EMRFD Message Archive 12354

Message Date From Subject
12354 2016-02-14 18:49:08 Ashhar Farhan measuring oscillator noise
hans has been experimenting and designing the Si5351a (substantially improved version of Si5351) boards. they are an excellent buy if you are inclined to fooling around with them.

in a chat, he mentioned this page http://www.qrp-labs.com/synth/synthnoise.html.

this makes me think, it might be easier to evaluate the phase noise of a local oscillator by using it as the local oscillator in primitive direct conversion receiver with a low pass filter at about 100 Hz. a low noise source can be used as signal. the phase noise of the oscillator could be indirectly measured at audio frequencies by a PC or a oscilloscope. 

This might be quite useful as our main concern with phase noise is the resultant dynamic range of the receiver. 

- f
12355 2016-02-14 19:15:21 billw77aaz Re: measuring oscillator noise
If I understand your proposed use, you couldn't measure phase noise of a DUT that was much better than the Si5351 itself.

Bill  W7AAZ
12356 2016-02-14 20:45:36 Ashhar Farhan Re: measuring oscillator noise
yes, i think a low noise, well adjusted VFO or a crystal source could be used to make a comparitive measurement.this is not going to be exact, but relative.

- f

12357 2016-02-14 21:30:00 hanssummers2000 Re: measuring oscillator noise
Hi Bill

As I understand Farhan's proposal, he is suggesting measuring the phase noise of the local oscillator itself, relative to the crystal. In this case, the local oscillator is the Si5351A itself. So it is a way to measure the phase noise of the Si5351A... assuming a good clean crystal oscillator and good clean power supplies, it seems like a nice idea, I think. 

FYI I have these "breakout-boards" (Si5351A chip is pre-soldered) here price $7.50 http://www.qrp-labs.com/synth and a rotary-encoder tuned, 16 x 2 LCD readout, VFO using it, price $29 at http://www.qrp-labs.com/vfo . Mine works from 3.5kHz to approx 300MHz, though that's violating the datasheet spec maximum (200MHz) so your mileage may vary if you insist on UHF ;-)

73 Hans G0UPL

12358 2016-02-14 22:48:56 Bill Carver Re: measuring oscillator noise
OK Fahran and Hans.........I thought the proposal was to use the Si5351A to test   ANOTHER  LO. Beating against a clean crystal oscillator should give you a clear picture of the noise sidebands of the Si5351A.
Phase noise is a relative measurement. So if you can measure the noise sidebands in the present of the beatnote itself, then you could get a db/Hz number at different offsets. It would be great if you could characterize the Si5352A v.s. Si570.

73 - Bill  W7AAZ




12359 2016-02-15 04:50:32 Steve Dick Re: measuring oscillator noise
When using mains for power, best results may be obtained by using a low noise and high power supply rejection ratio  voltage regulator rather than the standard 3 terminal regulator like the 780X. An example is the TI LP5907  or similar regulators from Maxim or STM specifically designed for low noise and high power supply rejection ratio. If a standard 3-terminal regulator is used, then very good post-power supply filtering should be used with heavy R-C filtering as these regulators are fairly noisy.
 
Regards,
“Digital Steve” K1RF
 
12362 2016-02-15 05:20:21 Jim Davey Re: measuring oscillator noise
I just posted an EDN article that demonstrates how to substantially reduce noise from 3-terminal regulators, plus some other circuits which will reduce noise.  I’m now in the process of retrofitting these ideas into some power sources I use for MF preamps I use for weak signal work below 500 kHz.  

Jim  K8RZ


12363 2016-02-15 05:38:14 Chris Trask Re: measuring oscillator noise
This is done simply by beating the oscilltor with itself with one path shifted by 90 degrees. Then measure the baseband noiose with an audio spectrum analyser. This sounds simplistic, but it was the basis for a very expensive piece of test equipment sold by HP back in the 80's. It is not suitable for measuring noise attributed to short-term drift.

>
>this makes me think, it might be easier to evaluate the phase noise of a
>local oscillator by using it as the local oscillator in primitive direct
>conversion receiver with a low pass filter at about 100 Hz. a low noise
>source can be used as signal. the phase noise of the oscillator could be
>indirectly measured at audio frequencies by a PC or a oscilloscope.
>
>This might be quite useful as our main concern with phase noise is the
>resultant dynamic range of the receiver.
>



Chris

When the going gets weird, the weird turn pro
- Hunter S. Thompson
12364 2016-02-15 06:43:23 Tayloe, Dan (Noki... Re: measuring oscillator noise

I remember a long time ago when I used to do this measurement for Motorola, we had custom “suck out” box that was a high pass filter to get rid of everything below several hundred Hz.  I never looked inside of it, but I imagine that it was just a couple poles of L/C high pass filtering.

 

A spectrum analyzer has limited dynamic range (70 dB?), and when you measure the noise skirts, you need to remove out the high amplitude part that you aren’t really concerned about so that you can amplify and measure what is left over.  This keeps from overloading the test equipment with stuff you don’t care about so that you can measure the things you are really interesting in.

 

You need to calibrate the set up without the filter and amplifier, but then with the filter in and amplifier in place, what you really want to see is what is left such as 1 KHz, 10 KHz, or 100 KHz away.

 

-          Dan, N7VE

 

12365 2016-02-15 06:43:24 Ashhar Farhan Re: measuring oscillator noise
Chris,

I would imagine that the closer you wish to measure the phase noise, the longer the delay line ought to be. 
For instance, to measure the phase noise at 20 KHz distance from the center, we will need to delay the signal by 50 usecs and the phase. Is my understanding correct?

Intuitively, I would imagine that a trivial phase shifted oscillator fed through a Fisher style hybrid will be of no value as the same noise will appear at both ports. 

I am not sure at all that I understand how to implement this. What kind of phase delay method would work?

- f

12366 2016-02-15 06:48:31 Tayloe, Dan (Noki... Re: measuring oscillator noise

I think he meant that the 90 degree shift is done on the RF side.  See the 90 degree RF shifter (L/C/L/C 90 degree splitter) used Rick’s R2 receiver.  You probably need to buffer the two output puts, then send one side to the RF input and the other side into the LO side of a diode mixer.

 

-          Dan, N7VE

 

12367 2016-02-15 06:53:47 Chris Trask Re: measuring oscillator noise
No. The delay line is 90 degrees long at the frequency of the oscillator. You can fine-tune the delay line by measuring the DC from the mixer, adjusting for a minimum.

>
>I would imagine that the closer you wish to measure the phase noise, the
>longer the delay line ought to be.
>

Chris

When the going gets weird, the weird turn pro
- Hunter S. Thompson
12368 2016-02-15 07:08:38 Ashhar Farhan Re: measuring oscillator noise
Oh ok. that would mean, getting an RG58/U of 5 meters for measuring it at 14 Mhz. It should be do-able. Let me get this done over the next weekend. 

- f

12369 2016-02-15 07:15:13 Tayloe, Dan (Noki... Re: measuring oscillator noise

I guess I looked at the R2 on line and it did not have Rick’s splitter that I was thinking of.  Go into the files section of the group and find the MicroR2mods pdf from I7SWX.  In the schematic it shows T1.  This is basically a “bridge” circuit that produces a +45 degree shift on one side, and -45 degree shift on the other for a net delta of 90 degrees between the two delta. 

 

I am sure details on this circuit are in emrf. I may be wrong, but I think the main constraints are that the two “L”s in this bridge a wound on the same core (like a T37-2 iron core) and that the reactance of the Ls and Cs are equal and should match in input and output impedance.

 

I modeled this and was very impressed about how flat the 90 degree phase shift was maintained over a relatively wide range.

 

-          Dan, N7VE

 

12370 2016-02-15 08:25:19 Lasse Moell Re: measuring oscillator noise
Farhan,
if you want (need) to learn a bit more about phase noise measurements, check these links:

Also look for HP Product Note 11729B-1
and 11729C-2

These HP documents should keep you going :) And it is a lot of fun measuring oscillators!

Cheers
Lasse SM5GLC


15 februari 2016 16:08:36 +01:00, skrev Ashhar Farhan farhanbox@gmail.com [emrfd] :
 


Oh ok. that would mean, getting an RG58/U of 5 meters for measuring it at 14 Mhz. It should be do-able. Let me get this done over the next weekend. 

- f

12371 2016-02-15 08:55:12 Bill Carver Re: measuring oscillator noise
Fahran,as Chris says if you split the DUT signal, phase shift one 90
degrees, and fed the two into a perfect mixer the mixer output will be
the phase noise spectrum. Amplitude noise is ignored.

But amplitude noise is still there. If the oscillator has no odd
harmonics and drives a perfect hard limiter or a perfect switching
mixer.........circuits which switch at precisely the V=0 point on the
oscillator waveform....... amplitude noise will be scrubbed off.
Schmitt triggers which intentionally have different switching points for
rising and dropping oscillator voltages will convert amplitude noise to
phase noise. Even with a perfect V=0 switching point even harmonic
energy in the oscillator signal will convert amplitude noise to phase
noise.

Bill W7AAZ
12373 2016-02-15 09:44:39 Chris Trask Re: measuring oscillator noise
>
>I guess I looked at the R2 on line and it did not have Rick’s splitter
>that I was thinking of. Go into the files section of the group and find
>the MicroR2mods pdf from I7SWX. In the schematic it shows T1. This is
>basically a “bridge” circuit that produces a +45 degree shift on one
>side, and -45 degree shift on the other for a net delta of 90 degrees
>between the two delta.
>
>I am sure details on this circuit are in emrf. I may be wrong, but I
>think the main constraints are that the two “L”s in this bridge a wound
>on the same core (like a T37-2 iron core) and that the reactance of the
>Ls and Cs are equal and should match in input and output impedance.
>
>I modeled this and was very impressed about how flat the 90 degree phase
>shift was maintained over a relatively wide range.
>

The L's can be a bifilar winding on a single toroid, which makes things simpler. There is also a method of tweaking the quadrature split, which consists of placing a variable capacitor across the isolated load resistor. Using a varactor there you can integrate the DC from the mixer so that the quadrature is adjusted automatically. There was a brief article about that in one of the trade journals decades ago. I know I have it in my files somewhere.


Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
12375 2016-02-15 10:46:18 Dan Rae Re: measuring oscillator noise
12380 2016-02-15 13:19:56 Dennis Anderson Re: measuring oscillator noise
Further to Farhan & Hans comments

Paul M0XPD with Pete Juliano N6QW further developed the Si5351A IC on an Arduino Shield.

The m0xpd Si5351 Shield is a flexible, 3-channel RF generator sub-system, partnering the Silicon Labs Si5351 Device with a pair of broadband amplifiers and a quadrature divider.

See the following link for futher details.


These are available through Kanga UK

Hopefully adds to the ever growing list of boards/shiels that are increasingly using the Si5351A IC.

Regards

Dennis G6YBC



12381 2016-02-15 14:07:23 Bill Carver Re: measuring oscillator noise
If you remove the 90 degree phase shift you get a DC voltage from the mixer that is "0 dB".
Compare the spectral plot you get to that DC voltage and you have "dBc" noise. Almost, anyway: the noise you measure is from both upper and lower sidebands. Assuming they are equal you'd just subtract 3 dB from the measured noise power, then compute how many dB that was from the DC voltage you'd measured.

Bill


12383 2016-02-15 21:56:12 victorkoren Re: measuring oscillator noise
Chris and Dan you are wrong. You certainly need a very long delay line to measure the phase noise at close to carrier frequencies. If delay is short than slow phase fluctuations will appear at both mixer ports and will not be detected.
I did such measurements approx. 25 years ago using a power splitter with one output connected directly to the mixer and the other through a long RG-213 coax. This took care of the signal levels as a strong signal was connected to the LO port and a weaker signal (attenuated by the long coax) connected to the RF port of the mixer.
You calibrate it by connecting a commercial signal generator instead of the tested oscillator and modulating it with FM or PM with known modulation that causes sidebands with known DBc level and correlating that with what you see at the mixer output.
Victor - 4Z4ME
12384 2016-02-15 22:15:00 victorkoren Re: measuring oscillator noise

Look for an old HP application note "Phase noise characterization of microwave oscillators".
In page 13 under "system sensitivity" you can see that the measurement is proportional to the delay value multiplied by a Sinc function that nulls at 1/delay, so you need a long delay line to get reasonable sensitivity on one side and youy get zero sensitivity at a frequency offset of 1/delay on the other side. So delay length is a compromise of what frequency offset is under measurement.
Victor - 4Z4ME

12386 2016-02-16 08:09:37 AncelB Re: measuring oscillator noise
Hmm, would there be an option for an HP8753D VNA to do this via a 3rd
party program?
12388 2016-02-16 08:29:22 hanssummers2000 Re: measuring oscillator noise

Dennis reminded me to mention, regarding Arduino shield kits. QRP Labs also has a matching Arduino shield kit for the Si5351A break-out board. The Arduino shield kit also has onboard a small BS170 Power Amplifier (a watt or two, depending on band, how many BS170 fitted, and PA voltage). 

It also has compatibility with a lot of other plug-in modules:: it has provision to fit QRP Labs Low Pass Filter kit, and 6-band relay-switched LPF extension, as well as the OCXO (kit) version of QRP Labs' Si5351A module, and QLG1 GPS receiver for GPS disciplining the SI5351A oscillator. 

References:

Si5351A module, $7.50, http://www.qrp-labs.com/synth.html
OCXO/Si5351A module, $15, http://www.qrp-labs.com/ocxokit.html

LPF kit, $4.20, Low Pass Filter 

Relay-switched 6-band LPF kit, $15, Ultimate relay-switched LPF kit

QLG1 GPS receiver kit, $20.50, http://www.qrp-labs.com/qlg1.html

 

73 Hans G0UPL

12389 2016-02-16 08:31:33 Roelof Bakker Re: measuring oscillator noise
Hello all,

I believe that at least comparitive oscillator noise measurements
can also be done using a SDR. I have just looked at the Si570 tuned
to 7050 kHz and a 7050 kHz crystal oscillator with a PERSEUS SDR.
The crystal oscillator is much cleaner.

However you can also use this technique with a Softrock SDR, which
uses a Si570 as LO, when the noise specifications of the Si570 are
known.

I will order a Si5351 kit just out of curiosity and see what it does
compared to the Si570.

73,
Roelof Bakker, pa0rdt
12390 2016-02-16 08:56:59 Johan Bodin Re: measuring oscillator noise
I made a similar test some years ago:

https://sites.google.com/site/sm6lkm/lo-noise

It's amazing how clean a simple crystal oscillator can be compared to
most kinds of variable frequency sources. Even the good old HP8640B,
cavity osc' + dividers, is very dirty by comparison!

73
Johan SM6LKM

Roelof Bakker wrote:
> Hello all,
>
> I believe that at least comparitive oscillator noise measurements
> can also be done using a SDR. I have just looked at the Si570 tuned
> to 7050 kHz and a 7050 kHz crystal oscillator with a PERSEUS SDR.
> The crystal oscillator is much cleaner.
>
> However you can also use this technique with a Softrock SDR, which
> uses a Si570 as LO, when the noise specifications of the Si570 are
> known.
>
> I will order a Si5351 kit just out of curiosity and see what it does
> compared to the Si570.
>
> 73,
> Roelof Bakker, pa0rdt
>
>
> ------------------------------------
> Posted by: "Roelof Bakker" <roelof@ndb.demon.nl>
> ------------------------------------
>
12391 2016-02-16 09:48:14 agustintomasferra... Re: measuring oscillator noise
To Jim Davey

Hola, would you mind to post the link to that article in EDN? Interested.

Gracias.
12393 2016-02-16 10:58:40 Roelof Bakker Re: measuring oscillator noise
Hello Johan et all,

Screenshots the noise of a crystal oscillator and the Si570 can be
found here:

http://ndb0.home.xs4all.nl/Oscillator%20Noise/

The noise pedestal of the Si570 is not smooth, but rises again
further away from the carrier.

The output of the crystal oscillator is also taken via the crystal.
Power out is +6 dBm and the transistor (2N2222) draws about 25 mA at
9 v.

The noise of the crystal oscillator at 10 kHz from the carrier is
148 dBc/Hz.

73,
Roelof, pa0rdt
12395 2016-02-17 02:37:24 mosaicmerc Re: measuring oscillator noise
Jim, Thanks for the Vreg noise link.

I expect to be powering an IoT ESP8226 12e wireless access point with 3.3V off an LM317. It can pull 300mA.
Which approach would you advise to use for lowering the noise floor of the Vcc in this case?

Ancel
12396 2016-02-17 03:53:31 Jim Davey Re: measuring oscillator noise
Ancel:

For the pennies involved, bypassing the LM317 as shown in Figure 4 with 10uf on the C-bypass junction and something large on the output should be a best practice.  For the things I am doing, which are relatively low current and would be best done with a battery, the capacitance multiplier of Figure 1 may be beneficial.  I have not experimented with this yet.  For some of my applications, good voltage regulation is not a big concern as it would be for a VFO.  The LM317 is just a convenience.  

For what you are describing, given all the RF noise associated with Ethernet interfaces, I’d be more concerned with keeping the noise that will be generated localized so it did not find its way to other sensitive circuits.  Just put a portable SW receiver near any router or access point for a good lesson in how much noise they can generate.  

Jim K8RZ


12397 2016-02-17 04:20:53 Cecil Bayona Re: measuring oscillator noise
Why do hams keep using old parts like the LM317, LM305, LM309 etc, that family of Voltage regulators is over 35 years old, there are many other families of regulators that are cleaner, have less minimum voltage requirements and are not expensive?

12398 2016-02-17 05:09:46 Chris Trask Re: measuring oscillator noise
>
>Why do hams keep using old parts like the LM317, LM305, LM309 etc, that
>family of Voltage regulators is over 35 years old, there are many other
>families of regulators that are cleaner, have less minimum voltage
>requirements and are not expensive?
>

There have been others, but the LM317 has been the benchmark for decades. It's being made by a number of manufacturers, keeping them readily available and inexpensive. We used them for low-noise microwave sources without exception.



Chris Trask
N7ZWY / WDX3HLB
Senior Member IEEE
http://www.home.earthlink.net/~christrask/
12399 2016-02-17 05:35:10 Johan Bodin Re: measuring oscillator noise
Hello,

yes Roelof, that spectral pattern seems to be typical for PLLs.
I've seen much worse performance on some frequency multipliers, ICS501
and the x6 clock multiplier of AD9851 comes to mind. The distance
between the "bumps" depends on the PLL loop bandwidth. The PLL can
compensate for FM fluctuations that are slow enough to fall in this
region. Far out from the bumps, the curve approaches the true noise
trumpet of the oscillator, I guess, where the PLL can do nothing to help.

The Si5351A-B-GT looks very interesting! The current consumption is more
QRP-friendly than that of Si570.

73
Johan SM6LKM

Roelof Bakker wrote:
> The noise pedestal of the Si570 is not smooth, but rises again
> further away from the carrier.
12401 2016-02-17 07:31:14 Cecil Bayona Re: measuring oscillator noise
Interesting reason, something like an LM117 that does the same job has been designed as a more modern replacement, has lower noise, regulates better, is more accurate, and is an LDO regulator and also is inexpensive doesn't beat tradition? I wonder then why people gave up using vacuum tubes, it was a tradition also, at some point tradition or not people need to move forward, personally I would not spend a penny to buy outdated obsolete devices when I can get much better devices for the money. Sorry if I appear cranky but some traditions get in the way of progress for not exactly great reasons.

12402 2016-02-17 07:56:40 Vern Re: measuring oscillator noise
Familiarity . ..?  Unless you are in the industry ( I am not ), I suspect that your knowledge of 'latest and greatest ' will fall behind rather rapidly.

Here's your opportunity to share what you know with the community. .. :)

73,
-Vern
N7GTB 
Kennewick, WA 



Sent via the Samsung Galaxy S®6 active, an AT&T 4G LTE smartphone
-------- Original message --------
12403 2016-02-17 08:25:52 Johan Bodin Re: measuring oscillator noise
I believe that LM117 is exactly the same chip as LM317 and they both
appear in the same old data sheet.

http://www.qsl.net/n9zia/tracking/LM317.pdf

The '317 is the "consumer grade" version of '117, and as far as I know,
just screened to more relaxed spec's. With some luck, a randomly
selected '317 can even be better than a randomly selected '117 ;-)

73
Johan SM6LKM

Cecil Bayona wrote:
> Interesting reason, something like an LM117 that does the same job has
> been designed as a more modern replacement, has lower noise, regulates
> better, is more accurate, and is an LDO regulator and also is
> inexpensive doesn't beat tradition?
12404 2016-02-17 08:40:35 Roelof Bakker Re: measuring oscillator noise
Hello all,

The LM117 has a wider temperature range and can be used outdoors in
Arctic areas. I have seen it used in an active antenna design
intended for use outdoors.

73,
Roelof, pa0rdt
12405 2016-02-17 08:44:40 John Levreault Re: measuring oscillator noise
Did you perhaps mean the LM1117?

John NB1I

12406 2016-02-17 09:03:44 Chris Trask Re: measuring oscillator noise
It should seem odd to most people here that we are being told that there are better and quieter devices than the LM111/317, but we have yet to see any specific part numbers. This is typical of engineering workplaces where someone shouts down someone else's approach but has nothing better to offer, just getting visibility by casting doubt on others.

>
>Familiarity . ..?  Unless you are in the industry ( I am not ), I suspect
>that your knowledge of 'latest and greatest ' will fall behind rather rapidly.
>Here's your opportunity to share what you know with the community. .. :)
>73,-VernN7GTB Kennewick, WA 
>


Chris

When the going gets weird, the weird turn pro
- Hunter S. Thompson
12408 2016-02-17 09:46:28 Jim Strohm Re: measuring oscillator noise
Cecil,
As another poster mentioned, you may be thinking about a different regulator.  Regardless, one reason we use "outdated obsolete devices" is that they're widely used and easily available in "outdated obsolete" equipment for FREE.  Every couple of years I set up test jigs and test my latest batch of "outdated obsolete" voltage regulators, and toss out the ones that didn't survive the years or the removal process.

I'll also add -- when I can use one of these "outdated obsolete devices" that I already have on hand, I save the cost of purchase, the time to order, and the time to wait for my order to arrive.  And things being the way they are nowadays, I still have to do incoming device testing.  Since I dn't have room to store all the test jigs I may assemble, it's faster in the long run for me to batch my testing for once a year and test dozens or even hundreds of parts at a time.

And when I get around to sorting all my vacuum tubes (I have hundreds, easily) I'll batch test them too.  Need any 12AX7?  I know a lot of audiophiles who do ... who'll pay insane money for them.  New, they run about $10 a pop.  But why should I pay for something I already have?

73
Jim N6OTQ

P.S. What about "outdated obsolete" dual-gate FETs?  Don't you wish you'd bought a bunch a few decades back?

12409 2016-02-17 09:47:30 Jim Strohm Re: measuring oscillator noise
And add to that -- to the best of my recollection, from Day One the
LM317 datasheets included specific bypass requirements on both the
input and output pins, which when I started playing with '317s around
1983 (nearly 35 years ago...) cost more than the chip itself.

I know that this voltage regulator -- and apparently its 78xx and 79xx
cousins -- will work without the bypassing, as I've seen them all
utilized in commercial products with no bypassing. But I've never had
any noise issues with any of them because I always adhered to the
bypassing requirements.

Except for one 7808T I used in a mobile application. Its heatsink was
undersized, and its output current was far too small for the
transmitter turn-on current of the 1 watt HT it was connected to.
Still, it was properly bypassed.

Never trust specifications. Well, more like "Trust but verify."

73
Jim N6OTQ

12410 2016-02-17 10:00:10 Cecil Bayona Re: measuring oscillator noise
Yes, sorry for the mix up, LM1117.

12412 2016-02-17 10:06:48 Cecil Bayona Re: measuring oscillator noise
Free is one thing, new designs is something else. Also I'm talking about obsolete parts that have better replacements, the leaded dual gate MOSFETS have no replacement still I would not design something new with them as SMT versions are available.

12414 2016-02-17 11:27:45 mosaicmerc Re: measuring oscillator noise
Well,
Here are my reasons for using the LM317 in a new design.
1) getting the design assembled overseas means one less part 'substitute' headache based on local availability.
2) The LM317 can handle almost 40V input. Most 'new' regulators don't come close.
3) Cost.
4) Reliability, the thing is nearly bulletproof.
5) Low noise isn't always important, and if it is...there are means to handle it.
6) Easy to obtain as a TO-220 form factor....which is suitable when doing repairs on old eqpt.

I still use 2n3904/6 and 2n2222 and 1n4148 parts......they work, they're quite cheap and proper engineering doesn't mean rushing out for the latest part. It means applying engineering economics, supply chain logistics, serviceability and ease of design/prototyping. These 'old' parts are well modeled in all the simulators. New parts are often not.

Boeing went ahead and implemented Li Ion battery tech in their 787 Dreamliner....upstaging the venerable Lead acid avionics battery. That got their fleet grounded cuz some engineer(s) didn't estimate the MTBF for the environment properly and wanted the lighter weight feature of the tech. Further, all they have done to get the planes in the air is to fire proof the battery compartment ...as no one has a solution for the problem.

So if you're on a Dreamliner...it may be on fire and you would never know. Such is the march of progress without proper engineering.
12416 2016-02-17 16:35:00 Chris Trask Re: measuring oscillator noise
>
>I still use 2n3904/6 and 2n2222 and 1n4148 parts......they work,
>they're quite cheap and proper engineering doesn't mean rushing
>out for the latest part. It means applying engineering economics,
>supply chain logistics, serviceability and ease of
>design/prototyping. These 'old' parts are well modeled in all the
>simulators. New parts are often not.
>
>Boeing went ahead and implemented Li Ion battery tech in their 787
>Dreamliner....upstaging the venerable Lead acid avionics battery.
>That got their fleet grounded cuz some engineer(s) didn't estimate
>the MTBF for the environment properly and wanted the lighter weight
>feature of the tech. Further, all they have done to get the planes
>in the air is to fire proof the battery compartment ...as no one
>has a solution for the problem.
>
>So if you're on a Dreamliner...it may be on fire and you would
>never know. Such is the march of progress without proper engineering.
>

And then there was the brilliant idea of using microprocessors as a high-tech middleman between the driver and the brakes, throttle, steering, etc. If the engine in my 2000 Toyota ever gives up, I'm simply going to short-block it with a remanufactured engine from Japan.


Chris

When the going gets weird, the weird turn pro
- Hunter S. Thompson
12418 2016-02-17 18:56:33 Dana Myers Re: measuring oscillator noise
12419 2016-02-17 19:09:14 Jim Strohm Re: measuring oscillator noise
And on that note, Chris, most states now consider vehicles
manufactured in 1991 or earlier to be classic vehicles, and
substantially exempt them from pollution and some safety regulations.
I am thinking about replacing the twin-cam L4 in my '87 Porsche 944
with a Chevy LS, and trashing all the computer-crippled systems in
favor of old-school hotrod stuff.

I figure I can sell the two twin-cam engines I have for enough money
to do the swap, and there are already adapter kits available.
Drive-by-wire only belongs on fighter jets.

73
Jim N6OTQ

On Wed, Feb 17, 2016 at 6:25 PM, Chris Trask christrask@earthlink.net

> And then there was the brilliant idea of using microprocessors as a high-tech middleman between the driver and the brakes, throttle, steering, etc. If the engine in my 2000 Toyota ever gives up, I'm simply going to short-block it with a remanufactured engine from Japan.
12420 2016-02-17 20:34:37 Dana Myers Re: measuring oscillator noise
12421 2016-02-17 22:45:13 Bill Carver Re: measuring oscillator noise
I loved my old 911s, turbo RX7, etc. But I'll take the nasty 'ol
computers between me and the shocks, differential and engine in my Z51
Vette.
W7AAZ
12422 2016-02-18 05:07:03 i7swx Re: measuring oscillator noise
Regarding "old fashion" components, unless we have restricted needs/requirements, I believe she should use them. I am still using unmarked  components bought from Polypacks in Mass in 1974-76 (double quantity adding a 1 cent to the pice), I have a lot of diodes supposed to be 1N4148 that are electrically very similar to modern ones and works as should be.

Also, interesting the cars discussion, tehre is always soemthing to learn from off topics too. One of my sons is working as a mechatronic with a Peugeot service, previously with OPEL, including the US brands like Corvette and he tell me funny stories of electronics and mechanical conflicts and faults.

When I got my driving license il 1958 (at that time minimum age 18, mine) we had to learn everything about engine and cars and be able to fix the fault on the road, 1st class driving license... if one had to drive a diesel car he had to have the additional training on diesel engine (second class), then, if you had to drive a truck an additional test was required with knowledge on trucks engines and accessories. I was questioned on the truck as the engineer running the tests mistook my class ... but I was OK, I did not know what was going on with a truck trailer ... hi. Now it is all quitzes ... I can imagine doing now the same kind of courses and test with the electronics gadgets replacing those old two or three mechanical cords and metal pieces.

73

Gian
I7SWX
12423 2016-02-18 05:10:02 Chris Trask Re: measuring oscillator noise
>
>I figure I can sell the two twin-cam engines I have for enough money
>to do the swap, and there are already adapter kits available.
>Drive-by-wire only belongs on fighter jets.
>

Amen on the fighter jets. I flew C-130s for six years, and I really do not like the fly-by-wire of the new J model. It does not save that much weight to make it worthwhile. Lockheed did an earlier weight reduction when they replaced the aluminum spar box with one made of boron filaments and epoxy. It saved a mere 600 pounds. They made 31 of them as an interim model between the E and H models. They were eventually sold to Canadian RCAF. At least three were destroyed in crashes, and they were recently replaced with Js.

It does mitigate the maintenance problems with the control cable pulleys, but I much prefer the security and reliability of electro-mechanical instruments in front of me. Those glass panels are worthless if you have total electrical failure, and those do happen on rare occasions.


Chris

When the going gets weird, the weird turn pro
- Hunter S. Thompson
12424 2016-02-18 05:15:08 Chris Trask Re: measuring oscillator noise
>
> It's like talking about tossing out Chris'
>excellent KISS mixer and replacing it with an old-school 2N3904.
>

Thank you for that. I'm still looking for opportunities to apply the "Howard Cosell Corollary", which is to "Make things better, not more expensive." It's an old-school concept that seems to have been forgotten.

>
>Forgive me, but y'all went off-topic and I went with you :-)
>

We need to do that once in a while so that we don't take ourselves too seriously. :{D


Chris

When the going gets weird, the weird turn pro
- Hunter S. Thompson
12425 2016-02-18 05:18:41 Chris Trask Re: measuring oscillator noise
>
>Regarding "old fashion" components, unless we have restricted
>needs/requirements, I believe she should use them. I am still using
>unmarked components bought from Polypacks in Mass in 1974-76 (double
>quantity adding a 1 cent to the pice), I have a lot of diodes supposed
>to be 1N4148 that are electrically very similar to modern ones and
>works as should be.
>

UNGH! Poly-Paks! How I've missed them all these decades. I have a lot of RTL logic parts that came from them, including the elusive uL958 decade counter.

Chris

When the going gets weird, the weird turn pro
- Hunter S. Thompson
12426 2016-02-18 05:56:43 hanssummers2000 Re: measuring oscillator noise

All

> Here are my reasons for using the LM317 in a new design.
> 1) getting the design assembled overseas means one less part 'substitute' headache 
> based on local availability.
> 2) The LM317 can handle almost 40V input. Most 'new' regulators don't come close.
> 3) Cost.
> 4) Reliability, the thing is nearly bulletproof.
> 5) Low noise isn't always important, and if it is...there are means to handle it.
> 6) Easy to obtain as a TO-220 form factor....which is suitable when doing repairs on old eqpt.

I think that sums it up very very well. There's not necessarily anything WRONG with an old part, or any real reason to use a new one... it all depends on the circumstances, taking into account all of the above. In my opinion a good engineer doesn't necessarily use the latest component. He looks at what is necessary to fit the entire requirement (e.g. performance, spec), and do it efficiently (economically) overall. 

40 years later and we still use 8-bit processors, we don't necessarily throw a 32-bit 100+ MHz ARM into everything. It depends on how much processing power we need. Same with every component. 

Overseas assembly is a really important point - the older the part, the more likely it is going to be easy to source locally. I experienced this several times. 

The comparison with tubes isn't that relevant, because the gap in most respects was so huge - tubes are big and expensive and delicate and power-hungry. The move to semiconductors was such a quantum leap forward in most respects, it's far from the same as the comparison between an LM317 and an LM1117...

73 Hans G0UPL


12427 2016-02-18 08:33:34 Ashhar Farhan Re: measuring oscillator noise
I used the LM317 in the specan. specan proved to be an illuminating study. the spurs and the noise was literally 'visible'.

first, the LM317 needs extensive filtering on the input as well as the output to keep the digital spurs from the microcontrollers and other digital oscillators from climbing into the Si70 via the power lines. 
Apart from the power lines, I also had to bypass the Control lines I2C to remove noise from the Arduino controller. 

the LVDS version has substantially lower noise than the CMOS version.

I realize that I am making vague and unsubstantiated clams. I must make a simple notch filter a la W7ZOI design and do some actual measurements.

- f


12428 2016-02-18 14:54:28 Bill Carver Re: measuring oscillator noise
Notching the "carrier" leaves BOTH the phase noise and amplitude noise. The usual assumption is an oscillator (including a VCO) makes AM and PM noise at about the same amplitude. So the notched data might be 3 dB high on phase noise number.  Of course the real issue in most cases is the total noise, not just the PM component.

Bill