Designing a Class A/B Push Pull Power Amp

[buildafriend]

In addition to what's been already said, what transistors are you going to use that handle 300V Vceo?

This is my main consideration at the moment. What MOSFETS? Initially, all I would like to do is make a working output section with two transistors for up and two for down that I can use a similar design principle to to make into a larger amp. I'd like to worry about all of the transient suppression and brownout after I have a very basic building block. The class D amps use high speed MOSFET drivers which are a further consideration. I'm looking for simple to start and then I'd like to build out from there. Perhaps it's time for a HV bench supply and 2 very high wattage heat sunk 4ohm resistors to run in series or parallel as a load for test. From what i'm reading 2 ohm stable amps are tough to build since you're practically shorting them out with such low resistance.

To answer Ian - a lot of power is needed to move the air in such ways at a night club. It's obviously insane but it's experiential feeling that. The lower your LF cutoff is the easier it is to unload the amp allow it to spit it's self out. The loading of the cabinet on the cone is important in horn loaded bandpass cabinets. I know studio electronics work much better than this but I enjoy a good kick here and there.

Sticking to the amp thing here. Thanks so far to everyone who has chimed in So selecting some transistors...

 

[user 37518]

Have you designed or built a high power amplifier before? If you haven't I don't think this should be the way to start, start small, make some low power amps before you dwell into the high power world. In my experience, making a reliable power amplifier goes beyond just selecting the appropiate set of transistors or the heatsink, you should consider a lot of things with such high powered beasts, protection circuitry is one, fire hazard is another.

 

[JohnRoberts]

Have you designed or built a high power amplifier before? If you haven't I don't think this should be the way to start, start small, make some low power amps before you dwell into the high power world. In my experience, making a reliable power amplifier goes beyond just selecting the appropiate set of transistors or the heatsink, you should consider a lot of things with such high powered beasts, protection circuitry is one, fire hazard is another.

I designed my first power amp (4x250W) back in the early 1970s and I borrowed liberally from the popular amps back then.

As I suggested earlier find one or more schematics to study and be sure that you understand what and why the designers did what they did.

I identified strengths and weaknesses so mixed together circuit aspects from more than one design (hint I didn't copy the Flame linear protection circuitry that was obviously inadequate. )

FWIW my old amp operates class A/B but that was like half a century ago.  :

My transformer alone weighed 65 pounds.** 

JR

** I used to joke if somebody stole my amp and tried to run away with it, I'd hold the door for him.  ;D

 

[buildafriend]

besides availability which others might have already considered, what would make you opt for the one that is designed for CRT's over this one that is designed for driving motors?

 

[user 37518]

besides availability which others might have already considered, what would make you opt for the one that is designed for CRT's over this one that is designed for driving motors?

Did you find the PNP complement part number?

 

[abbey road d enfer]

From what i'm reading 2 ohm stable amps are tough to build since you're practically shorting them out with such low resistance.

Not really. You need to put enough in parallels for handling the current but you also need to size the wiring and decoupling caps accordingly. That's when you find that conventional PCB layout techniques need to be abandoned for heavy bus bars. The Crown 5000VZ had some p-to-p wiring with heavy gauge wire.
It's a long time I haven't dabbed in power amps (I hate replacing buckets of power transistors) but the real issue for me has always been secondary breakdown.
It seems transistor mfgrs have made a lot of progress in that respect since the 80's.

 

[buildafriend]

gotcha. thanks guys

I guess the largest compliment ive been able to find so far are this..

https://www.mouser.com/datasheet/2/308/FJL4315-D-1809120.pdf

 

[JohnRoberts]

Not really. You need to put enough in parallels for handling the current but you also need to size the wiring and decoupling caps accordingly.

not only 2x the current at 2 ohm vs 4 ohm, but 2x the power dissipation requiring 2x the heat sink.

That's when you find that conventional PCB layout techniques need to be abandoned for heavy bus bars.

I recall solder mask intentionally left off amplifier PCBs so solder would accumulate on the traces lowering trace resistance.

The Crown 5000VZ had some p-to-p wiring with heavy gauge wire.
It's a long time I haven't dabbed in power amps (I hate replacing buckets of power transistors) but the real issue for me has always been secondary breakdown.
It seems transistor mfgrs have made a lot of progress in that respect since the 80's.

Safe operating area, and secondary breakdown are understood and managed with protection circuitry often combined with the output current limiting.

I don't think bipolar power transistors have been a growth area of interest for device manufacturers since the wholesale transition from metal to plastic packaged power devices a few decades ago.

Most of the R&D effort for decades has been into high voltage/high current switching devices for use in class D amps (typically mosfets). The switching power devices do not parallel as nicely as in linear amplifiers so single power devices needed to handle both the full voltage and full current. The good news is that saturated output stages don't dissipate as much power but low thermal resistance (junction to case) was still a design issue in the early days. 

JR

 

[user 37518]

The problem with the OP's request of being able to deliver full power at even 2 ohms without sagging is not easy to solve, one of the the problems is transistor beta droop at higher currents, the other is that the power supply has to be a lot bigger. The problem can be aleviated by paralleling and using transistors which exhibit relatively constant beta until a certain point (this also reduces distortion BTW) so the good old 2N3055/MJ2955 will not be a good candidate since it exhibits a lot of beta droop. The high voltage transistor the OP suggested has constant beta up to about 2-3 Amps, so you would need to parallel transistors so at full load none of them sources or sucks more than that current.

Douglas Self has written about this in his book on audio power amps, he calls it the "Load Invariant Amplifier".

 

[buildafriend]

Page 264 in the sixth edition. Nice find! I'll have to read read read and re read these few pages.

I've been bouncing around this book for some time now. Mostly around the the first few chapters, the class D section, and the G/H section. This book should come after "Small Signal Design" imo. He should make lectures so I can buy them and listen during work.

Backing up so i don't miss crucial techniques here I'll start with two of the Q's I posted with a large heat sink and then try adding more.

As we probably knew, the heat sink is expensive.. for the millionth time i need a CNC machine

 

[user 37518]

Page 264 in the sixth edition. Nice find! I'll have to read read read and re read these few pages.

I've been bouncing around this book for some time now. Mostly around the the first few chapters, the class D section, and the G/H section. This book should come after "Small Signal Design" imo. He should make lectures so I can buy them and listen during work.

Backing up so i don't miss crucial techniques here I'll start with two of the Q's I posted with a large heat sink and then try adding more.

As we probably knew, the heat sink is expensive.. for the millionth time i need a CNC machine

Both Self and Cordell should be mandatory reading if you want to build your power amps

 

[JohnRoberts]

Both Self and Cordell should be mandatory reading if you want to build your power amps

I was lucky enough to be able to pick the brains of Jack Sondermeyer.. (RIP). His office was almost next to mine, we had a conference room between us.

Jack was the power transistor applications guy at RCA when Hartley Peavey was trying to make power amps that wouldn't blow up. After spending hours on the phone Hartley hired Jack to come down and work full time in MS.

Jack is perhaps best known for the CS800 an iconic power amp, responsible for >100 million watts of rock and roll sound reinforcement around the world. Before Jack was the bipolar power transistor guy at RCA he was a tube guy so knew his way around them too.

Back last century we would both teach a seminar to Peavey dealers and salespeople. I called our joint seminar the "Jack and Johnny show", we were tasked with explaining modern technology in the context of new Peavey SKUs so the dealers and salespeople could know more than their customers. I would introduce Jack by saying that he has forgotten more about electronics than I ever knew. An obvious tweak about his age, but only half joking. Jack was brilliant.

He never wrote a book, but probably authored many RCA transistor applications notes before moving down to Meridian to do the real thing.   

JR 

PS: +1 to Cordell's work on low distortion oscillator/test sets, and improved amplifier feedback topologies. 

 

[user 37518]

John, I would've loved to be in one of those seminars, lucky you!

 

[JohnRoberts]

John, I would've loved to be in one of those seminars, lucky you!

When I was hired to go work at Peavey in the mid 80s I arrogantly thought I would be teaching them how to do everything better... As the saying goes I didn't know what I didn't know, at that time they were firing on all 8 cylinders, enjoying several months backlog. During my 15 years there I learned a ton as the world changed around us all.

I got a number of patents and designed too many products to count them all. (While customers still find me...I just answered an AMR production series console question this morning.) 

JR

 

[buildafriend]

Does anyone have a suggested method for finding diodes that best match those in the power transistors to help them reach saturation more cleanly? Would this be as simple as matching the diode drop measurement or Rbe of each in the forward direction?

 

[user 37518]

What do you want to do? do you mean diodes to bias the output pair? its best to use a Vbe Multiplier rather than diodes.

 

[buildafriend]

I'm going back to BJT 101 and looking at how a pair of diodes between the bases sets the Q's to saturate as well as they are perfectly matched to the diodes in the power Q's. I think this will have to morph as more pairs are added but, starting small.

 

[user 37518]

The problem with the diode approach is that your output pair quiescent current is not that accurately set, and output distiortion is heavily dependant on the quiescent current. With a VBE multiplier you have precise control of this current.

 

[JohnRoberts]

Does anyone have a suggested method for finding diodes that best match those in the power transistors to help them reach saturation more cleanly? Would this be as simple as matching the diode drop measurement or Rbe of each in the forward direction?

If you are talking about the anti-sat diodes (diodes used to prevent power devices from saturating to support fast recovery from amplifier clipping), I don't recall any specific matching required.

Peavey vigorously avoided factory floor production trims (like Vbe multipliers) so they used tightly specified dual diodes that were characterized for voltage vs current to provide suitable output stage class a/b bias.

For DIY onsey-twosey amplifiers I would use the old school Vbe multiplier (note: configure the trimpot so that open circuit trimpot failure results in minimum bias, not max).   

JR

 

[buildafriend]

If you are talking about the anti-sat diodes (diodes used to prevent power devices from saturating to support fast recovery from amplifier clipping), I don't recall any specific matching required.

Peavey vigorously avoided factory floor production trims (like Vbe multipliers) so they used tightly specified dual diodes that were characterized for voltage vs current to provide suitable output stage class a/b bias.

For DIY onsey-twosey amplifiers I would use the old school Vbe multiplier (note: configure the trimpot so that open circuit trimpot failure results in minimum bias, not max).   

JR

Amazing thank you both JR and user 37518!

So, here is my concept for a dev board where I can toy with concepts. The idea is that I can jump a lot of parts and slowly move into making it more stable. Then I can add the second power section while logging every change, failure, and improvement I see.