See Chapter 13 in my book "Designing Audio Power Amplifiers", specifically pp. 264, 265. There I show a plot of measured distortion of a 1/4-W metal film resistor as a function of frequency when it is dissipating its rated 1/4-W of power due to the AC flowing through it. This measurement largely captures only the thermal-related distortion. I think this resistor had a rated TC of about 100 ppm/C.
I was first introduced to the importance of thermal resistor distortion around 1982 when I designed my MOSFET power amplifier with error correction. I was getting higher distortion than I should have at low frequencies. When I put two 2-Watt metal film resistors in series to form the feedback resistor the LF distortion went way down. This was only a 50-watt amplifier, but the feedback network inpedances were fairly low (the feedback resistor was only 4.3k).
Cheers,
Bob
Bob,
In the mid 60's resistor distortion got lots of attention. Standard testing methods, new test equipment etc.
What is interesting is that it took until the 80's for audio gear to get good enough that resistors began to be a problem.
I think it is now possible to design power amplifiers where the distortion products are -140 db or lower.
In the mid 60's resistor distortion got lots of attention. Standard testing methods, new test equipment etc.
What is interesting is that it took until the 80's for audio gear to get good enough that resistors began to be a problem.
I think it is now possible to design power amplifiers where the distortion products are -140 db or lower.
Isn't there a bipolar 'lytic that they somehow manage to make that has the foil etched on both sides? It seemed as if I had read about that somewhere as being distinct from back-to-back conventional lytics.
Maybe I just dreamt it. That's been happening a lot lately
Isn't there a bipolar 'lytic that they somehow manage to make that has the foil etched on both sides? It seemed as if I had read about that somewhere as being distinct from back-to-back conventional lytics.
Maybe I just dreamt it. That's been happening a lot lately
Nope I guess I did read it. From the wiki for electrolytic capacitor: "Special capacitors designed for AC operation are available, usually referred to as "non-polarized" or "NP" types. In these, full-thickness oxide layers are formed on both the aluminum foil strips prior to assembly [my italics]. On the alternate halves of the AC cycles, one of the foil strips acts as a blocking diode, preventing reverse current from damaging the electrolyte of the other one."
As I recall these were the lowest distortion, beating biased and biased back-to-back.
Greg Timbers, the venerable speaker designer at JBL, made some loudspeakers that required a battery to bias the 'lytics. People I believe argued that there was more distortion than some alternatives, and he insisted that they simply sounded better.
I still have seen no compelling evidence that this matters at all not even a little. That is the WAVAC remains a large (pink?) elephant in the room.
In late 70'th our laboratory in Tomsk worked on 100W thick film IC power amp. Feedback resistors had to be moved out of IC because of the same reason, despite of requirement to minimize number of external components.
How about batteries then, not practical? The battery powered Sutherland products receive good press.
Personally I don't care for their use in audio but they do seem to have their place.
My ac calibrator (Optimation from the early '70s) uses a 2watt Vishay metal foil feedback resistor in the 10v amp just to meet specs. It's not new knowledge that thermals can have a big impact on linearity. Especially when you are trying for less that 100 ppm. accuracy.
I was just thinking that adding heat would reduce the impact of the ac component.
Sent from my iPad using Tapatalk HD
I was just thinking that adding heat would reduce the impact of the ac component.
Sent from my iPad using Tapatalk HD
I think most of the time, unless the device temperature coefficient itself is a strong function of temperature, the notion that higher quiescent temperatures reduce the effects of signal-induced self-heating is a misconception, albeit intuitively plausible.
If one is working at very low temperatures where carrier freezeout starts to occur (say below 100K in conventionally-doped silicon) then it is a very different story. For example 77K (liquid nitrogen, hence very convenient as such things go) is a very poor temperature for operation of silicon JFETs.
Brad Wood
If one is working at very low temperatures where carrier freezeout starts to occur (say below 100K in conventionally-doped silicon) then it is a very different story. For example 77K (liquid nitrogen, hence very convenient as such things go) is a very poor temperature for operation of silicon JFETs.
Brad Wood
Now for temperature stabilization against ambient changes, the 199 was a nice realization, albeit rather consumptive of power. And the buried zener was pretty quiet as zeners go.
But if the voltage reference portion were itself subject to significant dissipation changes I don't know how well the temp servo would follow them.
The $10K that Mr Sutherland charges for a twin case phono stage with low part count is more mind-boggling, imo.
+20 years ago, Avantgarde in Switzerland came out with an add-on battery PS for the MAS (Metaxas) preamp they distributed in Europe.
Bumped up the price of the pre considerably, and base retail wasn't exactly cheap to begin with. *
PS design has evolved in 20 years time, and batteries by no means deliver clean DC.
For a phono stage that has to stay under $1k retail, batteries may be a/the cheaper solution.
(* couple of years later, Mr Gassmann of Avantgarde, came out with an all-battery phono/pre/power amp combo under his own brand name. The powersupply of the power amp carried an automatic charging circuit, plus 8 Sonnenschein batteries of a size used for motorcycles. Total cost ~45K Swiss Francs in the early '90s. Afaig, Mr Gassmann still has the hots for batteries, his latest XA preamp, in a truly beautiful cast-alloy case, runs on can juice also)
"If you build it they will come..."
But has SY compared a -120 db amp to a -140 db one?
Feed-forward temperature stabilization: 2 layers of resistive conductor, and opposite signal on the 2'nd layer. Or 2 bifilar wndings, 2 wires, noninductive. No, 3 wires: 3'rd to sense average temperature.
Reminds me famous joke about modern engineer, light bulb, and 8 technicians.
- Status
- Not open for further replies.