How to drive 8 pairs of MJL21193/94?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I want to build 900W/2ohms bridged amplifier, which will use 8 pairs of MJL21193/94, in each amplifier module, with +/-37V PSU.
My only problem, how to drive this amount of output devices? There are 3 options:
1, Drive with same type of transistor. One more pair MJLs with 150-200mA bias, or
2, Parallel connected MJE15032/33, with 70-80mA bias in each pair. Two or three parallel, with base stopper, and small series emitter resistors, or
3, Separate the output devices to 4-4pairs, and drive them with separated driver (MJEs)


I don't know which would be better.

sajti
 
Yes, he (sajti) can do this... i mean to get 900Wrms/2Ù with +/-37Vdc supply rails per channel, connected to load IN BRIDGE MODE as he reffers.
In bridge mode, the peak-peak voltage in output will be:
[ +37 - ( -37) ] X 2 = 74 X 2 = 148Vpp
The peak voltage will be: 74Vp
And the rms voltage will be: 74 / 1,414 = 52,33Vrms.
So, in 2Ù load the theoretical power will be:
( 52,33 X 52,33 ) / 2 = 1369Wrms.
It is obvious that sajti has made right calculations... proof for this it is the reffered requirement of 8 pairs of output transistors per channel.
As for the output scheme, of course the better is the tripple darlington arrangement for so high power. There is also the option of using: As predrivers the MJE15030-31 and as main drivers the MJL3281/1302 (both have very good bandwidth).

Regs
Fotios
 
fotios said:
As for the output scheme, of course the better is the triple darlington arrangement for so high power. There is also the option of using: As predrivers the MJE15030-31 and as main drivers the MJL3281/1302
I like this suggestion, don't know if it will work, never having done anything like this.

900W into 2r0 bridged is the same as 450W into 1r0 and that is equivalent to 30Vpk and 30Apk.
But, check hFE of the outputs when Ic=10.7Apk/device.
 
It is obvious that sajti has made right calculations... proof for this it is the reffered requirement of 8 pairs of output transistors per channel.
As for the output scheme, of course the better is the tripple darlington arrangement for so high power. There is also the option of using: As predrivers the MJE15030-31 and as main drivers the MJL3281/1302 (both have very good bandwidth).

Regs
Fotios [/B]

Not 8 pairs per channel! 8 pairs per amplifier module. It means 16pairs for one channel (2 modules bridged). I'm not 100% sure about the MJL2119X, because I get tons of FJL4X15.

I have lot of NJW0281/0302 (about 70pairs). They would be OK as driver. I think to use 200mA bias, or higher.

Sajti
 
AndrewT said:
I like this suggestion, don't know if it will work, never having done anything like this.

900W into 2r0 bridged is the same as 450W into 1r0 and that is equivalent to 30Vpk and 30Apk.
But, check hFE of the outputs when Ic=10.7Apk/device.

30Apk output means only 3.75Apk for the output devices. So hFE must be over 50, and this means 0.6Apk for the driver.

Sajti
 
In bridge mode, as we know, the one channel amplifies the initial signal of source in phase and the other channel the same signal inverted. If the output of source it is single ended, then the invertion is done within the amplifier. If the source gives a balanced output, then we can use directly the hot (+) and cold (-) signals to drive directly the two channels of amplifier. From the one channel that amplifies the hot (in phase) signal we get the plus (+) output to load, and from the other channel that amplifies the cold (inverted) signal we get the minus (-) output to load.
This scheme is used in the named "Fully Differential" Hi-End amplifiers. Technicaly this output scheme, it is called "Horizontal Push-Pull".
I had always in my mind to build a monoblock amplifier, but with the usual way of using two heatsinks as sides of amplifier case. I had not idea how could this be done, untill a very known designer-member he told me the way for this, which is the above described. When i asked him, how can do this with single ended signal in input, he replied: simply you have to use an inverter of signal inside the amplifier to drive the second channel, like this used for bridge mode operation.
This method, offers some signifficant beneffits of which the main, it is the lower voltage of supply rails needed. With a power supply of +/-30Vdc we can get an output of about 200Wrms/8Ù. Another one, it is the non-use of power supply ground as return path for the load, avoiding thus a gnd loop.
And of course far better noise immunity.

Fotios
 
sajti said:


30Apk output means only 3.75Apk for the output devices. So hFE must be over 50, and this means 0.6Apk for the driver.

Sajti
30Apk into 1r0.
30/0.35 Apk into a 1ohm speaker when driven by a fast starting or fast stopping transient.

There is evidence that the 0.35 factor be reduced to 0.25 and maybe 0.2 for severe reactance speakers.
 
sajti said:
I think to use 200mA bias, or higher

Do you mean 200mA Iqs per output device?!!
That is amazing if it is true! You will exceed the "extreme class AB" operation, for which - according to my experiments - a 50mA Iqs per output device it is enough.
I am wondering where you want to go. Near class A?
If yes, you have to use a ton of supply xformers and heatsinks!

Fotios
 
fotios said:


Do you mean 200mA Iqs per output device?!!
That is amazing if it is true! You will exceed the "extreme class AB" operation, for which - according to my experiments - a 50mA Iqs per output device it is enough.
I am wondering where you want to go. Near class A?
If yes, you have to use a ton of supply xformers and heatsinks!

Fotios

No. 200mA bias for the driver. For the output stage I planned to use some 100mA for each output device. I have 2 pcs of 350mm long heatsinks, with forced cooling, and 1500VA toroid transformer, for each channel.

Sajti
 
AndrewT said:
A pair of output devices with an Re of 0r1 requires 260mA for optimal ClassAB bias.
Reduces to 114mA with Re=0r22 and to 53mA with Re=0r47.

Multiple pairs need these bias currents per device.

This means:

308W dissipation with 0R1
140W dissipation with 0R22
65W dissipation with 0R47

but higher emitter resistance increase the takeover distortion.

But You find the reason why no values on the emitter resistors...

Sajti
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.