jeeptechfred 2/21/2018 12:12 PM
gonna to build clone of SWR750 poweramp, have some questions
[ATTACH]47260[/ATTACH][ATTACH]47261[/ATTACH][ATTACH]47262[/ATTACH] I am going to build a clone of a SWR750 power amp. I plan on eventually using it, but also I am hoping to learn something and bolster my solid state electronic knowledge in the process of getting it up and running. I hope to later design my own preamp to pair up with it. I have a suitable power supply. My first question is if a mj21194 would be an ok substitute for the 2sc3264 in the schematic. (I attached some schematics, the one I plan on using is the second/last one in the group). the voltages seem safe and hfe seems close. I am willing to use the 2sc3264, however I already own some mj21194s mounted on some nice heat sinks. the second question is could I run 4 output devices instead of 3 per rail? would I need to change or adjust the driver? what would be the pros and cons if this is doable? thanks in advance, Fred
J M Fahey 2/21/2018 1:07 PM
I would use 4 MJ**** instead of 3 2SC****
Collectors and bases would be in parallel to existing ones, emitters would need their own emitter resistors to evenly share current.
Short protection circuit should be slightly modded because they sense current acorss 1 of 3 resistors, now you´ll be using 4 but that´s not a big deal, we´ll cross that bridge when we get there.
jeeptechfred 2/21/2018 1:27 PM
ok cool... thanks for the response. being able to reuse those transistors and heat sinks will be helpful. so do you believe the extra pair of transistors will not load the drivers too much?
jeeptechfred 2/22/2018 9:35 AM
c20 and C26. Are these decoupling caps? And if so where is their best physical location to be? I’m guessing they go between the output devices and their drivers.
nevetslab 2/22/2018 11:17 AM
Quote Originally Posted by J M Fahey View Post
I would use 4 MJ**** instead of 3 2SC****
Collectors and bases would be in parallel to existing ones, emitters would need their own emitter resistors to evenly share current.
Short protection circuit should be slightly modded because they sense current acorss 1 of 3 resistors, now you´ll be using 4 but that´s not a big deal, we´ll cross that bridge when we get there.
I agree with Juan on the MJL21194's. They're a much stronger transistor than the 2SC3264. They are a bit slower which may require tweaking the miller caps at the driver stage so there's no oscillation birdies at full output drive, or approaching that. Usually see it driving low impedance.

Good luck on your project, and let us know how it comes out.
jeeptechfred 2/22/2018 11:54 AM
Thanks for your comments there greatly appreciated. Now I need to study to see what a Miller Is
jeeptechfred 2/22/2018 11:58 AM
And that is part of the whole idea of this project! (-:
nevetslab 2/22/2018 9:07 PM
Looking at the SWR 750 power amp schematic, it's the base-collector cap on the drivers....560pF is shown. It's added to swamp out the tendency to oscillate as you pull a lot of current out of the part to drive the output transistors. Often you'll find base resistors added at the driver, and even base resistors to each power xstr to keep them from trying to become common-base oscillators. Some designers will fine tune the circuits and use a properly selected inductor instead of a base resistor. There's no doubt a number of threads in the Theory & Design category over the years concerning stability techniques on solid state amp circuits.

I've attached a later generation SWR SM900 Power Amp schematic that's more involved...uses fully complimentary output stage, all with base resistors, and an intersting pre-driver/driver stage and current limiter circuit.

jeeptechfred 2/23/2018 4:40 AM
Ok, thanks for that explanation, are the base resistors acting like a high frequency filter or current limiter? Also why would a “slower” transistor be more prone to oscillate? I would have guessed it would be less.
nevetslab 2/23/2018 11:07 AM
I had to stop and download the data sheets on the drivers shown in the 750 schematic. 2SA1668 and 2SC4382 are both 200V/2A rated parts, meaning they can source up to around that current, though in designs, usually less so you have headroom. Each power transistor you have in the output stage is being driven from the driver. Their hfe comes into play in calculating how much current the driver stage has to deliver....and the minimum load impedance becomes part of all this. You work backwards from how much power you want to deliver into what your minimum load will be. Peak currents will be part of all that , as well as safe operating area and temperature there sufficient heat sink capacity under average operation as well as maximum output....which is usually short term in music conditions, but it all comes to be part of the equation in choosing what to choose for output transistors, driver transistors, power supply potentials & supply regulations..

When SWR improved their power amp designs...the SWR SM900 being a good example, they went from a 2-pair output transistor stage to a 4 pair complimentary design (using the 2SC3264 & 2SA1295 Sanken MT200 package pwr xstrs, then then later to a 7-pair output stage with MJW21193/21194 output xstrs. Their driver transistors were changed in the process, so there would be sufficient drive current to supply the demands of the output stage.

You'd have to do the calculations to see if the drivers are sufficient to source the current needed for 4 MJL21194's. In the schematic I sent you on the somewhat larger & more complex circuit, using still a simple complimentary output stage, they went to the MJE15034 & MJE15035 drivers, which are good for 4 amps. The MJE15032 and MJE15033 are good for 8 amps, being a 250V rated part.

And, all of this has to be derated by temperature. , since you're typically going to be operating at around 45-60 deg heat sink temperature, depending on how your thermal design is handled. On each transistor data sheet, there are curves showing current/voltage as a function of temperature. the hfe has linearity curves with respect to collector current, as well as temperature. Ft likewise has curves with respect to current. It's a real billiards game

The speed of the drivers and output transistors comes into play with regards to how fact the circuit can deliver current into a load within the desired operating frequency. But, it really is all part of the open loop gain-bandwidth and having enough phae margin when it all comes down to unity there still enough phase margin, or does it become an oscillator. selection of compensation capacitors in various stages are part of that design process, as well as installing small value base resistors or inductors to improve on the phase margin, preventing the closed loop gain of the circuit to behave stable unconditionally.

Working from known commercial designs, as you're plannig to do is the easiest way into process. And, after having laid it out mechannically to allow for four output pairs of xstrs, I'd start with 2 pair and the drivers selected, see if everything comes up working, before adding the extra pair of outputs.

In looking at the MJL21194 vs the 2SC3264, the Ft is 10 times faster (60MHz vs 4MHz) and the hfe is typically twice is high (25 typ for the MJL21194 vs 50 typ on the 2SC3264. So, the phase margin won't be as good, suggesting adjustments in teh compensation, and teh hfe is less, meaning it will demand more current from teh driver to deliver the same output current. That's whre you may need higher current drivers, such as the MJE15032 and MJE15033. They're not quite as fast as the 2SC/2SA parts, but will have the current driver capacity

It does get confusing, but, it's a learning curve. How are you physically building the circuit?
jeeptechfred 2/24/2018 5:08 AM
I have a dead crown CE1000 power amp that I am going to high jack for a bunch of its parts. The power supply components are all good. I also have the heat sinks complete with several batches of their output devices. 4 or the 16 were damaged 12 or them test good. Those ar the MJ.... devices I’m hoping to use.
My early thoughts on how I may lay it are basically I thought I would have small board (probably vero board) with the drivers, emitter resistors, collector resistors and probably the bias servo mounted to the top of the heatsink and connections to outputs somewhat point to point running off of it. It would be easy to add transistor pairs one at a time this way. The rest of the circuit I intended to build on vero board. I will likely redo it with a custom eched pcb if I like the results.
I have a few 2SC5242 transistors from th crown that test good that possibly could be drivers. That was their original duty.
One of my thoughts.....please let me know if they are valid..... is that when I get it running with 2 pairs of output devices I could measure the voltage drop across r22 and r25 under full power and then using ohms law determine the current flow through the drivers. I then could add the other pairs and check to see if I am working them too hard. In the meantime I will order some mje150032 and 33s.
nevetslab 2/24/2018 3:01 PM
While I don't have teh schematic on the CE1000, look at the power transformer/power supply of that. Crown's are a different animal, not using high voltage bipolar supplies (+/- HV), but a single supplly, and have the power amps configured in a bridge pairwith one of the bridged ampps drving 'ground' of the load. In essence, you don't have a center tap on the power transformer that you need for the SWR type of circuit topology. They've been using that grounded-bridge amp configuration for decades now.

Now, you do have two power transformers, and could phase the secondaries to create a larger center-tapped transformer secondary, but the voltage level would be way too high for your needs..

You might look around for non-working SWR amps or Eden WT-800's and scrounge a power transformer from them, as they are the traditional type thaty you'd need. The Eden uses a Toroidal, while the SWR amps use the more traditional EI core xmrs with sturdy mounting brackets. Shipping cost, of course always hurts when you find them on teh opposite coast from you.
jeeptechfred 2/25/2018 7:56 AM
I’ll take a closer look but I do believe it is a center tapped transformer. I have some access to other transformers if that’s a “no go”. Thanks for pointing that out as a possibility.
jeeptechfred 2/25/2018 12:46 PM
I did indeed confirm that my transformer is center tapped. 61-0-61v, should see about 85v rails. A little lower than spec.
jeeptechfred 3/4/2018 7:19 AM
Quick question, can I move the “emitter” resistor to the emitter of q14 and q13? The reason I ask is that then on those transistors I could leave them mounted to the heatsink without an insulator pad. The heat sink is insulated from the chassis. Q12 would probably still need insulated for the protection circuit
nevetslab 3/5/2018 2:08 PM
Not on a Quasi-Comp amp circuit like you have. Q12, Q13 & Q14 are all in parallel, each with their 0.10 ohm resistors. Only if you were to re-build the output stage as a complimentary circuit, having NPN's and PNP's can you do that. BUT....the transistors still MUST be insulated, as the collector is NOT at Ground, even though at idle, it is at 0V. That output bus swings from V+ to V- as the transistors get close to saturation at full clipping level..

Of course, you're asking a loaded question. Is this being built in the Crown chassis with this SWR 750 circuit, and the heat sink is electrically isolated? I'd have to really look at the Crown circuit before passing judgement on that. Usually on circuits like this, the heat sink IS grounded
jeeptechfred 3/6/2018 6:51 PM
The heat sinks will be insulated, the crown had them mounted to the pcb. I am not reusing the pcb but will insulating the heat sinks from the chassis. I attempted to add the schematic to the crown, might be crumby cause I’m useing my phone. I suspect it is good enough to get the idea.
nevetslab 3/7/2018 5:28 PM
Is that schematic for the Crown CE1000? interesting compound driver stages up and down.
jeeptechfred 3/8/2018 4:09 AM
Yes that is the ce1000 schematic, well actually it is the ce2000. The only difference I can see is that the rails are 99v instead of 85v and it uses 6 output devices in place of 4. I have the service manual with all the schematics and I uploaded the wrong one
jeeptechfred 4/22/2018 4:05 PM
I finally got the amp together and it is functioning. I can run the amp up until it clips with my signal generator. It seems very stable and sounds good. I used mje15033 and mje15032 for drivers and I used all four pairs of the crown outputs mje21194. None of the components appear to be over heating. I was able to do fair amount of testing without even noticing any significant temp increases. I used an 8 ohm load and got an easy clean 250 Watts. I want to run a 4 ohm load and I have an appropriate dummy load. Any ideas on how I should go about testing it for ruggedness? Are there any warning signs or tests I can perform to see if it is on the edge?
J M Fahey 4/22/2018 10:38 PM
1) CONGRATULATIONS on your build

2) sadly ruggedness tests are destructive.
A semiconductor is either alive or dead, and it takes 1 millisecond to go from one to the other.

Factories build , say, 10 prototypes and torture/short/cook/kill them one by one, and then perform an autopsy to find ways to correct them.
What they find then applies to thousands of amplifiers; sadly that is not practical when dealing with "just one" , so just build it according to "rules of the art", be conservative, use good parts, and pray.

First 3 factors being the significative ones.

Just a small detail: not too sure about the airflow in your heatsinks.
jeeptechfred 4/23/2018 5:25 AM
The heat sinks are basically arranged just as the original crown heat sinks are they are designed that allow air to flow through them. There are a series of louvers. I’m hoping that since they worked in the crown amp the work and this one
nevetslab 4/25/2018 9:27 PM
At BGW Systems, we did much like JM Faheysaid....a number of prototypes, ran them thru their paces, pushed at elevated temperaturs to see what failed. I had a 4X15" subwoofer (from our M2200 Powered Subs line), and used those to look at how the current limiter circuits worked under high current transients. The current limier circuit in what you built is unusual. sense current taken only from one of the positive half output xstrs, feeding the upper Current limiter circuit, while operating on both the positive and negative clamp circuits to remove drive from the upper/lower driver circuits.

Over the years, I've found 40Hz/50Hz & 63Hz 1/3 octave pink noise as an interesting test signal for driving large bass bins in checking for abnormal current limiter circuit behavior. The pink noise thru those LF 1/3 octave filters produces wide random-amplitude sinewave, with amplitude variations as much as 15-18dB. Looking to see how the amp handles back-emf from a loudspeaker load was one of the things we looked at....often an overlooked behavior in current limiters.
J M Fahey 4/26/2018 1:10 AM
Heavy cone+voice coil woofers driven at or near resonance frequency are *deadly* :
They absorb a lot of electrical energy to start moving and when fully stretched "springs" (suspension, edge, air elasticity) push moving mass back with fury, they *generate* electrical power , voltage and current peaks which get injected backwards into power transistors, stressing or damaging them.