Aleph-X prototype with problems

[sbear]

Ok, got my first aleph-x on the bench. Fired it up and no smoke, no fire, no blown fuses. Hooray!!! So I started taking measurements and found a couple problems:

First: Unloaded the power supply was putting out almost 22V, under load the supply sagged to 18V. This seemed a bit drastic, is this usual? I guess you all would need some more info to really answer this question, so now is the time to tell a bit about the amp and PS.

I'm building 100W Aleph-X monoblocks with 12 irfp240 for each channel using the schematic attached below (except I am using 0.47 ohm source resistors, not 0.43). I am shooting for 22V rails and 6.5A total bias which gives me 100W at 8 ohm load, maxing out at about 125W at 6 ohms and dropping from there as the load impedance drops. The power supply is a dual 17V sec. 475VA EI frame tranny (surplus), followed by dual discrete schottky bridges, feeding a CRCRC filter using 47,000uF/25V panasonic THA caps for each C, and .33 ohm for each R.

For the prototype I just powered up, I reduced the power transistors to the minimum of 4 and scaled back the Aleph current sensing resistors from the 6 in the schematic to just 2 (as should be proper). So this 4V sag in the PS is with just a third of the current the entire amp will be using! I'm afraid to imagine what it will sag to with all 12 of the power transistors going.

There is good news:
At turn on, I got exactly 20mA in the differential current source, and both legs of the diff were pretty darn close to perfectly matched. I other words, the front end is good to go.
The current through each side of the second stage was 1.17A and 1.19A according to the voltages I got from the source resistors under Q2 and Q6, pretty darn close to what I was expecting, also. And the relative DC offset was something like 260mV and went down to 248mV after about 15min. All this with absolutely no adjustment whatsoever. BTW, I soldered both the + and - inputs to ground on the pcb for this initial testing.

BUT. . .
Absolute DC offset was +11V at start up and didn't come down hardly at all (ended up at +10.14 or something close to that) After fiddling with the diff. current source pot, I got the offset down to about 7.5V if I turned the pot all the way to zero. (but this of course made the current through to diff go up to 60mA). Next, I fiddled with the "magic resistors" and reduced them in half, from 4.7KOhm to 2.35KOhm. This got the offset down to 5.5V. But all this seemed like too much tweaking without getting the enough result.

All this brings me back to the power supply; I believe I have enough capacitance (47,000x3=141,000uF per rail), but is a 475VA tranny enough for this job? As a note, the tranny did not get warm at all during the testing. I thought they would get me by until I could afford some better ones, but maybe not?
And regarding the DC offset, could the drooping PS cause the Aleph current source to work incorrectly, resulting in the large offset?

WOW! I just re-read all this before I posted and I'm sorry about the super-long post; I guess I just wanted to put in all the pertinent info.

Any help is appreciated.

[wuffwaff]

Hi,

I think your large voltage drop is caused by a not very "stiff" transformer. Also 17V with a CRCRC will never make 22V with 6A bias. You will loose over 4V by the resistors alone.

I get 22V with 18V secondaries (600VA toroid, 18V specified at 600 VA), schottkys and C-L(2.2mH/0.12Ohm)-C at 7.2A of bias

Did you also match the IRFP240´s? Especially the two connected to the negative side.
260mV rel. offset is way too much and you can´t really adjust it anyway. The only thing I could more or less adjust was the absolute offset. The relative offset was always between 5 and 20mV.

I would try to match another pair of IRF9610´s at 20mA. Measure for at least 10min and match to better than 0,002 V.

You could also try to leave out the R´s out of the supply for a bit more voltage until you get new transformers.

William

[Blues]

sbear,

Try adjusting your diff CS to about 20 mA and then adjust the diff Drain resistors until you get just about 9mA or so on each branch. I believe both absolute and relative DC offset can be controlled through these resistors and having them equally share the total current from the diff CS.

[sbear]

Hey, again. I can't seem to post my schematic, but it's identical to the one in the high-power aleph-x thread. (I'm not a computer genius and I haven't figured out how to do links yet).

Anyway, you guys seem to be worried about matching everything, which I plan to do as soon as I can get the absolute offset close to zero. I actually built this prototype so I could match all my transistors in circuit. But I don't believe the wildest mismatch imaginable among the transistors would cause such a drastic absolute offset at the outputs (+10V for the + and - outputs), since the way I understand it, any mismatch on each side of the circuit would cause a relative offset, but the absolute offset should be pretty stable, a lot closer than +10V above ground!!

Anyway, I tried putting both my transformers in the PS to see if it helped make things "stiff". But I started blowing fuses, so I guess I'll need to think about using better inrush protection. I was using one CL60 thermistor, but I guess one is not enough for two 475VA trannys. I'll try figuring that out tomorrow.

[wuffwaff]

Hi,

I use exactly the same schematic and can easily set the absolute offset to 10V.
Not having everything matched before you put it together will cause you a lot of problems, including absolute DC offset.

William

[sbear]

Ok, wuffwaff, your reply has humbled me, so I will go about matching and check my offset after I have got everything matched as close as possible.

One thing that I have wondered regarding the schematic of the 100W aleph-x vs Grey's original schematic was how to select the proper values for r13,14,31, & 32. In the High powered aleph-x thread, Grey suggested increasing the values of these resistors as one of the necessary modifications in order to make it work. And since my power rails right now are quite a bit lower than the 22V I was hoping for, I thought that maybe the increased values of these resistors was causing the aleph current sources to not work correctly. I'll try changing them to Grey's original values and see what kind of change that makes.

That is, when I get the time. . . .time, there just is never enough time. . .I need more time. . . time, time, time . . .where does the time go. . .

BTW, if anyone ever asks you to run a full service restaurant, say NO!!!! and run. Run and don't stop. Keep running. If you feel as you can't go on, crawl. . . (sorry about the rant-I've worked about 80 hrs a week for the past month, with no relief in sight)

[wuffwaff]

Hi,

I used 1k5 for R14,15,31 and 32. Came up with this value after looking at the Aleph30&60 schematics.

Hope this helps,

William

[sbear]

OK, I solved all of the preliminary problems and I have completed one monoblock. Everything is working, I've got my offsets within reasonable tolerances and I hooked up a square wave (1Khz, 0.5V) with no load. Perfect!!!! I got a nice sharp wave at the output. Next, I hooked up an 8 ohm dummy load and. . .

I got an ok wave (see picture). The first picture is a zoomed in view of just one rise and the second pic is the same wave where you can see several cycles. I guess I should tell you that for these pictures I have 5pF caps across the feedback resistors. The wave looked too rounded for me so I removed the feedback caps and tried again. I thought I took a picture of the no cap trial, but I must have deleted it on accident, but I will describe it for you. The no cap wave looked exactly the same except for one very sharp overshoot spike. But after the spike the wave still wasn't flat but followed the slow gradual curve that didn't flatten until about halfway through the pulse as in the pics. What I'm am trying to say is that this sloping of the wave does not seem to be dependent on the the feedback caps. Any ideas?

[wuffwaff]

Hi,

did you have a look at the square wave input? Often this is also not perfect due to the signal generator.

What caps did you use in the active current sources?

William

[jarek]

hi sbear,

Look at my pictures http://www.diyaudio.com/forums/showt...329#post201329
Looks the same. IMHO it not depend on caps. This is effect of load resistance vs. amplifier output resistance.
I am living with it and very happy.