megajocke said:
You make a very convincing argument. Unfortunately it does not agree with measured results or with my simulations which DO match measured results. Perhaps my real units are a little short on the applied mathematics of electronics.
Your points:
#1- Sorry, that is incorrect. Both my simulation and my measured results show otherwise. And they agree.
#2- See #1.
Hint #2- And yet it does.
I do believe you think you understand what is going on in this circuit. I also believe I know what is going on. At this point I have measured over 100 units and confirmed voltages and operation. Since these results also match my simulations (and quite closely I might add) I will go with what I believe is happening.
I do appreciate you pointing out what you perceive as a design flaw. It simply does not show up in the real units.
Steve
EDIT: by the way, what I'm really interested in is the behaviour of the output bjt collector currents with respect to output current. I would be very much interested how small these (remaining, "keep on") collector currents can become in real circuits. If they indeed become microamp as in simulation I do not see much advantage to switch-off.
Have fun, Hannes
Hannes:
My measurements and simulations show collector current flowing in the "non conducting" transistor throughout the cycle. The minimum is at the peak of the waveform on the conducting side. It does get very low especially as clipping is approached but it does not go to zero. Even if it did, this still beats having it happen at 0.5V to 1V in the output swing.
Have fun, Hannes
Hannes:
My measurements and simulations show collector current flowing in the "non conducting" transistor throughout the cycle. The minimum is at the peak of the waveform on the conducting side. It does get very low especially as clipping is approached but it does not go to zero. Even if it did, this still beats having it happen at 0.5V to 1V in the output swing.
Hi Steve,
thanks again for your reply!
Please don't get me wrong, I would not bother if the current becomes microamperes near clipping; in simulation this happens very quickly depending on DC-bias current. If the current in the non-conducting bjt stays in general in milliampere-region I'm happy
All the best and thanks a lot for your great support, Hannes
thanks again for your reply!
Please don't get me wrong, I would not bother if the current becomes microamperes near clipping; in simulation this happens very quickly depending on DC-bias current. If the current in the non-conducting bjt stays in general in milliampere-region I'm happy
All the best and thanks a lot for your great support, Hannes
On "1) Ubc is about 2Vd, set by Q7,Q8" <= numbers and node names as in my schematic
You answered:
"#1- Sorry, that is incorrect. Both my simulation and my measured results show otherwise. And they agree."
But in post #54
http://www.diyaudio.com/forums/showthread.php?postid=1683679#post1683679
You said exactly what I stated in point 1:
"The voltage between the emitters of Q7 and Q10 is 1.2V Both transistors are on and conducting. "
So which way is it? Or maybe I should have been more clear that the transistors are numbered differently in my drawing, sorry for that.
"#2- See #1."
What is the bias voltage between driver stage bases then if it's not ~4 diode drops?
"Hint #2- And yet it does."
Not in the simulation you posted earlier, but it will under the right circumstances, for example if current sources and those biasing transistors are matched (maybe not by hand, maybe they are close enough from the factory) for example.
If it is not splitting 50-50 at idle it will at some other amplifier output current. If this is at a pretty low current and, with signal and load connected, voltage is read over transistors with a multimeter (which shows average) then it will read the same voltage over both too.
In this file you posted Q8 has 0.66V Vce while Q11 has about 0.2V
http://www.diyaudio.com/forums/attachment.php?s=&postid=1681062&stamp=1228927989
Or is it another simulation you are referring to? I'd like to know the measured voltage in the real circuit over those two, Q8 and Q11 in your diagram.
You answered:
"#1- Sorry, that is incorrect. Both my simulation and my measured results show otherwise. And they agree."
But in post #54
http://www.diyaudio.com/forums/showthread.php?postid=1683679#post1683679
You said exactly what I stated in point 1:
"The voltage between the emitters of Q7 and Q10 is 1.2V Both transistors are on and conducting. "
So which way is it? Or maybe I should have been more clear that the transistors are numbered differently in my drawing, sorry for that.
"#2- See #1."
What is the bias voltage between driver stage bases then if it's not ~4 diode drops?
"Hint #2- And yet it does."
Not in the simulation you posted earlier, but it will under the right circumstances, for example if current sources and those biasing transistors are matched (maybe not by hand, maybe they are close enough from the factory) for example.
If it is not splitting 50-50 at idle it will at some other amplifier output current. If this is at a pretty low current and, with signal and load connected, voltage is read over transistors with a multimeter (which shows average) then it will read the same voltage over both too.
In this file you posted Q8 has 0.66V Vce while Q11 has about 0.2V
http://www.diyaudio.com/forums/attachment.php?s=&postid=1681062&stamp=1228927989
Or is it another simulation you are referring to? I'd like to know the measured voltage in the real circuit over those two, Q8 and Q11 in your diagram.
Hey, I just found the other thread and this post:
I assume you by this mean Q8 and Q11 in your circuit is working in the saturation region? In that case the discussion is just about a misunderstanding because in post #41 (first one I posted) I wrote:
But when I there suggested the transistors had to be operating in the saturation region or in reverse you said I didn't understand the operation of the circuit. But I see you are referring to the region as "pinch off" so maybe that's why you didn't spot it.
Also, the files with voltage readings that showed non-equal voltage sharing didn't make it any clearer. I agree that the circuit will work when those two transistors are operating in the saturation region, that's pretty obvious. What I tried to explain is that the circuit cannot work as intended if the transistors were not run in the saturation region.
OT:
Pinch off usually refers to something entirely different, it's usually when the depletion region extends and "pinches off" the channel totally in FET transistors, making it as nonconductive as it can get. (the opposite of saturation) I've never seen it used for the saturation of BJT:s though. (what is it that is getting pinched off there?).
(It's not that you used pinch-off that made the issue unclear though, I hadn't seen the other thread before)
Steve Dunlap said:
I assume you by this mean Q8 and Q11 in your circuit is working in the saturation region? In that case the discussion is just about a misunderstanding because in post #41 (first one I posted) I wrote:
megajocke said:
But when I there suggested the transistors had to be operating in the saturation region or in reverse you said I didn't understand the operation of the circuit. But I see you are referring to the region as "pinch off" so maybe that's why you didn't spot it.
Also, the files with voltage readings that showed non-equal voltage sharing didn't make it any clearer. I agree that the circuit will work when those two transistors are operating in the saturation region, that's pretty obvious. What I tried to explain is that the circuit cannot work as intended if the transistors were not run in the saturation region.
OT:
Pinch off usually refers to something entirely different, it's usually when the depletion region extends and "pinches off" the channel totally in FET transistors, making it as nonconductive as it can get. (the opposite of saturation) I've never seen it used for the saturation of BJT:s though. (what is it that is getting pinched off there?).
(It's not that you used pinch-off that made the issue unclear though, I hadn't seen the other thread before)
Hello again megajocke:
First let me apologize for not replying sooner. I have satellite as my only high speed Internet option. When there is winter weather like now the snow and sleet block the signal. The ice isn't even reaching the ground here but the stuff in the upper atmosphere still messes with sending and receiving.
Now, it seems we have both realised we have a problem with understanding precisely what the other is trying to say.
So, starting 3 post back:
On "1) Ubc is about 2Vd, set by Q7,Q8" <= numbers and node names as in my schematic
You answered:
"#1- Sorry, that is incorrect. Both my simulation and my measured results show otherwise. And they agree."
But in post #54
http://www.diyaudio.com/forums/show...679#post1683679
You said exactly what I stated in point 1:
"The voltage between the emitters of Q7 and Q10 is 1.2V Both transistors are on and conducting. "
So which way is it? Or maybe I should have been more clear that the transistors are numbered differently in my drawing, sorry for that.
"#2- See #1."
What is the bias voltage between driver stage bases then if it's not ~4 diode drops?
I missunderstood what you meant by 2Vd and 4Vd. I took this to mean 2 volt and 4 volt when you meant diode drop. The first is a 2 diode drop as (we both) stated. Just not in the same post. The second (4Vd) or 4 diode drop is actually just under 2V. It can be adjuster from under 1.6V up to 2.2V With values shown. The adjustment is made with R12, which should be a variable resistor. Still short of a 4 diode drop, but not by much.
That is, the sum of the voltage drops over CE of Q5 and Q6 is 2 diode drops. But how does it divide?
Hint #2 = There's nothing in the circuit that will make it divide 50-50...
It does not divide 50-50 across the the emitters of Q7,Q10 but there is the 2 diode drop as we have both said. However, look at the voltages on the emitters of Q8,Q11. A split of 49.95% to 50.05% seems pretty close. Emitters of Q13,Q15 are even closer.
In this file you posted Q8 has 0.66V Vce while Q11 has about 0.2V
http://www.diyaudio.com/forums/atta...tamp=1228927989
These voltage drops reflect the differences on the collectors. The bias circuit forces matched voltages from the emitters onward. It does this by varying the resistance of the bias pair which is what causes the different collector voltages even with the same current into the emitters.
Steve
First let me apologize for not replying sooner. I have satellite as my only high speed Internet option. When there is winter weather like now the snow and sleet block the signal. The ice isn't even reaching the ground here but the stuff in the upper atmosphere still messes with sending and receiving.
Now, it seems we have both realised we have a problem with understanding precisely what the other is trying to say.
So, starting 3 post back:
On "1) Ubc is about 2Vd, set by Q7,Q8" <= numbers and node names as in my schematic
You answered:
"#1- Sorry, that is incorrect. Both my simulation and my measured results show otherwise. And they agree."
But in post #54
http://www.diyaudio.com/forums/show...679#post1683679
You said exactly what I stated in point 1:
"The voltage between the emitters of Q7 and Q10 is 1.2V Both transistors are on and conducting. "
So which way is it? Or maybe I should have been more clear that the transistors are numbered differently in my drawing, sorry for that.
"#2- See #1."
What is the bias voltage between driver stage bases then if it's not ~4 diode drops?
I missunderstood what you meant by 2Vd and 4Vd. I took this to mean 2 volt and 4 volt when you meant diode drop. The first is a 2 diode drop as (we both) stated. Just not in the same post. The second (4Vd) or 4 diode drop is actually just under 2V. It can be adjuster from under 1.6V up to 2.2V With values shown. The adjustment is made with R12, which should be a variable resistor. Still short of a 4 diode drop, but not by much.
That is, the sum of the voltage drops over CE of Q5 and Q6 is 2 diode drops. But how does it divide?
Hint #2 = There's nothing in the circuit that will make it divide 50-50...
It does not divide 50-50 across the the emitters of Q7,Q10 but there is the 2 diode drop as we have both said. However, look at the voltages on the emitters of Q8,Q11. A split of 49.95% to 50.05% seems pretty close. Emitters of Q13,Q15 are even closer.
In this file you posted Q8 has 0.66V Vce while Q11 has about 0.2V
http://www.diyaudio.com/forums/atta...tamp=1228927989
These voltage drops reflect the differences on the collectors. The bias circuit forces matched voltages from the emitters onward. It does this by varying the resistance of the bias pair which is what causes the different collector voltages even with the same current into the emitters.
Steve
quote:
Originally posted by megajocke
OT:
Pinch off usually refers to something entirely different, it's usually when the depletion region extends and "pinches off" the channel totally in FET transistors, making it as nonconductive as it can get. (the opposite of saturation) I've never seen it used for the saturation of BJT:s though. (what is it that is getting pinched off there?).
[/B]
Sorry, that was not right, pinch-off is an operating region in FETs where the depletion region pinches off the channel, but this does not imply zero current flow, just that current doesn't change much with Vds.
What is it that is getting pinched off? I should have said they are operating in the region near the off state. They are conducting but the resistance is fairly high. Small reductions in base current cause large increases in resistance. This is what keeps the output on when the other in the pair is conducting signal. The current drawn by the base of the "non conducting" output will increase with increased current flow in the signal conducting output. This slightly increases the base current in the bias pair keeping them from turning off.
Originally posted by megajocke
OT:
Pinch off usually refers to something entirely different, it's usually when the depletion region extends and "pinches off" the channel totally in FET transistors, making it as nonconductive as it can get. (the opposite of saturation) I've never seen it used for the saturation of BJT:s though. (what is it that is getting pinched off there?).
[/B]
Sorry, that was not right, pinch-off is an operating region in FETs where the depletion region pinches off the channel, but this does not imply zero current flow, just that current doesn't change much with Vds.
What is it that is getting pinched off? I should have said they are operating in the region near the off state. They are conducting but the resistance is fairly high. Small reductions in base current cause large increases in resistance. This is what keeps the output on when the other in the pair is conducting signal. The current drawn by the base of the "non conducting" output will increase with increased current flow in the signal conducting output. This slightly increases the base current in the bias pair keeping them from turning off.
Yes, it's pretty easy to get misunderstandings over the net...
I tried adjusting the simulation (diode string and resistor) to make Q8,Q11 work in the saturation region, about 200-250mV Vce. The bias current in output stage is around 40mA then, like you recommended.
The more bias is turned up though the more it misbehaves because Q8, Q11 gets enough Vce to work more like current sources instead of resistors - so I can see why a pretty low bias current compared to what is usual is optimal for the circuit configuration.
But as long as bias is kept low those are working more as resistors than transistors so it's not too bad. If there is 400mV to divide between them I'm sure it will divide equally, 500mV probably too, 600mV... maybe... but above that all bets are off. But it doesn't operate there I see now.
I tried adjusting the simulation (diode string and resistor) to make Q8,Q11 work in the saturation region, about 200-250mV Vce. The bias current in output stage is around 40mA then, like you recommended.
The more bias is turned up though the more it misbehaves because Q8, Q11 gets enough Vce to work more like current sources instead of resistors - so I can see why a pretty low bias current compared to what is usual is optimal for the circuit configuration.
But as long as bias is kept low those are working more as resistors than transistors so it's not too bad. If there is 400mV to divide between them I'm sure it will divide equally, 500mV probably too, 600mV... maybe... but above that all bets are off. But it doesn't operate there I see now.
megajocke said:Have you tried this type of circuit? Doesn't have issues at high bias currents if one wants that.
Maybe too boring though
Or how about the original Blomley?
Do one with split rails and the bias generator for the "steering" BJT pair would be especially easy to knock up (+/-0.7V from an other identical NPN/PNP pair connected as diodes).
But would it be better that a properly done double or triple EF?
Cheers,
Glen
Attachments
Interesting schematic...but thread is about Krill.. the little amp
Dunlap amplifier...not Blomlev or other...nice ideas, but maybe reduce focus.
Thread opening objective was Dunlap amplifier, simplified version offered to forum diyers to build, also to discuss and explain and teach his ideas about his topologie.... something alike may add or contribute or may diverge and reduce focus and attention to the main thread subject.
Well.... this is what i think... let's see what Dunlap thinks about.
The big damn reality, is that even having not bad will or bad intentions, those new ideas offered goes changing the main idea and schematic...improovements made will changing the face and main idea and finish creating a new amplifier, a clone inspired to a Dunlap by Dx or other guy.
You see.... i am interested to build... i felt the interest to build... so, interesting idea that should be object of another thread.... if people enter here and start to be interested to Blomlev ( I DO) will make this thread more and more empty with reference of the main topic.
regards,
Carlos
Dunlap amplifier...not Blomlev or other...nice ideas, but maybe reduce focus.
Thread opening objective was Dunlap amplifier, simplified version offered to forum diyers to build, also to discuss and explain and teach his ideas about his topologie.... something alike may add or contribute or may diverge and reduce focus and attention to the main thread subject.
Well.... this is what i think... let's see what Dunlap thinks about.
The big damn reality, is that even having not bad will or bad intentions, those new ideas offered goes changing the main idea and schematic...improovements made will changing the face and main idea and finish creating a new amplifier, a clone inspired to a Dunlap by Dx or other guy.
You see.... i am interested to build... i felt the interest to build... so, interesting idea that should be object of another thread.... if people enter here and start to be interested to Blomlev ( I DO) will make this thread more and more empty with reference of the main topic.
regards,
Carlos
You see... i am already interested..already simulating
and will build...i have appreciate the Klein.... idea
But i will not show it here, will no comment to avoid be off topic.
For sure those things, even beeing cooperation, helping hand, contribution, nice ideas to add and turn thread more rich in informations and options, represents also a danger to loose focus... we, thread visitors, should be carefull not to produce that effect.
Ahahaha... at least this hipnotizes uncle charlie.
bye.... busy doing new amplifier.
Carlos
and will build...i have appreciate the Klein.... idea
But i will not show it here, will no comment to avoid be off topic.
For sure those things, even beeing cooperation, helping hand, contribution, nice ideas to add and turn thread more rich in informations and options, represents also a danger to loose focus... we, thread visitors, should be carefull not to produce that effect.
Ahahaha... at least this hipnotizes uncle charlie.
bye.... busy doing new amplifier.
Carlos
Attachments
megajocke said:
In an earlier post I mentioned using a distortion analyzer to set the bias. I do realise not everyone has one. But, if you have access to one the easy way to set the proper bias is to start low and increase while watching for a null point. This is proper bias for the transistors used to build that amp. Setting the bias higher will increase the distortion (as you noted) and is simply a waste of electricity. You can make the amp run hot, for those that feel the need for heat, but you gain nothing and increase distortion.
Trust me, I have studied and examined every output stage I could get my hands on for the last 35 years. I simply wanted to do something original. Something that was mine. I have no quarrel with other designs, I just prefer my own. I have been fortunate enough to hear virtually every highly regarded amp brand currently in production and several that no longer are. I have also met the designers in most cases though I doubt many of them have reason to remember me. I build and listen to my own amps because I believe them to be as good as any I have heard, and I can afford mine. Of course the whole ego thing doesn't hurt either.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.