This is sound policy which people should use whichever SPICE model environment is being used. In fact, it should be a precursor to any simulation. Proper simulation takes a lot longer than some might think. If, at the end of a design exercise, something is wrong in a .model file, that effort will disappear in a flash.
Not only that but realistically there should be subtly different models to represent physical variations. Commercially, device models often have "fast, medium, slow" or other such extremes available.
So simulation is rather dependent on the user as much as the software.
Last edited:
It would be a good idea to include the critical device models in the .asc file so other people can see how it looks.
I had a quick play with this simulation using the first TIP3055 model I found and it was very unstable. (better with various 2N3055 models, although it was easy to upset with middling capacitive loads)
I had a quick play with this simulation using the first TIP3055 model I found and it was very unstable. (better with various 2N3055 models, although it was easy to upset with middling capacitive loads)
Subjective evaluation.
I flat out reject the validity of both technical and mathematical assessments. It can be dodgy to define what distortion is and how it relates to a large number of conflicting parameters in an immense complexity. LTSpice gives easily interpretable answers on an embarrassingly simple basis, you don't even have to roll up your sleeves. All you have to do is be confident.
It`s also worth mentioning that the Fourier transform only understands amplitude and time.
The Japanese have used far more sophisticated technologies to achieve desired device properties. Of course, you can think that the single diffusion technique provides low distortion and it is stupid to complicate things further.
English circuit designs have a strong propensity to produce a likable sound (so to say), but I have no preference for voltage followers.
professionals use rather expensive simulators . Not freeware. Things get really problematic for class D or "digital" audio amps. Timing errors become distortion there. Linear amps will disappear over time
You are free to disregard technical and mathematical assessments. Many engineers, however, find that simulation provides greater insight into circuits than would be revealed by simpler analyses. Have you published any papers explaining in your view why technical and mathematical assessments are wrong? Or how subjective assessment has led to better designs?
I would agree that the Fourier transform operates on quasi-continuous signals, and is incapable of evaluating distortion from transient effects (unless these are repetitive). But transient simulations can reveal problems inside an amplifier which may otherwise not be apparent. Many engineers find simulation indispensable as an aid to optimising a design. And yes it is easier today because of the development of SPICE. Before SPICE, calculations by hand were far more tedious. If a simulation indicates that a design has a high distortion it is unlikely to sound good.
This thread was about a 40 year old design. If the op is happy to have 40 year old quality, that's up to him. But your comment about "single diffused transistors" is moot. Transistors built with that technology are obsolete. The old 4MHz-ish transistors yielded circuits which can provide distortion levels of around 0.02%, and that was adequate and (subjectively) "as good as the best valve amplifiers" (Bailey). TI once described their transistors like the TIP3055 as "single diffused" but were in fact epitaxial base devices, as complementary devices are, and nothing like RCA's single diffusion process which was just that. A sort of copy of an alloyed germanium transistor process but higher temperatures, and resulting in devices barely able to make 1MHz. But were robust.
Higher frequency transistors with flat gain characteristics, optimised epitaxial layers and emitter geometry are much closer to the ideal transistor and yield lower distortion as a result. Older designs could benefit from such devices, but I've already commented in relation to this design that a quasi-complementary triplet is less stable than a complementary Darlington triplet. From simulations. But still has problems which need more refinement.
Like Cadence, for example? Or Mentor?
Out of reach of most people. Only commercial industry prices.
Hello N101N:
It`s also worth mentioning that the Fourier transform only understands amplitude and time.
Care to explain what else there is out there?
It`s also worth mentioning that the Fourier transform only understands amplitude and time.
Care to explain what else there is out there?
I got a notice from developer of - as far as i recall of German SAC amps - a product name and a price in the range of 50,000 US$. But don't have the notice anymore. That tool makes no sense to me because i don't have the test equipment to measure the amp ...software based spectrum analyzer...well.
If it sounds ok to me it is ok. That is the satisfaction that should come from diy.
My wife and daughter have their own hifi stuff they would never listen to my diy amp except if it comes in fancy looking rosewood cabinet
then it will sound marvellous. Hearing is very far from being objective
Last edited:
hahfran,
the restrictions do not lie in the software, but in the primitive language of mathematics and in its primitive logic. Logic ultimately sets the boundary of expressibility.
Boden,
analysis implies separating the whole into its component parts. Not being composed of parts, structureless entities, such as quantity, are not definable and not analyzable.
In the Fourier theorem, all characteristics are recklessly derived from the linear relationship between amplitude and time that can be continuous or discrete as needed. Logic is not about reason and rationality. A high level of generality is associated with this acclaimed concept.
the restrictions do not lie in the software, but in the primitive language of mathematics and in its primitive logic. Logic ultimately sets the boundary of expressibility.
Boden,
analysis implies separating the whole into its component parts. Not being composed of parts, structureless entities, such as quantity, are not definable and not analyzable.
In the Fourier theorem, all characteristics are recklessly derived from the linear relationship between amplitude and time that can be continuous or discrete as needed. Logic is not about reason and rationality. A high level of generality is associated with this acclaimed concept.
i am resistant to learning..😛
i own hi end amps but not in rosewood case.
Found nice looking heatsinks and job is to determine thermal resistance.
Using 10 Ohms resistor in metal case rated 50 watts. Thermal res 0f heatsink is 2.5 grades per watt. ok. Total thermal res junction to free air is then at best 4 deg/watt. Max junction temp 120 degrees ambient assumed 40 delta 80 max dissipation 20 watts. Max dissipation class B single rail 50 volt supply voltage
is about 30 Watts. Thus either take a risk or reduce supply voltage. I'll take a risk but provide temperature sensor at heat sink . Power off in case of overtemp. Case settled
Poor advice IMHO: ouputs can blow up long before getting too hot. I would use instead: first a light bulb testing device then a multimeter to measure current.
I would recommend resistors (on heatsinks) rather than light bulbs. Ordinary light bulbs (tungsten sort) are becoming scarce, but the point being that when cold, the filament resistance is about one tenth of the working (hot) value. As for LED bulbs, the electronics would get in the way of being a useful protection device.
Hahfran - only 120 C for Tjmax? Most (plastic) power devices can operate at 150C. Metal cans, generally 200C (one reason I prefer the old TO-3).
I suspect that you would be OK with your heatsink on music, as amps are rarely giving full power for sustained periods. 2.2C/W used to be a nominal heatsink size for a 30W amp., though I would recommend larger (with plastic devices) as that "rule of thumb" generally applied to metal cans.
Hahfran - only 120 C for Tjmax? Most (plastic) power devices can operate at 150C. Metal cans, generally 200C (one reason I prefer the old TO-3).
I suspect that you would be OK with your heatsink on music, as amps are rarely giving full power for sustained periods. 2.2C/W used to be a nominal heatsink size for a 30W amp., though I would recommend larger (with plastic devices) as that "rule of thumb" generally applied to metal cans.
Perhaps dd not make it clear that the resistors are on the low side of the transformer. Light bulbs are usually wired in the mains side. OK if you've still got a tungsten lamp around. I was assuming some might not have these days.
Having seen the magic smoke too many times 😱 I prefer not to rely on temperature for amp testing! Nowadays I'm using foolproof current-limiting power supplies instead of a light bulb tester.