ESP 3A output power at 4 ohms load

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Just measured the output power of my my modified ESP3A amplifier.. Using two pairs of 2SA1216/2SC2922.. Powersupply consists of a 500VA 2*38V toroid transformer, and two 47000uF RIFA caps.

The results into 4 ohms load was:

28,3V RMS, I = 7A (It starts to clip just above this voltage)

So output power is about 200W.. I would have expected slightly more.

Supply voltage (no load): +/-53,6V
Supply voltage (full load): +/-49,6V

So is this normal.. It seems like it only can swing +/-40V.

/Freddie
 
The collector resistors of the CFP outputstage is 0,47 ohms.
7A and 0,47R means that I will lose over 3V across those resistors.

Perhaps the output power is reasonable after all.
Think I will reduce those resistor to 0,15R.

Anyone else measured the output power of their ESP 3A?

/Freddie
 
Did you measured it into a resistive load ? Also for how long ? I would be interested in more specifics of what other components you had to change from the original p3a schematic .

Yes, I measured power into a 4 ohm resistive load for a couple of minutes. The 4R resistor got quite hot! Otherwise no problem at all.

Q1, Q2, Q3, Q9 = 2N5550
Q4 = MJE350
Q5 = MJE15030
Q6 = MJE15031

Outputstage transistors = 2 pairs of 2SA12116/2SC2922 (each with a 0,1R emitter resistor)

R9, R10 = 3k3

Q4: Needs a small heatsink
Q5, Q6 : Need a small heatsink

Q9 mounted (glued) at the top of Q5 to create a stable bias current.


/Freddie
 
Freddie

The collector resistors of the CFP outputstage is 0,47 ohms.
7A and 0,47R means that I will lose over 3V across those resistors.

Perhaps the output power is reasonable after all.
Think I will reduce those resistor to 0,15R.

Anyone else measured the output power of their ESP 3A?

I think you have to consider the voltage drop on the 0,47R is a function of the peak current (near 4,7v over the resistor). You have 49.6Vcc at full load minus 4,7v. Since you have 39.9v peak and 44.9 available to swing I think you´re reaching a reasonable power level in this case. Did you measure the Vcc at full load using the scope or the DMM?
 
I think you have to consider the voltage drop on the 0,47R is a function of the peak current (near 4,7v over the resistor). You have 49.6Vcc at full load minus 4,7v. Since you have 39.9v peak and 44.9 available to swing I think you´re reaching a reasonable power level in this case. Did you measure the Vcc at full load using the scope or the DMM?

I measured Vcc at full load using a DMM. But I think I will reduce the value of the 0,47R resistor, perhaps to 0R1.

/Freddie
 
Excuse me... from which u say the schematic have modifyed... how? ANd is there a substitute to MJ? Here in italy are very EXPENSIVE (the 2sc too... 17€!)

I haven't got the schematic left, but here are the changes I've done to the P3A amp. Yes the 2SA1216/2SC2922 are expensive. Try MJ15003/15004, they are cheap.

/Freddie


-----------------------------------------------------
Powersupply/channel:

One 500VA toroid transformer
Two 47000uF 63V RIFA capacitors.

Amplifier:

Q1, Q2, Q3, Q9 = 2N5550
Q4 = MJE 350
Q5, Q6 = MJE1503x

I use two pairs of Sanken 2SA1216/2SC299 transistors each with a 0R1 5W emitter resistor.

R9, R10 = 3k3

Q4, Q5 and Q6 need to have small heatsinks.

I also placed Q9 (Vbe multiplier) at the top of Q5 to have a very stable bias current.
 
I suspect the 200W figure @ 4-ohms is rather close to SOA envelope for the output devices. If it were me, I would go to three pairs if the space to mount them were available. This is partly a personal preference of mine - I like my equipmen (audio, auutomotive or comutational) to run well with in limits rather than to push them to the max.
 
"I suspect the 200W figure @ 4-ohms is rather close to SOA envelope for the output devices. If it were me, I would go to three pairs if the space to mount them were available. This is partly a personal preference of mine - I like my equipmen (audio, auutomotive or comutational) to run well with in limits rather than to push them to the max."

Use TO3 outputs for best SOA. Even the old MJ15003/04 have 250W safe area at 50V, and de-rate to zero at 200*C junction.

Plastic parts are a joke at high voltage, high temperature, and high current.

Speed?

Yeah, those plastic 30Mhz outputs are fast at low currents.

But this ESP3A example here is a good case in point: on program material the collector current at full output will be on the order of 12A peak, 6A per device (±48V into 4R).

At 6A the speed of the 30Mhz plastic NPNs drops to about 3Mhz, about the same as the older MJ15022, and not as fast as the newer MJ21194 (the PNPs are all faster).

A quick price check at:

http://www.future-active.com/Comergent/en/US/adirect/future?cmd=search

Shows the MJ21193/94 to be USD$1.95 each.
 
Plastic parts are a joke at high voltage, high temperature, and high current.

Idon't agree with this. Plastic packages suffer in relation to metal cans due to lack of contact surface area for heat dissipation but it's still silicon inside.

Metal cans like the mj15003/04's are expensive compared to plastics so paralleling of the plastic makes them much more appealing.

As for speed Fairchild's fjl4315 (to-264) are rated at 30Mhz @ 1a 5v's and have a 10th of a second repetitive peak of 350watts.
And 30a 10ms collector peak.

Mj15003/04 are rated at 2mhz @ .5a 10v's with a collector current peak of 30a for 5ms. Try matching a comp pair of mj's by there GBP, your head will start to spin. The mj's are more heat tolerant but who wants there heat sink to be almost 400 degrees fahrenheit.

And with plastics ease of mounting and heat sink considerations, I"ll have plastic please.
 
This is an output module, single channell, I'm building. These tip2955/3055 are pretty dated but cheap. I'm shooting for 200~250w's a channel into 4ohms. I'll post the complete, of course, when I'm done.
 

Attachments

  • 300watt.jpg
    300watt.jpg
    33.6 KB · Views: 406
I think you will find you can get pretty good results with tip2955/3055 if you under-rate them, i.e., use more than you really need so as to keep them well within their SOA. I did a simulation using Onsemi models that seemed to indicate that if you used enough so that the worst case emmitter-collector current stayed below 2/3 to 1/2 the max rating, the noise and distortion figures would be in the same ball park as much newer and exotic devices. Since these are relatively small and cheap your approach could be quite sensible.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.