John Curl's Blowtorch preamplifier part II - Page 197 - diyAudio
Go Back   Home > Forums > Member Areas > The Lounge

The Lounge A place to talk about almost anything but politics and religion.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 28th October 2009, 01:21 PM   #1961
syn08 is offline syn08  Canada
Account disabled at member's request
 
Join Date: Aug 2005
Location: Toronto
Quote:
Originally Posted by john curl View Post
SY, as a moderator, yourself, would you like to fire up the HP3581 in your possession, and look at how the meter responds to random noise at different bandwidth settings? You might be able to break this 'stalemate' and we could all learn something new.
John, let go please. There's nothing to learn from this exercise.
  Reply With Quote
Old 28th October 2009, 01:29 PM   #1962
diyAudio Member
 
john curl's Avatar
 
Join Date: Jul 2003
Location: berkeley ca
Not to appear rude, Syn08, but we were addressing the measurement of noise with the HP3581 Wave Analyzer. I also have an HP3562(3) FFT based analyzer, but that is not what Ikoflexer was referring to, as he apparently doesn't have an HP 3562 to measure with, he has an HP3581. Some test equipment has a more limited setting range than other equipment, and this is the problem that I am addressing.
  Reply With Quote
Old 28th October 2009, 01:38 PM   #1963
diyAudio Member
 
john curl's Avatar
 
Join Date: Jul 2003
Location: berkeley ca
Everyone, it should be clear that we are trying to address the problem of measuring noise coming from different sources. The Norton equivalent Zener has been measured, in order of magnitude, at least. The LM329 Zener diode substitute, may be measured in future, although the graph of noise vs frequency already exists in the app note. The biggest problem is worst case, and its probability with this device, since the device could be more than 10 times noisier than typical, yet be within spec.
In any case, either approach could be used to establish a low noise reference. I leave the details to the individual designer.
  Reply With Quote
Old 28th October 2009, 02:19 PM   #1964
iko is online now iko  Canada
diyAudio Moderator
 
iko's Avatar
 
Join Date: May 2008
Location: Toronto
syn08, I'd hate to make you work just to satisfy my curiosity. Please don't worry about it if you don't think it's worth it. If you do measure though, others might find this exercise very helpful. If indeed the lm329 has a mean X and variance Y noise, then that's an important fact to take into consideration for the regular DIYer who doesn't have the necessary measurement instruments to make sure they got the "typical" lm329. As I said before, horses for courses.

Also, syn08, I don't put much weight on "absolute" value measurements. Is there such thing? One can get very picky about such things and demand mean and variance, instrument calibration, etc. AFAIC what I was trying to do with the scope is relative measurements, for comparison purposes. Perhaps I should omit the units when reporting. My scope self noise, or whatever you'd call it, shows as a flat line of perhaps a few micro volts when on 100uV/div settting. Relative to that, the jfet+R reference shows clearly a wiggly trace, with peak-to-peak values as I mentioned before. If I add a capacitor in parallel with the resistor, the wiggly trace becomes flat but thicker than a when nothing is measured with the scope.

Now, your point b) don't know the bandwidth. No need for that in relative measurement; I'll just choose the wave/wiggle with the larges peak-to-peak value. As I said, no paper gets published here, I just want to compare the waves that one component shows on the scope, with the waves the other component makes. This also addresses your point a) hard to estimate average value. One can compare max peak-to-peak as well, not only average.

What I'm saying is, one can directly compare two devices this way if the instrument has enough resolution to show their relative noise difference. I agree with anyone who says that this method does not give any absolute values.
  Reply With Quote
Old 28th October 2009, 02:37 PM   #1965
diyAudio Member
 
john curl's Avatar
 
Join Date: Jul 2003
Location: berkeley ca
Ikoflexer, at the risk of 'butting in', your measurement is reasonable, if not very accurate. However, if you found real improvement by adding the cap across the resistor, why not just use it and be done with it? Of course you can add an RC of equal value to the LM329 and get good results there too. My point is that it is not cheaper or better to do so. In fact, it can be limiting, as the Zener substitute is only a fixed voltage and some way normally has to be used to increase it, in order that it can give higher voltage power supplies.
Salas, the 1000uF cap should do no harm, but it might not do any significant good either. This is because you are trying to filter from a virtual voltage source, rather than having a resistor to work against. This doesn't work very well, because a 10 ohm source (for example) is 100 times less efficient with a cap, as a 1000 ohm source would be. RC is the filter, not just C. Therefore, a 10uf cap with 1K in front, is equal to 1000uf with 10 ohms in front. I hope that this is clear to everyone. It is a usual oversight, and I see it all the time, even with the engineers that I work with, on occasion.
  Reply With Quote
Old 28th October 2009, 02:41 PM   #1966
Salas is offline Salas  Greece
diyAudio Chief Moderator
 
Salas's Avatar
 
Join Date: Oct 2002
Location: Athens-Greece
Quote:
Originally Posted by syn08 View Post
Measuring noise with a scope is extremely imprecise and that's because a) it's not easy to estimate the average value b) you don't know exactly the bandwidth and c) you don't know the scope Y amp noise.

I'll do some measurements myself tonight (time permitting). I have all the needed devices and a 89410A plus a 3562A to compare with. I can do 0.001Hz RBW (overnight) without any "bouncings", in fact all the noise measurements on my web site are at 0.1Hz RBW or less.
Hey syn, thanks for bothering. Will you do the 4 generic leds with big cap bypass fed from a GR or BL CCS too while at it? Cheers.
  Reply With Quote
Old 28th October 2009, 02:45 PM   #1967
Salas is offline Salas  Greece
diyAudio Chief Moderator
 
Salas's Avatar
 
Join Date: Oct 2002
Location: Athens-Greece
Quote:
Originally Posted by john curl View Post
Salas, the 1000uF cap should do no harm, but it might not do any significant good either. This is because you are trying to filter from a virtual voltage source, rather than having a resistor to work against. This doesn't work very well, because a 10 ohm source (for example) is 100 times less efficient with a cap, as a 1000 ohm source would be. RC is the filter, not just C. Therefore, a 10uf cap with 1K in front, is equal to 1000uf with 10 ohms in front. I hope that this is clear to everyone. It is a usual oversight, and I see it all the time, even with the engineers that I work with, on occasion.
That is why I will go 10-20x the cap usual moderate value when with the relatively low impedance 100-200R LEDs Vref. Got that. Thanks.
  Reply With Quote
Old 28th October 2009, 03:06 PM   #1968
Salas is offline Salas  Greece
diyAudio Chief Moderator
 
Salas's Avatar
 
Join Date: Oct 2002
Location: Athens-Greece
P.S. The LEDs are going to be far less governed by the IDSS Tc wandering of the Jfet CCS feeding them than when feeding a resistor, their Vf is much stiffer to plateau current variation, so I think this is a good thing for keeping the cascode base Vbias the same for both channels at all times and with different nearby heat radiations to the Jfet CCS per side. I mean I don't know it yet, I just presume.
  Reply With Quote
Old 28th October 2009, 03:08 PM   #1969
iko is online now iko  Canada
diyAudio Moderator
 
iko's Avatar
 
Join Date: May 2008
Location: Toronto
OK, so how about this. My concern as an average DIYer is a good Vref solution. We are considering so far the following:

1. the Norton zener, i.e. an appropriate jfet in series with an R||C filter. Please note that a resistor in parallel with a capacitor DOES act as a band filter IF it is being fed (in series) with a current source. That's the job of the jfet. The CCS can be improved by two cascoded jfets, but they have to be chosen appropriately (see figure). I already posted a few pages ago a list of jfets which are recommended in that book for this purpose.

Advantages: flexibility in Vref value. With suitable C across R, low noise (how low to be established). Cheap.

Disadvantages: to be determined.

2. lm329 or equivalent low noise voltage reference.

Advantages: low noise, less affected by temperature variations, specifically created for this purpose.

Disadvantages: large variation in noise specs (John pointed this out), but noise can be reduced with an RC filter.

3. LEDs in series, possibly with a capacitor across them. Clearly some LEDs have less self noise than others. Issues that are not IMHO well addressed: unless one can measure it, how can one choose from the multitude of LEDs? How effective in reducing noise is the capacitor across them.

4. Zener reference, possibly with RC or C filter.

IMHO it would be best if we could characterize more completely these four solutions in terms of what they offer and what they lack, instead of claiming one is best. If the solutions are well defined then everyone can choose what is "best" for them individually.

Please correct any issue I might have gotten wrong here. Comments are welcome.
Attached Images
File Type: png jfetccs-casc.png (20.3 KB, 249 views)

Last edited by iko; 28th October 2009 at 03:12 PM.
  Reply With Quote
Old 28th October 2009, 03:13 PM   #1970
SY is offline SY  United States
diyAudio Moderator
 
SY's Avatar
 
Join Date: Oct 2002
Location: Chicagoland
Blog Entries: 1
Quote:
Originally Posted by john curl View Post
SY, as a moderator, yourself, would you like to fire up the HP3581 in your possession, and look at how the meter responds to random noise at different bandwidth settings? You might be able to break this 'stalemate' and we could all learn something new.
How about if I do that just as a participant, a measurement geek, and someone interested in low noise design?

Have you checked your pm? I sent you one a few days ago. You should have a little notification in the upper right hand corner.
__________________
You might be screaming "No, no, no" and all they hear is "Who wants cake?" Let me tell you something: They all do. They all want cake.- Wilford Brimley
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



New To Site? Need Help?

All times are GMT. The time now is 01:16 PM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2