BTW here some interesting facts derived from the work of Cyril Bateman as measured in 2002. Please check info on MKS/MKT versus MKP/FKP. Also differences between foil and film. What already was known by many... There is no technical reason NOT to choose a polypropylene capacitor over a polyester one despite what the nay sayers tell us. On the contrary. Post 360:
http://www.diyaudio.com/forums/head...2-based-se-class-headamp-without-heat-36.html
And some more:
https://linearaudio.nl/cyril-batemans-capacitor-sound-articles
http://www.diyaudio.com/forums/head...2-based-se-class-headamp-without-heat-36.html
And some more:
https://linearaudio.nl/cyril-batemans-capacitor-sound-articles
Last edited:
Yes, the original articles by Cyril Bateman. Read them then but it is nice to have those for reference.In the second link in my post you can see Cyril Bateman and Jan Didden measured some more.
Last edited:
I've built a stereo pair of 909 clones. One works perfectly, the other oscillates. I'm yet to find why only one of them works.
Maybe the lesson here is that the design is a bit temperamental. Both use components as originally spec'd except for the output transistors.
Maybe the lesson here is that the design is a bit temperamental. Both use components as originally spec'd except for the output transistors.
DC servos can be really useful, not just an exercise in complexity. High value coupling caps are huge when they're films, and electrolytics can leak and wear out. A DC servo can be built in a very small size - under 1.75mm height and a few square centimeters, and built with components that will never wear out and can handle surface mount assembly.
They can be designed with dead flat passband response, essentially unmeasurable distortion and low noise. While the circuit is certainly more complex, the total cost is relatively low, especially compared to a nice, large film cap.
Coupling caps used on inputs can generally be made much smaller, but then, the offset of the amplifier that follows them is still uncorrected. With a power amp, you need a huge cap to couple to a 4 ohm speaker - not small, cheap or high endurance. There are modern polymer caps used for solar panel inverters, but they're approaching the size of motorcycle batteries, and that adds its own PITA factor.
Someone mentioned before that the TLC271 is a reasonable DC servo amplifier, and while you can do better, something like the LME49710 is a really bad idea. The input bias and offset current of that amplifier is huge compared to a CMOS op amp, and that can ruin the final DC offset of the integrator, even with equal source resistances on the inputs.
The TLC271 has a pretty high offset of 10mV max, it's pretty noisy, and given that it's a CMOS op amp, its extremely low frequency voltage noise is huge. This noise is actually most important due to the fact that the integrator will pass low frequencies with pretty high gain. Sure, it might not be audible, but it's a fault that can be avoided. Nobody wants popcorn-like noise in the output, even if it is pretty small.
The Analog ADA4522 series of amplifiers is a much better choice for a DC servo - it has extremely low offset (5µV) low noise, even at low frequencies, and low distortion at low frequencies. Distortion isn't so important, but if the integrator isn't pre-filtered, or if it barfs on the wideband input presented to it, distortion can still be generated. It's not dirt cheap, but only about $1/channel, so it's more cost effective than some other choices.
Personally, I'd leave the Quad 909 DC servo alone and be happy about it 'as is', but if you must roll some op amps, the ADA4522 series will make a probably meaningless improvement to the circuit. You'll get extremely low overall offset and less noise with the swap, but it's not clear that you need this.
As was said earlier, leave the DC servo alone unless you know how to design one. They are not difficult circuits to design, but you need to pay attention to some specific issues. A simulation can be very helpful, since tuning the corner frequency and loop gain can seem counter-intuitive - it's nice to have the overall design verified before you order parts.
Actually i think i'll go with SMR which i think may be better again than MKP.
One question here though, you guys are saying up to 470nF is ok, but do i keep the voltage at 100V?
Or use a higher voltage?
One question here though, you guys are saying up to 470nF is ok, but do i keep the voltage at 100V?
Or use a higher voltage?
Last edited:
Try it out. Use higher rated caps as long as they as they are not too large. Keep the size of the original as reference.
SMR measures fine. You can do the reading yourself:
https://linearaudio.nl/sites/linear...an EW 11 2002 mar 2003 caps 100 nF to 1uF.pdf
OK, what is next ? The mains connector ? 🙂
SMR measures fine. You can do the reading yourself:
https://linearaudio.nl/sites/linear...an EW 11 2002 mar 2003 caps 100 nF to 1uF.pdf
OK, what is next ? The mains connector ? 🙂
Last edited:
Like you I am technically educated. Still I know the virtues of using different parts even if my knowledge should prevent me (at least from the extremes). I know the comments are sometimes over the top but even the most hard headed will hear differences when using for instance Black Gate NX HiQ caps.
Like the measurements of Bateman and Didden show us there are differences. It can be debated if the measured differences can be heard. Or just give in and try out. I choose for the latter. Simplest proof is using film caps instead of electrolytic caps in the signal path (coupling cap). No religion or strict technical knowledge needed, just listening.
When polypropylene caps measure (and sound) better than their polyester brothers why not simply use them ?
Like the measurements of Bateman and Didden show us there are differences. It can be debated if the measured differences can be heard. Or just give in and try out. I choose for the latter. Simplest proof is using film caps instead of electrolytic caps in the signal path (coupling cap). No religion or strict technical knowledge needed, just listening.
When polypropylene caps measure (and sound) better than their polyester brothers why not simply use them ?
Last edited:
Because polyester are smaller and cheaper than polypropylene.
If both are up to the duty, then what's wrong with using the smaller/cheaper capacitor?
Polypropylene caps are usually larger than polyester so could suffer more from stray capacitance. Their lower distortion is likely to be irrelevant for coupling cap duty, except perhaps if the cap setting the LF rolloff has wrongly been placed at the output end of a system where signal voltages are high.
Bateman's tests are helpful, provided people read them carefully and understand that they are much more relevant to speaker crossovers than coupling caps.
Bateman's tests are helpful, provided people read them carefully and understand that they are much more relevant to speaker crossovers than coupling caps.
As discussed earlier Wima MKP2 are as small as MKS or slightly larger. There is no physical barrier anymore. Keep using MKT/MKS if you like, I will opt for MKP anytime.
MKS2-XL are just somewhat larger than electrolytic caps with the same value. No reason to use electrolytic caps anymore till 10 µF 50V in 5 mm pitch. I use these for many years with great results. Price is no object when one buys them in large numbers. Anyone that uses these will likely tell the same: they are superior to most electrolytic caps.
I am not in this hobby for cheapest, I am in this hobby for best results at affordable costs.
MKS2-XL are just somewhat larger than electrolytic caps with the same value. No reason to use electrolytic caps anymore till 10 µF 50V in 5 mm pitch. I use these for many years with great results. Price is no object when one buys them in large numbers. Anyone that uses these will likely tell the same: they are superior to most electrolytic caps.
I am not in this hobby for cheapest, I am in this hobby for best results at affordable costs.
Last edited:
They look better too, nice square red boxes. I've just built Doug Self's Preamp and they and muse caps are used everywhere, size is not an issue anymore..😉
Do the test and please report back. I prefer MKS2-XL-10/50/10 over any electrolytic cap anytime (in the right circuit of course). I am not so sure if the looks make them sound better but I have seen uglier caps 🙂
Once there were Ropel MKP caps that were coffee brown and plain ugly. They were very good. Looks can be deceiving. Just look at the sleeves of chinese bulging electrolytic caps with their fine gold lettering 😀
Once there were Ropel MKP caps that were coffee brown and plain ugly. They were very good. Looks can be deceiving. Just look at the sleeves of chinese bulging electrolytic caps with their fine gold lettering 😀
Last edited:
In perspective to Bateman's figures: TI uses ceramic x5r input coupling capacitors on last evaluation board of a classD chip, publishes total ampboard distortion + noise down to 0.0005%. Slightly earlier almost identical board with polar electrolytics as coupling capacitors has almost the same performance, maybe a tiny little worse (as in maybe closer to 0.0006% minimum). With ~20kOhm inputimpedance TI did pick 10uF's for both types, x5r dropped maybe 20% because of DCbias ?
Measuring what you hear could be hard. I remember Bruno Putzeys posting BC 136type was only low impedance electrolytic that sounded better than standard types used on his Hypex ampboards, all other low impedance electrolytics he tried sounded much worse than the standard types. Measurements showed no difference, but on his private ampboards he used BC 136. Maybe it was the only blue electrolytic he tried 🙂
Measuring what you hear could be hard. I remember Bruno Putzeys posting BC 136type was only low impedance electrolytic that sounded better than standard types used on his Hypex ampboards, all other low impedance electrolytics he tried sounded much worse than the standard types. Measurements showed no difference, but on his private ampboards he used BC 136. Maybe it was the only blue electrolytic he tried 🙂
If we go from electrolytic to film and then to ceramic SMD (even X5R which is bad!) any discussion very likely will derail 😉
https://www.maximintegrated.com/en/app-notes/index.mvp/id/5527
Let's keep it here to TH caps as the thread was intended for replacement parts for the Quad 909. That one has TH parts. Another "which cap is best" thread can be started but it will be an eternal one I guess.
https://www.maximintegrated.com/en/app-notes/index.mvp/id/5527
Let's keep it here to TH caps as the thread was intended for replacement parts for the Quad 909. That one has TH parts. Another "which cap is best" thread can be started but it will be an eternal one I guess.
Last edited:
The ampboard with very bad x5r inputcoupling capacitors distorts in total(input to output all in between included in 0.0005%) much less than total Quad909. Single part contribution to thd, when boardgain was 10x, would be...? In output TI used MKP and COG, (probably not to make us feel better) because they think/know it makes a difference there.
(Mentioned Hypex electrolytics were bulk-decoupling, not coupling. Bruno's remarks resembled earlier quotes here)
(Mentioned Hypex electrolytics were bulk-decoupling, not coupling. Bruno's remarks resembled earlier quotes here)
SMR (Polyphenylene Sulphide) is readily available in the correct 10mm size, so i will probably go for that.
I'd also like to consider changing resistors R16, R17 (560R 4W 5%)
Many of the resistors with these specs seem to be wirewound.
What would be the best thing to use here?
Any opinions?
I'd also like to consider changing resistors R16, R17 (560R 4W 5%)
Many of the resistors with these specs seem to be wirewound.
What would be the best thing to use here?
Any opinions?
You like to create chaos on purpose isn't it ? 🙂 It is nice if you understand your own post though. Please let's not make this thread like the TPA thread.
Comparing apples and pears with too many variables and pulling in any extra item really leads nowhere.
Last edited: