I was thinking in a diy lock-in. I expect its build to be extremely time-consuming as it approaches an infinite-Q filter.
I have many experiments in progress and very little free time, but my idea is to start using a sound card as sine wave generator (at first only to try) and a small time constant to get less selectivity because it would allow for automated scanning and then improove everything to discover how far can i go keeping everything cheap.
It would be based on a 7.1 soundcard with one channel outputting the reference signal and two other channels outputting the harmonic to meter at 0 and 90 degrees.
My idea is that this would boost the soundcard sensitivity up to extraordinary levels since noise would be canceled due to long time integration and to use a parallel port adc i built to meter large electrolytics by fitting theyr discharge curve.
Uh, i forgot the worst problem. Do you thing of a good analog multiplier for that purpose?
I have many experiments in progress and very little free time, but my idea is to start using a sound card as sine wave generator (at first only to try) and a small time constant to get less selectivity because it would allow for automated scanning and then improove everything to discover how far can i go keeping everything cheap.
It would be based on a 7.1 soundcard with one channel outputting the reference signal and two other channels outputting the harmonic to meter at 0 and 90 degrees.
My idea is that this would boost the soundcard sensitivity up to extraordinary levels since noise would be canceled due to long time integration and to use a parallel port adc i built to meter large electrolytics by fitting theyr discharge curve.
Uh, i forgot the worst problem. Do you thing of a good analog multiplier for that purpose?
Hello all,
Just an idea.
http://www.novotone.be/fr/index.php?option=com_content&task=view&id=43&Itemid=34
This is a good site too.
Max
Just an idea.
http://www.novotone.be/fr/index.php?option=com_content&task=view&id=43&Itemid=34
This is a good site too.
Max
ionomolo said:
This is indeed the most difficult problem if you would decide to build a lock in amplifier by yourself. I doubt that any integrated analog multiplier implementation will match the linearity requirements for a good noise rejection.
A digital switching multiplier is a much better solution, however I am not aware of any good IC or chipset that would implement such.
Building a really good lock-in should be way easier than it used to be. I think there's even a circuit in the LT app notes, probably using the LT1043 switch. I'd check but my books are put away and my memory is fuzzy. Only problem is I'm not sure how fast it is.
Conrad Hoffman said:
There is a 0.01% multiplier application in the LT1043 datasheet. But I doubt that's good enough and, as you said, the speed is also pretty low.
This is really interesting, i will do some research on those multipliers and pray to find a good one.
Lock In amp
Just to throw a curve at an old thread the Shibasoku distortion analyzer (725 series) has that lock in amp as part of it for harmonic analysis. It measures harmonics 2-5 and THD with a residual of .00005 or so. And the matching oscillators are good enough to get that low.
Just to throw a curve at an old thread the Shibasoku distortion analyzer (725 series) has that lock in amp as part of it for harmonic analysis. It measures harmonics 2-5 and THD with a residual of .00005 or so. And the matching oscillators are good enough to get that low.
Any opinions on the HP Model 8903A Audio Analyzer, 20 Hz, 100 kHz:
Hewlett Packard 8903A Audio Analyzer, Refurbished HP Agilent Hewlett Packard - Used Test Equipment at Test Equipment Depot
Seems to be rather old from the early 1980s.
Hewlett Packard 8903A Audio Analyzer, Refurbished HP Agilent Hewlett Packard - Used Test Equipment at Test Equipment Depot
Seems to be rather old from the early 1980s.
I have one. It works OK. The a version doesn't have a real differential input. The b version does. I find the Boonton's more useful myself.
I posted this to the HP/Agilent Yahoo group in response to queries on the 8903B. A comparison of the 8903B, Tektronix SG5010 and AP generators:
The Boonton 1120 which I have works just as well as the 8903B.
An externally hosted image should be here but it was not working when we last tested it.
The Boonton 1120 which I have works just as well as the 8903B.
Hey, at least the HP doesn't shortchange you on any of the higher harmonics like some of those other units do! You should get what you pay for. 😀 It would be interesting to see the output of the Cordell analyzer under the same conditions. Mine is about 0.0002% mid-band and the oscillator board is reasonably easy to build.
When Bob came to the NJAS DIY gathering last October, I proudly showed off the copies of the Audio Magazine article -- with the prices of the parts pencilled in. I never built the full blown analyzer, however.
To fill out the roster of instruments that play in the same league, it would be nice to have the Tek SG505, and Krohn-Hite 4400, represented in this data set.
It certainly looks like it's still a worthwhile goal to build a tunable signal source with total corruption of 120 dB (or better) below the fundamental. I'm hoping that Jim Williams' article scheduled for publication in August might give a few hints for how that can be done . . . with Jim's untimely passing I guess that article will be the last opportunity to glean whatever Jim knows about the topic.
Dale
Your KH4402 looks like it needs a tuneup. My KH4400 is consistently .0003%. Its box stock. Every time I feel like hot-rodding it I realize I have 2 ShibaSoku oscillators (earlier versions of this http://www.shibasoku.com/download/avc/ag15c_e.pdf ) that are 20 dB better (less than .00001%) and what's the point? Any of these are better than anything we build that is useful.
Really good deal on a distortion analyzer Krohn-Hite 6880 Distortion Analyzer # | eBay
I have one and it works very well. Totally automatic and just works. The residual is below .005% but not way below. I think I have some docs as well.
I have one and it works very well. Totally automatic and just works. The residual is below .005% but not way below. I think I have some docs as well.
"The difficult we do immediately. The impossible sometimes takes a bit longer."
Thanks - you must collect audio test equipment.
If Demian is correct, and that 2nd harmonic from the KH4402 IS an anomaly, then the most notable distinction among these instruments is not the harmonic distortion, but rather the low-frequency power line residuals.
Dale
p.s. - the KH4402 also shows a greater spectral width for the fundamental component - which makes me think "phase noise". Putting that with the 2nd harmonic, I'd be tempted to look for something not-quite-right in the AGC circuit. Maybe a leaky electrolytic in the error amp's integrator section, or (for a JFET used as a variable resistor) a problem with the components that are supposed to inject a bit of signal from the drain, into the gate circuit.
Thanks - you must collect audio test equipment.
If Demian is correct, and that 2nd harmonic from the KH4402 IS an anomaly, then the most notable distinction among these instruments is not the harmonic distortion, but rather the low-frequency power line residuals.
Dale
p.s. - the KH4402 also shows a greater spectral width for the fundamental component - which makes me think "phase noise". Putting that with the 2nd harmonic, I'd be tempted to look for something not-quite-right in the AGC circuit. Maybe a leaky electrolytic in the error amp's integrator section, or (for a JFET used as a variable resistor) a problem with the components that are supposed to inject a bit of signal from the drain, into the gate circuit.
You can find the complete manual on-line. What I found surprising, is that it achieves that performance using a simple twin-T notch filter - and tunes the filter with Vactrol-style optocouplers (LED's shining on CdS light-dependent resistors)!
Dale
I have a 4400 which wasn't working when it came in - blown power supply and easy to fix. The 4400 does better (0.00033%) than the 4402 (which may need a bit of TLC @0.00055%). The 4400 is enclosed in a metal clase while the 4402 is plastic, and (I believe) the AGC circuitry of the 2 are different.
I also had a 6800 which I unloaded. It was very good, but only 1kHz.
