Help needed: Firing up the Aleph-X100

[Netlist]

Originally posted by Blitz
OK, found another faulty insulator

Whenever I replace transistors that should be isolated from the heatsinks with silpad or mica I measure with the ohmmeter if all is well.
It happens sometimes a faulty one is detected. Saves a lot of trouble afterwards.
Carry one, you'll make it

/Hugo

[Blitz]

So, Checked now all insulators and exchanged quite a lot of them with micas plus CPU-thermal grease from my computer (lessons learned: forget Silpads). Exchanged the 9610 again...and nearly made it: 0,1V Offset (without adjusting anything during warm-up)...

All mosfets showed 0,25V across the source resistor exept of one. This one had 0,4V. Ok, exchanged it. Same result. it was the one which acts as a current source and is connected to the BC550C ( the other 044n on that side is not connected to the BC550C, same principle as in the big Alephs).

OK, Just for testing, detached the BC550C from this mosfet and connected it to the other one: Bingo ! Now the other one showed the 0,4V. OK, exchanged the BC550C again.

Powered it up and start to measure across the source resistor of the current source mosfet which is now connected to the BC550C. Measure nicely, does the expected 0,25V. Slowly then starts to loose Voltage...strange as usually the voltage raises when the amp warms up....OK, let's measure the voltage across the source resistor of the other mosfet (same side, current source, but not connected to the BC550C): ooooooops: 1,2V ! Raising ! So the mosfet must be broken, but how come ?

So, clearly this time all insulators are OK ! Nevertheless I still have the same phenomen as before: When the Rail-Voltage has been raised to 17-18V, The mosfet which is not connected to the BC550C somehow breaks and starts to draw extreme high current.

So, any clues ? Is the BC550C not suitable for this job ? Oszillations ?


Heeeeelp !

[Netlist]

Could you provide a link to the used schematic?

/Hugo

[Blitz]

The schematic comes in pieces: As I used hifizen's board, the input piece is Hifizen's with the LM329 option going to the negativ rail and the ZV3306. Instead of MPSA18 I used BC550C. No HF-Caps used across the BC550C. As I get nicely my 3,9V across the 392 source resistor of the 9610, the current source seems to work nicely.

The mosfets are then connected using 0,2mm stranded wire with teflon insulation, star-wired. I know that I could have used thicker cable, but this cable is actually specified up to 6A and as I star-wired everything, it should be no issue, right ? So, there is no mosfet directly conneted to the board, but through p2p-starwire.

The connection from the BC550 is made only to one Mosfet like shown in the schematic.

[Blitz]

OK, here is now Hifizen's schematic.

Comment to the schematic above: I use finally only 8 devices to have more current through each device. Source-Resistor to get 100W is than 0,33 Ohm.

[Netlist]

http://www.semiconductors.philips.co...C549_550_3.pdf
Unless some connections in your Aleph current source are wrong, I can think of nothing else then the BC550 is wrongly connected.
I assume you use the 100k pot to set the bias. Start with turning that one fully clockwise (or anticlockwise) so it measures 0 Ohm (no voltage across it).
This will limit the bias to an almost harmless current.
Now you should be able to raise the voltage to the target 22V.
If all is well you could slowly start upping the bias with the 100k's.
What bias will you use? Around 5A?

Something I discovered when building mine is that you should properly discharge the supply caps (use something like a big 50 to 100ohm resistor) before you make changes to the circuit. There was always something like +/-5V remaining on them after powering down the amp. This was enough to ruin several mosfets.

/Hugo

[Blitz]

OK, found some errors in wiring where loose connections shortened with others, solved this. Started to fire up again: 9610 are perfectly matched, very same voltages of the source resistor.

After raising to 1,2 A per mosfet same problem again: The mosfet on one side which is not connected to the BC550C draws to much current (3 A), raising...it is always to same one (place) which makes trouble...as the other side simply work wonderful, it is a miracle...well maybe not: I will short the 200k potentiometer next time after having replaced the mosfet again and let's see if this one broke (I turned it already down to 0 Ohm, but you never know I guess)

[Blitz]

Oh, by the way: I received quite a bit of mails around this threat as many people seem to be eager to understand my errors to avoid them themselfs. So here is another finding: Obviously a variac is essential, otherwise I would have already blown two douzen mosfets more (one dozen though is already killed...)

Heatsinking the rectifier is as well essential...right now they are only mounted on a L-Aluminium-profile and they become hotter than the mosfets themselfs, so I will mount this pieces of metall to the big heatsink before starting any long-term tests.

[Blitz]

OK, the poti was dead, I bridged it and now the amp stopped killing mosfets.

Right now I have at 20V:

- 3,701 V across the 9610 left side
- 3,723 V across the 9610 right side
- 0,15 Volt DC at the output
- 13,3 Volt absolute DC measured to ground
- The 200 Ohm Poti at the current source does not do really a difference when playing with it.

So, I guess there are still some issues. I will post some resistor value which I changed as suggested by prominent DIYler of the forum for a high power version of the Aleph-X. MAybe some of these changes are not so wise, so please comment (I refer to Hifizen's schematic above):

- Current source with ZV3306, so R24=392, R26=158, R48=10, R17= 5Mohm, oooops, this one seem to dead as well (was a 0,125W Dale), OPtion J1b
- R14 and R31 = 1k5
- R12 and R34 = 1K
- R1+R4 100Ohm in sum instead of 31 Ohm.

I have as well still variations between mosfets, especially in the current sources (typical 10-20% different). How important is that ? What are your actual values in your amp across the source resistors ?

Overall I have as well one surprise: I have for the actual Amp-Mosfets (so the upper ones) 6A current through them right now at 20 V already. From my calculations this should be the current at 22,7V actually, so how can I get the additional Voltage up without having to exchange the source resistors ?

And: Is 0,15V DC at the output good enough ? How important is the absolute DC-offset of 13 Volt ?

Best REgards

[grataku]

Hi Blitz,
I didn't check this thread in a while, the 13 VDC to gnd at the output is NO GOOD. It means that a group of output mosfets are not properly turned on. From your voltage figures I would guess the gain mosfets are to blame. Don't quote me on that. I don't know what mods you made to the circuit and I don't feel like going thrugh the posts. Simply measure the voltage on the gate output CCS and note the difference between them and the output gain mosftets.

Normally you need to adjust the input CCS to bring it down to by bringing up the voltage from 3.7 to say 4. If the mosfets are selected for equal Vgs and transconductance everything should conduct the same with the same gate voltage.
After the first adjustment let the temp equilibrate and adjust the ccs pot again. If all the output mosfets are matched, adjusting the CCS should produce the same effect on the DC at both outputs. If that doesn't happen you have a matchin problem. In Grey's circuit there were two additional pots available. These provide some room for adjusting the unbalance in the output transistors but you better not rely on them too much.

You are almost there. We'll talk about AC gain adjustment later, you are not quite there,yet. Meanwhile you should have a little talk with the quality control department.
I used silpads in my 50W Aleph X without a problem. Actually I will never use mica again.