Hey, Nick - My sympathies on the shipping versus actual part cost. It's ridiculous. Since the op-amp is the only active component in there, and you've installed one of the best....I'd be tempted to look at that output with a spectrum analyzer, and see what the worst harmonic(s) is/are. Perhaps those could be filtered. (And all my DIP sockets are *ancient*. Thanks for the reminder to de-ox them before trying to solder them in place.) -Joe
Hi Joe, I had a look yesterday evening and the biggest harmonics are the 3rd and 5th at around 45 dB down. Maybe I need to build a low pass filter. If it was at RF would be easy. But at these frequencies I've never attempted it. Will look for designs.
@@ptronix The caps and inductors are just BIG by comparison, sir. Crossover designs might even be a good baseline. Active filters are physically smaller, but they might just introduce more distortion of their own (especially if they have any gain). Having typed that last sentence, I'm not so sure. Probably worth investigating, at least.
@@matambale hi Joe, I don't even want to go there with inductors and capacitors at 1 kHz! Yes active filters would introduce distortion. I think diego has the answer. Even though it has 4 digits of resolution it's not designed to measure below 99 db when set into log mode. The fundamental notch is not good enough for anything lower! In the calibration it says adjust notch for over 90 db using the internal source. Well my box reads just over 100 db so about 8 db better than the internal source and that's the limit it seems of what I'm able to measure but interesting project all the same and I'm happy with my little circuit. Best wishes nick
The HP8903B datasheet show a 3% measurement error at RMS levels used in the video. Also, the distortion measurent does not seem to go below the higher of -80dB (100uV) or 15uV (-96dB) (I do not understand that specification the way it is written) A reading of 0.001% is -100dB distortion. The testset display does go down to 0.0001% which means -120dB (1uV) This makes the "the higher of -80dB (100uV) or 15uV (-96dB)" statement weird once again. As for the PCB and parts, the shorter the leads, the better the circuit. I do wonder what a spectrum analyzis of the output could show. Some oscillators have a FET in the loop used to kill most of the second harmonic. I bought a Chinese low distortion oscillator and, through adjusting it, I was able to get: -96dB THD+N (0.0016%) at 999.95 Hz. - 102 dB THD (0.0008%) at 999.95 Hz. The board was sold with a claim of -124dBFS distortion. With an output of 16.3dB (6.5VRMS) the measured THD is -118.3dBFS, which is not quite the claim. Driving the board with two 9V batteries, the SNR got 3 dB better. SO did the THD with -105 dB (0.0006%) and 119 dBFS (output = 14V RMS) The 3 x 10 turns potentiometers were not optimally adjusted, so some tewaking was needed. Measurements were made with a QA403.
@@yveslegault6825 sorry for the delay in replying. That QA403 looks like an interesting piece of equipment but a bit beyond my price range. Its been a while now since this project so I'm a bit hazy on my results. But it seems the 8903 will measure down to -100 db although it's specification is not that low. But the whole point of all this is I've reached a level where my confidence in my measurements are getting unreliable at such low levels of distortion. The other meter I have is the Keithley 2015 and that is fairly hopeless doesn't go anywhere near as low as the 8903. I did look at the spectrum first of all my scope which has a built in spectrum analyzer (not FFT) also later on using software and an external sound card box 'thing'. The results between the two were completely different, so what do you belive in! Think there's a video showing the computer spectrum analysis. I still think I've produced an oscillator with lower distortion than anything else I have in my workshop. OK its fixed at 1 KHz but that's OK for most audio amplifier tests. Best regards Nick
Hey Nick I had a quick look at the HP specs and found that the display range goes from 0.001% up to 100% even though you’ve got a 4 digits resolution? Perhaps you reach the bottom of the reading you can get from it? And maybe again it needs some alignment to have a proper notch? Or another thing could be noise pick up? Have you tried to connect the box directly to the input from the HP with a double female BNC adapter and see if there’s any improvements? Just random thoughts.. D.
I think you have the answer diego and thanks for reading the manual. Its not designed to read anything lower than 99 db in log mode. The notch is not good enough. In the notch calibration it says adjust for 90 db or better using the internal source. Well at great stress to my back I dragged it down again and using my box as a source I adjusted the notch pot again. The best I could get was just over 100 db which is about 8 db better than internal signal. So in other words I've reached the limit of what I can measure. But without you spotting that I would still be searching for the unobtainable! Thanks again
@@toooldtodie I think diego has the answer, the hp is not designed to measure below 99 db in log mode. And my box measures just over 100 db about 8 db better than the internal source. So I've reached the limit of my measuring equipment!
What about PCB's layout? It doesn't seem to be particularly optimized and there may be some parasitic capacitances and there may be oscillations at higher frequencies.
Yes it's possible but I'm not sure how I would make the PCB any better. I may at some time make another PCB. There are no oscillations at other frequencies at other frequencies that I can see. There are harmonics as you would expect but the biggest one is 45 dB down.
