It seems that I missed the importance of a well-implemented current source for my Aleph-X's stability of offset: I have some problems with absolute offset stability, which is greatly temperature dependent. As the temp of the fet of the current source (CCS) rises the offset gets more negative and tha bias (measured on the positive rail) decreases.
If I touch the fet I feel it warm and, acting as a heatsink, I slightly lower its temperature, making the absolute offset rise a little.
Adding a small heatsink things got more stable, but before trying a different current source (I think I'll try the one with TL431 and a BJT because I already have the parts) I wanted to know if someone experienced something similar.
Cheers
Andrea
If I touch the fet I feel it warm and, acting as a heatsink, I slightly lower its temperature, making the absolute offset rise a little.
Adding a small heatsink things got more stable, but before trying a different current source (I think I'll try the one with TL431 and a BJT because I already have the parts) I wanted to know if someone experienced something similar.
Cheers
Andrea
Absolute offset voltages less than a volt, are not generally
a problem, so let's assume that you are talking figures as
high as this or higher.
This is why we typically load the output of the amp to ground
at each side, and also why we resistively feed both outputs
back to the Source pins of the input diff pair.
a problem, so let's assume that you are talking figures as
high as this or higher.
This is why we typically load the output of the amp to ground
at each side, and also why we resistively feed both outputs
back to the Source pins of the input diff pair.
jh6you said:
I have heard about the "magic" resistor in the AX thread. Are you indicating it? Thanks for the info.
I don't remeber NP being ambiguous about that in the past.
Andy,
in a rather crowded assembly (like my prototype) proximity of the ccs mosfet to heat sources makes abs DC adjustment a little bit tricky but very do-able nonetheless.
I tried using a TL431 CCS but I noticed something strange on the output halves of the waveform (referenced to GND), while the differential output remained the same. So I switched back to the original one.
While torture-testing it
I noticed that if I higly overload the amp the offset jumps to about -10Vdc (differ. offset stays within 200 mV) and idle current drops to almost zero. Removing input signal things come back to normality after a few seconds.. i think that during normal listening this will never occour, but I wonder if this situation is normal or not.
Any feedback?
Cheers
Andrea
While torture-testing it
I noticed that if I higly overload the amp the offset jumps to about -10Vdc (differ. offset stays within 200 mV) and idle current drops to almost zero. Removing input signal things come back to normality after a few seconds.. i think that during normal listening this will never occour, but I wonder if this situation is normal or not.Any feedback?
Cheers
Andrea
Andrea,
What you'll see depends on where you're measuring it. You'll see different things if you measure from one side's output to ground or across both sides' outputs. The Aleph-X, being a bridged amp, should be measured from output to output, not output to ground. Otherwise, you'll only see half the output signal.
As far as the "magic" resistor goes, my hat's off to Ian for picking up on it when I dropped the ball. I had it on a list of ideas to try when I was putting the Aleph-X together, but managed (wrongly, as it turned out) to convince myself that it wouldn't work. Silly me.
Grey
What you'll see depends on where you're measuring it. You'll see different things if you measure from one side's output to ground or across both sides' outputs. The Aleph-X, being a bridged amp, should be measured from output to output, not output to ground. Otherwise, you'll only see half the output signal.
As far as the "magic" resistor goes, my hat's off to Ian for picking up on it when I dropped the ball. I had it on a list of ideas to try when I was putting the Aleph-X together, but managed (wrongly, as it turned out) to convince myself that it wouldn't work. Silly me.
Grey
GRollins said:
Grey, first of all I take the occasion to thank you for your great work.
I know that the "real" output signal is between the two halves, but it seemed strange to me that with the TL431 CCS the single ended outputs were somewhat "modulated" by the rails' ripple, since with the FET CCS this doesn't happen.
The "magic" resistors are probably the answer to my previous question, as also Nelson pointed out, but I think that this behaviour (absolute offset jumping negative overloading the amp) is caused by slightly unsymmetrical clipping, that makes the absolute offset go negative, then the resistors "steal" current from the CCS making the absolute offset even more negative... a sort of positive feedback in that sense.
Maybe I'll try raising their value (now 4k7) and see if things get better, even if they are fine enough now.
Cheers
Andrea
Yeah, the basic Aleph circuit clips asymmetrically; the current source and output just don't behave the same out at the limits. The ratio between the two can be adjusted, but it's still likely to act a bit funny if you punch it hard.
But you wouldn't overdrive the poor thing, would you?
Lawsy me...current sources...now there's a topic guaranteed to raise a ruckus around here. The first stage of my no-holds-barred tube amp design uses a differential 6SN7 as a phase splitter. In order to get good matching between the two polarities of signal coming out you need a lot of resistance under the tail of the thing. I was in a hurry the day I set that up and used an LM317 with a resistor to cobble together a CCS. Now, the thing is, the LM317 CCS gets seriously non-linear when you're asking it for that small an output--in this case about 15 mA for the pair. Needless to say, I paid for my laziness. Backed up and did a proper CCS...ta da! The amp thanked me and did its job properly.
Grey
But you wouldn't overdrive the poor thing, would you?
Lawsy me...current sources...now there's a topic guaranteed to raise a ruckus around here. The first stage of my no-holds-barred tube amp design uses a differential 6SN7 as a phase splitter. In order to get good matching between the two polarities of signal coming out you need a lot of resistance under the tail of the thing. I was in a hurry the day I set that up and used an LM317 with a resistor to cobble together a CCS. Now, the thing is, the LM317 CCS gets seriously non-linear when you're asking it for that small an output--in this case about 15 mA for the pair. Needless to say, I paid for my laziness. Backed up and did a proper CCS...ta da! The amp thanked me and did its job properly.
Grey
Say Grey, are we going to see some of these tube designs you've been tinkering with? You never know, I just might be convinced to build one someday.
I think you mentioned the ladies really go for the romantic setting of a room lit only by filaments, with a nice glass of wine...who needs a fireplace and bearskin rug anyway?
(oh, welcome back to the forum again...)
I think you mentioned the ladies really go for the romantic setting of a room lit only by filaments, with a nice glass of wine...who needs a fireplace and bearskin rug anyway?
(oh, welcome back to the forum again...)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.