Replacement Parts

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I recently acquired 2 old Carvin FET 1000 Amplifiers for my PA on EBay. They were both defective I had to replace the OPS, as suggested by some board members I replaced the original 15D20 20N20 FET's with IRFP240 and IRFP9240 HEXFETs. I completed one installation and biased according to the manufactures spec, when I went to test it at full volume the OPS blew.

The original owner of this amp had the amplifier repaired and the he also replaced the parts with the IRFP series named above. I have to assume the same thing happened.

In any event I have a show on Saturday and I need to get this sucker working. I was wondering if the OPS might need to be rebiased as a result of the replacement parts, any thoughts?


  • fet1000new.gif
    21.8 KB · Views: 275
Joined 2004
Paid Member
You need to get the specifications for both the old power fets and the new ones. Vds for the new ones should be equal to or greater than the originals. Rds should be less than or equal to the old ones. Drain current rating continuous and pulsed (or look at SOA curves - some fets do have them) should be greater than or equal to the originals. Pd should be greater than or equal to the originals. Gate capacitance and transconductance should be comparable or loop compensation may need to be tweaked.

There are no ballast resistors in the sources of your output devices, hence equal current sharing is not assured. In this design you probably need to use matched fets and/or add a small (.1 ohm) resistor in series with each source connection.

Finally make sure that the thermal resistance of the package is no more than the originals and that you have low thermal resistance to the heatsink. (Use thermal grease with mica insulators or silpads with the lowest thermal resistance you can find.)

All of this presupposes that the design is stable into a speaker load and doesn't also oscillate at VHF. Fit 1K gate stopper resistors right at the gates of the power devices if you see high frequency oscillation. (use a scope.)

This design uses a transconductance amplifier output stage - loop gain is a function of the load impedance on the output. Not noted as a paragon of stability. I would cautiously suggest you add a zobel network right at the output of perhaps 10 ohms and 0.1uF to define the load impedance at high frequencies and thus the loop gain. Be very careful though doing this, as depending on where the dominant pole is located this might make it less stable not more. In such a case you would decrease the value of the 0.1uF until it was stable - both with speaker load and without. Caveat emptor - if you are not comfortable using a fast scope and doing some analysis of what you observe I wouldn't attempt this.

Also look at the minimum load impedance - are you below it?

Testing at full power is ill-advised until you know that the amplifier is rock stable and can actually safely operate at the required power levels with the new devices.

You can add reversed biased high current diodes across the rails to output to provide some protection against misbehaving inductive loads at high power levels as well. (Recommended for PA use!)


Thanks so much for your input. Very insightful. I discovered that the reason the OPS was so instable was because two traces from the drain to the OP rail were severed. As a result only one OP FET was conducting on the negative rail, hence when the amp was turned up if it fried the device causing the whole OPS to go with it. I have since replaced the OPS and hard wired the Drain to the OP rail. Unfortunately I am now having difficulty adjusting the bias. I really don't know if I want to spend more time with these units. I may keep them for parts. I was an interesting project however. Thanks again for your assistance.
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.