It's as sweet and fine as ever, thanks.
Clarify, the sequence actually was:
On-Off-On (metallic screech!!) then very quickly OFF!! (I think some of the neighborhood dogs began barking at that point)
+1 on V1 or V2. There are some mute circuit and open loop recovery improvements in v2.0.
All of National's, er TI's, chip amps have UVLO (undervoltage lockout), meaning they turn off below a certain supply voltage. In nearly all composite implementations---including, I believe, Neuro's---the controlling op amp does not have UVLO. This results in the outer control loop opening and usually causes the control device to swing to its rails, which are falling too. Normally this doesn't matter since the power stays off until everything discharges fully. However, if you turn the power back on quick enough what happens instead is the chip amp starts back up with the control device in some funky state. This is an extremely difficult condition to design for and the result is usually some amount of rude noise. In particular, the 3886 tends to respond to this sort of overdrive from the control device by oscillating.
If you're curious to read up about this a good starting point for understanding the phenomena is integrator windup in PID controllers.
Perhaps true, sort of, if you are driving low sensitivity speakers using the upper end of the amplifier's power range. Both measurable and audibly not true with a high sensitivity horn driver at "normal" listening levels. The lower noise and distortion (particularly crossover distortion, which while low in the "bare" 3886 is not completely absent) in the 1-100 milliwatt range can make a discernible difference. If you're driving a SEAS T25CF-002 you probably won't hear much of a difference between the Modulus-86 and a "bare" 3886 . . . if you're driving a B&C DE250-8 I'd venture that you probably will. The 20dB difference in driver sensitivity makes a difference in where you fall on the amplifier noise and distortion curve for the same acoustic output.
Or, to clarify your point, ostensibly (since I can't state with confidence, but measurements thereof suggest so) it behaves as nicely in the lab as it does deployed (which most 3886 applications cannot state). How much of that would be realized by putting a THAT1200 in front of a 3886 I do wonder, and how much the extra loop gain's salutatory effects incur that advantage. Obviously, you won't get the PSRR from a THAT1200, but input buffering and CMRR are much better.
Bill--you can turn it down a notch. We DIY for our enjoyment and to appease our many psychoses, so let's let people do that!
If you do not have DC in your mains then you have not listen any difference. I have more and more DC, three connected before AV Marantz!!!
But other function of this DC blocker is to filter the ripple. The EPCOS 18,000 microF 105 ºC series have "Outstanding ripple current capability" and Low ESR.
Aleksandar did not believe in it hence the prototype uses proposed by him. After, in the second DC blocked, I wanted to check for myself if it was worth the extra cost 10x2 = 20 € + shipping with better specifications, and YES.
In other forums (Dark web) I strong recommend DC blocker x2 with these big EPCOS, slow fuse and varistor = "Maty Edition".
- The end about DC blockers in this three, I hope -
Last edited:
The speaker delay and isolation relay when fitted should have an instant OFF capability.
When that is activated it should restart the delay ready for the next ON.
Quickly going from ON to OFF to ON should cut the speaker relay and the speaker should be isolated for the full delay period.
I don't know if mod or par has a speaker delay and/or instant off.
the LM3886 has a mute function. That is normally wired with a very short delay and generally does not incorporate an instant mute.
The lm3886 also has a under-voltage OFF. But this is tied into the supply rail -ve voltage, it has no facility for instant OFF.
When that is activated it should restart the delay ready for the next ON.
Quickly going from ON to OFF to ON should cut the speaker relay and the speaker should be isolated for the full delay period.
I don't know if mod or par has a speaker delay and/or instant off.
the LM3886 has a mute function. That is normally wired with a very short delay and generally does not incorporate an instant mute.
The lm3886 also has a under-voltage OFF. But this is tied into the supply rail -ve voltage, it has no facility for instant OFF.
THAT 1200 series are bal to unbal receivers.Or, to clarify your point, ostensibly (since I can't state with confidence, but measurements thereof suggest so) it behaves as nicely in the lab as it does deployed (which most 3886 applications cannot state). How much of that would be realized by putting a THAT1200 in front of a 3886 I do wonder, and how much the extra loop gain's salutatory effects incur that advantage. Obviously, you won't get the PSRR from a THAT1200, but input buffering and CMRR are much better.
Bill--you can turn it down a notch. We DIY for our enjoyment and to appease our many psychoses, so let's let people do that!
If you have a balanced impedance source, then the bal to unbal receiver allows the full balanced impedance CMRR characteristics to be obtained.
If you have an unbalanced source, then you cannot obtain the balanced impedance CMRR.
Did your tweeters survive?
The muting circuit on an LM3886 works by having a largish resistor charging a capacitor to the point where the amplifier turns on. A fly-back diode over this resistor could prevent the kind of problem you mentioned.
Dewardth, you may be right, but I will find out once I have assembled the boards I have on order with Tom. I have a pair of Tannoy K3838 with about 98 dB/Watt iirc.
Derfnofred, as to the RFI immunity, that is the That´s work as you point out. For single ended a JFET front end, opa213x or such, works nicely too.
Ummmm... ?
I appreciate that, but still the THAT1200 on a "pseudo-balanced" connection seems to be pretty beneficial.
I think Walt Jung provides an insight to slew rate related distortion.
http://www.waltjung.org/PDFs/SID_TIM_1.pdf
The key is measurement voltage levels. So it would be interesting to see how measurements show.
Actually I was wondering what Tom's view is on this regarding the Mod 86, and whether this was a design consideration or not.
The MyRef has a form of relay control. I don't think I have ever experienced any problem with fast switching of power on any of the LM3886 chip amps.
Be aware that that article is from 1979, from the days of the "TIM hysteria". I think understanding of feedback and control theory has made some progress in the 35 years since that article was written.
My reference to the Wikipedia article was in response to your comment "Slew rate is an interesting parameter, I am wondering whether it might have any relation with any form of inter modulation distortion". The first paragraph of the wikipedia article states, rather clearly, "Limitations in slew rate capability can give rise to non linear effects in electronic amplifiers".
Of course we now know how to (completely) avoid those non linear effects (distortion) - it is not rocket science.
THD and IM are both results of the same thing - non-linearity, and we will never achieve absolute linearity in the general case - but we are at levels that are orders of magnitude below audible limits. Slew rate limiting is different - it can (easily) be avoided completely.
The B&C DE250-8 is 108dB/Watt . . . for "home" use it will never see more than a couple Watts . . . the amplifier is always operating in the milliwatt range. That's a whole different world from where amplifiers are usually spec'd, and Class A territory really. But I couldn't find a really good low power Class A design that bettered, or even matched, the Modulus-86.