Richard,
i can follow the arguement for film and foil caps for high current crossover applications. The literature for the Multicaps also refers to filter networks like servos, what performance inprovement in that type of application over a cap like Wima FKP film/foil or MKP metalized end attachment?
probably no difference with servo use. Depends if one stage or two stage servo and cut-off freq chosen. Minor, IMO. However, in other filter apps, it matters a lot more.
THx-RNMarsh
The FFT measured on an AP 2722 on 3 generators:
I need better generators !!
For comparison, here is a plot from the R&S UPV with the B1 Low Distortion Generator option. 1kHz at 0dBV, with H2 and H3 almost at -126dB.
Attachments
Last edited:
For another reference do you still have Victor's oscillator?
Sorry, I no longer have the Viktor oscillator.
Only see 3H at -130dB for victor 1KHz
Looks like a pretty steep notch filter at 1kHz. Here is the UPV with its analog notch and digital 20kHz LPF applied. H2 is now below -132dB.
Attachments
While not sure when we shall get a proper version of the balanced Viktor, I messed around with a quick & dirty solution.
A LME49990 inverter using 4.7k 0.1% SMD resistors.
Works first time. No oscillations.
Patrick
.
A LME49990 inverter using 4.7k 0.1% SMD resistors.
Works first time. No oscillations.
Patrick
.
Attachments
While not sure when we shall get a proper version of the balanced Viktor, I messed around with a quick & dirty solution.
A LME49990 inverter using 4.7k 0.1% SMD resistors.
Works first time. No oscillations.
Patrick
.
Patrick you come up with the nicest looking enclosures. How do you do it?
Passive Twin T (balanced), still to be trimmed to the Viktor.
Patrick
.
Any boards for sale?
> Patrick you come up with the nicest looking enclosures. How do you do it?
Finished product from Chinese Ebay.
https://world.taobao.com/item/36032...&spm=a312a.7700824.w4002-5699239201.40.WLEfCO
(You can order blank front plate, but need someone to do it for you from Hong Kong or Singapore.)
> Any boards for sale?
If just you, you can just send me a PM and I'll let you have 2x.
If more than 1 person then better open a new GB thread.
Else they will move this entire thread to the GB forum, like they did e.g. with Scott's ULNA and quite a few of my year-old threads.
http://www.diyaudio.com/forums/grou...1000-low-noise-measurement-amp-ikoflexer.html
I am sure we don't want that.
Cheers,
Patrick
Finished product from Chinese Ebay.
https://world.taobao.com/item/36032...&spm=a312a.7700824.w4002-5699239201.40.WLEfCO
(You can order blank front plate, but need someone to do it for you from Hong Kong or Singapore.)
> Any boards for sale?
If just you, you can just send me a PM and I'll let you have 2x.
If more than 1 person then better open a new GB thread.
Else they will move this entire thread to the GB forum, like they did e.g. with Scott's ULNA and quite a few of my year-old threads.
http://www.diyaudio.com/forums/grou...1000-low-noise-measurement-amp-ikoflexer.html
I am sure we don't want that.
Cheers,
Patrick
Looks like a pretty steep notch filter at 1kHz. Here is the UPV with its analog notch and digital 20kHz LPF applied. H2 is now below -132dB.
The Victor 2H (H2) is non-existent in my tests.... < -140dB. But 3H is there at -130dB. No filtering used. Is it from the Victor or the analyzer? checking.....
A passive twin-T for the 3H or a LP filter would help. As would a third pcb layer as shield.

THx-RNMarsh
Last edited:
The second is there just below you noise floor.
I would say the third is the analyzer residual.
If this is the Shibasoku 725D then most likely the third is from the first band eliminate (notch) Twin T filter.
I would say the third is the analyzer residual.
If this is the Shibasoku 725D then most likely the third is from the first band eliminate (notch) Twin T filter.
Last edited:
The Victor 2H (H2) is non-existent in my tests.... < -140dB. But 3H is there at -130dB. No filtering used. Is it from the Victor or the analyzer? checking.....
A passive twin-T for the 3H or a LP filter would help. As would a third pcb layer as shield.
When I had Viktor's 1kHz, with input at -9.6dB, H2 was at about -126dB and H3 at -128dB running off AC. This is not close to what you are getting. Perhaps it was the analyzer used - a 1616m recapped and modded, run at 192kHz on ARTA. But I did not have a notch filter. Since I had bought the 1kHz, 10kHz and 50kHz oscillators from Viktor, he told me the next improvements would only come from tuned Twin-T's at each frequency. With the effort and fixed-frequency oscillator limitations stacking up, I went looking for an "all-in-one" solution.
The last plot showed H2 at -132dB and H3 at -136dB. This is with the built-in -50dB notch filter in the UPV, no weighting. I did not see any difference with the 20kHz LPF in or out. Attached is the 20Hz - 22kHz noise floor with input shorted using an RF shorting plug so all the gremlins are likely from the generator.
The APx555 has a generator that can do -115dB THD+N at 1Vrms, and with their "special analog notch filters, additional ADCs and DSPs", they can get to -122dB THD+N. This is demonstrated on their YouTube video and I believe is likely the current state of the art, both for generator and analyzer performance. They certainly think so.
Attachments
What you will find amongst the analyzers is that they are very input level sensitive when looking at very low levels of harmonics. In other words, their non-linearity at low levels becomes apparent. I am looking to keep a 1dB tolerance on the harmonics of accuracy in measurement.
At levels below -100dB, the shielding is paramount to keep unwanted freqs from entering. I have resorted to extream measures in order to see the levels Davada has produced in his generator. But if the levels are not too bad and the extraneous field in the room isnt a total disaster..... I can drop the Victor pcb into an IC anti-electrostatic bag and get significant reduction in noise pickup (emi/rfi).
Victor's pcb in the bag.
THx-RNMarsh
At levels below -100dB, the shielding is paramount to keep unwanted freqs from entering. I have resorted to extream measures in order to see the levels Davada has produced in his generator. But if the levels are not too bad and the extraneous field in the room isnt a total disaster..... I can drop the Victor pcb into an IC anti-electrostatic bag and get significant reduction in noise pickup (emi/rfi).

THx-RNMarsh
Last edited:
Never thought of using a "IC anti-electrostatic bags" for shielding. I just threw out two quart size bags. Now to go dig them out. Thanks.....
This is what I get from Victors oscillator (early example running on batteries) at 1V into the Shibasoku 725 (modified). At 3 V the Victor distortion seems to go up. I have a passive 10 dB per step attenuator on its output so the input to the analyzer is constant but the harmonics increase significantly. Still more stuff to degrade your confidence in measurements below -100 dB.
Attachments
- Home
- Design & Build
- Equipment & Tools
- Low-distortion Audio-range Oscillator