Problem with mosfet amp

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi
I am a student at a university in Sweden. And in one course we made a pcb of a mosfter amplifier. I guess the design is old but still sounds good. And also is good for learning what feedback and input/output filters and dclevel (to gate) is.
http://d106.ryd.student.liu.se/~kurv/Audio/Amp1.jpg
This is the board we made but not much. Just adding a better filter to the mosftets like this
http://d106.ryd.student.liu.se/~kurv/Audio/Amp1a.jpg
If someone want to build I can scan the whole article.

This desging has been used for about 3-4 years at school and it works good. And the pcb I got is tested and works.
The mosfets we have used are the 2SK413 and 2SJ118. Now theese are quite hard to find so I had to use the 2SJ162 and 2SK1058.
Theese have alot lover gate voltage about 1V instead of about 4V for the 413/118 ones. So I just alterded R27 and the pot to be able to set the current to 100mA.

Now to the problem.
It sound good at low levels but when I turn the volume up I get some kind of oscillation. I put a 33 Ohm in series with the speaker and now I had to increase the input level to get the same oscillation. So it seems to occur at a sertain output voltage/current/wattage?
One strange thing was also if I had long wires from the powersource I got less of this but when I mounted the board in the box I then have short wires and then it got worse again.

Can it be theese new other mosfets that are doing it?
Should I play with the gate resistor?

Glad for any help on this.
Jonas
 
The N-channel oscillates

I have hade almost the same trouble. The N-channel device oscillated over a certian audio level, only positive halfperiod of the signal.

Try to increase the gate resistor. 220-680 ohms is normal, don't use higher than 1 kohms because the speed of the fet gets too redused.

You can also try ferrite beads or combinations.

http://www.elfa.se/elfa/produkter/se/20/2011489.htm text only in swedish
 
Hi,

Per-Anders: The inductor seems to be there already, and the cap+res to ground as well. Hmmm... Prof Leach recommends to put the cap+res at the output terminal instead of your recommended "before the inductor". Maybe it varies case by case.(?)

The design has some similarities to the "old version" of the Crescendo amplifier, which is known to have oscillation problems.

Do you have any pictures of your board? It would be nice to have a look on it.

Q11 should have thermal contact/the same temperature as Q12/Q13 to increase stability, but I'm not sure if this can cause your problems.

How is your powersupply wired? (I assume you have a simple "line in to transformer to rectifier to big bank of caps" powersupply) There is another thread (http://www.diyaudio.com/forums/showthread.php?threadid=2633) on that subject.

Did you do any measurements on the amp? (with an oscilloscope + sine generator) If so, did you see any distortion on the collector of Q8?

In yet another thread (http://www.diyaudio.com/forums/showthread.php?threadid=3539) someone claimed the MJE340/350 to have less current gain above 150mA and therefore not recommended as drivers (hence recommending the MJE15031/32 as replacements).

My pennies...

//magnus

P.s by the way... I know what Per-Anders can do with a 220V 50Hz sinewave and a loudspeaker...
 
Ah, I found it...

From Prof. Leach's description on the output stage of the Leach Amp (http://users.ece.gatech.edu/~mleach/lowtim/output.html):

R50 and C25 mount on the loudspeaker output binding posts. On the first amplifier I built, I had these on the circuit board where they caused violent oscillations because the current through R50 and C25 generated positive feedback when it flowed through the circuit board signal ground lead. When the oscillations occurred, the heat sinks would get very hot. Moving R50 and C25 to the loudspeaker output binding posts solved this problem.

He argues to put the cap+res on the output terminal mainly to give eventual oscillations in the zobel network another path to ground than on the circuite board.

//magnus
 
Hi people

This oscillation is weird. One thing it depends on is how long the cables are from the powersource/caps to the amp. If they are short it is worse. Anyway I have got rid of the oscillation. I have put a 108pF cap from ground to R33 (not the gate side) and a 68pF cap on Q12 from drain to source. The oscillation is gone but I have no idea how this has effected the design. I think I am gonna loose theese caps and try the other tips you have given me.

As for Q11 the article says this amp has not temperature compensation and therefore Q11 must not be mounted near or on the heatsink, it should work in ambient temperature.

Powersupply I use a 40V 470VA transformer and two ricifiers and 40.000uF for each side.

One thing I saw when comparing the two current vs. gain for MJE 340 and 15030.
For the 340 the gain graph goes from 1mA to 500mA and for 15030 it goes from 0.1A to 10A. I have set my amp to have about 35mA through Q8, idle current or whats it called. I have no idea how high this current will be when music is played.
But lets say it gets up to 50mA and the graph for the 15030 seems to fall when you look at it down at 0.1A. I looks like 15030 is happy when it has at least 0.2A and I dont think my design get up to that?

The amp by the way is from Erno Borbley I think. At least he wrote the article.
 

Attachments

  • 340_1503.gif
    340_1503.gif
    20.6 KB · Views: 466
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.