Aleph-X offset problem

[Andypairo]

I have finally managed to build and test a channel of my Aleph-X, but I encountered a problem I couldn't solve: the offset between speaker terminals is always > 250 mV.

Supply rails are 15V, I use the standard Grey's schematic, output fets are IRFP 140. I made provision to install 8 of them (instead of the regular 4), but even with only 4 I never get to lower this offset.
I tried swapping different input pairs, trimming the drain resistors of the diff pair (R23-25) and even the source resistors of the lower fets (R6-41), but even if the absolute offset varies a lot the offset at the speakers terminals stays more or less the same.

My setup procedure was to set first the current of the two halves (in two steps since I noticed they slightly influence each other, probably via R46 and R47) and then use VR2 to minimize the offset.

Did I made something wrong?
Suggestions on how to get rid of this offset?

Cheers

Andrea

[Netlist]

Andrea, congrats with your AlephX.
Did you measure with shorted inputs or something connected to them?
Did you wait long enough to let the amp warm up?
Did you look at this post? : http://www.diyaudio.com/forums/showt...644#post225644

Keep us informed.
Ooh, and post some picture too please.

/Hugo

[Edwin Dorre]

I had the same.... When you did not make any mistake in putting the circuit together, it comes down to one thing; better matched mosfet's!

The Aleph-X is way more picky to the matching of the IRF9610 pair and the 'lower' two IRF240's (the ones NOT in the current source) . I have matched about 8 pairs (with the procedure Nelson's describes in the A40 or A75) and swapped them in and out to get two good working pairs! I even swapped in and out the IRF240 to get a even measurement of the voltage over the bias resistor. Result: I have a steady 15mV without connected amp and about 30mV with a Aleph P connected in balanced mode!

See the other threads:

Another Aleph-X coming up!

Edwin

[Andypairo]

Hugo, I have the inputs AC coupled via 2,2 uF caps (one on each input) and shorted to ground.

For pictures I'll have to borrow the digital camera fron my workplace, as soon as things will allow me to.
I looked at the post, I'll try to swap some output Mosfets...

Edwin, the circuit works well, apart from this problem.

Will try swapping some other fets or their source resistors...

Cheers

Andrea

[jarek]

Hi Andypairo,
Try to short inputs to the ground before capacitors - not from the circuit side.

[Andypairo]

I tried both ways, before and after the capacitors (DC coupled) with no effect.

Yesterday I remembered that on the schematics of the Aleph 30 there were some "misterious" resistors, called "R8" and "R1" with a asterisk instead of a value.
In some thread it was said that they were used to trim the offset, thus their value was to be trimmed in circuit.
I used this method (R between D and S of one of the input fets) and it seemed that was finally possible to lower the offset close to zero.

Does this method produce some serious drawbacks?

Cheers

Andrea

[jarek]

Did you try to trim currents using VR1 and VR3? (100k pots). At the start there is nice to have exactly the same currents on both sides of the amp.

[Andypairo]

First I set the currents of both sides to be equal, then with VR2 I can trim the absolute offset, but the differential (to the speakers) remains the same.
Only puting a resistor between the sources of the differential pair and the drain of one of them i can lower it.
Putting it on the drain of the other things got worse.

But the question remains the same: does this method have serious drawbacks?

The word to the experts (or to the )

Cheers

Andrea

[Andypairo]

I swapped the lower fets (even if all 8 were matched within 10 mV), and trimmed one of the 392 ohm resistors at the drain of the diff input pair. This way the differential offset has been lowered to < 10 mV (even if varies a bit with temperatire), but the behaviour of the absolute offset puzzles me: it starts at about a couple of volts, then slowly decreases to about 300 - 400 mV. It stays around there jumping now and then to 500 mV; after that is keeps going negative, stops for a while at about - 300 mV and behaves as when of the positive side.
If I leave the amp for a long time without signal it seems to go in a sort of "negative thermal runaway", absolute offset becomes more and more negative and stops at about -7V.
In this situation it doesn't amplify any more, while if this doesn't happen output sinewawe is perfect.

I use 0R22 source resistors for the outputs and about 5A total bias, i saw Edwin uses much higher resistors (0.56), did I miss the relationship between them?

Cheers

Andrea

[Edwin Dorre]

I use 0R22 source resistors for the outputs and about 5A total bias, i saw Edwin uses much higher resistors (0.56), did I miss the relationship between them?

Perhaps you use less mosfet's... I use 16 mosfet's per channel. So 8 mosfet's per side so I have 4 times 4 mosfet 's parallel. To get to about 7A of total bias I use 0.56 ohm resistors (0.5V / 0.56 ohm = 0.89A * 4 mosfets * 2 sides = 7.1A). When you use half the amount of mosfet's you can use about 0.22 ohm!

My amplifier keeps running around but nevers gets higher to about -1V and +1V of absolute offset... It is very very heat dependend. I had about -500mV and when I opened the case 1 minute later I had about +400mV ... I read of simulair behaviour of Aleph-X; Peter Daniels had the same...

Edwin

P.S. I am actually running at a total of 6.7A I trimmed the bias a bit down...