Figuring SS safe load impedance

[nashvillebill]

Bear with me, I'm still trying to figure out how theory is applied to real-world examples.

How--I'm looking for practical ways here-- does one figure out whether a given SS output stage design should run into 8 ohms minimum only, or whether it can safely provide enough current to run a 4 ohm load (or even lower)? Assuming of course the power supply is capable of providing enough current, and there's sufficient heat sinking. This has had me curious for some time, whether looking at Rod Elliott's projects or some of the projects in G. Randy Slone's books.

And I really got curious yesterday when poking around in a mid-90's Peavey Basic 112 (bass combo). This has a complementary pair MJ15015/15016 (well, Peavey's in-house numbers xref to these) with 39.5 volt rails, and is hooked to an 8 ohm speaker. Heat sinking is decent. Does the amp's designer just go by rules-of-thumb, or...? It appears that the calculation examples I've seen with hfe just kinda get you in the ballpark and it's up to the designer to figure whether one is getting too far from the SOA. (In that amp for example, it made me wonder why Peavey didn't push for higher power ratings and stick a 4 ohm speaker in there.)

Hopes this makes sense, my coffee hasn't kicked in.

[Jazz P Bass]

Rod Elliott is the guy to study on this topic.
He will walk you through the why & wherefore of amp output.
All amplifiers basically modulate the power supply.
In design, I would assume that you start with the output.
What load is driven. What wattage.
These points describe the output stage drive circuit & power supply & heat sink size.
Peavey is pretty good with there specs, as far as being truthfull.
Yeah, they could have made that amp to run a 2 ohm load.
In design, you have a price point.
Where it is exceded, you draw the line.

[Wilder Amplification]

All amplifiers basically modulate the power supply.

Actually they modulate the load current, which in turn modulates load voltage drop. If the power supply is designed properly and load current draw does not exceed max available current, the power supply voltage should remain constant.

Minimum safe load impedance on SS amps is determined by how much current the power supply itself can supply along with the current handling rating of the current carrying transistors. You take the power supply voltage and divide it by the load impedance to know how much current the load will draw at that voltage, then spec the power supply's max available current to meet or exceed that figure. The product of the power supply voltage and load current is your maximum output power.

[Jazz P Bass]

Very concise reply, Wilder.
Thanks.
As an aside, I had a Yamaha powered mixer , 1970 something, that needed a STK "brick" output IC.
Yamaha specifically stated in there literature that you must use the Yamaha xxx speakers.
Looked them up, 8 Ohm.
O/k.
Looked up the IC specs, 35Volts max power supply.
Yamaha ran these puppies right up to the wall at 40 volts.
Everything is fine @ 8 ohms.
Customer takes the mixer to a club & some yahoo hooks it up to 4 ohm cabinets.
" Man this thing got hot" was what I was told. Right before it blew both IC's.
"It sounded great" right before it went poof.

[Steve Conner]

Transistor amp output power is limited by two things: Overheating and second breakdown. (Well, three if you count a wimpy power supply, but Wilder already covered that.)

As far as overheating is concerned, they're the same as tubes, except that the dissipation rating given in the datasheet is for a hypothetical infinite heatsink that can keep the case temperature at 25'C. No such heatsink exists, so the transistor can never dissipate the amount it proudly claims on the datasheet. This is the best explanation of the thermal design process I've seen. http://ludens.cl/Electron/Thermal.html

Second breakdown is a non-ideal failure mode where the transistor can die at high voltage and low current, even when the dissipation is within safe limits. The safe operating area chart shows which combinations of voltage and current are likely to cause it. Quality audio transistors have a large SOA, and are 100% tested at the factory to make sure that they do in fact meet it.

The way I do it is to get the safe operating area chart provided in the transistor datasheet, add a line to allow for the finite-sized heatsink, and draw load lines on it for 4 and 8 ohm resistive and reactive loads at various rail voltages.

Douglas Self's book on power amps describes the process in gruesome detail. Here's one that I designed according to his recipe. scopeblog » Blameless short circuit protection

In short: Two good TO3 or TO247 transistors on a decent heatsink will provide about 60W RMS of audio all day, no problem. For marketing purposes you can probably push it to 100.

It doesn't really matter whether the impedance is 8 or 4 ohms: an amp designed for 100W @ 4 ohms will have lower voltage rails (by a factor of 1.4) than one designed for 100W @ 8 ohms, but it could use the same transistors and heatsink, because the amount of power dissipated is the same.

[nashvillebill]

Thanks, I had forgotten Rod Elliott's aricle on SOA.

The Peavey I mentioned has TO-3 transistors, and they do indeed claim 60 watts into the 8 ohm load.

As a logical extension, suppose the power supply has the aforementioned 39.5 volt rails (no-load) and the designer wants to safely drive 4 ohm loads. The projects I see usually accomplish this by adding output transistors in parallel. But wouldn't merely substituting a higher-performance output transistor pair (15024/15025 instead of the 15015/15016) accomplish the same goal? Again, assuming the heat-sinking would be adequate (or is that the "brick wall" that designers are up against?)

[J M Fahey]

Fully agree with above posts and add:
*Commercial* design is a compromise. You get carried away too much and the competition blows you out of the water by quoting just 1$ wholesale price below yours.
Peavey has always over-rated parts, specially very fragile output transistors, good for them.
I think your actual question is : "may I load this amp with 4 ohms , get more "free" power, and get away with it?"
Let's see what you need:
Max. transistor current: [peak voltage/min load impedance] = [40V/5.5] ohms = 7.3A
Some will say that it's not realistic, let's see:
The transistor will not pass that current *all* the time, because as soon as you start pulling current from the PSU its voltage will drop, probably to +/-32V DC or thereabouts (plus having 2 or 3V ripple), but if you are playing at low power (so the PSU is still close to +/-40V) and Slap or Thumb Tap, you may easily saturate the amp for some fractions of a second , supplying that current.
Transistor voltage: It will always see at least those 80V; you may often have *more* than your rated wall voltage, get one rated at least 20% more, say 100V Vce.
Power dissipation? : Rule_of_thumb: 30% dissipation at 40% full power .
*What* power to consider ? Good question.
You might use the FTC approved version: continuous power with no distortion, amplifier preheated for an hour, nominal resistive load, exact rated voltage at the wall plug, etc.
You know the power transformer will be hot, resistive losses at maximum, real output voltage quite lower, "real PSU power" at its lowest.
It provides a *minimum* power output, good for consumer ratings.
In this case you would calculate based on the (correct) peavey specs: dissipation 30% of rated 50W RMS= 15W between both devices; say 8W each.
Looks very small compared to the "120W" rating of each power transistor.
Read that datasheet again:
1) Those 120 W are specified with the case at 25ºC which is impossible to achieve. Even if you bolted that transistor to a 1 Ton (yes, 1 Ton, it's not a typo) monster finned heatsink, placed under Niagara Falls to cool it, achieving almost 0ºC/W dissipation, you would still have the case_to_heatsink thermal resistance (0.6 to 0.8 ºC for a TO3 case, with grease and without mica) to consider, plus chip_to_case (0.3 to 0.4ºC/W).
A real world heatsinking solution might be: 0.8ºC/W chip to case + 1ºC/W case to Heat Sink + , say, 2,5ºC/W from the accountant approved heat sink (it may even be bolted to some cheap flat sheet aluminum)= 4.3 ºC/W
As I said before, the FTC rating is benign, because it deals with a "tired" PSU. Your "heating" power conditions with a just turned on one might reasonably be : 40V+10% ; load 5.5 ohms.
44V PSU x .707:= 31V RMS. ; into 5.5 ohms= 175W RMS .
Will this amp provide 175W RMS on stage? No way!! but the transistors , in certain "very easy to happen" conditions might be overheating and dissipating power *as if* they were mounted in a "perfect" 175 W RMS amplifier.
And how much will they dissipate?= 0.3 x 175 = 52W (both) , meaning 26W each, meaning 26Wx4.3W/ºC=112ºC *above* ambient temperature.
@25ºC it means 137ºC, that might be true in a Spring open air Country show; but in a smoky club that chassis might easily be at , say, 40 to 45ºC (you translate it into ºF), meaning 157ºC chip temperature.
Silicon shorts at 200ºC ; spec sheets suggest do not go over 150ºC , now you see that transistor is not *that* overrated at all; in fact Peavey often used *four* TO3 deviced for 100/150W amps; Fender uses 4 TO218 for 100W; six for 160W (Stage 160) and so on.
SWR makes an acoustic 100W amp with only two TO220 MosFets .... and they blow with some regularity.
I use 2 x TO3 or TO218 devices for 100W but my short protection is a very paranoid relay which cuts speakers off on the slightest suspicion.
I do not even want to get into explaining Second Breakdown failure mode, Google it.
Now you see why sometimes amps blow for no apparent reason, the owner says "I did nothing wrong" ... and he's right, just that day he got unlucky and the Stars aligned in a bad way.
As Enzo often says, just repair the amp and continue using it.
When an anxious customer asks "do whatever's necessary to make sure it will *never* happen again", well, you can't promise that; only that if he's not careless it's very unlikely it happens again.
As to the original question: yes, very probably you can add a second 8 ohm speaker with no problems, you will get, say, 20% or 30% "free" power because Peavey rates well the parts they use, yet you won't get more power than the PSU can deliver, avoid shorting that extra speaker cabinet and don't obstruct the air flow around that chassis.
A small fan *always* helps (or a regular home type, pointing in the general direction of your backline amps, from behind).

[R.G.]

As a logical extension, suppose the power supply has the aforementioned 39.5 volt rails (no-load) and the designer wants to safely drive 4 ohm loads. The projects I see usually accomplish this by adding output transistors in parallel. But wouldn't merely substituting a higher-performance output transistor pair (15024/15025 instead of the 15015/15016) accomplish the same goal?

It's possible, depending on the internal construction of the higher-performance devices. In the case of the MJ150xx series, the real (as opposed to counterfeit!) ones are well constructed in general, and it's unlikely you'll get more out of changing to a different one in the series.

The internal differences are largely thermal, and how well the heat gets spread out of the transistor chip into the heat transfer plate that the bottom of the transistor forms. Copper heat spreader platforms are good, but with copper becoming priced like a precious metal, all the second tier makers are trying to go cheap on it. Oddly enough, if we could just do diamond heat spreaders, it would be a huge advantage. Diamond has a much higher thermal conductivity than copper, which is the next best material as I remember. You have to go to heat pipes or other fluid heat transfer to do better than diamond.

Again, assuming the heat-sinking would be adequate (or is that the "brick wall" that designers are up against?)

It's kind of a brick wall for economic reasons. Technically, the heat transfer is a heat spreading problem, or an "impedance matching" problem, depending on how you look at it. Air is a great insulator, as witness the efficiency of fur and goose down. It's also not very dense compared to solid materials, so it has a high thermal impedance. We have to get heat out of a solid material of low thermal impedance (metal) into the air. A heat sink is a kind of thermal impedance transformer. It takes heat on a thick, low impedance place and spreads it out into a number of ever-finer fin variations to increase the surface area so heat can transfer into the high-impedance air. Still air has a very high impedance, so making the air move by fans or thermal effects is a big issue in heat sink design.

The more concentrated the heat is, the higher the temperature. Incandescent lamps get to thousands of degrees by having almost no surface area for the filaments to transfer heat from, so the surface temperature rises very high. A 25W light bulb filament is quite hot, while a 25W dissipating transistor is overheated on no heat sink and probably under 150C heat death if it's bolted to a flat metal plate to spread the heat out some.

So with heat sinks, one can spend a lot of money making large, finely crafted sinks, buying more transistors to parallel, or letting a few very tough (and resumably expensive!) transistors get hot.

[nashvillebill]

Actually, the amp works fine now, and I really don't intend on running a 4 ohm load on it--it's just a small practice amp, and i've got bigger and better amps to use on stage. (Though I'll admit, for a second or two the thought of sticking a better 4 ohm speaker in there did cross my mind, I can't help myself--but instead I stuck a better 8 ohm speaker in that I had laying around)

But in my wilder moments I have thought about using the preamp from it and sticking in a bigger power amp design. Which drew me back to the "cookbook" amp designs I had in my books, and I noticed that they weren't clear at all on the loads that could be driven. Since all of my big 2x15 cabs are 4 ohm, if I did build a high-power amp it'd need to drive 4 ohms dependably.

That led me back to wondering why the original design had stuck with an 8 ohm load, it seems like their marketing department would've jumped on the chance to market it as a 100 watt amp instead of a 60.

[J M Fahey]

That led me back to wondering why the original design had stuck with an 8 ohm load, it seems like their marketing department would've jumped on the chance to market it as a 100 watt amp instead of a 60.

Well, to begin with it probably won't put out anywhere close to 100W RMS if loaded with 4 ohms, because the transformer fitted won't allow it (the diodes and capacitors would have no problem); and if you increase the PT now it's not the same amp anymore, it will also dissipate more heat.
In sum: let's go to the Technically and Phylosophycally important point: it would cost MORE (booooooo !)
Besides that, Peavey prefers to build its 60W amp *as if* it had to put out 100W; I see nothing bad in that concept.
They are justly famous for their reliability and toughness, that helps sales so they are not wasting $$$$.
Besides, broken amps are expen$$$$ive (warranty service) , not to mention the bad mojo.