I think it is important to get feedback from different people. Some people try to convince me they are right. I just explain that a final judgement cannot be made unless there is some data or technical reasoning which seems to explain a problem, and the problem goes away when you fix it based on such. Sometimes you are quite limited in what you can do in analog.
It is very clear. The ideal amplifier should add nothing and take nothing away. Just increase the signal level.
Thanks for this maxidcx.
Alas, like the TI model, this doesn't model susceptibility to the rails.
In da early 80's, I wrote my own linear circuit analysis programme to work on power amps (and other stuff). I spent many happy (??) hours translating the then crude SPICE models to work with my programme.
One of the biggest takes from that exercise is that the power rails are a very important part of the stability eqn. for a power amp .. one that is usually ignored in classical texts. Realising this allowed me to extract the maximum from very simple circuits in real life.
Tom, I think the TINA LM3886 model is probably good for voltage amps .. and you seem to have covered the stability of such beasts most thoroughly.
I'm not sure, with its immunity to PS stuff, it models stability at the limit with little or no Zobels .. as you'd need to do with a Current Drive amp.
This beach bum will have to wait until he can sensibly melt solder and look at scopes to investigate this properly.
_________________
That's absolute rubbish. The Zobel must be CRC!soongsc said:
And the caps must be hand carved from solid Unobtainium by Virgins.
Otherwise, all your clarity & definition will be unbalanced cos the 2 resistors will have impeded the flow of electrons.
I was first introduced to SPICE back in 1988. It was pretty clunky. A friend from Uni joined TI in 1990 and all SPICE runs were still submitted as batch jobs to the mainframe. Scary how far things have moved forwards, but amazing what is available to the hobbyist these days.
Models are good within certain limitation. Not made by Virgins though. The need for experience to obtain more complete modeling is really a very complicated.
I thought microsim was already running on Windows about that time. Certainly workstation solutions like Dazix, MentorGraphics etc were around.
Regardless of your feelings on the topic, the impedance of a resistor in series with a capacitor is Z = R + 1/sC. The impedance of a capacitor in series with a resistor is Z = 1/sC + R. You'll notice that the two impedances are equal. If you split the R in two halves, you'll have Z = R/2 + 1/sC + R/2 = R + 1/sC. s is complex frequency, by the way. It's a shorthand for j*2*pi*omega, where omega is the frequency in radians/second.
In other words, it makes no difference if you use R+C or C+R or R/2+C+R/2. These networks all have the same impedance. This is fundamental circuit theory.
As one of my professors in college would say, the current doesn't know where in the wire it is. All that matters is the total impedance.
Now, I don't doubt that you, soong, will hear a difference between the different networks. However, this perceived difference is not because the circuit is any different or performs any different. The perceived difference is purely the result of your cognitive biases.
Why bring it up then? Was the SNR of the thread getting a bit too high there?
~Tom
Yay... Signal. Thank you for increasing the SNR of the thread.
The main difference is power. A PAR86 amp will deliver 60 W in 8 Ω, 120 W in 4 Ω on a ±35 V supply. 120 W into 2 Ω if you lower the power supply to ±28 V. The PAR86 is also the best candidate for a bridged amp.
Tom
Thanks for your reply. My Mod 86's have been done for weeks (umm, well months actually) and I was very pleased with the sound when I had them running back in January, but I could benefit from more power. Unfortunately my custom wood/alloy enclosure has proceeded very slowly (new puppy in the house) so they are not "done" just yet.
So clarifying my query, I am interested in Parallel 86 vs two sets of Mod 86 configured as monoblocks. Obviously if I were starting from scratch 2xMod is roughly twice as expensive, but I already own the first set, so, input on merits of these two configurations is what I was fishing for. Thanks again.
You can connect two MOD86 Rev. 1.0 in parallel and arrive at a circuit that is almost as good as the PAR86. Two MOD86 R1.0s in parallel will have slightly higher THD than the PAR86 but provide the same amount of power.
MOD86 Rev. 2.0 is not suited for parallel operation. Only bridged operation is supported, but due to the current limit on the LM3886, it's only realistic to use a bridged MOD86 Rev. 2.0 to drive 8 Ω loads.
Going forward, my recommendation will be to use the MOD86 as a single board solution. 38 W 8 Ω, 60+ W 4 Ω. Want more power or bridged operation, go with the PAR86.
~Tom
Basically I just felt it a good idea to share the experience with those whom see improvement. The explanation is too complicated, but if you consider more realistic modeling of each device, and look at both forced and natural responses under realistic loads, it should be quite understandable.Regardless of your feelings on the topic, the impedance of a resistor in series with a capacitor is Z = R + 1/sC. The impedance of a capacitor in series with a resistor is Z = 1/sC + R. You'll notice that the two impedances are equal. If you split the R in two halves, you'll have Z = R/2 + 1/sC + R/2 = R + 1/sC. s is complex frequency, by the way. It's a shorthand for j*2*pi*omega, where omega is the frequency in radians/second.
In other words, it makes no difference if you use R+C or C+R or R/2+C+R/2. These networks all have the same impedance. This is fundamental circuit theory.
As one of my professors in college would say, the current doesn't know where in the wire it is. All that matters is the total impedance.
Now, I don't doubt that you, soong, will hear a difference between the different networks. However, this perceived difference is not because the circuit is any different or performs any different. The perceived difference is purely the result of your cognitive biases.
Why bring it up then? Was the SNR of the thread getting a bit too high there?
~Tom
I know you are against many things, and there was a thread where a moderator posted different recordings to see what people can hear. I took the challenge and pointed out some differences where moderator admitted some phasing error inserted. I cannot find which thread it was in but the recording was a simple vocal recording.
Last edited:
If you really want to get into why people are hearing things that seem like their imagination, it is necessary to understand how acoustics and modulation combinations humans may find more objectable or sensitive to. This is in the field where they study noise, and area where B&K seems to specialize in.
I'm just a newbie at DIY, but so much is clear from years of involvement with audio foolery. There are three areas that this pursuit encompasses:
1) Money
2) Labor
3) Knowledge
Of the three, knowledge is the most difficult to acquire. Not so much because the subject matter is so hard, but because most people don't have the background. It is more difficult to start from scratch. Pseudo-science offered by people they trust, gives them a short cut. Generally down the wrong path, but they feel good about knowing something, even if common sense would say otherwise.
For a zobel network, what you're after is low inductance. This means the RC needs to connect to the amplifier output with as short and fat a trace as possible. I use a copper pour for that. It also means the R and C will need to be located as close to each other as possible. Finally, the inductance of the ground trace (or plane, preferably) will need to be minimized. All of these things are obtainable by putting some thought into the component placement on the PCB.
Tom
In most cases, what needs to be considered are generally correct, but how it is applied can be controversial. This applies to many industries as well. In past, we had a project where a hardware supplier wanted to do the software, and they had some ideas that would significantly reduce system test and verification time; but they only got to supply the hardware. A different team got to do the software, but they did not catch on to the idea until about two years into the program when there was a schedule pressure. As systems integrator, it was hard when you have to teams each proud of their own experience. When we were finally able to get the team thoughts more aligned, things went quite well. I think everyone learned lots of things out of it, even cultural differences.