I thought those were too expensive for what they are. I believe they are much more expensive than the larger ones. Many years ago a Vishay sale rep told be those were invented for hearing aid volume controls and had really low contact resistance variation (noise) but I could never get an update on that story.
I have used the 3/4" rectangular versions for many things. They seems to work really well. I could provide a sample or two if you want to try. PM me and let me know.
If you use the same part type in each circuit, they will be cancelled as well. External to the output... layout et al is another story.
Was still thinking about app for further improving the 339A oscillator.
I personally dont have too much issue finding best passive parts. Worked that one to death already. Thx-RNM
Was still thinking about app for further improving the 339A oscillator.
I personally dont have too much issue finding best passive parts. Worked that one to death already. Thx-RNM
Last edited:
Well, you need the oscillator thd a LOT lower before worrying about R thd. At least with the R's i would use. Still for the 339A. Victor's are so good already, even with sm parts, that a change in parts seems like the least of the thd issues with oscillators. Trimmer contacts are a biggy for controls and contacts from switches and the like. But just pure R of most thin film R's have thd that is extreamly low. I'm not worried about them just yet as being the limiting factor for the oscillator. Thx-RNMarsh
Just to put some numbers on the trimmer issue I've re-measured 3 random samples of the Vishay metal foil 1248W parts. The ones I have are 5 kOhm, and for the test setup they were wired as rheostat (variable resistor, two terminal) and set to ~1 kOhm. Distortion measurement was carried out with a resistor bridge (an improved version of the one shown in figure 2 here: http://www.linearaudio.net/images/letters.pdf/volume1ltees.pdf) and +20 dBu across the trimmer. Observed distortion changed from specimen to specimen, with a varying mixture of 2nd and 3rd harmonic. Worst one was at -93 dB, best at -115 dB (individual harmonics, not THD).
Unless I've not considered everything or messed up the math in the hurry, this result suggests that to keep the trimmer distortion contribution negligible at the -140 dB level, the voltage across it would need to be below -30 dBu. That is for the rheostat directly in series with a fixed resistor, +20 dBu across the total resistance, and assuming that the 2nd harmonic falls proportional with fundamental level. This means that the achievable trimming range is less than 1%. If more is needed, lower fundamental levels must be chosen.
Comparison with other trimmers will have to wait...
Just saying: if you want to dive towards -140 dB I'd be careful with this sort of assumptions. Verifying everything and assuming nothing is the most promising approach in this territory.
Samuel
Unless I've not considered everything or messed up the math in the hurry, this result suggests that to keep the trimmer distortion contribution negligible at the -140 dB level, the voltage across it would need to be below -30 dBu. That is for the rheostat directly in series with a fixed resistor, +20 dBu across the total resistance, and assuming that the 2nd harmonic falls proportional with fundamental level. This means that the achievable trimming range is less than 1%. If more is needed, lower fundamental levels must be chosen.
Comparison with other trimmers will have to wait...
Just saying: if you want to dive towards -140 dB I'd be careful with this sort of assumptions. Verifying everything and assuming nothing is the most promising approach in this territory.
Samuel
Last edited:
My previous measurement was about other optocoupler - Vactrol VTL5C7. They showed bad results.
You can see Silonex NSL-32SR3 data:
http://www.google.lv/url?sa=t&rct=j...QhWZNgnzVJmNq4g&bvm=bv.41642243,d.bGE&cad=rja
OK. Here is NSL-32SR3 photoresistor set to 1k. Results are no perfect as previous, but not bad:
For comparison, here is 1k linear resistor in the same place:
Victor.
Like Samuel, I had seen worse measurements of the Silonex optocouplers.
-120dB is the noisefloor for my system, so a bit of 3rd above that is not a big issue.
OFC for an analyzer that aims for -140 it is, unless lower "values" can be used.
Maybe in little different conditions... My condition is only 17,7mV RMS (50mV p-p) on photoresistor.
Dont put words in my mouth. I clearly stated I was refering to R's (fixed resistors). And clearly said contacts on controls and switches were a biggy -- meaning a big issue for thd.
If you followed closely the 339A upgrades, pots and switches had to be changed and cleaned to get H2 and H3 down to the levels we now have. Also capacitors. Which BTW thd at the same level as Victor's.
Yes, we are and have been aware of the pot contacts for decades now. Demian has seen this also for as many years. These things show up in emplimenting amplifier circuits -- preamp and amps. Stuff not shown in SIM of circuit topologies.
This is an area John Curl has explored a lot and has used the best parts and controls that dont cause added harmonics. Its been a High-End issue for a long time. For myself, I said before, i can add to it my noise analyzer to find R's/pot etc that not only have the lowest thd but lowest noise - both.
The real goal: Once we get a system that can both measure and produce frequencies at -140, then we can develope audio circuits and finished product that measure what the SIM say they are capable of doing.
Thx-RNMarsh
Last edited:
Hi Samuel,
Do you have a list of preferred passive components on your web pages?
I've seen the one for cap types.
Resistors are on the same page as capacitors, but the text is pretty dated. I'm working on a new revision, but progress is slow...
IMHO the key trick to get very low distortion from a resistor (or passives in general) is not so much about picking a special part, but to keep the AC voltage and AC power dissipation sufficiently low by using series (and possibly parallel) connections. With this approach, ordinary metal or thin film parts (I'm often using MiniMELF ones) seem to work well, and even in the sum they tend to cost less than many precision parts (bulk metal, wirewound). The so far open question is how many series/parallel combinations are necessary for a given total resistance value, AC voltage and distortion goal. I'm currently researching this.
Samuel
Hi all,
I was just wondering if an SVF oscillator might have lower distortion with series / parallel capacitors?