How to drive 8 pairs of MJL21193/94?

[sajti]

Here You can find the driver stage for the bridged amplifier. It will be fully balanced from input to output.

Sajti

 

[roender]

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.

Andrew, you seems to forgot the internal parasitic emitter resistance. Also, the base stopper has a influence too.
With 0r1 Re and 2r2 base stopper the optimum bias current is 180...190mA

 

[roender]

This is the output stage linearity in relation with various values of Re, for a single output pairs (ref D. Self)
Of course, the figure is much more better as the number of pairs is increased, but the 0r1 re value is always better if the Vbe variation among output BJTs is in the range of 1..2mV
Re=0r47 is only for the cases when no OPS BJT matching is performed

 

[sajti]

Andrew, you seems to forgot the internal parasitic emitter resistance. Also, the base stopper has a influence too.
With 0r1 Re and 2r2 base stopper the optimum bias current is 180...190mA

But how can I count with the internal parasitic emitter resistance?

Sajti

 

[sajti]

This is the output stage linearity in relation with various values of Re, for a single output pairs (ref D. Self)
Of course, the figure is much more better as the number of pairs is increased, but the 0r1 re value is always better if the Vbe variation among output BJTs is in the range of 1..2mV
Re=0r47 is only for the cases when no OPS BJT matching is performed

The distortion is much better with more output pairs, but the load impedance is lower, because of the bridge connection.

And the 0R1 emitter resistor gives less protection against thermal runaway.

Sajti

 

[roender]

But how can I count with the internal parasitic emitter resistance?

Sajti

You can't ...
You should measure the minimum distortion vs bias current.
The numbers posted above are from measurements and not from theory.

 

[AndrewT]

Hi Roender,
I had not forgotten about the internal part of Re nor the effect that Rb has on it.
Your and others' earlier lessons are well remembered.

 

[sajti]

First, I will try to reduce the number of the output devices, to reduce the capacitances, and get smaller size. I think that 4pairs for each amplifier module (8pairs for 900W/2ohms)will be enough. The peak current will be 8A for each pairs, 1ohm load for each module, with 32Vpeak output.
I put together some test circuit to measure the bias. 2pairs of FJL4X15 with 0.1ohm resistors, driven by MJE15032/33 biased to 80mA, predrivers are BC639/640 with 8mA. I use two transistors bias circuit with BD139/140.
The heatsink is about half sized I planned for one module.

The bias looks overcompensated. 26mV with hot heatsink, and if I cool down the heatsink it increase up to 38-40mV

Sajti

 

[Coastben]

Replacements for MJ15003/15004

Just checking something here...


Can I use an MJL21193/21194 to replace MJ15003/15004's


They only have a rating of 140V and wouldnt mind bumping it up to 250V. I would obviously have to compensate for a slightly lower W rating though...


Any thoughts?

 

[wg_ski]

Re: Replacements for MJ15003/15004

Can I use an MJL21193/21194 to replace MJ15003/15004's

Not if you want them to fit in the same socket. The MJ is a drop-in, the MJL is not.

In reality, in audio apps the MJ and MJL handle the same power. You just can't run the temp as high with the plastic pacakge.

 

[roender]

Re: Re: Replacements for MJ15003/15004

... You just can't run the temp as high with the plastic pacakge.

Which mean that they can not handle the same amount of power (SOA)

 

[wg_ski]

Pulsed SOA in the s/b limited region where Tj<<150C is going to be very very close. That's limited by the die and the heat spreader which is copper in either case.

 

[roender]

Did you ever sow how big is the heat spreader in TO3 capsule? It has almost two time the thermal inertia than TO3P capsule ...

 

[wg_ski]

The copper coin in a TO-3 is actually a bit smaller than the copper TO-264 base. The TO-247 (3P) base is the small one.

 

[sajti]

Here is the prototype, under test:

An externally hosted image should be here but it was not working when we last tested it.

Sajti

 

[sajti]

One more picture, contains the drivers.

An externally hosted image should be here but it was not working when we last tested it.

Triple darlington stage, 1st: MJE15032/33 with 25mA bias, 2nd:NJW0281/0302 with 120mA bias 3rd:8 pairs FJL4315O/4215O.

Sajti

 

[tinitus]

I run a single pair of MJL91193/94 in 50watt monos
Sound quallity of the finest kind

They have driven a complicated 5way 2ohm with no problem
In 1ohm they got pretty hot, and obviously didnt like that, but survived

It was remarkable that despite the very heavy load, amps didnt loose anything in sound quality

Is your supply really only +/- 37Vdc?
I reckon that would be below 100watt, which means you will have a very tough amp

UPS, balanced thats very different
Though I thought balanced wasnt very good fore driving low impedance, as it will see 2ohm as just 1ohm
Which would mean that it will deliver the 900watt/2ohm into a 4ohm load, but might get into trouble below 4ohm
Sorry, maybe I have no clew what so ever

 

[sajti]

The rails are +/-36V. 2 modules will be bridged, and I will use tube driver.
2ohms load means 32A peak, which means 4A/MJL pair. I think, there will be no problem with that.

Sajti

 

[nigelwright7557]

Hello sajti,

First you won't get 900W/2ohms with +/-37VDC rail Voltages.

Have said that, then I tell you that I have several times with succes used the MJL21193/94 as drivers in large output stages.
Use the MJE15032/33 as predrivers and either BD139/140 or MJE340/350 as pre predrivers.

I made that mistake on my first amp build and used +/-45v
The amp worked but at any decent volume clipped the transients.

Music is a very complex waveform.
Drums especially cause very large spikes.

I now use +/-60V and that seems to be a lot better.
Sadly the price of electrolytics rockets above 63V.

 

[AndrewT]

at any decent volume clipped the transients.

Sadly the price of electrolytics rockets above 63V.

that's precisely why we specify 100W amplifiers and use just 1W of output power. It allows transients ten times the average signal voltage to pass unclipped.

His 900W into 2r0 proposal will allow 9W of average output and still retain the 20dB overhead for transients.

Those 63V caps allow one to use a 40Vac transformer.
170W into 8r0, or 310W into 4r0. I don't know what you would get into 2r0, but it will surely be much less than the 900W he can get from +-36/7Vdc.