No, I wouldn't think twice about it. It is just plain silly.
I did a bit on microphones and it was quite difficult to get the differences between a $10 knock-off and a $300 good quality microphone to show in the final production.
My OPINION is even a comparison between carbon composition and metal film voltage dividers would not be useful in that manner.
As to cheap but surprisingly good resistors would be Mouser's Yaego line from my tests.
I think the stated problem, a 100K stepped attenuator for an ~1V line level signal in front of a PA, sets sufficient boundary conditions to compute the worst case signal degradation from the resistors themselves. Vishay has been playing the audiophile card for decades.
Last edited:
Bulk metal resistors have decreasing performance with decreasing frequency. However they normally are higher wattage than the resistors they are compared to.
In a volume control, a standard potentiometer has 45 dB of range control. So somewhere around 1-2 dB steps should be fine. There are two ways to arrange the resistors. One is a sting of resistors and the switch selects the connection point to the string. As a normal value for the string would be 100,000 ohms and the typical maximum voltage would be 2 volts RMS the power dissipation would be .04 milliwatts. Even with a 10K string value the power would only rise to .4 milliwatts.
Using 1/4 watt resistors in a string would give a minimum power dissipation at .16% of capacity.
The other method is to have multiple 2 resistor voltage dividers and select which pair to use. If each is 100,000 ohms and 24 pairs are used the load to the audio source would be 4,167 ohms, a bit low for my taste.
In a volume control, a standard potentiometer has 45 dB of range control. So somewhere around 1-2 dB steps should be fine. There are two ways to arrange the resistors. One is a sting of resistors and the switch selects the connection point to the string. As a normal value for the string would be 100,000 ohms and the typical maximum voltage would be 2 volts RMS the power dissipation would be .04 milliwatts. Even with a 10K string value the power would only rise to .4 milliwatts.
Using 1/4 watt resistors in a string would give a minimum power dissipation at .16% of capacity.
The other method is to have multiple 2 resistor voltage dividers and select which pair to use. If each is 100,000 ohms and 24 pairs are used the load to the audio source would be 4,167 ohms, a bit low for my taste.
Last edited:
The whole string has 1/4 watt total capacity? I'm not sure but wouldn't the worst case be the first step? At 2dB per step that would be ~20K or .008mW
Last edited:
That depends on how you switch. You can also have 24 pairs each of 10k ohm. And you only connect one pair at a time. The source will only see 10k load at any time. (And you can still connect all the heads and all the tails together, respectively.)
https://www.tubecad.com/2011/04/26/Ladder Attenuator.png
http://www.diyaudio.com/forums/head...s-zgf-headphone-amplifier-20.html#post4347276
Patrick
Last edited:
Interesting statement.
Can you please define "performance" as used here?
What would "decrease" actually mean here?
Ed, you are my favorite guy to quote.
(I couldn't find the dancing banana, which my Tiffany Loves). Now you are in for it. Would you be kind enough to bless us with your report or research on micrphones? Either link or attribution that non members of IEEE can find and download? Please?
Now you got me thinking again about resistors...I recall a Demian post where he used some cheapo carbon film resistors for some test he was going and found they were "relatively" distortion free. He was surprised. He was even surprised that even the china load resistors for for audio amp testing showed surprising lack of parasitics and did a good job when immersed in water for derating.
Finally, I had another question but now I'm having a senior moment.
Cheeks,
I’ve been getting the resistor packs from Digikey which I believe are the Yageo resisters you are referring to. Really good value. 480 pieces MFR series 1/4W for ~$30. Unit price ~ 6 cents. http://www.yageo.com/documents/recent/Yageo LR_MFR_2013.pdf
Mouser no longer sells through hole assortments unfortunately. On this topic does anyone have a line on some 1/2W MF through hole packs? It would be really useful for diyers to be able to source full tables of let’s say Beyschlag MBB or something comparable.
The Chinese assortments floating around eBay are not the best solution.
I was also surprised, I went to repeat some old measurements that I did on original AB carbon comp resistors with cheap carbon films and could not get anywhere near the noise or distortion I had before.
BTW AB still runs the worlds largest 4 sided clock, I can see it from my daughter's living room window.
It is much better to use a make before break rotary switch. The tubecad design will give pops if there is any DC voltage present or stored charge on the coupling capacitors.
No the first resistor would be in series with a combination of the rest of the chain and the next stage input resistor, so it might see half of the signal.
As the frequency decreases there is a phase shift between the mechanical strain correction mechanism and the normal thermal errors, so the distortion goes up.
Ed, you are my favorite guy to quote.
Would you be kind enough to bless us with your report or research on micrphones? Either link or attribution that non members of IEEE can find and download? Please?
Cheeks,
YouTube
Last edited:
I use the modified ALPS (break before make) in multiple headphone amps without any pops.
No such issues either in my system. For sure no caps. In addition, the wiper has a 100k resistor to Gnd connected permanently. The shunt resistor value takes that into account.
Works like a treat for me. Best attenuator I have.
Patrick
Last edited:
I guess you missed my point a chain of series resistors each 1/4W has a lot more dissipation capacity than 1/4W, here the largest one was a far smaller number than 0.16% of 1/4W. I was taking the case of input impedance >> than the ladder for simplicity.
So you are not doing what that schematic shows!
Scott, as usual. I figured absolute worst case of a high impedance attenuator into a lower impedance load. In proper use thermal distortion might be measurable with a bit of technique. However voltage coefficient might show up. Voltage coefficient is usually worse when thermal ain't so good.
I am, if you consider the 100k resistor to be part of the amplifier downstream.
Most amplifier will have such a Gnd resistor at the input anyhow.
Patrick
Back to the original point the problem has boundary conditions now use the published specs to compute the expected answer. Grading by measuring resistors at 1/2 or 1/4 rated power is not applicable to this problem.
OK, so recompute the resistors to get reasonable steps with a 100k input resistor. Maybe the 8uW becomes 16uW. Vishay only specs voltage coefficient from 10% rated voltage and up so I don't know the answer.
Last edited:
Scott,
I think the real issue is comparing a carbon or cermet film variable resistor to a switched attenuator.
Biggest issue for the stepped attenuator is probably the switch quality. In my listening tests some switches really do have perceptable differences. But I am not sure what the measured difference is that causes the issue. However I haven't really spent much time on the issue, just use better switches.
On a potentiometer I can see differences on wiper noise.
As to spending money on resistors, my point was measurements have shown for most applications a few cents per resistor is enough. But for a quarter you can do better. I have only spent money on the bulk film ones to have test samples.
I believe we disagree on what really happens with low level signals near zero volts. I think that is where component differences show up. I have seen some evidence of a dead zone but my measurement technique does not rule out other possibilities like a rise in signal to noise as the signal increases.
I think the real issue is comparing a carbon or cermet film variable resistor to a switched attenuator.
Biggest issue for the stepped attenuator is probably the switch quality. In my listening tests some switches really do have perceptable differences. But I am not sure what the measured difference is that causes the issue. However I haven't really spent much time on the issue, just use better switches.
On a potentiometer I can see differences on wiper noise.
As to spending money on resistors, my point was measurements have shown for most applications a few cents per resistor is enough. But for a quarter you can do better. I have only spent money on the bulk film ones to have test samples.
I believe we disagree on what really happens with low level signals near zero volts. I think that is where component differences show up. I have seen some evidence of a dead zone but my measurement technique does not rule out other possibilities like a rise in signal to noise as the signal increases.
Last edited:
Still waiting to take that on. A Nobel is there for the taking. We only disagree because one result rewrites physics.