A subject that causes to me questions also is the Vbe multiplier. This except the idle current that offers to output transistors, it ensures also thermal compensation to keep the bias in a safe margin unless the output devices will be in danger of destruction due to overheating. Those from you who have read the D.Self book, you have seen that he puts in this place the MJE340. From my experiments I have realised that MJE 340 is a lot slow in the recovery of the preset idle current after the pause of amplification procedure during that the heatsink it is heated significantly. I have tried other devices of common type such as BD137 and BD139 and I have realized that these are by far much faster from MJE340 in thermal compensation and thus in the recovery of the original idle current after heating. Concretely, I realized that as much higher the Vce of transistor as much slower its thermal reaction. In an amplifier with a supply of +/-70Vdc (thus 140Vpp) the peak of signal across the C-E contacts of transistor it touches many times upon the 140Vpp. Thus the use of lower Vce transistors such as BD137 (60V) or BD139 (80V) it is a risk. From the other hand I fear the use of MJE340 (Vce=300V) because his slowness (a thermal runaway watch for in output transistors? I am not sure)
Also and this time I wait for your answers with your thoughts, your experience and your opinion on this subject.
Thanks in advance for your help
Fotios
Also and this time I wait for your answers with your thoughts, your experience and your opinion on this subject.
Thanks in advance for your help
Fotios
Fotios, are you aware there is a whole thread extensively dealing with this issue?
Somehwere around here:
http://www.diyaudio.com/forums/showthread.php?postid=1277243#post1277243
Jan Didden
Somehwere around here:
http://www.diyaudio.com/forums/showthread.php?postid=1277243#post1277243
Jan Didden
Hi Janjanneman said:
No i never have seen this thread at whole (it is verry long and my time it is allways limited due to work, many times i spend my sleep to read or write posts) and i had not idea that in this thread there is also a discussion around the idle current (Vbias if you prefer) generated by the Vbe multiplier. I taked a look in the post that you addressed me, but i have not find anything with clear and practical documentation to give an answeer in my bothering. Sorry, but the formulation of my thoughts it is clearly of a practical implementation of Vbe multiplier with 2 or 3 concrete types of transistors (MJE340 vs. BD137 and BD139 in speed of thermal recovery of preset Ic and the Vce sustain). In this thread of B.Cordell "BJTs vs. FET" the only that i found are some exchanges of long discussions and equations and not a practical conclusion (as much as i have read). I will frame with other words my bothering. From a first view i suppose that the use of a MJE340 it is reasonable because his Vce breakdown it is 300V thus it is in position to sustain with safe large excursions of signal of 140Vpp. But his sloweness to recover his preset Ic - or idle current - after a significant heating of his case (which increase his Ic significatly) it is enough in time to cause a thermal runaway in output transistors? This is my query exactly. Also i excluded the use of diodes in this place because i know VERRY WELL that they needed at least two doudle diodes "special selected" in series with two simple diodes to achieved a voltage of +1,75V to -1,75V at the base of the two opposite driver transistors for a class AB bias level of output devices. Also i excluded the use of the curious feedback schemes with two transistors because they have proved in D.Self book only in amplifiers of small supply. Also i prefer the traditional methods of managing of idle current. I need direct answeers in this, and not long theories and equations. For the last we have enough time after 2 or 3 months to pass our winter with amusement. To the present, i request you not spoil my thread.
With all respects
Fotios
Nordic said:
MikeB said:
Thanks craftsmen for your advices. Also me i supposed the same as you. A confirmation it is a very good thing. I use a BD139 as Vbe multiplier in my monster "Dirty Harry" (i don't remember where it is found in the forum because have passed 5 months from his publication in diyforum but you can view it in my web page by clicking on the button down left) of +/-82V supply. In the plan presented, the MJE340 appears was substituted with a BD139 and the trimer with 5K.
Hey, MikeB BTW i taked a look in your web page in your symasym amplifier and i discovered that our thougts around a circuit architecture are concverging at big part. Nice work man!
My best regards also to my friend Nordic.
You are colleagues to me.
Regards from Greece
Fotios
Confirmation in practice
To be sure, just now i made an experiment with the aid of "Dirty Harry". First i connected accross the C-E pins of BD139 the test leads of a RMS voltmeter. Then i supplied the amplifier with +/-80Vdc and i injected in his input a sinus wave of 1KHz. By increasing the drive signal untill i see clipping in the scope connected in the output, the maximum value that i saw in voltmeter was 26Vrms thus: 26 X 1,414 = 36,77Vp so the maximum voltage swing was 36,77 X 2 = 73,54Vpp. I thing this extremelly condition voltage can sustained from a BD139 of Vce=80Vdc. The conclusion it is that, the Vbe multiplying transistor sees only the half of the full voltage swing allways across his C-E contacts. Thus in my case of the amplifier with supply +/-70Vdc which translated in: 70Vp : 1,414 = 49,5Vrms as much and thus 49,5 : 2 = 24,75Vrms which is translated in 24,75 X 1,414 = 35Vp X 2 = 70Vpp (the voltage level of one of the supply rails). The use of a MJE340 then it is useless. Attention! we speak for voltage peaks and not for power dissipation, thus the same is in effect for a signal of any frequency.
Fotios
To be sure, just now i made an experiment with the aid of "Dirty Harry". First i connected accross the C-E pins of BD139 the test leads of a RMS voltmeter. Then i supplied the amplifier with +/-80Vdc and i injected in his input a sinus wave of 1KHz. By increasing the drive signal untill i see clipping in the scope connected in the output, the maximum value that i saw in voltmeter was 26Vrms thus: 26 X 1,414 = 36,77Vp so the maximum voltage swing was 36,77 X 2 = 73,54Vpp. I thing this extremelly condition voltage can sustained from a BD139 of Vce=80Vdc. The conclusion it is that, the Vbe multiplying transistor sees only the half of the full voltage swing allways across his C-E contacts. Thus in my case of the amplifier with supply +/-70Vdc which translated in: 70Vp : 1,414 = 49,5Vrms as much and thus 49,5 : 2 = 24,75Vrms which is translated in 24,75 X 1,414 = 35Vp X 2 = 70Vpp (the voltage level of one of the supply rails). The use of a MJE340 then it is useless. Attention! we speak for voltage peaks and not for power dissipation, thus the same is in effect for a signal of any frequency.
Fotios
lumanauw said:
I am sure for this. As i said, from experiments it proved that the smaller Vce BD137-BD139 are by far much faster in thermal reaction from the higher Vce MJE340. In Vbe multiplier the task of transistor it is to keep constant his Ic under all thermal conditions as we know; as the junction temperature it is increased the Ic must be going decreased and vice versa. The only problem it is the Vce of the transistor used must be a little bigger than the voltage of one of the two supply rails. For me, i have changed already the dull MJE340 with a nice BD139 (Vce=80Vdc). So, dear lumanauw, i think we must to point out this topic in mr. D.Self. Maybe he can draw for us a graph with the time delay of Ic decrease vs. junction temperature for different types of transistor and another one with the types of these transistor vs. the danger of output devices thermal runaway or anythig else bad result. We may don't forget that a idle current of 20mA per output device it is translated in a voltage (respect to gnd) of 1,74V at the base of each of the two driver transistors, and consequently in a voltage of 0,6V approximatelly (respect to gnd) in the base of output transistors. As you can understand during operation each half of amplified signal from VAS vary between +1,74V and the positive supply level and between -1,74V and the negative supply level at the base of drivers under normal conditions. If the Ic of Vbe multiplier goes increased then also the voltage across his C-E junction goes increased and so on in the base of drivers and what does means that for the preset voltage at the bases of output transistors.
Fotios
Re: Confirmation in practice
Fotios,
No worry, I am not out to spoil anybody's thread, of course not.
But there is something wrong here. If you really measure across the Vbe multiplier C and E, you cannot see 26VRMS! You can see only a DC voltage of a few volts. If this would be more your output stage would explode!
What *may* have happened is that you had your multimeter on AC instead of DC? (It should be on DC). Also, sometimes what happend to me is that I connect the meter as you say but I forgot that the meter itself is also grounded and that the input ground is not floating, so I really was mesuring C to ground. In that case, you would indeed have the 26VAC or something.
Whatever it is, your measurements are incorrect.
Jan Didden
fotios said:
Fotios,
No worry, I am not out to spoil anybody's thread, of course not.
But there is something wrong here. If you really measure across the Vbe multiplier C and E, you cannot see 26VRMS! You can see only a DC voltage of a few volts. If this would be more your output stage would explode!
What *may* have happened is that you had your multimeter on AC instead of DC? (It should be on DC). Also, sometimes what happend to me is that I connect the meter as you say but I forgot that the meter itself is also grounded and that the input ground is not floating, so I really was mesuring C to ground. In that case, you would indeed have the 26VAC or something.
Whatever it is, your measurements are incorrect.
Jan Didden
Hi,
the Vbe multiplier only sees a few volts across it.
single pair of outputs with no intervening drivers would need no more than 1.8Vdc
A two stage EF needs a bit more, upto about 3Vdc.
A three stage EF can use upto 4.5Vdc.
Vertical FETs need yet more.
The AC signal across the Vbe is near zero due to the impedance of the bypass capacitor.
Any transistor of Vce0>12V will do for this Vbe duty.
I agree that speed of response is important.
That was the purpose of that link to the BJT vs FET thread.
They are discussing in detail the need for speed and the disadvantages of using a Vbe, which is too far removed from the junction temperature that it is correcting for.
Have you seen the posting that suggested using a sot23 transistor wired as a diode glued to the collector leg of the output transistor.
Three of the "diodes" located in the output stage could be wired up like a Leach multiplier. I bet those will have a quicker response than any plastic to plastic pair of packages. But maybe not as quick as the internal diodes that ONsemi has started to release.
the Vbe multiplier only sees a few volts across it.
single pair of outputs with no intervening drivers would need no more than 1.8Vdc
A two stage EF needs a bit more, upto about 3Vdc.
A three stage EF can use upto 4.5Vdc.
Vertical FETs need yet more.
The AC signal across the Vbe is near zero due to the impedance of the bypass capacitor.
Any transistor of Vce0>12V will do for this Vbe duty.
I agree that speed of response is important.
That was the purpose of that link to the BJT vs FET thread.
They are discussing in detail the need for speed and the disadvantages of using a Vbe, which is too far removed from the junction temperature that it is correcting for.
Have you seen the posting that suggested using a sot23 transistor wired as a diode glued to the collector leg of the output transistor.
Three of the "diodes" located in the output stage could be wired up like a Leach multiplier. I bet those will have a quicker response than any plastic to plastic pair of packages. But maybe not as quick as the internal diodes that ONsemi has started to release.
I'm bored, may as well be usefull....
from wikipedia
A VBE multiplier voltage source is shown in the image on the left. It works by virtue of the fact that as long as the transistor Q has a high enough current gain (hFE), the base current is negligible, and the output voltage depends only on the transistor's VBE and the ratio of the resistors R1 and R2. Analysis of the circuit is as follows:
Vout(=VCE)=VR1 +VR2 (simple resistor voltage divider)
Since base current of the transistor is negligible, therefore IR1 = IR2 = IBB and so,
Vout=(VCE)=IBB(R1+R2)
but IBB=VBE/R2 (equal current over series resistance rule)
So, Vout(=VCE)+VBE/R2 (R1+R2)
and, Vout=VBE (1+R1/R2)
From the above equation, it follows that the output voltage of this circuit depends only on VBE and the ratio of R1 and R2. The circuit is known as a "VBE multiplier", since the above equation shows that the VBE is multiplied by (1 + R1/R2). This circuit provides a constant output voltage that is set by the ratio of R1 and R2, if VBE is constant. Also, R1 (or R2) can be made variable to compensate for VBE variations due to device tolerance. A VBE multiplier is also known as a rubber diode or a rubber zener.
Uses of VBE multiplier
Because it does not require a ground connection (that is, it is a floating circuit) and gives a predictable and easily adjustable voltage drop, this circuit is frequently used in biasing the class-AB output stages of power amplifiers. R1 (or R2) is varied till the required voltage is achieved. Sometimes R1 and R2 are replaced by a potentiometer for easy adjustment. Since VBE decreases with increasing temperature (thereby reducing the VBE multiplier's output voltage) this circuit also acts to compensate for temperature induced changes of VBE in the output devices. This tends to counteract the effect of reduction in VBE of the output devices and helps prevent thermal runaway of the output stage. This is called bias temperature compensation in the electronics industry at large, but a "bias servo" in the field of audio amplifiers.
from wikipedia
An externally hosted image should be here but it was not working when we last tested it.
A VBE multiplier voltage source is shown in the image on the left. It works by virtue of the fact that as long as the transistor Q has a high enough current gain (hFE), the base current is negligible, and the output voltage depends only on the transistor's VBE and the ratio of the resistors R1 and R2. Analysis of the circuit is as follows:
Vout(=VCE)=VR1 +VR2 (simple resistor voltage divider)
Since base current of the transistor is negligible, therefore IR1 = IR2 = IBB and so,
Vout=(VCE)=IBB(R1+R2)
but IBB=VBE/R2 (equal current over series resistance rule)
So, Vout(=VCE)+VBE/R2 (R1+R2)
and, Vout=VBE (1+R1/R2)
From the above equation, it follows that the output voltage of this circuit depends only on VBE and the ratio of R1 and R2. The circuit is known as a "VBE multiplier", since the above equation shows that the VBE is multiplied by (1 + R1/R2). This circuit provides a constant output voltage that is set by the ratio of R1 and R2, if VBE is constant. Also, R1 (or R2) can be made variable to compensate for VBE variations due to device tolerance. A VBE multiplier is also known as a rubber diode or a rubber zener.
Uses of VBE multiplier
Because it does not require a ground connection (that is, it is a floating circuit) and gives a predictable and easily adjustable voltage drop, this circuit is frequently used in biasing the class-AB output stages of power amplifiers. R1 (or R2) is varied till the required voltage is achieved. Sometimes R1 and R2 are replaced by a potentiometer for easy adjustment. Since VBE decreases with increasing temperature (thereby reducing the VBE multiplier's output voltage) this circuit also acts to compensate for temperature induced changes of VBE in the output devices. This tends to counteract the effect of reduction in VBE of the output devices and helps prevent thermal runaway of the output stage. This is called bias temperature compensation in the electronics industry at large, but a "bias servo" in the field of audio amplifiers.
YOU ARE ABSOLUTELY RIGHT ALL
Sorry for my delayed response but the internet was down from 2 p.m. lockal time because the big fires in forests in all of Greece region.
I am feeling realy idiot because my ignorance and my hasty conclusions. Jan you are right as for the voltmeter measurement. Nordic, MikeB and Andrew your remarks are absolutelly right. The mistake it is my own. To prove this in myself i made a draw which is much better from thougths. And the result from the calculations with its aid proved that i have mistake. Indeed the voltage accross the Vbe multiplier never exceeds the limits of voltage preset. I quote bellow my drawing to understand you the way that of finding my mistake.
From this it appears that the voltage swing it has the same phase always in the C-E junctions of Vbe multiplier, thus the only difference it is the sum of the voltages at each junction. I apologise to all of you for my mistake, i am ashamed realy for this gross mistake. For drawing purposes i comprise a single form of the Vbe multiplier circuit from the CCS to the VAS. I supposed a signal amplified by VAS at +/-60Vpp for clear viewing in the draw.
Thanks to all of you for your exhortation to open my mind.
Regards
Fotios
Sorry for my delayed response but the internet was down from 2 p.m. lockal time because the big fires in forests in all of Greece region.
I am feeling realy idiot because my ignorance and my hasty conclusions. Jan you are right as for the voltmeter measurement. Nordic, MikeB and Andrew your remarks are absolutelly right. The mistake it is my own. To prove this in myself i made a draw which is much better from thougths. And the result from the calculations with its aid proved that i have mistake. Indeed the voltage accross the Vbe multiplier never exceeds the limits of voltage preset. I quote bellow my drawing to understand you the way that of finding my mistake.
An externally hosted image should be here but it was not working when we last tested it.
From this it appears that the voltage swing it has the same phase always in the C-E junctions of Vbe multiplier, thus the only difference it is the sum of the voltages at each junction. I apologise to all of you for my mistake, i am ashamed realy for this gross mistake. For drawing purposes i comprise a single form of the Vbe multiplier circuit from the CCS to the VAS. I supposed a signal amplified by VAS at +/-60Vpp for clear viewing in the draw.
Thanks to all of you for your exhortation to open my mind.
Regards
Fotios
Re: YOU ARE ABSOLUTELY RIGHT ALL
Fotios,
I like your drawing, it gives a very clear indication of all the voltages in the output stage. Thanks for taking the time to create it. This is the spirit of this forum: we ALL learn from the interaction.
I hear about the fires in your country on the radio. Such a bad happeniong! My experience of Greece is limited to Athens and the island of Crete, but I very much regret all the destruction that fire causes. Hope your people get it under control quickly.
Jan Didden
fotios said:
Fotios,
I like your drawing, it gives a very clear indication of all the voltages in the output stage. Thanks for taking the time to create it. This is the spirit of this forum: we ALL learn from the interaction.
I hear about the fires in your country on the radio. Such a bad happeniong! My experience of Greece is limited to Athens and the island of Crete, but I very much regret all the destruction that fire causes. Hope your people get it under control quickly.
Jan Didden
Re: Re: Confirmation in practice
I hope to you forgive me if you perceived a cynicism in my post. I confirm you that i had no aim to reproach you. Some times i am exhortative and as you know already from our previous discussions in other threads i am also a bit nagger. But i don't want to spoil my relations with persons that i respect (by no way). I hope to stay friends. If you are angry with me, please accept my apologises or tell me your address to send you a bouquet in token of friendship.
My best regards
Fotios
Dear Janjanneman said:
I hope to you forgive me if you perceived a cynicism in my post. I confirm you that i had no aim to reproach you. Some times i am exhortative and as you know already from our previous discussions in other threads i am also a bit nagger. But i don't want to spoil my relations with persons that i respect (by no way). I hope to stay friends. If you are angry with me, please accept my apologises or tell me your address to send you a bouquet in token of friendship.
My best regards
Fotios
Nordic said:I'm bored, may as well be usefull....
from wikipedia
A VBE multiplier voltage source is shown in the image on the left. It works by virtue of the fact that as long as the transistor Q has a high enough current gain (hFE), the base current is negligible, and the output voltage depends only on the transistor's VBE and the ratio of the resistors R1 and R2. Analysis of the circuit is as follows:
Vout(=VCE)=VR1 +VR2 (simple resistor voltage divider)
Since base current of the transistor is negligible, therefore IR1 = IR2 = IBB and so,
Vout=(VCE)=IBB(R1+R2)
but IBB=VBE/R2 (equal current over series resistance rule)
So, Vout(=VCE)+VBE/R2 (R1+R2)
and, Vout=VBE (1+R1/R2)
From the above equation, it follows that the output voltage of this circuit depends only on VBE and the ratio of R1 and R2. The circuit is known as a "VBE multiplier", since the above equation shows that the VBE is multiplied by (1 + R1/R2). This circuit provides a constant output voltage that is set by the ratio of R1 and R2, if VBE is constant. Also, R1 (or R2) can be made variable to compensate for VBE variations due to device tolerance. A VBE multiplier is also known as a rubber diode or a rubber zener.
Uses of VBE multiplier
Because it does not require a ground connection (that is, it is a floating circuit) and gives a predictable and easily adjustable voltage drop, this circuit is frequently used in biasing the class-AB output stages of power amplifiers. R1 (or R2) is varied till the required voltage is achieved. Sometimes R1 and R2 are replaced by a potentiometer for easy adjustment. Since VBE decreases with increasing temperature (thereby reducing the VBE multiplier's output voltage) this circuit also acts to compensate for temperature induced changes of VBE in the output devices. This tends to counteract the effect of reduction in VBE of the output devices and helps prevent thermal runaway of the output stage. This is called bias temperature compensation in the electronics industry at large, but a "bias servo" in the field of audio amplifiers.
Congratulations friend Nordic for your detailed explanation also from me
Fotios
Re: Re: YOU ARE ABSOLUTELY RIGHT ALL
Regards
Fotios
Indeed dear Jan it is an unprecedented destruction in lifes (about 60 deaths untill now!) in natural environment in estates (about 200 house was burned) and in economics at the end. I am living in a town at north eastern Greece where we don't have problems with fires. Usually the fires break out in southern Greece. But this moment all of us in Greece we relieve our wounded people with any way. BTW my mother she comes from Crete.janneman said:
Regards
Fotios
Hi Fotios,
We meet again.
The time constant of the collector and die is probably what makes the difference. I normally use small TO-92 transistors. A small hole is drilled into the heat sink that the transistor just fits into. Some thermal compound helps minimize air pockets.
Interestingly Denon, on some models, installs a pair of diodes between the collector and emitter leads of the driver transistors to sense their temperature. This seems to be effective as well. The thin Zetex parts in their modified TO-92 case may fit in between very nicely.
-Chris
We meet again.
The time constant of the collector and die is probably what makes the difference. I normally use small TO-92 transistors. A small hole is drilled into the heat sink that the transistor just fits into. Some thermal compound helps minimize air pockets.
Interestingly Denon, on some models, installs a pair of diodes between the collector and emitter leads of the driver transistors to sense their temperature. This seems to be effective as well. The thin Zetex parts in their modified TO-92 case may fit in between very nicely.
-Chris