amp whit k1058/j162

[minnolella]

Unless, are new so I apologize if I make mistakes.
I also apologize for my English is very poor.
1 thing I ask you, where can I find a diagram of an amplifier at least 750Wrms-1000Wrms on 4hom using couples like k1058/j162 final.
I have 10 k1058 and 10 j162 toshiba and I would use them.

[AndrewT]

1000W into 4r0 is the same as 89.4Vpk into 4r0.
The losses through the amplifier would require the loaded supply rails to be ~+-95Vdc.
The PSU would have to run at >=+-100Vdc.
This requires devices that can survive under temperature and current stress to 200V!
Forget that idea and think about something achievable.

as a point of interest, I ran some numbers through Bensen's spreadsheet
14pair of 1058/161 can do 1000W into 4ohm 60degree phase angle @ the 100mS 65degreeC Temp derated SOAR. The +-45mF PSU is supplying +-99.2Vdc when quiescently loaded to 1.05A.
If each FET is biased to 75mA then the output stage has a quiescent dissipation ~208W. Now try to keep 28FETs below 65degC when stressed to well above their guaranteed Vds.

[megajocke]

Those transistors only have 160V Vds so you are limited to about 72V rails at idle and nominal line voltage.

You might be able to get about 500W in 4 ohms or 700W in 2 ohms with a beefy supply.

Two channels bridged with 5 pairs per channel might be able to do a bit over 1000W with a 4 ohm load.

[AndrewT]

5pair cannot look at pushing 500W into a 2ohm 60degree phase angle load.
10pair runs into Rds on when trying to drive current into 1r0. This is normally the easy to achieve requirement for an amp designed to drive reactive loads.

However, 10pair into 4ohms just avoid Rds on limitations and can easily drive the 500W into 4ohm 60degree phase angle load.
It appears that a 40FET output stage can achieve 1000W into 8ohms.

[megajocke]

1000W -> 1 ohm (equivalent to 4 ohm bridged with same number of transistors) is 45A and 45V peak.

4.5A/transistor -> maybe 10V drop over transistors.

Requires 55V rails under load, this allows for 25% power supply droop. Seems perfectly doable, but it's not very efficent of course! Power output will probably be higher at a higher load impedance.

Worst case average power dissipation per transistor will be about (((Urail/2)^2 / (Rload*N)) / 2 Which gives about 50W average per transistor if we are conservative and calculate with only 10% rail voltage sag. This is not going to be a problem.

There are commercial amps with only three pairs of the 200V rated laterals at 93V idle rails that can drive 4 ohms without reliability issues. That has 50% more dissipation per transistor than this.

But I agree this is not the optimal way to do this An amp with BJT:s will give much more power for this load for the same number of transistors.

[AndrewT]

Originally posted by megajocke
1000W -> 1 ohm (equivalent to 4 ohm bridged with same number of transistors) is 45A and 45V peak.

as load impedance drops, so current demand rises.
With a bridged design for 1000W into 4ohms each half of the amp must be designed to deliver 500W into 2ohms.
A minimum requirement for this to be successful is current delivery into a 1r0 load.
When modeling this easy loading the FETs ran into Rdson limitations. It is not a Power Dissipation problem.
Doubling the number of devices nearly solves the Rds on problem but doubling the impedance is what finally gets us out of Rds on limitations.

That's why I said 10pair (40FETs in total) feeding 1000W into 8ohms is workable but at an implied tremendous cost.

[megajocke]

"A minimum requirement for this to be successful is current delivery into a 1r0 load."

Do you mean that it should be able to drive full voltage swing into a 1 ohm load for those freak currents on rare (do they ever happen?) freak waveforms?

If you have that requirement then lateral mosfet output stages will need to be pretty overdesigned BJTs is probably the better way then.

[minnolella]

you could not possibly make 2 amplifiers from 500W to 2 hom, which are bridge so as to have 1000 over 4 hom

[megajocke]

Why not? Or is it a question?

I believe it's possible but probably not if you have the requirement of an amp being able to drive its half nominal impedance load to the near full voltage swing.

I'd say possible but not optimal.

[AndrewT]

That is the standard way to determine the ability of a bridged amplifier.
Double the power into double the load impedance. This holds for any bridged pair of amplifiers. If it fails to deliver the predicted power then there is something seriously wrong with the amplifier.

A conventional ClassAB push/pull amplifier that can deliver 100W into 4r0 and about 180W into 2r0, must if competently designed be able to deliver 200W into 8r0, when used as a bridged pair.

Note that the power delivered into 8ohms is MORE than a single amp can deliver into 2ohms. This also applies regardless of the design.