Edmond Stuart said:
NO!I said: "using the same resistors from the same batch". That also means using resistors of the same value. So for a 10:1 divider you will need ten (10) resistors, I repeat, of the same value, and of course with the same nonlinear (if any) behavior. Then nil distortion from the resistors.
Edmond,
Wouldn't this configuration also reduce noise? Multiple resistors of the same value would add noise in RMS fashion (they are uncorrelated).
For the 10:1 ratio and a 100K ladder: Nine 10K resistors would RMS to 30K (for noise figure calculations, not resistance) while one 90K resistor would RMS to... 90K and resistor noise is proportional to resistor value.
It seems to me that your configuration would reduce noise by 9.5dB (but I might be misremembering something).
Jeff
PHEONIX said:
Hi Arthur,
Apparently the guys at Halcro were afraid of the self heating effect and used a bunch of larges MK8 (2W) resistors.
I'm also surprised about using carbon R's. According to Cyril Bateman, who did a lot distortion of measurements on C's and R's, one specific brand of carbon film R's produce ~0.00013% distortion when loaded to 25% rated power.
Welwyn RC55 (metal film), on the other hand, performs much better: 2nd harm. -127dB, 3rd harm. -125dB.
As for SMD thick film, you are right, totally unsuitable for ultra low THD. I had to replace those 'sic film' types (603) on my ESI sound card by professional MELF types. (those ******* idiotes at ESI realy know how to spoil a good sound card)
Cheers,
Edmond.
Christer said:
Hi Christer,
Sure. Making, for example, a 30:1 divider this way is not very practical.
The only reason I put forth this trick was to show that it's theoretical possible to make a divider with far lower distortion than the distortion of the resistor itself. Also, that distortion figures below -130dB from a simulator are not necessarily fooled by the distortion of passive
components in real life.
Cheers,
Edmond.
Edmond Stuart said:
Yes, and even if not practical to follow to an idea to 100 %, it is still interesting to figure out ways to cancel errors. If voltage dependancy is a non-linear function, it would then still help to have a few resistors in series, than using a single larger one.
For thermal effects, using an oversized resistor is better than one based on straightforward power calculations, but that has been covered in earlier threads.
Resistors in series is a good way to make a precision divider, but again there is a lot of technique Precision divider Just stringing them in a line may not be the most accurate. Also looking at the construction of a high voltage scope probe will show that there are tricks to getting good frequency response accuracy. Also I ran across a reference that said voltage coefficient is a non-linear phenomena and points out a 2 PPM VC can be a bigger problem that one might think, especially in a high voltage application (like a tube amp).
Resistor noise is a direct function of the resistor value, doesn't change with material. Excess noise, the noise generated by the resistor from the application of power, is very different with different technologies. Wirewound will be the lowest and carbon composition the highest. However carbon composition will handle transient power better than any of the other types. I confirmed this attempting to make a controlled impedance in series with a surge generator. Fried some big Caddock resistors. A bunch of 2W carbon comps worked fine. There is no single good answer for all applications.
Resistor noise is a direct function of the resistor value, doesn't change with material. Excess noise, the noise generated by the resistor from the application of power, is very different with different technologies. Wirewound will be the lowest and carbon composition the highest. However carbon composition will handle transient power better than any of the other types. I confirmed this attempting to make a controlled impedance in series with a surge generator. Fried some big Caddock resistors. A bunch of 2W carbon comps worked fine. There is no single good answer for all applications.
Thin film chip resistors are manufactured using tantalum nitride (TaN), ruthenium dioxide (RuO2), lead oxide (PbO), bismuth ruthenate (Bi2Ru2O7), nickel chromium (NiCr), and/or bismuth iridate (Bi2Ir2O7) as resistive material. Via a sputtering process, a very thin metallic film is deposited on a substrate under vacuum conditions. The deposited film is continuous and defect free conferring stability, very low TCR, very low current noise and negligible non-linearity.
In contrast, thick film resistors normally use resistance metal oxides with glass-based materials to form a paste for printing on the substrate. After burning off the solvents and after few other heat treatments the glass filled resistive material has a sponge like structure with random inclusions and defects. These films will absorb gases and moisture (if not hermetically sealed) causing drifts and long term instability. The random defects are an obvious source of noise and detrimentally affect the temperature coefficient of resistance, TCR.
Both thin film and thick film resistors use laser trimming to adjust the resistance value during final steps of the manufacturing process. However, the heat generated during laser trimming causes micro-cracks on a thick film resistor and therefore affects the short term and long term stability. Thin film resistors need less laser power than that needed for thick film resistor. Thin film resistors do not exhibit micro-cracking during laser trimming. Consequently thin film resistors show superior stability and noise performance.
In contrast, thick film resistors normally use resistance metal oxides with glass-based materials to form a paste for printing on the substrate. After burning off the solvents and after few other heat treatments the glass filled resistive material has a sponge like structure with random inclusions and defects. These films will absorb gases and moisture (if not hermetically sealed) causing drifts and long term instability. The random defects are an obvious source of noise and detrimentally affect the temperature coefficient of resistance, TCR.
Both thin film and thick film resistors use laser trimming to adjust the resistance value during final steps of the manufacturing process. However, the heat generated during laser trimming causes micro-cracks on a thick film resistor and therefore affects the short term and long term stability. Thin film resistors need less laser power than that needed for thick film resistor. Thin film resistors do not exhibit micro-cracking during laser trimming. Consequently thin film resistors show superior stability and noise performance.
1audio said:
Correction: Thermal resistor noise is a direct function...
The excess noise (which comes on top of the thermal noise) is indeed material dependent but it's not necessary only related to the application of power. Wirewound is the lowest because the conduction mechanisms in metals are the closest to the ideal thermal noise limit.
zigzag
Hi Christer,
What about putting that bunch of equal resistors side by side. With a zigzag layout the total length will be reduced by about a factor of four. Also the inductance will be smaller. Using 30 pcs. of 1/8W resistors will occupy ~8cm, roughly equal or less than the length from the output back to the inverting input.
Cheers,
Edmond.
Christer said:
Hi Christer,
What about putting that bunch of equal resistors side by side. With a zigzag layout the total length will be reduced by about a factor of four. Also the inductance will be smaller. Using 30 pcs. of 1/8W resistors will occupy ~8cm, roughly equal or less than the length from the output back to the inverting input.
Cheers,
Edmond.
Re: zigzag
Will it help to cancel out inductance? Isn't the main inductance internat to the resistor rathet from from approximating is as a conductor?
On second thought, does it matter? The relative inductance ratio for the divider should be the same as the relative resistance ratio, thus nog affecting the frequency response much.
Edmond Stuart said:
Will it help to cancel out inductance? Isn't the main inductance internat to the resistor rathet from from approximating is as a conductor?
On second thought, does it matter? The relative inductance ratio for the divider should be the same as the relative resistance ratio, thus nog affecting the frequency response much.
In a resistor array such as has been discussed here the inductance will be a very small issue if its even present, the dominant issue will be the shunt capacitance.
I have read that capacitance is the dominant reactance in resistors over 100 Ohms, inductance below. However general rules like that mean very little when you have the odd part.
I have read that capacitance is the dominant reactance in resistors over 100 Ohms, inductance below. However general rules like that mean very little when you have the odd part.
However its the RC ratio that matters. All the caps in are in series and the resistors are in series. Probably would not affect things to model them as separate strings instead of as a ladder. The result is that the reactance of the caps follows the resistance of the resistors EXCEPT for the external stray aspects. And those can be significant in the layout.
i recently read an article about carbon composition resistors introducing distortion in tube amps. i always knew they were noisy, but when you apply voltages in the range of 100-600 volts to them, carbon composition resistors begin to have a nonlinear reaction to the voltage, so that as you vary the voltage, the resistance changes slightly, causing the current to be nonlinear with respect to the applied voltage. this distortion is subtle, but in a guitar amp with many of these resistors in it (as a "ferinstance") the distortion adds up. heating effects also play a role. it is well known (at least when i was growing up) that carbon button microphone elements were very nonlinear, not only in their response to sound, but their resistance in general. they were also exceedingly noisy. one good thing about them is that they can take abuse from momentary excessive currents. i used to work for a company that had 8 ohm 2 watt carbon comp resistors for discharging large electrolytics as part of a test procedure. well one day a tech built a new one from parts from the parts room. he began his test procedure, and when he got to the part of the test procedure where the caps had to be discharged, he got just a very weak spark on one cap, none on the other. he measured the voltage on the caps and one of them was reading 55V (only slightly discharged from 65V) and the other one at 65 volts. he measured the resistor and it was open. so he built another discharger, same thing, the resistor burned open when attempting to discharge the caps. we found out that a parts manager had made a decision to buy carbon film resistors instead of carbon comp (i.e. he cheaped out), thinking they were cheaper AND better. big mistake, since the carbon comp could handle the current surge from the caps and the new resistors could not, and he had to order carbon comps anyway.