And the question is, is it relevant to Audio amplifiers? If so I need to learn more.
Irrelevant, of course. Unless you are one of those that can hear the grass grow.
Good to know. My SMD project continues. Although from Edmund's experience, it may be worth using MELF type resistors.
Grass grow. Unfortunately, no. Have had slight tinitus for as long as I can remember.
> "Although from Edmund's experience....."
That was only an "anecdotal" experience. I'm not claiming that all SMD's are that bad.
FYI, it was a ESI Waveterminal 192X sound card that was equipped with horrible resistors. On the outside I couldn't see what kind of SMD's were used, of course.
Cheers, E.
That was only an "anecdotal" experience. I'm not claiming that all SMD's are that bad.
FYI, it was a ESI Waveterminal 192X sound card that was equipped with horrible resistors. On the outside I couldn't see what kind of SMD's were used, of course.
Cheers, E.
It does sort of make sense though. The MELF looks to me like a Through hole but without the leads. Which without further research looks ideal for my application.
Paul
on the level of discussion of 0.1 to 1 ppm/V resistor nonlinearity pcb material limitations are relevant to audio frequency circuits, impedances often encountered in audio amp circuits
that doesn't mean "audible" - we left that behind a long time ago
my point was that there are always tradeoffs - smt has a different set than through hole, sometimes you have to think about the differences and allow for them in design, layout rather than just think you reduced all of your problems
that doesn't mean "audible" - we left that behind a long time ago
my point was that there are always tradeoffs - smt has a different set than through hole, sometimes you have to think about the differences and allow for them in design, layout rather than just think you reduced all of your problems
MELF packaging is no guarantee for quality, metal film or thin film is (and even that doesn't guarantee "perfection", RTFDS). Are you aware they do carbon film MELF devices?
Reason why metal film devices are rare in SMD is because the limited range of achievable values, and of course spiraling is not an option.
Fortunately, thin film (high resistivity) layers like titanium nitride are as good as the pure metal films, in all compartments.
Did you mean tantalum nitride?
https://www.youtube.com/watch?v=chF7SI2b95s
http://www.ttelectronicsresistors.com/pdf/application_notes/ThinFilmCircuits.pdf
https://www.youtube.com/watch?v=chF7SI2b95s
http://www.ttelectronicsresistors.com/pdf/application_notes/ThinFilmCircuits.pdf
Both are used, there have been patents issued for combined TiN and TaN resistors (the temperature coefficients can cancel.)
Neat trick, anyone know if these are actually available?
Best wishes
David
Hi Waly,
This is what I have found... But didn't know MELF come as carbon films. RTFDS
At least there is hope. My searches for SMD resistors in the usual places has resulted in many thick films and thin films. The thin films are preferable in my humble opinion.
Will look for these titanium nitride films...
Thank you for this information.
Paul
Hi Waly,
This is what I have found... But didn't know MELF come as carbon films. RTFDS
At least there is hope. My searches for SMD resistors in the usual places has resulted in many thick films and thin films. The thin films are preferable in my humble opinion.
Will look for these titanium nitride films...
Thank you for this information.
Paul
Tantalium nitride are very expensive. Nichrome thin film components have the same performance but are ferro magnetic (I believe) therefore some interactions and induced distortion in magnetic fields. Tantalium nitride is then better.
Nichrome can make subppm voltage coef; What about Tantalium nitride ?
Tantaliu nitride must be important in Telecom designs but is it important in audio? Test results are welcome.
MELF have high reliability with temp , dissipation and pulses; is this important in audio ?
JPV
Your calculation is for a 100w amp with 1ppm feedback resistor. Another way is 100w/8 ohms is ~28V on the resistor at full power. this gives :
HD3 = 1ppm x 28/5.9 = 4.7 ppm which is 0.00047% your result.
In my example I used a 50V output and 10ppm resistor so about 20 times more which gives 0.01%
(We should also use the peak voltage and not the rms).I read the pages that Self wrote in his 'Small signal audio design' book. He gives there a table of resistor's THD with different ppm' s under different voltages;
the results correspond to B Hofer's formula This table was made using Spice ( so the FFT) with a model that is given in another book ( Active Crossover)
JPV
Yes, this is so very true. Often, the very first step for me is figuring out what I don't know.
Cheers,
Bob
Your calculation is for a 100w amp with 1ppm feedback resistor. Another way is 100w/8 ohms is ~28V on the resistor at full power. this gives :
HD3 = 1ppm x 28/5.9 = 4.7 ppm which is 0.00047% your result.
In my example I used a 50V output and 10ppm resistor so about 20 times more which gives 0.01%
I read the pages that Self wrote in his 'Small signal audio design' book. He gives there a table of resistor's THD with different ppm' s under different voltages;
the results correspond to B Hofer's formula This table was made using Spice ( so the FFT) with a model that is given in another book ( Active Crossover)
JPV
I guess I was assuming that Bruce was using rms values all around in his example where he came up with that formula. Did I miss something that he said that would remove the apparent ambiguity of using peak signal or rms signal in this formula?
I used 1ppm/V as what I thought was a reasonable compromise between thick film and thin film resistors. He had 1ppm/V as the high end of thin film and the low end of thick film. I think using 10ppm/V is overly pessimistic for the situation we are talking about, since for feedback resistors the starting reference point for any designer that cares is a discrete metal film with adequate wattage.
If I was designing an amplifier with SMT resistors, I would not hesitate to go with a through-hole, larger resistor, in the few important places where the effects of the voltage coefficient are not mitigated by feedback - most particularly in the feedback network itself.
I would love to know where in his analyzers Bruce uses SMT vs. through-hole and why.
Cheers,
Bob
Guys, I may be dense today, but:
- A typical 1-3W power metal film resistor made by Vishay has no less than 250ppm/K thermal coefficient.
- A typical 10W metal oxide resistor made by Yageo has no less than 300ppm/K thermal coefficient.
- One of the best and most expensive resistor series (except for some military and/or space applications) available today, the Vishay PHP series, has 25ppm/K and 0.1ppm/V thermal respective voltage coefficients (SMD, 1W, BTW).
Now assume a peak voltage of 40V, this would amount to 40ppm (4ppm for the best) variation due to the voltage coefficient. It would take only 40/250=0.4/25=0.15 degrees to have the same variation due to temperature. Of course, there is the 5.9 H3 factor but still, in such applications, thermal coefficients are masking far and beyond the voltage coefficients, even for the best of the best resistors available today.
So what's the point discussing about the voltage coefficient in audio applications? As far as I know, this can be an issue for very high voltage dividers, very high crest factor attenuators, and other specialty applications. I'm afraid you are ghost hunting.
When Richard Marsh got his AP2700 he post some pics of the inside.
I noticed on one side of the analyzer is all the digital stuff done in SMT. On the other side of the analyzer is all the analog stuff done in through hole bent wire.
May this helps to answer your question.
Hi Way,
I have lots of dense days with the occasional brain fart. But I don't think you're having a dense day. These are the sorts of informal sanity checks I always like.
One thing I would point out is that nonlinear distortion resulting from the temperature coefficient will depend on the frequency of the audio signal, due to the thermal low-pass filtering in the resistor. At higher frequencies, the thermal effect will just approach a static resistance change. But if we are talking about distortion down to 20Hz, then we need to take the possible dominance of the thermal effect seriously.
So then one question that remains is which phenomenon dominates in Bruce's application where I think he is mainly dealing with voltages less than 10V rms.
Cheers,
Bob
And that's probably because good quality passive SMD parts (thin film resistors, 1% C0G ceramic caps, etc...) are, for good reasons, much more expensive than the through hole parts.
Reasons include: SMD resistor value laser adjustment range is limited (no spiraling) so the film thickness is more critical, caps are, due to the size constraints, always thin multilayered (a more expensive technology) etc...
I don't think there is any gap in performance between the through hole parts and the SMD parts, if one is prepared to pay the price. Of course, in the same price range, through hole parts may always win the game.
Of course, the relevance/audibility of resistors is questionable, in fact I believe no human could double blindly identify thick vs. thin film resistors in a power amplifier. Perhaps the excess noise could be audible in a MC phono preamp, but even that is IMO questionable.