Ok, I'll take a bite...

The PE magazine design undoubtedly works to some degree. To some that might be enough and they may be the same people who appreciate the seeming simplicity.

But then there are inherent issues...

Bias. Well, the bias string probably tracks thermal changes to some degree if diodes are thermally coupled to output devices. But it probably has to be somewhat "cold" bias, though, because... well... in all its simplicity it's a sub-optimal design and more susceptible to thermal overruns. This issue is enhanced by the lack of emitter resistors in the output stage. So we can expect more crossover distortion and compromised reliability.

Gain. As is, the stage is configured as inverting amplifier. Very similarly to opamps (this is a generic NFB circuit after all), gain is determined by feedback impedance and total input impedance. So, consistent performance will require an additional buffering stage to keep source impedance to power amp stage constant. Added complexity.

Reliability. Obviously this amplifier is not protected against exessive load currents in any efficient manner. A momentary short circuit or a complex load may destroy the entire amplifier in less than an eyeblink.

Overload performance. If this was a small signal stage, operating at low currents, similar circuit might provide rather musical, though somewhat hard and asymmetric, clipping. Being a high current stage it's much more susceptible to transistor "deep saturation", which leads to much less musical momentary "rail sticking", which will sound nasty. In worst cases rail sticking may become permanent and destroy the amplifier instanteniously.

Stability. As is, it looks poor and is highly affected by layout and overall design. A practical circuit may require local high frequency negative feedback in VAS (and possibly drivers), and the output probably should have a Zobel filter (remember those complex loads?). Added complexity.

I agree with the opinion that addressing these issues will require a complete redesign of the amplifier (I built a similar circuit myself and vouch for that). In the end, picking a more adequate design as a starting point may prove more effective. Unless you want to use this as a learning experience of course.

Anyway, as I said, to some degree it works. It's a moderately powerful power amp, makes noise and all. I would just be constantly worried about its reliability. No thanks.

Series cap+pot in parallel to R5...?

http://www.auditorium-23.de/MagAmp/MagAmp.html
http://www.acousticplan.de/html/magamp_english.html
http://www.teslapress.com/magamp.pdf

The JFET input is actually a circuit idea conceived by Behringer but why, why, did they put it to input where it has least effect? It could have served as a decent "power amp" emulator in the later stages.

Yeah, otherwise pretty much a Sans Amp. Isn't Tech 21's Trademark 60 different from the usual SansAmp, though? I always thought these were nice amps for the money, and never really cared too much if someone was using circuitry similar or nearly identical to someone else's, unless there were legal issues involved. Marshalls, Dumbles, Mesas, whatever, have all started as "Fender circuits" anyway. Fender, on the other hand, adopted circuits from Gibson and so on.... What does it matter? In the end someone get's a "Tech 21"-ish sounding amp more inexpensively. Sounds suspiciously similar to Marshall's plan regarding Fender amps he sold in his shop. When Behringer does it they are big, bad evil corporation. Sigh.

Anyway, sorry for the "off topic".

"Factory" schematic:

http://wiki.talkbass.com/images/d/d5/%2769_pcb_BT-15_Schem.jpg

Few examples.... (See attachments).

And for comparison, an example of typical Fender SS amp preamp:
http://www.amparchives.com/album/Fender/Schematics/slides/Fender%20Deluxe%20112%20Schematic.jpg

QuoteI've heard they are based upon the old Irvine made models??

Not sure about the newest.... but for quite long period (At least from RGII series and alike, if not longer) all the amps they produced in Asia had preamps downright cloned from the SS Fender amps. And very blatantly I would say. Makes Behringer look like quireboys.

I recall they recently made some noise about returning to preamp designs with discrete JFETs, you know, what those famous RG-series amps were about. I would assume that at the moment there would be only a handful of these amps, rest would have "Fender preamps".

That said, Randall's been making "RG" series since day one so basically I would just regard "RG" as an abbreviation of "Randall" and "Guitar", not as much as as a suggestion that all "RG" series amps are even remotedly alike. They are not.

I'm somewhat familiar with early BBE "Enhacers" and later BBE "Sonic Maximizers".

The phase shift idea... I think it's nothing but common "time alignment" thing: Introduce electrical phase shift to compensate the acoustic one created by the loudspeaker system.

I'm not too familiar how all this "phase" stuff behaves in practice so I can't comment on whether BBE's "enhancer" idea works like it's claimed to do, but I do know that time alignment with phase shift or phase lag filters is extremely common design practice and in practice one can make it work effectively.

...But I would think you need to match it to specific drivers though, not offer it in some mysterious "black box" that approximates some generic phase lead/lag behaviour.

Edit: Refreshed my memory and took a glance too:
- Amplifier load correction system, US 4260954, is a generic mixed-mode feedback scheme. Note's are made about the scheme correcting phase errors at certain frequncies. Never investigated that.
- Method of correcting for variations in a load driven by a power amplifier, US 4482866, is a scheme somewhat "external" to power amp. Instead of sampling loudspeaker load current the circuit senses an approximated load current from a reactive circuitry that emulates characteristics of a real loudspeaker. Now current feedback can be applied in earlier, low current stages, indenpendent to the load. At least that seems to be the idea by quick glance...
- Reference load amplifier correction system, US 4638258, continues based on earlier patent's ideas.
- Notch filter system, US 4496859, is a resonant filter (with gyrator) that notches deep at 60 Hz.

On that note... I never got to try any of these but I always wondered about the "Accent" controls: How exactly do they work? Low-band / High-band distortion? Post-distortion tone controls in addition to other tone controls? Something else...?
There's not much information around about these amps so every little bit helps all others who are interested in them.


Edit:

http://patents.justia.com/inventor/robert-c-crooks

Bob Crooks definitely did design for them.

I have been somewhat interested in design details of these amps because likely related Barcus Berry patents feature some neat inventions : "Load correction" - more specifically, several embodiments of it - obviously had something to do with reactive speaker loads, but the patents are somewhat obscure in whether BB tried to compensate/eliminate the reactive nature of speaker loads or simply enhance it similarly to modern "mixed mode feedback" schemes. Then there are others too, like a nice notch circuit for feedback or hum elimination.

This in an era when most solid-state guitar/acoustic amp designers were still unaware that such features should perhaps even be developed.

Haven't seen schematics. Circuitry patented by Bob Crooks / Barcus Berry, however, is something I have never encountered in Crook's earlier work with Standel, Randall, SG Systems, Gibson, etc....

"Limiter" can practically mean two things.

You have limiting by clipping, which is basically about instantenous gain compression that is so drastic that it seems to "clip" away all signal peaks at some predetermined threshold.

Or you have limiting by gain compression, where amount of gain is controlled by external signal. This is typically not instantenous limiting, the driving signal always tends to introduce some attack and decay characteristics. Also, it's more about dropping the overall gain in magnitude that prevents traditional signal clipping.

In nutshell, clipping limiting limits by clipping off signal peaks. Feed sine wave to it and it clips off top portions of the wave and introduces significant amount of quite audible distortion. Clipping action can be either soft or hard, but in practice the difference is very, very subtle.

Externally driven limiter limits by reducing gain. Feed a sine wave to it and the output signal basically just has gradually decaying magnitude but overall retains the sinusoidal shape, which means much less audible distortion. This type of limiting tends to portray clipping distortion only at very first signal peaks, at which the "detector" circuit, driving the limiter stage, is too slow to instanteniously react.

So which one is it?

Yes, they will BOTH reduce headroom because they work by lowering overall signal magnitude. Both are clearly audible but in different manner: Clipping limiter in operation pretty much just sounds like clipping distortion, gain compressing limiter makes the overall sound quieter. If not implemented properly, it makes the amp too quiet, or the volume pumps up and down annoyingly.

Quoteapparently the STA540 (that is used in the orange microterror) has a built-in limiter...

QuoteThe Orange Micro Terror build may make such use of the clip signal out of pin 10, but the chip amp itself sure don't. 

Pretty much this. The datasheet makes no remarks of other features but the "dignostics" terminal, which indicates both clipping, overcurrent, over temperature and standby and requires some additional circuitry to determine when the signal is actually clipping instead of other alternatives that also toggle the flag.

Secondly, when chips and such do feature "soft clipping" it may simply refer to decent behaviour during overdrive. The signal may hard clip but it doesn't "stick to rail", portray hysteresis, etc. which are effects that result into really nasty type of clipping distortion. As said, the audible difference between traditional soft and hard clipping is extremely subtle.

QuoteAre there any commercial guitar-amps in class-d?

Yes.

They have seemed to gain a lot of popularity within last five years. Powerblock for a while was almost only "compact" class-D guitar amp out there but now plenty of products have flooded the market: Bad Cat Unleash is class-D, Quilter amps are class-D, some DV Mark amps are class-D, Matrix just released a Marshall clone with tube preamp and class-D power amp.... Plenty of them now. Of course they've been a mainstay in bass amps for almost a decade by now. Guitar scene is always at least a few years behind of everything, if not few decades. ;-)

That said, few years ago it would have been customary to loathe the idea of a class-D power in a guitar amp, and (gulp) a switchmode supply. Now everyone seems to be raving how great they are.

It's funny how things change.

QuoteDo they have reliability issues?

Everything does, especially when not implemented right. I think many of these modules still have a lot to work on generating their rated output power continuously. The TDA8950 is probably aimed for "domestic" duties because it couldn't achieve that rated output power reliably when operating at different crest factors associated with pro audio. ...Not to say that similar issues wouldn't apply to even generic linear integrated chips like TDA7294, LM3886, TDA2050, etc.

QuoteHT3 voltages take a long time to get up to steady state and the simulation time become too long.
I would like to replace these volteges with ideal generators, but first I need to know the right value, how can I do this?

There are several options to overcome the particular issue:

1) Run the simulation for enough time for all DC voltages to "settle". Use those values for ideal voltage sources. Usually you only need just one or two DC sources because with DC voltage sources in the supply the solver doesn't have to plot the time it takes for all capacitors to get charged by the pulsating DC.

Naturally the rectified AC voltage source will fluctuate according to current draw while the "ideal" DC voltage source won't. One of compromises between simulation speed vs. simulation accuracy. This also introduces an issue whether you want to derive and use "unloaded" or "loaded" supply voltages, which will be somewhat different.

2) Introduce a time delay for "plotting". For example, LTSpice transient analysis has the "time to start saving data" option. This option can be used for the plotter to skip over the period that it takes for voltages to settle. The solver will run but it won't plot anything, so it's a tad bit faster that way. If you also "mute" the input for this time (put a delay to input signal source as well) the voltages will settle faster as the circuit runs "unloaded" for a moment and draws less current.

3) I don't remember the SPICE directives from top of my head but refer to very good built-in "Help" documentation. There is an option that allows saving all circuit voltages at given time of simulation to an external file, and then restoring them from this file for new simulations. This way you only have to run the "voltage settling" period of simulation only once. Later you simply restore the voltage condition from file and rest of the simulation runs to end from that point. Much, much, much faster....

I use this method if I really want to include effects of rectified and fluctuating power supply to the simulation. It's not always that important but can make simulations of say, tube power amps, more realistic since effects of voltage sag and such are included. If you use an external audio file as input signal source and include a rectified power supply simulation then doing this is pretty much a "must". ...unless you want to start adding those muted periods in the beginning of each waveform. I prefer not to, not to mention the far greater speed to run the simulation.

The drawback is that if you make any bigger modifications to circuit you'll have to run the initial "solver" again to get results that match the modified circuit.

Yes, there are limits. One can either accept them or completely ignore them. Don't forget that you can always parallel or bridge amplifiers or increase the current handling with additional discrete transistors. You can compensate the limits of the small die area by means of design.

But of course if you just stick a single TDA7293 into a way too small heatsink located inside a chassis with moderately hot ambient temperature, and then claim a 100W output power rating you can just blame yourself for frequent chip failures.

And overall, I do believe the Rod Elliot article paints a somewhat unoptimistic image of chip thermal handling. Those chips are used in countless applications ranging from cheap car radios and computer speakers to professional active monitors and yes, guitar amps. Picking a burned Marshall board as an example is kinda misleading because Marshall for some reason seems to be pretty good on making amps where power chips fail while thousands and thousands other manufacturers seem to get the same chips to function just fine and with decent reliability.

Pritchard amps, I believe, use LM3886 chips and their designer actually preferred them over the discrete circuits due to wide range of built in protections. His former discrete design had to be ridiculously overbuilt to meet equal performance regarding overall reliability.

As a side note, the die area issue mostly refers to modern power chips where everything is integrated on a single die. Old chips, such as those old Sanyo's, were built differently: sometimes they even housed the power transistors as discrete devices so overall die areas of the critical points were usually larger. Such details are of course hidden by the chip enclosure.

QuoteI'm guessing caps or transformer? 

Why so? Seems like an automatic assumption for people these days but if the sound simply gradually faded away with slight, fizzy distortion it sounds very much like there is simply an intermittent contact somewhere which gradually got worse and worse up to the point the signal just got muted. So, there may not even be any failed parts in that thing but probably just a solder joint gone bad.

Quote...the send into another amp works 100%; full tone, its all there.  I will do the other side this evening and see if the loop works all the way to rule out the preamp.

If the send from preamp works fine with another amp then you have ruled out the preamp, at least up to that "FX return" jack.

QuoteOn a side note, I noticed 2 out of the four larger capacitors have a bulge to them on the top of the cap, one in particular is very noticeable.  The circuit board with those on it is small and fairly easy to get out.  If it is the capacitors, as I suspect a bulge on the top is no good, then I can replace those no problem as long as I can find replacements to spec.

Bulging could be an indication of an issue, or just result of sloppy manufacturing. Real question is that if you suspect capacitors to be the problem why haven't you taken any measurements to rule them out? If these are, for example, power supply filtering caps have you measured the DC voltage they filter to find out whether that shows any obvious signs of issues...? Just looking at the caps and pondering if they are ok ain't going to answer the question nor fix the amp by itself.

Are you sure you are up to fixing, or trying to fix, this amp? Because quite frankly, you don't seem to be. If issues with a 300W power amp is your first attempt to fix things then may I kindly suggest that the job is likely better to be left for someone who knows what to do to that amp properly. Amateur attempts to fix things can often lead to problems just escalating for the worse.

Quoteit does end up being the power amp (which I have found something suspicious I will detail below)...I am wondering if it will even be worth it as I will have to take it to a shop.  These are fairly rare, but usually go for ~$600, so not sure the work would be worth it if I can just get another one.

You could take it to some shop that repairs these sorts of things, explain the issue and its symptoms, and ask for an estimate of how much they are probably going to charge. They will most likely take a fee for just spending their time to look at it closer and later charge more for additional work and parts needed. So, it's going to be an estimation for the start but hopefully they can give you a ballpark value of what they are probably going to charge, which let's you decide whether it's worth it or not. Possibly you can settle some marginal for expenses after which they should consult you if its still worth it to continue with the repair. Like, if they look at it and find one bad resistor or a solder joint then the work and parts cost likely isn't going to be astronomical in comparison to them looking at it and finding it needs a replacement transformer and a circuit board and is going to cost more than a new, equivalent amp. So, contact some suitable outlet and find out what they can do to it and at what estimated price. They are the only ones who really can tell you this.