Ask a simulator the wrong question and it will gladly predict nonsense.
Feed a simulator nonsense and it will gladly ........
at these levels of sub ppm THD you cannot expect the models to give simulator any chance of an accurate prediction.
Wake up, and stop trying to con yourself into believing you have uncovered a new "truth".
Feed a simulator nonsense and it will gladly ........
at these levels of sub ppm THD you cannot expect the models to give simulator any chance of an accurate prediction.
Wake up, and stop trying to con yourself into believing you have uncovered a new "truth".
At any distortion level, decreasing nonlinearity will decrease distortion. We don't need a simulator to know this. So I'm not skeptical that adding a diode increases distortion. We don't need a simulator to know that the output distortion is the sum of all the distortions in the signal path, all of which can in fact be calculated without the help of a simulator, given sufficient knowledge (the simulator just does it for you). Of course I can say this because I've tried to learn more how to predict what will happen without needing the simulator, so that I'm not so reliant on it.
Actually, at these low levels the simulator doesn't necessarily have such a problem predicting nonlinearity, given sane models.
The problem is not always simulator accuracy, but the user's ability to approach simulator conditions in real life... Usually, either a change of simulator conditions or a change of real life conditions is necessary. There is a limit to what you will be able to deliberately achieve if you can't tell which is appropriate.
I'm sure that distortion levels that low are possible, although not very practical because they will be buried in noise. How to make a power amp with an SNR close to -160db?
Actually, at these low levels the simulator doesn't necessarily have such a problem predicting nonlinearity, given sane models.
The problem is not always simulator accuracy, but the user's ability to approach simulator conditions in real life... Usually, either a change of simulator conditions or a change of real life conditions is necessary. There is a limit to what you will be able to deliberately achieve if you can't tell which is appropriate.
I'm sure that distortion levels that low are possible, although not very practical because they will be buried in noise. How to make a power amp with an SNR close to -160db?
What do you know about simulator, have you simulated anything, and please don't talk nonsense, and don't comment my text in the future?
read post442 for confirmation of limitations to simulations
And yes, I have used computers since the late 1960s
As a user who is aware of the limitations in modeling, I never claim unrealistic answers for any predictions.
I have written some of my own simulation programmes, some of which were life (my life) dependent on reasonable predictions.
And yes, I have used computers since the late 1960s
As a user who is aware of the limitations in modeling, I never claim unrealistic answers for any predictions.
I have written some of my own simulation programmes, some of which were life (my life) dependent on reasonable predictions.
Also, emitter reverse-bias will cause transistors to get worse in the quasi-saturation. It makes sense to me if this has caused the worse sticking behavior. If this is true then replacing the transistors that were reverse-biased may reduce the sticking to normal levels. This of course assumes they were simply turned back around rather than being replaced in the first place.
Kean, you mean those predriver transistors could suffer some damage wile connected backwards, and that is not possible to detect with simple ohmmeter? In my amp I had some sticking but not that oscillations I saw in Terry's clipping test.
Sticking and/or oscillation is caused by a dramatic reduction in Ft and Early voltage during quasi-saturation, and how it interacts with the feedback loop. Transistors damaged from a reverse-biased emitter can exhibit this much worse.
Transistors that have suffered this damage can actually be repaired by heating up the junction, although I don't know the details. Weird trivia.
Transistors that have suffered this damage can actually be repaired by heating up the junction, although I don't know the details. Weird trivia.
See ... it works.
I'm just using a 4148 for my sub amp ( similar LIN topology).
I would use the better reverse leakage device for HF/critical listening.
These amps are very dependent on layout , especially grounding.
For even double digit PPM , just a CM deviation in NFB takeoff or asymmetrical
decoupling returns will make for high double digit (ppm)THD.
5'th element made a SMD condensed version of my wolverine - actually
bettered the simulation with a real tested 3-5ppm - 20k. At this level , even
where he had the probe leads routed would add /subtract PPM.
OS
Last edited:
Yes I think that simulation can show quite realistic distortion numbers, of cause if model are good. In real life it is not easy to get those numbers, a layout and a cabling are being some of main reasons, not a simulator.
Damir
Right after I released the wolverine , he built it. You seem to have scattered posts
around that time - I can't find it with search - but I know it is after #599.
Too many posts.
3-5ppm is just typical for this type amp with TMC. The only tested amps I've
seen that go below are E. Stuart's and SYN08's 40 transistor TIS topologies.
OS
Hi Terry,
This board contains so many jumpers as I wanted to try different possibilities but finally this is how it worked for me:
Not needed jumpers: J1, J2, J4, J5, J6, J7, J8.
Not needed components: Q8, R35, C22.
Needed jumper: J3.
If you like to try without C3, C4 and D1 you can insert jumper J8 instead and adjust output DC offset with P2.
You can use following SMD TO243 mosfets:
DN3135, product marking DN1S or DN2535/2540, product marking DN5D
As it was long time I built this amp, please don't hesitate to ask, I could've forgotten something.
best regards
Damir
As I said, you will find some more. Q1 is 2SC3503, TO-126 transistor because of power dissipation, and it will easy fit if you bend a bit collector pin.
Terry, please just ask anything you want to know or it's not clear, and that doesn't mean I know all the answers, far from it.
Q1 dissipates a bit more then 500mW.
I was going by the spice file. I see 11mv there. I know on my working boards, Q1 runs pretty warm. Not hot enough to warrant a heatsink. I will populate these tomorrow but had to order the MOSFETs so I won't have those for a few days. I have three extra pair of boards. If someone is interested in a pair or two please PM me.
Blessings, Terry
Blessings, Terry
Terry, you mean probably 11mA not 11mV as I don't see there 11mV at any place. To see transistor dissipation just place the cursor (pointing hand) on the transistor and see the dissipation in the lower left corner, or if you want to plot the dissipation place the cursor ans press Altgr and left button on the mouse.
By the way, I think that I mentioned it, instead mosfet you can use 15K resistor(increase R42 to 15k and use a jumper to connect to +, there are two holes but the jumper is not numbered), it will not degrade the performance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.