hi Allison - Thank you very much for your lengthy explanation! Is there any substantial relaxation if I go to higher bands and using a relative low-gain and small bandwidth antenna? - Henry -- www.ehydra.dyndns.info Allison Parent schrieb: > [..] |
| 3042
| 2009-05-16 10:53:53
| Chris Trask
| Re: Mixer Purity, JFET+SA602+JFET
|
>
> >
> > > Meanwhile I googled around and found that message:
> > > http://qrp.kd4ab.org/2003/030316/0019.html
> > >
> > > So why not test it against SA602 ?
> > >
> >
> > It would be a waste of time, like comparing a tank to a skateboard.
>
> Hm. I don't understand how do you mean it!
> What is the tank, what is the skateboard?
>
> Or is it just personally?
>
I think I just found the problem. Chris ,----------------------. High Performance Mixers and / What's all this \ Amplifiers for RF Communications / extinct stuff, anyhow? / \ _______,--------------' Chris Trask / N7ZWY _
| 3043
| 2009-05-16 11:03:29
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Would it be ok for you that I forward that to Nick, the author of the circuit? If so, I will do that and translate his response for here. regards - Henry Allis
| 3044
| 2009-05-16 11:05:17
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
bobtbobbo schrieb: > I agree with Allison suggesting the AD831. I have been very satisfied with them. One key point is to keep the VFO level not more than -25 dBM. Over that and you will get birdies. Certainly its performance may not be as good as a switch-mode mixer like the one in the Elecraft K3, however the surrounding circuitry is much simpler. A problem, for some, is that the AD831 comes in a 20 pin PLCC package, slightly more difficult to work with physically than a DIP part. That aside, the results are rewarding. If you decide to use a single power supply I would recommend 9 volts. On 10 volts I have had overheating problems. Because of its limitations I have only used the SA602 for product detectors. Good luck with your experiments!
>
> Bob, K1AO
>
Thank you Bob! The AD831 is a high-power device not very useful for a mobile circuit. Otherwise that may be is an option. - Henry -- www.ehydra.dyndns.info |
| 3045
| 2009-05-16 11:12:15
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
I don't have a glue where the problem is. Surely it is not technical. Can I ask how old are you? I'm 43 years old. - Henry Chris Trask schrieb: >>>> Meanwhile I googled around and found that message:
>>>> http://qrp.kd4ab.org/2003/030316/0019.html
>>>>
>>>> So why not test it against SA602 ?
>>>>
>>> It would be a waste of time, like comparing a tank to a skateboard.
>> Hm. I don't understand how do you mean it!
>> What is the tank, what is the skateboard?
>>
>> Or is it just personally?
>>
>
> I think I just found the problem.
>
> Chris
>
> ,----------------------. High Performance Mixers and
> / What's all this \ Amplifiers for RF Communications
> / extinct stuff, anyhow? /
> \ _______,--------------' Chris Trask / N7ZWY
> _
| 3046
| 2009-05-16 11:22:20
| Chris Trask
| Re: Mixer Purity, JFET+SA602+JFET
|
>
> >
> > >
> > > >
> > > > > Meanwhile I googled around and found that message:
> > > > > http://qrp.kd4ab.org/2003/030316/0019.html
> > > > >
> > > > > So why not test it against SA602 ?
> > > > >
> > > > It would be a waste of time, like comparing a tank to a
skateboard. > > > Hm. I don't understand how do you mean it!
> > > What is the tank, what is the skateboard?
> > >
> > > Or is it just personally?
> > >
> >
> > I think I just found the problem.
> >
> I don't have a glue where the problem is. Surely it is not technical.
>
> Can I ask how old are you? I'm 43 years old.
>
Sorry, you're not my type. Chris ,----------------------. High Performance Mixers and / What's all this \ Amplifiers for RF Communications / extinct stuff, anyhow? / \ _______,--------------' Chris Trask / N7ZWY _
| 3047
| 2009-05-16 11:47:16
| Thomas S. Knutsen
| Re: Mixer Purity, JFET+SA602+JFET
|
Hello Henry. The bias pin on AD831 is used to adjust the current through the mixer. I have done some experiments with AD831 and found it to be the ideal mixer when need for an Gilbert cell. Best 73's de Thomas LA3PNA. 2009/5/16 ehydra < ehydra@arcor.de> >
>
> The AD831 is a high-power device not very useful for a mobile circuit.
> Otherwise that may be is an option.
>
> - Henry
>
>
[Non-text portions of this message have been removed] |
| 3048
| 2009-05-16 12:03:02
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Chris Trask schrieb: >> I don't have a glue where the problem is. Surely it is not technical.
>>
>> Can I ask how old are you? I'm 43 years old.
>>
>
> Sorry, you're not my type.
>
> Chris
OK Chris. I don't understand why but I will respect it. Sorry. - Henry -- www.ehydra.dyndns.info |
| 3049
| 2009-05-16 12:14:41
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Thomas S. Knutsen schrieb: > Hello Henry.
> The bias pin on AD831 is used to adjust the current through the mixer. I
> have done some experiments with AD831 and found it to be the ideal mixer
> when need for an Gilbert cell.
>
Hello Thomas - Oh. Then I must reread the DS. That will left price and single-sourced. It is hard to get control of rf technics if one is coming from the digital domain. But at least my first design works better than most circuits from China. So I'm satisfied at the moment ;-) Thank you! - Henry -- www.ehydra.dyndns.info |
| 3050
| 2009-05-16 12:31:17
| Allison Parent
| Re: Mixer Purity, JFET+SA602+JFET
|
| 3051
| 2009-05-16 12:34:24
| Allison Parent
| Re: Mixer Purity, JFET+SA602+JFET
|
| 3052
| 2009-05-16 12:51:16
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Allis
| 3053
| 2009-05-16 12:55:53
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Allison Parent schrieb: >> The AD831 is a high-power device not very useful for a mobile circuit.
>> Otherwise that may be is an option.
>>
>
> Mobile covers a lot of ground. Usual use here means auutomotive battery power and that is considerable. If you mobile as in
> battery portable hand held then the dimensions change.
>
> However without a statement of what the needs are in a clear way
> all suggestions have the same value, none.
>
Yes Allison! I'm thinking of a handheld device with NiMH or Li-battery. But that can easily change if I find a new field of using the design. Part of it even works as a very capable power-line modem. The nice thing about such a design philosophy is that I can modularize it and connect it to a new system. But read my other post for birds... - Henry -- www.ehydra.dyndns.info |
| 3054
| 2009-05-16 19:11:56
| Allison Parent
| Re: Mixer Purity, JFET+SA602+JFET
|
| 3055
| 2009-05-18 11:21:11
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Allison Parent schrieb: >> I'm thinking of a handheld device with NiMH or Li-battery. But that can
>> easily change if I find a new field of using the design.
>> Part of it even works as a very capable power-line modem. The nice thing
>> about such a design philosophy is that I can modularize it and connect
>> it to a new system.
>>
>> But read my other post for birds...
>>
>> - Henry
>>
>
> Understood, now it is clear. Yes, several things will aid you, increasing frequency helps as the transmitting antennas can be more efficient and it's possible to use a directional antenna on receive.
> It also sounds like FM is the planned modulation and that works in your favor as well. For that the 612 is fine and possibly the better
> part to use or something similar such as some of the complete radio
> on chip systems. Some of the more highly integrated radios offer significant power savings over a discrete design.
>
Thanks Allison! - Henry -- www.ehydra.dyndns.info |
| 3074
| 2009-05-21 11:42:23
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
|
|
| 3077
| 2009-05-21 19:44:03
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
ehydra wrote: > To continue the mixer discussion, here I found an interesting statement
> that active mixers tolerate more LO phase noise than passives:
>
> http://embedded.eecs.berkeley.edu/Alumni/mehrotra/ps/thesis.pdf
>
> page 114
There are no significant qualifications to those assertions (unless i read the whole thesis to get a clear picture). Gilbert cells and diode rings have a similar switching/mixing action. They differ most in second-order characteristics such as transistor Cbc capacitances acting as varactors and generating extra distortion. Blanket statements using "noise figure" can be misleading. Imagine a room-temp Gilber cell mixer connected to a low-noise source such as a radio telescope dish pointed to a quite area of space. The added noise from the Gilbert cell can be 1000x more than the noise floor from the source, whereas with a diode ring of the same noise figure, the added noise would be a fraction of that. The reason for this difference is that for a diode ring connected to a room-temp source (in a noise-figure test), most of the output noise is that from the input source. -- Russell Shaw, B.Eng, M.Eng(Research) Thesis: "Analysis and Design of Radio Frequency Mixer Circuits". (You won't find my noise-figure analysis there, because i figured out all this stuff, but it didn't get included for various non- technical reasons. I haven't found these statements in the literature, but i was concentrating more on HF than microwave at the time) |
| 3081
| 2009-05-22 11:07:49
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Hi Russell - Before I answer you: "Thesis: "Analysis and Design of Radio Frequency Mixer Circuits" I tried to read your paper but cannot find your pages at the university?? Strange. http://www.rmit.edu.au/browse?STYPE=PEOPLE&QRY=%2Brussell+%2Bshaw&CLOCATION=&Search.x=8&Search.y=9 Is it possible to email it? regards - Henry -- www.ehydra.dyndns.info Russell Shaw schrieb: > ehydra wrote:
>> To continue the mixer discussion, here I found an interesting statement
>> that active mixers tolerate more LO phase noise than passives:
>>
>> http://embedded.eecs.berkeley.edu/Alumni/mehrotra/ps/thesis.pdf
>>
>> page 114
>
> There are no significant qualifications to those assertions (unless i
> read the whole thesis to get a clear picture).
>
> Gilbert cells and diode rings have a similar switching/mixing action.
>
> They differ most in second-order characteristics such as transistor
> Cbc capacitances acting as varactors and generating extra distortion.
>
> Blanket statements using "noise figure" can be misleading.
>
> Imagine a room-temp Gilber cell mixer connected to a low-noise
> source such as a radio telescope dish pointed to a quite area of space.
>
> The added noise from the Gilbert cell can be 1000x more than the
> noise floor from the source, whereas with a diode ring of the same
> noise figure, the added noise would be a fraction of that.
>
> The reason for this difference is that for a diode ring connected to
> a room-temp source (in a noise-figure test), most of the output noise
> is that from the input source.
> |
| 3082
| 2009-05-22 20:34:14
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
ehydra wrote: > Hi Russell -
>
> Before I answer you:
>
> "Thesis: "Analysis and Design of Radio Frequency Mixer Circuits"
>
> I tried to read your paper but cannot find your pages at the
> university?? Strange.
> http://www.rmit.edu.au/browse?STYPE=PEOPLE&QRY=%2Brussell+%2Bshaw&CLOCATION=&Search.x=8&Search.y=9
>
> Is it possible to email it?
I couldn't find it on the RMIT web site. It was done in 1994, and i know that it was copied on to a CD back then. I'll email it. You won't find any worthwhile useful break-throughs in it. If i was supervising myself now, i'd tell me to study up more on noise figure and really understand what the results mean in terms of added noise in a system when the noise temperatures are way below 273K. Then i'd tell me to study up on bezier polynomials, splines, and stuff involving optimal curve-fitting. Then i'd guide myself on writing circuit simulators. Obscure mathematics outside the scope of the supervisors expertise and the course was one of main problems, as was computer and programming expertise. I did the thesis after a 4 year B.Eng degree. Although i was way ahead of everyone else in the course when it came to RF and analog, i still had quite a way to go to fully comprehend every detail needed to do decent analog design. I'd been playing with, fixing, and studying radios and audio stuff from when i was 10 or less, and was very self-taught before starting uni. I wouldn't consider my own thesis to be very worthwhile, and didn't then when i finished it. The results i got didn't correlate well with predictions anyway. I think the overall procedure would work if implemented better. The main idea was this: Measure the active mixing elements (transistors) and determine a 2D or 3D polynomial that fits closely, using an optimization program that fit the polynomial to a spice model that i determined from measurements. Using the polynomial coefficients, one can then determine the mixer distortion products using a closed-form formula. Seems simple. Problems i had is that quite high order polynomials were needed for the whole Vce-Ic-Vbe characteristics. Had i known about splines and piece-wise joining of polynomials, i would have applied that. I only had a slow 386SX pc with an extra $200 maths coprocessor plugged in. I lived 80km from the uni and it was a hassle going everyday just to do a few measurements, and running students labs. I only had windoze and DOS, so i had limited computer ability (my supervisor had none, other than being quite good at running spice and doing spice models). I only knew turbo-pascal because that's all that was taught to comm-eng students then. Despite that, i wrote the core of a spice simulator that computed the gummel-poon DC model so that i could make my polynomials give the same transfer characteristic as spice would. Ignoring capacitances limits verification to lower HF frequencies only. Spice was not good for determining third-order mixer products because of numerical noise. It's really the wrong tool, but that's all we had then. The whole idea of predicting distortion products from polynomials was my own. I didn't find it in other literature at the time. An idea i had since then is this: Measure the IF output amplitude Vs RF input amplitude. Determine a transfer function using a polynomial that fits this. From the coefficents, one could then determine 3rd-order distortion products using a simple formula. Although it sounded like a waste of time, i got far more value out of the spare time i spent thinking about all kinds of analog and RF design and reading all the interesting stuff in the library. I did the masters at a time just when a major recession hit and there were no real jobs. It was an interesting time when everyone was saying analog is dead and everything will be digital. 99% of the students did all the digital electives and out of 200 i was the only one heavily into analog. I just didn't understand how CPUs and computers worked well, even though i'd been through that part of the course. I hated them, seeing them as useless, until i figured out you could do useful number crunching for designing analog stuff. I only really got heavily into decent unix programming in the last 5 years. I'd missed the whole early-computing era, and stood on the outside seeing everyone spending their time hacking software and hardware that would eventually be obsoleted many times over (commodore64s, vic-20s, sinclair-spectrums, BBCs, apples, XT PCs, ZX80s, amigas, all through high-school and uni). Now i'm writing compilers, compiler-compiler tools (having made my own breakthroughs), and network client-server programs. A fair bit of microcontroller stuff currently pays bills part-time. I'm still good at analog/RF stuff and think about it every day. |
| 3083
| 2009-05-22 20:45:26
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
Russell Shaw wrote: > ehydra wrote:
>> Hi Russell -
>>
... I really wanted to study oscillator circuits and predict their noise characteristics, but my supervisor thought they were too trivial. > I only had a slow 386SX pc with an extra $200 maths coprocessor plugged in.
It was 25MHz with 4MB ram, 20MB HDD, and one 5-1/4" floppy. |
| 3087
| 2009-05-23 07:19:29
| Chris Trask
| Re: Mixer Purity, JFET+SA602+JFET
|
>
> I really wanted to study oscillator circuits and predict their noise
> characteristics, but my supervisor thought they were too trivial.
>
< >
Oscillator noise trivial? Of course, there was little in the way of
things like GPS pushing issues such as oscillator noise back in 1994. An
employee of Rhode & Schwartz (sp?) devised a clever method of reducing
oscillator noise by using a low-noise audio transistor in a closed loop
within the oscillator to suppress the 1/f noise of the oscillator
transistor:
Hagemeyer, F., "Low-Noise Oscillator Circuit Having Negative Feedback,"
US Patent 5,900,788, 4 May 1999.
Rohde, U.L. and F. Hagemeyer, "Feedback Technique Improves Oscillator
Phase Noise," Microwaves & RF, November 1998, pp. 61-70.
I use a similar method to reduce the NF of common-base and lossless feedback
(aka Norton) amplifiers:
Trask, C. "Active Loop Aerials for HF Reception, Part 2: High Dynamic
Range Aerial Am-plifier Design," QEX, Sept/Oct 2003, pp. 44-49.
Chris
,----------------------. High Performance Mixers and
/ What's all this \ Amplifiers for RF Communications
/ extinct stuff, anyhow? /
\ _______,--------------' Chris Trask / N7ZWY
_
| 3090
| 2009-05-23 08:03:22
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Chris Trask schrieb: >> I really wanted to study oscillator circuits and predict their noise
>> characteristics, but my supervisor thought they were too trivial.
>>
>
> <>
>
> Oscillator noise trivial? Of course, there was little in the way of
> things like GPS pushing issues such as oscillator noise back in 1994. An
> employee of Rhode & Schwartz (sp?) devised a clever method of reducing
> oscillator noise by using a low-noise audio transistor in a closed loop
> within the oscillator to suppress the 1/f noise of the oscillator
> transistor:
>
This is http://www.rohdeschwarz.com/ It makes sense to read all papers from this genius man. Even I don't understand all ;-) - Henry -- www.ehydra.dyndns.info |
| 3093
| 2009-05-23 09:00:31
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
Chris Trask wrote: >> I really wanted to study oscillator circuits and predict their noise
>> characteristics, but my supervisor thought they were too trivial.
>
> <>
I could see a lot more complexity in oscillators than he, but there wasn't much hope of going far if i already knew more. > Oscillator noise trivial? Of course, there was little in the way of
> things like GPS pushing issues such as oscillator noise back in 1994. An
> employee of Rhode & Schwartz (sp?) devised a clever method of reducing
> oscillator noise by using a low-noise audio transistor in a closed loop
> within the oscillator to suppress the 1/f noise of the oscillator
> transistor:
>
> Hagemeyer, F., "Low-Noise Oscillator Circuit Having Negative Feedback,"
> US Patent 5,900,788, 4 May 1999.
I think that patent is a little out of context. I think you could get the same noise reduction by simply putting an equivalent regulator in the "B" power supply lead. I think the the noise they're trying to suppress is from the power supply in cheap and nasty mobile phone applications, because they mention the advantage of the circuit running from 3V. Most likely this circuit is aimed at being made into one of those VCO modules. Therefore, the patent is really all about incorporating a power supply regulator into the oscillator to suppress power supply noise, rather than making any difference to the noise level of the oscillator if the power supply was completely quiet. > Rohde, U.L. and F. Hagemeyer, "Feedback Technique Improves Oscillator
> Phase Noise," Microwaves & RF, November 1998, pp. 61-70.
Can't find a download. > I use a similar method to reduce the NF of common-base and lossless feedback
> (aka Norton) amplifiers:
>
> Trask, C. "Active Loop Aerials for HF Reception, Part 2: High Dynamic
> Range Aerial Am-plifier Design," QEX, Sept/Oct 2003, pp. 44-49.
http://home.earthlink.net/~christrask/Active%20Ferrite%20Rod%20Antenna.pdf I like reading RF articles. I've thought a long time about oscillator noise modelling and have various ideas i need to verify when i get around to it. I haven't kept up with new literature so i don't know how easy or accurate current methods are for predicting noise. I've determined lots of optimal ways of designing the overall large-signal setup for typical RF oscillators that i know isn't in any literature. I should build real circuits and test it some time. |
| 3095
| 2009-05-23 10:07:00
| victorkoren
| Re: Mixer Purity, JFET+SA602+JFET
|
| 3096
| 2009-05-23 10:13:37
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
Russell Shaw wrote: > Chris Trask wrote:
>>> I really wanted to study oscillator circuits and predict their noise
>>> characteristics, but my supervisor thought they were too trivial.
>>
>> <>
>
> I could see a lot more complexity in oscillators than he, but there wasn't
> much hope of going far if i already knew more.
>
>> Oscillator noise trivial? Of course, there was little in the way of
>> things like GPS pushing issues such as oscillator noise back in 1994. An
>> employee of Rhode & Schwartz (sp?) devised a clever method of reducing
>> oscillator noise by using a low-noise audio transistor in a closed loop
>> within the oscillator to suppress the 1/f noise of the oscillator
>> transistor:
>>
>> Hagemeyer, F., "Low-Noise Oscillator Circuit Having Negative
>> Feedback,"
>> US Patent 5,900,788, 4 May 1999.
>
> I think that patent is a little out of context.
>
> I think you could get the same noise reduction by simply putting an equivalent
> regulator in the "B" power supply lead. I think the the noise they're trying to
> suppress is from the power supply in cheap and nasty mobile phone applications,
> because they mention the advantage of the circuit running from 3V. Most likely
> this circuit is aimed at being made into one of those VCO modules.
>
> Therefore, the patent is really all about incorporating a power supply regulator
> into the oscillator to suppress power supply noise, rather than making any
> difference to the noise level of the oscillator if the power supply was
> completely quiet.
After thinking more about it, that extra transistor could make a difference if the power supply was completely quiet (i was assuming only shot noise before). An RF transistor would have a small area and thus have high 1/f popcorn noise. This noise looks like random step changes in the Vbe voltage, which modulates the DC current and thus oscillator amplitude. Varactor effects transfer that to some phase noise. If the auxillary transistor is a large-area low 1/f noise device, then its negative feedback counteracts the high 1/f noise of the oscillator transistor, resulting in more stable bias current and oscillator amplitude. |
| 3097
| 2009-05-23 10:30:53
| ehydra
| Re: Mixer Purity, JFET+SA602+JFET
|
Russell Shaw schrieb: >>> Hagemeyer, F., "Low-Noise Oscillator Circuit Having Negative
>>> Feedback,"
>>> US Patent 5,900,788, 4 May 1999.
>> I think that patent is a little out of context.
>>
>> I think you could get the same noise reduction by simply putting an equivalent
>> regulator in the "B" power supply lead. I think the the noise they're trying to
>> suppress is from the power supply in cheap and nasty mobile phone applications,
>> because they mention the advantage of the circuit running from 3V. Most likely
>> this circuit is aimed at being made into one of those VCO modules.
>>
>> Therefore, the patent is really all about incorporating a power supply regulator
>> into the oscillator to suppress power supply noise, rather than making any
>> difference to the noise level of the oscillator if the power supply was
>> completely quiet.
>
> After thinking more about it, that extra transistor could make a difference
> if the power supply was completely quiet (i was assuming only shot noise
> before).
>
Sennheiser makes fm modulated portable microphones. They run on low battery voltages. So I think this is just a LO core for such a device. Now I wondering where there is the difference to the Rohde circuit? I thought this low-frequency feedback was invited by Rohde. - It is known that the 7805-type voltage regulators are very noisy. The lower-current 78L05 is even worse. The LM317 is much better! Does somebody know better devices/circuits with a 'comparable' price? Seems that especially the low drop-out regulators tend to have bad noise characteristics. - Henry -- www.ehydra.dyndns.info |
| 3098
| 2009-05-23 10:43:47
| ehydra
| Re: Driving Mixers, was: Re: Mixer Purity, JFET+SA602+JFET
|
Thanks Russell for the lengthy history. Yes, I think most of us find themself in such a personal history scheme. - You mentioned in your paper a non-saturated driving of a mixer. The paper miss the figures. Is it possible that you explain the benefit of using it? I think Chris Trask wrote that it is essential to drive the mixer hard. I'm not able to find the core of difference here. Does it depends on mixer architecture? (I exclude here the H-mode mixer) If it is a fact that the mixer must be driven hard in limit to let the diodes work essential as switches, then, a digital drive waveform should work as good as a clipped sinuoid. And that means, we can close the circle by using a clock generator out of the many intended for the PC industry mainboards. I already mentioned the CY22393. Unfortunately it is not very well specified as a rf signal source and Cypress was not be willing to give me details. But I can say that it works! Again, I have no values because I don't own sophisticated measurement equipment. So I'm interested to hear from other what they think about integrated clock generators beside the famous Si57x (single-sourced, expensive, power-hungry). Such a device is cheap, compact, simple to use (program). regards - Henry Russell Shaw schrieb: > [..]
-- www.ehydra.dyndns.info |
| 3101
| 2009-05-23 15:16:12
| Chris Trask
| Re: Mixer Purity, JFET+SA602+JFET
|
>
> > Oscillator noise trivial? Of course, there was little in the way of
> > things like GPS pushing issues such as oscillator noise back in 1994.
An > > employee of Rhode & Schwartz (sp?) devised a clever method of reducing
> > oscillator noise by using a low-noise audio transistor in a closed loop
> > within the oscillator to suppress the 1/f noise of the oscillator
> > transistor:
> >
> > Hagemeyer, F., "Low-Noise Oscillator Circuit Having Negative
Feedback," > > US Patent 5,900,788, 4 May 1999.
>
> I think that patent is a little out of context.
>
> I think you could get the same noise reduction by simply putting an
equivalent > regulator in the "B" power supply lead. I think the the noise they're
trying to > suppress is from the power supply in cheap and nasty mobile phone
applications, > because they mention the advantage of the circuit running from 3V. Most
likely > this circuit is aimed at being made into one of those VCO modules.
>
You obviously did not read the patent. Hagemeyer placed a low-noise transistor in a feedback circuit with the oscillator base-emitter junction in the negative feedback path. With the low-noise transitor having a significantly lower 1/f noise profile than the oscillator transistor, the oscillator phase noise assumed the 1/f noise profile of the low-noise transistor. You cannot at any time ever overcome the 1/f and Nyquist noise of a transistor by way of adjusting the power supply. >
> Therefore, the patent is really all about incorporating a power supply
regulator > into the oscillator to suppress power supply noise, rather than making any
> difference to the noise level of the oscillator if the power supply was
> completely quiet.
>
No, it is not. The patent as well as the cited journal article are specifically and clearly about the addition of the low-noise transistor to control the 1/f noise of the oscillator. It has nothing at all to do with power supply regulation, and there is absolutely no way that you can infer that from either the patent or the journal article. Chris ,----------------------. High Performance Mixers and / What's all this \ Amplifiers for RF Communications / extinct stuff, anyhow? / \ _______,--------------' Chris Trask / N7ZWY _
| 3102
| 2009-05-23 19:04:08
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
Chris Trask wrote: >>> Oscillator noise trivial? Of course, there was little in the way of
>>> things like GPS pushing issues such as oscillator noise back in 1994.
> An
>>> employee of Rhode & Schwartz (sp?) devised a clever method of reducing
>>> oscillator noise by using a low-noise audio transistor in a closed loop
>>> within the oscillator to suppress the 1/f noise of the oscillator
>>> transistor:
>>>
>>> Hagemeyer, F., "Low-Noise Oscillator Circuit Having Negative
> Feedback,"
>>> US Patent 5,900,788, 4 May 1999.
>> I think that patent is a little out of context.
>>
>> I think you could get the same noise reduction by simply putting an equivalent
>> regulator in the "B" power supply lead. I think the the noise they're trying to
>> suppress is from the power supply in cheap and nasty mobile phone applications,
>> because they mention the advantage of the circuit running from 3V. Most likely
>> this circuit is aimed at being made into one of those VCO modules.
>>
>
> You obviously did not read the patent. Hagemeyer placed a low-noise
> transistor in a feedback circuit with the oscillator base-emitter junction
> in the negative feedback path. With the low-noise transitor having a
> significantly lower 1/f noise profile than the oscillator transistor, the
> oscillator phase noise assumed the 1/f noise profile of the low-noise
> transistor. You cannot at any time ever overcome the 1/f and Nyquist noise
> of a transistor by way of adjusting the power supply.
Hi, I did read every line of that patent, but the whole thing only once. I didn't initially comprehend it all. >> Therefore, the patent is really all about incorporating a power supply regulator
>> into the oscillator to suppress power supply noise, rather than making any
>> difference to the noise level of the oscillator if the power supply was
>> completely quiet.
>>
>
> No, it is not. The patent as well as the cited journal article are
> specifically and clearly about the addition of the low-noise transistor to
> control the 1/f noise of the oscillator. It has nothing at all to do with
> power supply regulation, and there is absolutely no way that you can infer
> that from either the patent or the journal article. |
| 3103
| 2009-05-23 19:34:16
| Russell Shaw
| Re: Driving Mixers, was: Re: Mixer Purity, JFET+SA602+JFET
|
ehydra wrote: > Thanks Russell for the lengthy history. Yes, I think most of us find
> themself in such a personal history scheme.
>
> -
>
> You mentioned in your paper a non-saturated driving of a mixer. The
> paper miss the figures. Is it possible that you explain the benefit of
> using it? I think Chris Trask wrote that it is essential to drive the
> mixer hard.
Hi, On what pages are the figures missing? > I'm not able to find the core of difference here. Does it depends on
> mixer architecture? (I exclude here the H-mode mixer)
>
> If it is a fact that the mixer must be driven hard in limit to let the
> diodes work essential as switches, then, a digital drive waveform should
> work as good as a clipped sinuoid.
Any switching mixer where the main elements are driven by the sum of the RF and LO sinewave sources across two terminals have the same problem. Ideally, the switching elements should have their impedance modulated only by the LO, independently of the level of RF. For the sum of LO and RF hard-switching a diode however, the zero-crossings of the LO are phase-modulated by the RF, resulting in non-ideal mixing products. The larger the ratio LO/RF current through the diodes, the better. The ideal active elements to use instead of diodes (in a ring mixer) would be two-port (4 terminal) devices, where the impedance between the RF terminals is modulated by the LO applied to the other two terminals. This means the impedance across the RF terminals is a *linear* time-varying characteristic. A LO modulated light source pointed at a really fast light-dependent resistor would work (if one existed). Parametric mixing. Another way to minimize zero-crossing LO modulation is to simulate a really large sinewave by using a square-wave LO. Then the dominant distortion products are caused purely by the dependence of diode on-resistance on the sum of LO and RF current. That is minimized by increasing the amplitude of the squarewave LO. > And that means, we can close the circle by using a clock generator out
> of the many intended for the PC industry mainboards.
>
> I already mentioned the CY22393. Unfortunately it is not very well
> specified as a rf signal source and Cypress was not be willing to give
> me details. But I can say that it works! Again, I have no values because
> I don't own sophisticated measurement equipment.
> So I'm interested to hear from other what they think about integrated
> clock generators beside the famous Si57x (single-sourced, expensive,
> power-hungry). Such a device is cheap, compact, simple to use (program).
The main thing to look out for in any digital device is that there is no other asynchronous processes such as state-machines running in them. If there are, noise is induced into the PLL elements via ground-bounce spikes in the bond wires of the IC, and could also be via internal stray couplings within the IC. With the proper design in an IC, multiple PLLs can be isolated better by adding extra bond wires and ground/supply pins. This can preclude running more than one PLL in a three-PLL device at a time. Of course, it depends on how much noise your application can tolerate. Apart from that, the only noise that remains (apart from the power supply) is phase jitter which just comes from the transistors used in the dividers. |
| 3104
| 2009-05-23 19:42:54
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
ehydra wrote: >
> Russell Shaw schrieb:
>>>> Hagemeyer, F., "Low-Noise Oscillator Circuit Having Negative
>>>> Feedback,"
>>>> US Patent 5,900,788, 4 May 1999.
>>> I think that patent is a little out of context.
>>>
>>> I think you could get the same noise reduction by simply putting an equivalent
>>> regulator in the "B" power supply lead. I think the the noise they're trying to
>>> suppress is from the power supply in cheap and nasty mobile phone applications,
>>> because they mention the advantage of the circuit running from 3V. Most likely
>>> this circuit is aimed at being made into one of those VCO modules.
>>>
>>> Therefore, the patent is really all about incorporating a power supply regulator
>>> into the oscillator to suppress power supply noise, rather than making any
>>> difference to the noise level of the oscillator if the power supply was
>>> completely quiet.
>> After thinking more about it, that extra transistor could make a difference
>> if the power supply was completely quiet (i was assuming only shot noise
>> before).
>>
>
> Sennheiser makes fm modulated portable microphones. They run on low
> battery voltages. So I think this is just a LO core for such a device.
>
> Now I wondering where there is the difference to the Rohde circuit? I
> thought this low-frequency feedback was invited by Rohde.
>
> -
>
> It is known that the 7805-type voltage regulators are very noisy. The
> lower-current 78L05 is even worse. The LM317 is much better! Does
> somebody know better devices/circuits with a 'comparable' price? Seems
> that especially the low drop-out regulators tend to have bad noise
> characteristics.
I've had noise problems with LDO regulators. Many have a slow control loop, letting power supply noise pass through. My favourite thing is one or two BJTs, a largish cap or two (like 10uF), and some resistors that simulate an LC low-pass filter (gyrator with large simulated inductor) in the oscillator power supply. By adding diodes, you can make it quickly charge when power is applied if neccessary. Low-freq power supply noise still gets through, but it's well within the PLL bandwidth and so is highly attenuated. |
| 3105
| 2009-05-23 20:00:04
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
Russell Shaw wrote: > ehydra wrote:
>> Hi Russell -
>>
>> Before I answer you:
>>
>> "Thesis: "Analysis and Design of Radio Frequency Mixer Circuits"
>>
>> I tried to read your paper but cannot find your pages at the
>> university?? Strange.
>> http://www.rmit.edu.au/browse?STYPE=PEOPLE&QRY=%2Brussell+%2Bshaw&CLOCATION=&Search.x=8&Search.y=9
>>
>> Is it possible to email it?
It appears the figures don't appear in open-office. Now i remember why. The figures were hand-drawn and the photos weren't digital. (no digital cameras back then) This is the first time i've sent anyone the thesis since last century. I've only ever showed the paper copy, which was done by photocopying the diagrams and photos. I don't know what RMIT did to copy the thesis onto CD. I'll have to do something about it some time. |
| 3106
| 2009-05-23 20:04:03
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
Russell Shaw wrote: > ehydra wrote:
>> Hi Russell -
>>
>> Before I answer you:
>>
>> "Thesis: "Analysis and Design of Radio Frequency Mixer Circuits"
>>
>> I tried to read your paper but cannot find your pages at the
>> university?? Strange.
>> http://www.rmit.edu.au/browse?STYPE=PEOPLE&QRY=%2Brussell+%2Bshaw&CLOCATION=&Search.x=8&Search.y=9
>>
>> Is it possible to email it?
It appears the figures don't appear in open-office. Now i remember why. The figures were hand-drawn and the photos weren't digital. (no digital cameras back then) This is the first time i've sent anyone the thesis since last century. I've only ever showed the paper copy, which was done by photocopying the diagrams and photos. I don't know what RMIT did to copy the thesis onto CD. I'll have to do something about it some time. Scan the paper thesis figures on a scanner. |
| 3107
| 2009-05-23 20:22:28
| Russell Shaw
| Re: Mixer Purity, JFET+SA602+JFET
|
ehydra wrote: >
> Russell Shaw schrieb:
>>>> Hagemeyer, F., "Low-Noise Oscillator Circuit Having Negative
>>>> Feedback,"
>>>> US Patent 5,900,788, 4 May 1999.
>>> I think that patent is a little out of context.
>>>
>>> I think you could get the same noise reduction by simply putting an equivalent
>>> regulator in the "B" power supply lead. I think the the noise they're trying to
>>> suppress is from the power supply in cheap and nasty mobile phone applications,
>>> because they mention the advantage of the circuit running from 3V. Most likely
>>> this circuit is aimed at being made into one of those VCO modules.
>>>
>>> Therefore, the patent is really all about incorporating a power supply regulator
>>> into the oscillator to suppress power supply noise, rather than making any
>>> difference to the noise level of the oscillator if the power supply was
>>> completely quiet.
>> After thinking more about it, that extra transistor could make a difference
>> if the power supply was completely quiet (i was assuming only shot noise
>> before).
>>
>
> Sennheiser makes fm modulated portable microphones. They run on low
> battery voltages. So I think this is just a LO core for such a device.
>
> Now I wondering where there is the difference to the Rohde circuit? I
> thought this low-frequency feedback was invited by Rohde.
>
> -
>
> It is known that the 7805-type voltage regulators are very noisy. The
> lower-current 78L05 is even worse. The LM317 is much better! Does
> somebody know better devices/circuits with a 'comparable' price? Seems
> that especially the low drop-out regulators tend to have bad noise
> characteristics.
I've had noise problems with LDO regulators. Many have a slow control loop, letting power supply noise pass through. My favourite thing is one or two BJTs, a largish cap or two (like 10uF), and some resistors that simulate an LC low-pass filter (gyrator with large simulated inductor) in the oscillator power supply. By adding diodes, you can make it quickly charge when power is applied if neccessary. Low-freq power supply noise still gets through, but it's well within the PLL bandwidth and so is highly attenuated. If using a DC regulator, ones with zener references can be noisy compared to band-gap reference ones. With a zener reference type, look for one that has a pin for adding a filter cap to the reference if neccessary. You could make a simple DC regulator with a well bypassed zener too. | | | | | | | | | | | | | | | | | | | |