Aleph-X builder's thread.

PS Ripple Measurements

Finally got some time in the EE lab to do some measurements. First up, power supply ripple. With just the most basic arrangement for the power supply (Toroid -> Bridge -> 290,000uF Caps) on 21.4v rails at about 8A bias I get about 200mv of ripple. Using clip leads and various power resistors between the 70,000uF cap and the 220,000uF cap for a CRC filter, 0.2 ohms dropped ripple to about 40mv and 0.075ohms resulted in about 50mV of ripple.

Just caps: PS Ripple C Filter.jpg
CRC with 70,000uF, 0.2 ohms, followed by 220,000uF: CRC 0.2 ohms.jpg
 
Sine Wave Clipping Behavior

With 21v rails and just about 9A bias (adjusted bias so drop across mosfet source resistors measured 0.500 on both sides), I was able to drive 35.0v into a 4ohm load with a 2.0v (peak-to-peak) input sine wave before the onset of clipping. This looks works out to about 153w into 4ohms.

For an 8ohm load, I measured a voltage swing of 36.8v, yielding about 84w into 8 ohms before clipping occurs. This is pretty close to the prediction by the AXE1-2.xls spreadsheet of 90w into 8ohms and 160w into 4ohms. Its always nice to see reality live up to the simulations! :) :D

Overall, sine wave behavior looks pretty good (though the 1kHz and 2kHz look a little funny to me). The top trace is the input signal and the bottom trace is the output signal.

1kHz: Sine 1kHz.jpg
2kHz: Sine 2kHz.jpg
5kHz: Sine 5kHz.jpg
10kHz: Sine 10kHz.jpg
15kHz: Sine 15kHz.jpg
20kHz: Sine 20kHz.jpg

Sinewave at onset of clipping (Approx 153w into 4 ohms): 0.980v 4ohms .470bias clip.jpg
 
Last edited:
Square wave behavior

I then ran a series of square waves into the amp. The results don't look quite as nice as the sine waves. They start off rather square at lower frequencies (500Hz and 1kHz look pretty good), but as the frequency increases, they look less and less like squares on the output.

500Hz Square: Square 500Hz.jpg
1kHz Square: Square 1kHz.jpg
2kHz Square: Square 2kHz.jpg
5kHz Square: Square 5kHz.jpg
10kHz Square: Square 10kHz.jpg
15kHz Square: Square 15kHz.jpg
20kHz Square: Square 20kHz.jpg
40kHz Square: Square 40kHz.jpg

Are these OK as square wave output? Or do I need to being tweaking to try to make these higher frequency waves more square? These square waves were produced with a 1.5v input signal (peak to peak) into a 4 ohm load generating approximately 128w of output into the dummy load.

Thanks!
Eric
 
Last edited:
When I first got things up and running last week, I drove the amp with a 7v input sinewave from my cd player and preamp into the dummy load to make some measurements. After using the scope today, its pretty clear that I way over-drove the amp. Could this have damaged the output devices and led to these distorted output waves? Or is this something that playing with output coupling capacitors can help fix?

While I have measured the resistance of the dummy load (wire wound resistors), I have not measured its capacitance or inductance. Could this account for the output wave forms?
 
Hi Eric,

did you measure between the two outputs (differential) or from one output to ground? The latter will not give you the right results. You can measure each output to ground and ADD the signals when using a two channel scope.

I can see my famous spike too.....

At what voltage did you measure the square waves? At too high voltages something in the active current source can´t follow anymore and you get some nasty distortion.
I´ve got some pics of square waves in my Aleph-X, and Aleph J X threads.

Did you check the frequency response into 8 and 4 ohm?

William
 
William,

Thanks for the links - I'll read them more thoroughly tonight and tomorrow.

The setup for measuring output was signal generator for input and the dummy load resistors connected to the output. The leads for the scope were connected directly across the dummy load (set to 4ohms). What I didn't notice until I was done was that I was running a rather large signal into the amp, so the amp was running close to full output with the square wave input. With a 4 ohm load, clipping starts when the input signal is close to 2.0v (peak-to-peak) and with the square waves, the scope indicates the input signal was at 1.5v, so the amp is fairly close to running out of steam with these signals.

Is the setup correct? I just assumed that I should be using the dummy load while running the signal generator.

How do I measure the frequency response of the amp? I was thinking that I should be doing this while I was working, but wasn't sure what the procedure was. The signal generator I was using would only go down as low as 100Hz or 500Hz (I forget which...). Do I just run a sine wave in and look for output attenuation at either end of the frequency spectrum?
 
Hi,

try running the tests at 1V and 10V output. The sinusses should look undistorted from 10Hz to >200kHz
try to measure the gain (looking at the voltages on the scope pictures this can´t be right)
Try to measure with two scope channels differentialy. The way you measured it now can make a lot of problems when the scope and the amp are connected to ground.
One probe on each side of the dummy load, ground to amp chassis.

For frequency response just turn up the frequency until the voltage goes down to aprox. 0,7 times the voltage at 1kHz. That´s your -3dB point.

William
 
Thanks, William! I'll give this a try. The lab tech indicated he had some equipment I could bring home to use which will make things much easier than lugging this heavy amp around and getting time between classes.

The voltages on the scope are reduced by a factor of 10, we used a 10:1 probe since the scope was unable to display the 2v input signal and a 40v output signal simultaneously - the 40v signal was too large for some reason. So, where it indicates 3.56v (or whatever it was) across the bottom, it is really 35.6v that is being output.

We thought about the ground loop between the chassis and the scope possibly causing a problem with the output waveform, so I lifted AC Mains ground from the PSU 0v point on the caps before making these measurements. The AC Mains ground had been connected to the PSU 0v point through a thermistor. The chassis remained grounded throughout testing.

I'll reduce the input signal for further testing. The measurements of the square waves indicate that the amp was producing nearly 103w into 4ohms at the time (max output into 4 ohms is near 150w, so the amp was working pretty hard at the time).

If I use two channels on the scope (one channel connected to speaker pos and chassis, the other connected to speaker neg and chassis), should I then see two normal looking sine waves on the scope? I was using the RCA input while I conducted these measurements - is should I be using the RCA or XLR input?

It will probably be early next week before I can get my hands on a scope and function generator again.

Thanks!
Eric
 
Hi William,

I've just read through the threads that you linked above -thank you. Based on the waves that you posted, mine are definitely out of whack. Until I am able to do more testing with a scope, my guess is that my trouble is related to 1) testing at much too high of an output level and 2) from the thread you referenced my bias settings are likely not to be perfectly balanced across the two sides of the amp. I'll have to try the technique posted by Hugo for setting the bias with a scope.

Replacing the 9610s in the input differential with JFets sounds like an interesting and worthwhile modification to make. I'll have to read through the discussion a few more times before I understand enough be able to being asking you lots of questions ;)

As for testing output on the scope with two leads, I'm not 100% clear on the procedure (the first time I've used a scope was last week, so I still have a lot to learn here). When you make reference to testing each side of the amp with respect to the chassis ground, are you driving the amp through the XLR input or the RCA input? I'm guessing the XLR input connecting the signal generator to the pos signal in and neg signal in. Is this correct?

I have finally gotten tired of listening to my buzzing transformer and ordered a custom 1500Va with 19v secondaires. This ought to provide a little extra voltage for me to burn off with my CRC filter (to reduce ripple) and a bit less rail droop under load. I am also going to rebuild the chassis using 6 heatsinks rather than the 4 that I am using now. Based on my calculations, this should reduce the final chassis temp from just over 50c (which is quite hot) to closer to 40c.

Sometimes I think I should have majored in EE instead of computer science and information systems...
 
Last edited:
JFET front end

I think I found it. I spent a 6 hours last night reading through what seems like a dozen never-ending threads...

The last discussion I read included a PDF schematic from Graeme where he uses the 2SJ109BL JFET: View attachment Graeme ax100j-01.pdf

Another discussion identifies the 2SK170BL as a direct drop-in replacement for the discontinued 2SJ109BL. The 2SK170BL seems to be available from a few places still...

I want to work out the troubles with my output sine and square waves before moving onto a JFET front end, though.
 
Official Court Jester
Joined 2003
Paid Member
I think I found it. I spent a 6 hours last night reading through what seems like a dozen never-ending threads...

The last discussion I read included a PDF schematic from Graeme where he uses the 2SJ109BL JFET: View attachment 159120

Another discussion identifies the 2SK170BL as a direct drop-in replacement for the discontinued 2SJ109BL. The 2SK170BL seems to be available from a few places still...

I want to work out the troubles with my output sine and square waves before moving onto a JFET front end, though.


2SJ109BL = 2 x 2SJ74BL

2SJ isn't 2SK !!!!