Marantz 4400 Scope: Testing modern substitutes for 2SK30

[TimTheTech]

Hello. I am repairing a Marantz 4400 oscilloscope circuit.
I would like to report the result of testing some modern FETs in place of some shorted 2SK30s on the 'scope board.

The horizontal channel FETs were shorted while the vertical channel was fine.

Rather than trying to find NOS or trustworthy 2SK30s, for this repair I let what is in stock at DigiKey determine what was used.
KSK30 was available but is now obsolete.
The FET selection is dominated by some fine devices from Linear Integrated Systems.
I decided to try to replace the two 2SK30s with a single dual FET device.
After studying datasheets, which is hard when parameters on one datasheet do not appear on another, I ordered these suitably spec'd devices:
LSK489A
LS3954A
U401

I saw the LSK489A mentioned a few times in threads but no further study.

Tested: All three seem to be close replacements, although there are slight differences.
It is nice that a scope is being used to fix itself. "Scope, heal thyself!"

Two non-scientific observations, regarding complete circuit frequency response, and scope deflection plates DC bias:

U401 had the closest scope deflection plates DC bias match, a few volts different than the other original working deflection channel.
But the circuit's frequency response appeared to be a bit higher than the other channel, resulting in slightly elongated test circles on the scope
when approaching 18KHz - 20Khz.

LSK489A resulted in the closest frequency response match, with the test circles appearing to match quite well all the way up to 20KHz.
But it resulted in scope deflection plates DC bias being about 5 to 10 Volts lower than the other original working channel.

LS3954A resulted in a frequency response similar to U401, and scope plates DC bias similar to LSK489A.
Worst of the three, but still quite close.

I think I am going to choose LSK489A.

Sure, the scope plates DC bias are a little bit lower than spec, meaning slightly more collector current in the two 2SC1507 plate drivers
which seems to be responsible, I assume, for heating them slightly, almost imperceptibly higher than the other channel.
But it seemed to have the closest specs on paper, and gave the circuit a close frequency response match.

There is a later circuit of concern connected to one of each of the H and V plates.
The circuit is capacitively coupled to these two plates and appears to be a heater modulator.
I reason I am OK as long as the changes in voltage, rather than absolute plate voltage, are the same as before.

Hope that helps.
Opinions? Good choices here?
Thanks.

 

[rBuckner]

I appreciate your research and reporting on this! I've used a Linear Systems dual FET before and it worked perfect and I presume it still does, having not heard back from the owner. On yours, I wonder if a resistor change could cool down that plate voltage but I'm too lazy, er, ah, busy to look at the schem and think about it.

 

[TimTheTech]

Update: I think I will go with LS3954A.

The above tests were done while the unit was plugged through a 300W safety load lamp.
This affected the scope B+ slightly, down to 180V from 200V.

Then I plugged the unit directly into AC, such that the scope B+ was back to normal 200V, and I tested all three FETs again.

In the end, all three FETs resulted in a very slight elongation of test circles on the scope as the swept-frequency test signals
approached 20KHz. I believe this is because the frequency response of the repaired horizontal circuit is slightly
higher than before, higher than the original working vertical channel.
The difference among the three FETs was practically indistinguishable, from what I can see.

However, in the end it was the LS3954A that resulted in the closest match of the scope plates DC bias.
Both sets of plates now have virtually the identical 112V DC bias now, after one hour of swept-frequency test circles
followed by no-signal idling.

And, both sets of plate driver 2SC1507s have virtually identical warmth.

So no worries so far... I'll post if anything changes.
Now time to solder that LS3954A FET in place for good and be done with it, I think...

Thanks.