You should read Dick's trial and errors with this analyzer on his web page.
High noise and distortion.
Cheers,
Its for these kinds of applications that I use my Distortion Magnifier described in vol. 0 of Linear Audio. It works for DUTs with a flat frequency response, like amplifiers. It just subtracts a properly-scaled and phased version of the source signal (e.g., oscillator). It is basically as good as the op amps it is built with.
Phase adjustment is just achieved with a frequency-variable single-pole approximation to the DUT's HF rolloff. As a result, it provides the fundamental rejection function over a fairly wide bandwidth with only minimal adjustment for phase and amplitude. A null of 60dB is not difficult to achieve. Then some source signal at -20dB or -40dB is put back in, so that the distortion magnification is 20dB or 40dB. This keeps enough fundamental in to allow a standard THD analyzer to lock on, and also provides a fundamental reference level. Of course, one can operate without any source signal added back in and get the deep 60dB or so notch.
The distortion magnifier also works well for making two-tone measurements, like 19+20 kHz CCIF, since its cancellation of the source signal is reasonably broadband.
Compared to a variable-frequency notch, it requires adjustment of two things, namely amplitude and phase, instead of a single critical frequency adjustment. There is usually a little bit of iteration between the two as well.
Cheers,
Bob
I will do that. I expect to replace the active circuits with equivelent opamps. But I havent seen the schematic yet. I have the passive B&K notch filter and this would be an active equal to it. Well, I'll see if it can be made up to-date and useful. because its cheap and dont need the analyzer portion.
Thx-RNMarsh
Last edited:
Hi Bob.... guess what. I have your Distortion Magnifier. Great circuit. And, a very useful product. Highly recommended.
Thx-RNMarsh
Hi Richard,
Thanks! I'm glad it has worked well for you.
Back in the early '80s when I did my MOSFET power amplifier with HEC, I had to use an early version of that circuit to measure the amplifier - that was the only way I could get that low with the equipment I had at the time.
Cheers,
Bob
IIRC it is a Wien bridge so any info on Wiens apply. Can the RC ratios be altered for higher Q like with the Bridged T?
Dick and I tried all sorts of ways of doing this with large amounts of positive FB. Although it worked and was stable the distortion was out of control.
If you can reorganize the actives like the dual inverted stage oscillators it could work. The tuning capacitor was the limit because of the way HP set it up.
Last edited:
Probably easier to leave the high Q and have a separate path for mixing back the fundamental. More precise and stable as well. I tried tweaking the notch filter options to control the q some time ago. Nothing worked well enough in simulation to even try in real parts.
The HP331-334 all have air caps for tuning (as I remember) meaning high Z. High Z means higher noise floor. That will be the real limit of the performance of it as a notch filter (or much of any other application) unless you jack up the voltage a lot. No good roadmap from their to here.
@RNMarsh -- maybe the better way is to make the B&K an active filter. I think it is a Twin-T, yes? Easy, non-destructive addition. I haven't seen my active Twin-T add any THD that is clearly identifiable as coming from the filter, and noise seems to be very low as well -- that said, the measurement limits are set by the oscillator and the analyzer, so it's an open question. But as I said, an easy mod, and you have the Shiba Soku stuff as a guage of performance....
RE HP 331-4 notch filter -- Demian nailed the basic problem as the very high circuit Z, but I never found an opamp configuration that didn't end up with a lot of 2nd H in the output. I hope you're successful, so that I can tailgate on your work.
RE HP 331-4 notch filter -- Demian nailed the basic problem as the very high circuit Z, but I never found an opamp configuration that didn't end up with a lot of 2nd H in the output. I hope you're successful, so that I can tailgate on your work.
That is a good idea also -- B&K mod. I thought about that awhile back. As with the 339A though, I am pressed for time -- a multi-tasking life style, it seems. If the bulk of it is done then i can modify it and be good to go. I am looking for the easy way. But if it isnt easy or adds a lot of distortion then I'll dump it and move on.... I will take a peek at it and run some tests on ideas and see if it seems to be do-able. -- wish me good luck. [I dont have a 334 yet. on its way. ... just an idea to see how feasable and what will make easy, quick improvements -- and if that will be enough for my application to the ADC. It wont be used as a distortion analyzer so deep, deep notch isnt needed and can have lower distortion and noise -- maybe]
Thx RNMarsh
Thx RNMarsh
Sounds like a very good approach. The idea of a continuously variable notch - if clean and quiet enough - is powerful. maybe the twin-T which many have built, can be made variable, easily?
-RM
Richard, you can turn auto-tuning off on the 334 and tune the notch depth to whatever you want. A 40dB notch will be stable enough for most uses, and even a 60dB notch can hang in there for quite a while. But I like Demian's idea of using the full notch and adding fundamental back in, too, though noise and other effects seem uncertain to me.
Then this may be all I need... just tune it to a 40dB null and I'm done. Simple, easy, cheap enough.... I have seen a 334 asking $50 on eBay. Not even worth my time to build a twinT/ box/power supply/calib dial etc for similar application. I can also use it to attenuate any residual harmonics from an oscillator - like 2H or 3H.
Is there a circuit for Demian's suggestion to try? And, how would it be better for this kind of app?? Would it be quieter? Less noise? More stable?
Can a conceptual sketch/schematic to view of summing/mixing back in the fund.?
-RM
Last edited:
Well, be aware that the 334 is all-discrete circuitry with +/- 25V supplies and that just replacing the discrete amps with opamps is far harder than you think, with possible performance issues remaining in any case, as I found to my sorrow. Not saying you can't do it, just that it's no day at the beach.
