What transistors do you recommend for a good LTP input?

[EWorkshop1708]

Originally posted by sajti
I agree with Eva. In 99% of the lifetime the input LTP can see only the half of the rail to rail voltage. In case of failure, You have to plan some "budget", say 20-25%. So I recommend to use 90-100V transistors for LTPs.

sajti

Yea, I see how that works. The other half of the total rail voltage is usually dropped across the current source, or a resistor.

90V or so makes sense. I have a bunch of old complementary 60V transistors, and I doubted that they'd be enough (without cascode) for a 100W amp, since rails can be 50V or so.

[sajti]

Yes! For 100W/8ohms +/-50V rails looks good, and 60V transistors in the input stage are good enough. But You can use them in Your current design with cascode connection.

sajti

[quasi]

I just bought 26 units of 2SC1845. Not cheap though @ almost $2.00 AUS each.

The specs look good though

http://www.necel.com/discrete/images/pdf_e/TC-3236D.pdf

[AndrewT]

Originally posted by Eva
In normal circumstances the input stage only has to withstand the voltage of one supply rail plus half of the amplitude of the maximum input signal

In faulty circumstances like a open base connection, input connector transients when signal makes contact before ground does, shorted output devices causing severe output DC, etc..., the LTP transistrors may get exposed to the full rail to rail voltage. They can be easily protected by creating +-15V zener+resistor references and clamping the LTP bases to these references with small signal diodes [ie: 1N4148]

Then you can cascode a 45V device like BC550C/BC560C at 30V leaving 15V of margin in case the 15V clamps have to do their job

As a cascode device you may use something like BC546B/BC556B that are still small signal devices and are rated at 65V, so maximum supply rails are +-95V

For higher rails an alternative is 2N5401/2N5551 that feature high Ft and are rated at 150V. These may be also used as LTP devices directly without cascode

There are also a lot of good small signal high voltage transistors with excelent specs from Toshiba and Sanyo, but these are pretty hard to find

I don't recommend MPSA42/92 for an input stage since they suffer from low Ft thus reducing phase margin, and low gain thus causing large input offset voltages

A really old thread, but Eva has called up a spike protection for the LTP front end.
Are there any downsides to this zener + diode solution? like noise? leakage current? anything?

The Leach has a pair of dual 20V Zeners creating +40V or -40V on the cascodes.
The +20V & -20V references are already fitted, just waiting for a pair of diodes to be added.

Any thoughts?

[Bigun]

I would have thought that leakage currents were negligible, so no impact on noise performance or linearity.