batteries again
After a little more digging on this topic, I'm forced to agree that battery operation would be prohibitively expensive here. My initial cost estimate simply multiplied the rate of consumption by the desired playing time to arrive at a total battery capacity requirement.
In reality, this is way off base if you need the battery voltage at the end of play to be within 10% or so of the starting voltage. In that case, according to sources I've read, you have to multiply that basic capacity figure by a factor of 20 or more, at which point cost goes through the roof.
As a side note, this exercise has given me a new respect for the spring motor in my 77-year-old Victrola, from which I can get several minutes of reasonably pitch-stable play from a few turns of the crank.
After a little more digging on this topic, I'm forced to agree that battery operation would be prohibitively expensive here. My initial cost estimate simply multiplied the rate of consumption by the desired playing time to arrive at a total battery capacity requirement.
In reality, this is way off base if you need the battery voltage at the end of play to be within 10% or so of the starting voltage. In that case, according to sources I've read, you have to multiply that basic capacity figure by a factor of 20 or more, at which point cost goes through the roof.
As a side note, this exercise has given me a new respect for the spring motor in my 77-year-old Victrola, from which I can get several minutes of reasonably pitch-stable play from a few turns of the crank.
Grey,
Did you check the phase behaviour of your ALEPH-X?.
I am getting some strange results when running sims. But then it could be that I am doing something wrong running the sim.
Running the same sim on a normal ALEPH 3 gives good results. At least what I would expect.
I hope the real world circuit is not plagued by any of this.
regards
Did you check the phase behaviour of your ALEPH-X?.
I am getting some strange results when running sims. But then it could be that I am doing something wrong running the sim.
Running the same sim on a normal ALEPH 3 gives good results. At least what I would expect.
I hope the real world circuit is not plagued by any of this.
regards
phase behaviour
"I am getting some strange results when running sims. But then it could be that I am doing something wrong running the sim." Or it could be that the amp has insufficient phase margin.....
Phase margin perhaps? Remove C104 in your Aelph 3 simulation and see what happens. Also try adding 0.047uF cap from collector to emmiter of Q105 and tell us what happens with and without C104.
http://www.engin.umich.edu/group/ctm/freq/freq.html
H.H.
"I am getting some strange results when running sims. But then it could be that I am doing something wrong running the sim." Or it could be that the amp has insufficient phase margin.....
Phase margin perhaps? Remove C104 in your Aelph 3 simulation and see what happens. Also try adding 0.047uF cap from collector to emmiter of Q105 and tell us what happens with and without C104.
http://www.engin.umich.edu/group/ctm/freq/freq.html
H.H.
John,
That's the sneaky part about X stuff, it doesn't look like much once you strip the cascodes and current sources off. It's nothing more than a feedback strategy for balanced circuits. Pretty simple really. Get out the patent, ignore all the fancy stuff, and trace the signal phase relationships. You have no idea how many times I looked at that patent before the light came on over my head.
Okay, folks, I finally got a chance to play cap games with this circuit. I put caps in here and there and...nothing happened. Zippo. Nada.
So I says to myself,"Self, what's going on here?"
I mean, rtirion has this thing run up in his simulation program and says something looks funny. Harry is being Harry--even to the point of spewing vituperation at me about this in an unrelated thread. And, yes, there are one or two caps (depending on the model) in Nelson's Alephs that I don't have in this circuit. So what's the deal?
I'm sitting here looking at the circuit and most of it's assembled on three of those experimenter boards that you push parts into, and it occurs to me to wonder what the capacitance between the rows of contacts might be. I've always assumed that it's low...on the order of a few pF. But what if it's higher? The total distributed capacitance could add up to a fair amount, depending on whatever the individual numbers might be. If so, that might be a nice--albeit unpredictable--bandwidth limiter.
Short of doing a complete layout, etching a board, and trying that, does anyone have any suggestions for solving this conundrum?
Grey
P.S.: Harry, I'm more than willing to try to work this out to everyone's satisfaction, but your arrogance, condescension, and negativity wears thin. If you have something constructive to add, by all means do so, otherwise please refrain from burning bandwith on the Net. You have alienated members of this site to the point that people are leaving and many others dread your posts in their threads. My e-mail tells me that the ratio is about ten or twenty to one against you.
This is <i>not</i> a good sign.
If you don't intend to help, and I mean <i>real</i> help, not stunts like tossing an admitted newbie with a non-functioning Aleph a URL suited more suited for 2nd year EE students (what the dickens were you thinking, man?), then don't bother posting.
And, yes, I did make an attempt to explain current sources (both standard and Aleph) in layman's terms...more than you could be bothered to do. If this is what getting an EE does to someone's attitude, I'm glad I didn't get one (I considered it briefly before doing geology and psychology instead).
Or, in the vernacular, buzz off.
That's the sneaky part about X stuff, it doesn't look like much once you strip the cascodes and current sources off. It's nothing more than a feedback strategy for balanced circuits. Pretty simple really. Get out the patent, ignore all the fancy stuff, and trace the signal phase relationships. You have no idea how many times I looked at that patent before the light came on over my head.
Okay, folks, I finally got a chance to play cap games with this circuit. I put caps in here and there and...nothing happened. Zippo. Nada.
So I says to myself,"Self, what's going on here?"
I mean, rtirion has this thing run up in his simulation program and says something looks funny. Harry is being Harry--even to the point of spewing vituperation at me about this in an unrelated thread. And, yes, there are one or two caps (depending on the model) in Nelson's Alephs that I don't have in this circuit. So what's the deal?
I'm sitting here looking at the circuit and most of it's assembled on three of those experimenter boards that you push parts into, and it occurs to me to wonder what the capacitance between the rows of contacts might be. I've always assumed that it's low...on the order of a few pF. But what if it's higher? The total distributed capacitance could add up to a fair amount, depending on whatever the individual numbers might be. If so, that might be a nice--albeit unpredictable--bandwidth limiter.
Short of doing a complete layout, etching a board, and trying that, does anyone have any suggestions for solving this conundrum?
Grey
P.S.: Harry, I'm more than willing to try to work this out to everyone's satisfaction, but your arrogance, condescension, and negativity wears thin. If you have something constructive to add, by all means do so, otherwise please refrain from burning bandwith on the Net. You have alienated members of this site to the point that people are leaving and many others dread your posts in their threads. My e-mail tells me that the ratio is about ten or twenty to one against you.
This is <i>not</i> a good sign.
If you don't intend to help, and I mean <i>real</i> help, not stunts like tossing an admitted newbie with a non-functioning Aleph a URL suited more suited for 2nd year EE students (what the dickens were you thinking, man?), then don't bother posting.
And, yes, I did make an attempt to explain current sources (both standard and Aleph) in layman's terms...more than you could be bothered to do. If this is what getting an EE does to someone's attitude, I'm glad I didn't get one (I considered it briefly before doing geology and psychology instead).
Or, in the vernacular, buzz off.
My e-mail tells me that the ratio is about ten or twenty to one against you
OH PLEASE! I have never gotten an Email complaining about my post and you have gotten enough to discern 10 to one ratios on for and against? I guess we will have to take your word on this one. I have seen you lobby to keep 17 year olds that with unquestionably abusive language on the forum and yet I am driving people away? Isn't there a feature where one can block certain posters? I'm sorry that some what I post is over a few people heads. Guess what, alot of it isn't and some people read it and learn something new. It seems to me that I see more "thank you, Harry"s than "buzz off Harry"s in reponse to my post....
H.H.
OH PLEASE! I have never gotten an Email complaining about my post and you have gotten enough to discern 10 to one ratios on for and against? I guess we will have to take your word on this one. I have seen you lobby to keep 17 year olds that with unquestionably abusive language on the forum and yet I am driving people away? Isn't there a feature where one can block certain posters? I'm sorry that some what I post is over a few people heads. Guess what, alot of it isn't and some people read it and learn something new. It seems to me that I see more "thank you, Harry"s than "buzz off Harry"s in reponse to my post....
H.H.
Email
Thanks Geoff I wasn't aware. I will rectify the situation. All mail pro or con can be sent to Fdieck@hotmail.com
H.H.
Thanks Geoff I wasn't aware. I will rectify the situation. All mail pro or con can be sent to Fdieck@hotmail.com
H.H.
Hi grey and john-china,
I think john-china had a good question .. I did some simulations one the circuit and I think I understand the way this topology came up.. but I do have another question:
My simulation shows more linear behavior of the folded-cascode X-preamp topology compaired to the BOSOZ topology, so why not use the folded-cascode X-preamp and a Aleph active current source output stage?
greetings,
Thijs
I think john-china had a good question .. I did some simulations one the circuit and I think I understand the way this topology came up.. but I do have another question:
My simulation shows more linear behavior of the folded-cascode X-preamp topology compaired to the BOSOZ topology, so why not use the folded-cascode X-preamp and a Aleph active current source output stage?
greetings,
Thijs
Grey,
Please do not get angry with me. I am not going to criticize your ALEPH-X. Not before I have built one. I am just trying to understand what is going on in the circuit.
And to me it is a little easier to try out the circuit in a simulator. Although a simulator is a good tool to try and get some extra information about a circuit, I know they are not flawless.
The correct interpretation of the results is also a different matter.
So, as I stated in my previous post,
- Could be I am doing something stupid.
- I will post results and my interpretation later.
For now I am still happy to see an actual implementation of an X and we have you to thank for that.
Harry,
This is my understanding of the bode/nyquist stuff:
An amplifier using feedback must, lower it's open loop gain from a frequency A to unity (gain=1).
At untity the phase shift should be less than 180 degrees. (Open loop gain lowers 12dB/octave).
You could see this as a bare minimum if you want a stable amplfier.
Since a reactive load (speaker) will have a bad influence on the phase shifting, you are aiming for less than 90 deg open loop phase shift. (Open loop gain lowers 6dB/octave).
Adding compensation capacitors is the usual way to influence the open loop gain, to get the required stability.
To minimize the effect of phase shift (minimal phase shift in 20-20K) you want the frequency A as high as possible
without sacrificing stability.
According to my sim, noticable phase shift for the ALEPH-X starts between 1Khz and 2Khz. This is what puzzles me.
I hope this explanation is somewhat correct and that this story survived my tanslation skills.
If not, I hope this gets corrected.
Regards
Please do not get angry with me. I am not going to criticize your ALEPH-X. Not before I have built one. I am just trying to understand what is going on in the circuit.
And to me it is a little easier to try out the circuit in a simulator. Although a simulator is a good tool to try and get some extra information about a circuit, I know they are not flawless.
The correct interpretation of the results is also a different matter.
So, as I stated in my previous post,
- Could be I am doing something stupid.
- I will post results and my interpretation later.
For now I am still happy to see an actual implementation of an X and we have you to thank for that.
Harry,
This is my understanding of the bode/nyquist stuff:
An amplifier using feedback must, lower it's open loop gain from a frequency A to unity (gain=1).
At untity the phase shift should be less than 180 degrees. (Open loop gain lowers 12dB/octave).
You could see this as a bare minimum if you want a stable amplfier.
Since a reactive load (speaker) will have a bad influence on the phase shifting, you are aiming for less than 90 deg open loop phase shift. (Open loop gain lowers 6dB/octave).
Adding compensation capacitors is the usual way to influence the open loop gain, to get the required stability.
To minimize the effect of phase shift (minimal phase shift in 20-20K) you want the frequency A as high as possible
without sacrificing stability.
According to my sim, noticable phase shift for the ALEPH-X starts between 1Khz and 2Khz. This is what puzzles me.
I hope this explanation is somewhat correct and that this story survived my tanslation skills.
If not, I hope this gets corrected.
Regards
the bode/nyquist stuff
The number that is important is the phase shift between the input
and the output when amp is at unity gain. This number is refered to as the phase margin. The frequency where the open loop response rolls off is not directly a factor in this. Many op amps open loop gain start rolling off at 10 Hz to 100Hz and still have good phase margin. Spice is an excellent tool to play with high frequency compesation with.
H.H.
The number that is important is the phase shift between the input
and the output when amp is at unity gain. This number is refered to as the phase margin. The frequency where the open loop response rolls off is not directly a factor in this. Many op amps open loop gain start rolling off at 10 Hz to 100Hz and still have good phase margin. Spice is an excellent tool to play with high frequency compesation with.
H.H.
janey,
Given +-20V rails (and sufficient current), you're looking at an amplifier that will deliver about 150W into 4 ohms. Your problem is going to be device dissipation, which with one pair of ouput devices per side (the way I drew the schematic) will be on the order of 63W per device.
Ouch!
Okay, so we need to back up and look at two, or perhaps three pairs per side. With two pairs of outputs per side, you're looking at half that, or about 31W per device. Okay, that's doable, especially if you've got good heatsinks. Or you could go with three pairs per side and suddenly things start getting entirely reasonable--about 21W per device.
For two pairs of devices per side, shoot for a bias around 1.5A...say, .33 ohm resistors (a little scant actually, maybe try .30). For three pairs per side go for a little over an amp--.47 ohms should do the trick.
It wouldn't hurt to recalculate the resistor for the front end current source while we're at it--3.3k will do there.
I'm at work, running this off on the little calculator on the PC. I'll try to remember to double check these figures later, but they should be sufficient to get you started. Just bear in mind that you're going to be running a fairly heavy amount of current and plan your power supply accordingly. As soon as we get this stability question settled so that everyone's happy, I'll be addressing the power supply requirements in detail later on.
Harry,
I could respond at length, but I will simply note that some of the more self-assured members address you directly, but there are a large number of people who are not so bold. They have no intention of contacting you directly. Why jump from the frying pan into the fire? Instead, they contact me, either on their own behalf or because they are offended on behalf of others (for whatever reason, the latter is more common than the former, dunno why), and have me deal with the situation. This leads to things like my post to you in Marc's Official DIY Preamp thread which appeared to come out of the clear blue sky...unless you knew about the e-mails behind the scene. I blame myself for not getting involved in the preamp thread earlier, but I can't cover things here the way I used to; there are simply too many threads and I have too little time.
Yes, the member you refer to used language that I'd prefer he hadn't used on this site. But you've upset a lot more people than he did, and you show no signs whatsoever of realizing what a mess you've made. At least he seemed to realize that he'd overdone it.
Thijs,
Yeah, historically speaking, I started with the full X front end, straight from the patent. That just about drove me crazy. Complete fiasco. I kept stripping things away until I ended up back where I started--just the simple differential. The thing is, the Aleph back end has voltage gain, so you don't have to jack up the voltage with this long, strung-out front end. Just 20 dB or so will do the trick (one of these days I may get around to trying a tube front end, just for fun). At that point it simply became a question of trying to get the phase relationships right so that it would be "X." I was concerned whether crossing over to the other side was 'allowable' by the rules of X. I was prepared to try to run each side completely independent of the other except for the linked Sources in the front end differential (the way the X patent goes, where the two halves are linked only by the one resistor), but I didn't want to go that route if I didn't have to because the criss-cross front end was so elegant...if it satisfied Nelson as being X.
Trust me, it was with great trepidation that I sent him this unorthodox schematic that went from one side to the other, when the patent clearly showed the two sides remaining separate. Fortunately, Nelson didn't smite me with a lightning bolt for being so impertinent as to mis-read his patent on purpose.
rtirion,
Good heavens, I'm not mad at you! If there's a problem with the circuit, let's figure what it is and fix it. I no longer model things. Had two different modelling programs on my computer until Mr. Virus came to visit this past winter. However, I was continually frustrated by a lack of Spice models for devices I wanted to use, and the inaccuracy of the models I did have compared to real life. After I rebuilt my PC, I just never got around to putting all that stuff back on there. It can be useful sometimes, that I freely grant. But to me it's more fun to just go ahead and build the silly thing and be done with it. Using push boards, I can change components in about the same amount of time that I can model something.
Maybe the push boards have something to do with the difference between what you see when you model things and what I'm seeing here. I have no idea. I advanced the possibility about stray capacitance, but that doesn't mean that I'm convinced that's what's going on. It could be any number of factors.
By the way, just for fun I ran the amp without the caps across the 100k feedback resistors (no closed loop compensation). There was a slight rise in gain around, I forget, maybe 125kHz? but otherwise it <i>still</i> refused to oscillate. (What does it take to make this thing misbehave, anyway?) Resistive load, by the way. I still haven't had time to scrounge up a cap and create a torture test load with a couple of uF. I'll try that next, but whether I get to it tomorrow or not remains to be seen.
Harry's post about phase shift at unity gain is quite correct.
The real killer for a negative feedback circuit is 180 degrees, which means that you suddenly have positive feedback instead of the negative feedback you thought you had. At that point, the circuit begins to radio Mars, requesting to be relieved of duty, since clearly all Earthlings are sadists who want to torture innocent circuits.
The oddity is that...under those conditions (only) you can actually hook the circuit up as though it was a positive feedback circuit, and things will calm down immediately. Nice for a parlor trick, but not something you'd want to do in your system. In other words: Don't Try This At Home.
Grey
Given +-20V rails (and sufficient current), you're looking at an amplifier that will deliver about 150W into 4 ohms. Your problem is going to be device dissipation, which with one pair of ouput devices per side (the way I drew the schematic) will be on the order of 63W per device.
Ouch!
Okay, so we need to back up and look at two, or perhaps three pairs per side. With two pairs of outputs per side, you're looking at half that, or about 31W per device. Okay, that's doable, especially if you've got good heatsinks. Or you could go with three pairs per side and suddenly things start getting entirely reasonable--about 21W per device.
For two pairs of devices per side, shoot for a bias around 1.5A...say, .33 ohm resistors (a little scant actually, maybe try .30). For three pairs per side go for a little over an amp--.47 ohms should do the trick.
It wouldn't hurt to recalculate the resistor for the front end current source while we're at it--3.3k will do there.
I'm at work, running this off on the little calculator on the PC. I'll try to remember to double check these figures later, but they should be sufficient to get you started. Just bear in mind that you're going to be running a fairly heavy amount of current and plan your power supply accordingly. As soon as we get this stability question settled so that everyone's happy, I'll be addressing the power supply requirements in detail later on.
Harry,
I could respond at length, but I will simply note that some of the more self-assured members address you directly, but there are a large number of people who are not so bold. They have no intention of contacting you directly. Why jump from the frying pan into the fire? Instead, they contact me, either on their own behalf or because they are offended on behalf of others (for whatever reason, the latter is more common than the former, dunno why), and have me deal with the situation. This leads to things like my post to you in Marc's Official DIY Preamp thread which appeared to come out of the clear blue sky...unless you knew about the e-mails behind the scene. I blame myself for not getting involved in the preamp thread earlier, but I can't cover things here the way I used to; there are simply too many threads and I have too little time.
Yes, the member you refer to used language that I'd prefer he hadn't used on this site. But you've upset a lot more people than he did, and you show no signs whatsoever of realizing what a mess you've made. At least he seemed to realize that he'd overdone it.
Thijs,
Yeah, historically speaking, I started with the full X front end, straight from the patent. That just about drove me crazy. Complete fiasco. I kept stripping things away until I ended up back where I started--just the simple differential. The thing is, the Aleph back end has voltage gain, so you don't have to jack up the voltage with this long, strung-out front end. Just 20 dB or so will do the trick (one of these days I may get around to trying a tube front end, just for fun). At that point it simply became a question of trying to get the phase relationships right so that it would be "X." I was concerned whether crossing over to the other side was 'allowable' by the rules of X. I was prepared to try to run each side completely independent of the other except for the linked Sources in the front end differential (the way the X patent goes, where the two halves are linked only by the one resistor), but I didn't want to go that route if I didn't have to because the criss-cross front end was so elegant...if it satisfied Nelson as being X.
Trust me, it was with great trepidation that I sent him this unorthodox schematic that went from one side to the other, when the patent clearly showed the two sides remaining separate. Fortunately, Nelson didn't smite me with a lightning bolt for being so impertinent as to mis-read his patent on purpose.
rtirion,
Good heavens, I'm not mad at you! If there's a problem with the circuit, let's figure what it is and fix it. I no longer model things. Had two different modelling programs on my computer until Mr. Virus came to visit this past winter. However, I was continually frustrated by a lack of Spice models for devices I wanted to use, and the inaccuracy of the models I did have compared to real life. After I rebuilt my PC, I just never got around to putting all that stuff back on there. It can be useful sometimes, that I freely grant. But to me it's more fun to just go ahead and build the silly thing and be done with it. Using push boards, I can change components in about the same amount of time that I can model something.
Maybe the push boards have something to do with the difference between what you see when you model things and what I'm seeing here. I have no idea. I advanced the possibility about stray capacitance, but that doesn't mean that I'm convinced that's what's going on. It could be any number of factors.
By the way, just for fun I ran the amp without the caps across the 100k feedback resistors (no closed loop compensation). There was a slight rise in gain around, I forget, maybe 125kHz? but otherwise it <i>still</i> refused to oscillate. (What does it take to make this thing misbehave, anyway?) Resistive load, by the way. I still haven't had time to scrounge up a cap and create a torture test load with a couple of uF. I'll try that next, but whether I get to it tomorrow or not remains to be seen.
Harry's post about phase shift at unity gain is quite correct.
The real killer for a negative feedback circuit is 180 degrees, which means that you suddenly have positive feedback instead of the negative feedback you thought you had. At that point, the circuit begins to radio Mars, requesting to be relieved of duty, since clearly all Earthlings are sadists who want to torture innocent circuits.
The oddity is that...under those conditions (only) you can actually hook the circuit up as though it was a positive feedback circuit, and things will calm down immediately. Nice for a parlor trick, but not something you'd want to do in your system. In other words: Don't Try This At Home.
Grey
I could respond at length........
"But you've upset a lot more people than he did, and you show no signs whatsoever of realizing what a mess you've made." The only Email I have recieved is one of support so far. Come on we are all adults here, for anyone with a real beef my Email is fdieck@hotmail.com. I would really like to know if this is for real, about all I have now is Grey's word for it.
H.H.
"But you've upset a lot more people than he did, and you show no signs whatsoever of realizing what a mess you've made." The only Email I have recieved is one of support so far. Come on we are all adults here, for anyone with a real beef my Email is fdieck@hotmail.com. I would really like to know if this is for real, about all I have now is Grey's word for it.
H.H.
It's the same principle as our legal system in cases such as The State vs. So-And-So. The government serves as a proxy for the wronged party. It's really a pretty decent system, even when the wronged party is strong enough to stand on their own.
Harry, the part you cannot comprehend it seems is that others are not like you. You feel that you're being fair by giving people a chance to 'duke it out with you, mano a mano.' From your point of view, if someone offered you that chance, you would probably take it. But others feel more like you're the schoolyard bully, and that there's nothing to gain and possibly much to lose in the sense of more sarcasm, more putdowns, more whatever. They don't want it, don't need it, and have no intention of risking it.
Until you grasp the fact that the other viewpoint exists, you're not going to 'get it.'
Grey
Harry, the part you cannot comprehend it seems is that others are not like you. You feel that you're being fair by giving people a chance to 'duke it out with you, mano a mano.' From your point of view, if someone offered you that chance, you would probably take it. But others feel more like you're the schoolyard bully, and that there's nothing to gain and possibly much to lose in the sense of more sarcasm, more putdowns, more whatever. They don't want it, don't need it, and have no intention of risking it.
Until you grasp the fact that the other viewpoint exists, you're not going to 'get it.'
Grey