"Naked base-emitter junctions"?
In that case, I guess I'd better cover up my collector!
I must disagree with Waveboun's apparent fear of exposure in post #111.
In the circuit of post #7, Wavebourn appears to be referring to the base-emitters of drivers Q7 and Q8 as being "naked", and furthermore characterizing this as being a bad thing. He is apparently unfamiliar with the great Bart Locanthi's "T" circuit which used this approach as well. Welbourn seems to be distressed by the absence of a base-emitter resistor on Q7 and on Q8. These base-emitters are not "naked", either in the a.c. sense or the d.c. sense. Indeed the circuit would work just as well if R11 was replaced by a current source with the same standing current as R11. The only function of R11 is to pull enough turn-off current so that Q9 (or Q10) can be turned off without Q7 (or Q8) having to turn off.
By always having a certain amount of current flowing in the emitter of Q7, the dynamic impedance at the base of Q9 is quite low, and the base-emitter junction of Q9 is therefore not "naked". To understand what the a.c. impedance at the base of Q9 is, particularly at high audio frequencies, one can start at the collector of the VAS, Q4, and recognize that the Miller compensation local feedback provided by C3 keeps the impednace at this node relatively low (less than 20 kohm at frequencies above 10 kHz, even if dashed resistors R15 and R16 are not in place).
The impedance at the emitter of Q5 will be less than this by a factor of about beta of Q5 (say 50), or on the order of 400 ohms. The impedance at the emitter of Q7 will be lower still by another factor of beta plus 1/gm of Q7, or about 8 ohms plus 1/gm of Q7. If Q7's NET emitter current never falls below 1 mA, the impedance seen looking back from the base of Q9 is on the order of only about 30 ohms.
Note that the normal standing current through R11 is about 25 mA, thus a full 24 mA is still available to pull current out of the base of Q9 to shut it of quickly, when needed. The base-emitter junction of Q9 is very well-controlled, and is in no way "naked".
Cheers,
Bob
In that case, I guess I'd better cover up my collector!
I must disagree with Waveboun's apparent fear of exposure in post #111.
In the circuit of post #7, Wavebourn appears to be referring to the base-emitters of drivers Q7 and Q8 as being "naked", and furthermore characterizing this as being a bad thing. He is apparently unfamiliar with the great Bart Locanthi's "T" circuit which used this approach as well. Welbourn seems to be distressed by the absence of a base-emitter resistor on Q7 and on Q8. These base-emitters are not "naked", either in the a.c. sense or the d.c. sense. Indeed the circuit would work just as well if R11 was replaced by a current source with the same standing current as R11. The only function of R11 is to pull enough turn-off current so that Q9 (or Q10) can be turned off without Q7 (or Q8) having to turn off.
By always having a certain amount of current flowing in the emitter of Q7, the dynamic impedance at the base of Q9 is quite low, and the base-emitter junction of Q9 is therefore not "naked". To understand what the a.c. impedance at the base of Q9 is, particularly at high audio frequencies, one can start at the collector of the VAS, Q4, and recognize that the Miller compensation local feedback provided by C3 keeps the impednace at this node relatively low (less than 20 kohm at frequencies above 10 kHz, even if dashed resistors R15 and R16 are not in place).
The impedance at the emitter of Q5 will be less than this by a factor of about beta of Q5 (say 50), or on the order of 400 ohms. The impedance at the emitter of Q7 will be lower still by another factor of beta plus 1/gm of Q7, or about 8 ohms plus 1/gm of Q7. If Q7's NET emitter current never falls below 1 mA, the impedance seen looking back from the base of Q9 is on the order of only about 30 ohms.
Note that the normal standing current through R11 is about 25 mA, thus a full 24 mA is still available to pull current out of the base of Q9 to shut it of quickly, when needed. The base-emitter junction of Q9 is very well-controlled, and is in no way "naked".
Cheers,
Bob
Mike, you are right on target
I have struggled a lot over the years with exactly what you are talking about. I have resorted to building custom differential amplifiers and the like to measure noise between different ground points, or noise at a signal node uncorrupted by ground loops. Even that has not always been completely successful. It is a struggle, to say the least.
Bob
I have struggled a lot over the years with exactly what you are talking about. I have resorted to building custom differential amplifiers and the like to measure noise between different ground points, or noise at a signal node uncorrupted by ground loops. Even that has not always been completely successful. It is a struggle, to say the least.
Bob
Dear Bob;
all impedances you mentioned are dynamic in nature, non-linear, and depend on currents and voltages on transistors. Also they significantly depend on a frequency. What happens in case of capacitive load? Suppose, positive swing charged that capacitance, so during a negative swing this charge will be applied between bases and emitters of your transistors as a short peak of high current turning class AB amp into the B or even C amp for the short period of time (deep NFB will help to make it sharper). The higher is frequency, the higher is impact. Not crossover distortions per se are guilty in so called "transistor phenomenon", but crossover distortions with capacitive load. One my design does not have any bias in output followers, but transistor sound is absent.
Speaking of RFI, such topology is goof for rectifications of high frequency noises on emitter - base junctions of output transistors and to amplify the product of rectification causing intermodulation between RF signal and its rectified part. That;s why it always sound very distorted despite the original AM signal is very good.
PS: However, you may use this approach as well. On resistive loads.
all impedances you mentioned are dynamic in nature, non-linear, and depend on currents and voltages on transistors. Also they significantly depend on a frequency. What happens in case of capacitive load? Suppose, positive swing charged that capacitance, so during a negative swing this charge will be applied between bases and emitters of your transistors as a short peak of high current turning class AB amp into the B or even C amp for the short period of time (deep NFB will help to make it sharper). The higher is frequency, the higher is impact. Not crossover distortions per se are guilty in so called "transistor phenomenon", but crossover distortions with capacitive load. One my design does not have any bias in output followers, but transistor sound is absent.
Speaking of RFI, such topology is goof for rectifications of high frequency noises on emitter - base junctions of output transistors and to amplify the product of rectification causing intermodulation between RF signal and its rectified part. That;s why it always sound very distorted despite the original AM signal is very good.
PS: However, you may use this approach as well. On resistive loads.
SPICE is nice
Wavebourn, of course those impednaces are dynamic. That doesn't mean that your postulation is right. I suggest you run some SPICE simulations. My amplifiers have no problems with capacitive loads, including in regard to either static or dynamic crossover distortion. I hate crossover distortion.
Cheers,
Bob
Wavebourn, of course those impednaces are dynamic. That doesn't mean that your postulation is right. I suggest you run some SPICE simulations. My amplifiers have no problems with capacitive loads, including in regard to either static or dynamic crossover distortion. I hate crossover distortion.
Cheers,
Bob
I'm among those who is skeptical about boutique/magical component accessories/exotic cables.
But once (1-2year ago?) I wanted to know what is the difference between cheap and expensive female RCA connector (for power amp). This thing is small, maybe about 2cm x 2cm, so I think it cannot do any difference.
But there is a difference seen on scope. Various priced female RCA's generates different noise aritifact on scope. I don't understand this, how come a 2cm x 2cm brass make a difference in scope.
The interesting part is that the more expensive ones are noisier ones.
I think it is the construction of this RCA connector that makes it an effective "noise collector".
lumanauw said:
Lumanauw,
I'm curious as to what the test setup was and what kind of difference you were seeing on the scope. The signal used, source electronics, load circuitry, signal levels, what the difference looked like. It's fine if it's been to long ago to remember.
There is most likely a logical explaination. In all cases there is, it's just depends on how hard you want to look and the resolution of your test setup.
Years ago I built a static discharge detector to sniff for the field emitted by a arc occuring to the chips in a IC test handler. Playing around with it I found I could detect, among other things, the signal emitted from the contacts carrying the test signals from the computer to the device. If the contacts had issues they were quite noisy. My take was that any energy that was being emitted was not making it to the chips being tested.
Regards, Mike.
jez said:MikeB,
Some very interesting points you have raised there
The huge RF spikes from the missile early warning system which I metioned earlier were only visible on a battery powered and therefore floating, handheld oscilloscope....hmm.
Interesting,
Grounding and return paths are an interesting topic. In this case I would think that the signal to the early warning system being referenced to the earth ground and the floating scope having a small capacitance to ground, The path for which it is hard to control, had a lot to do with it. It's all about the return path and whether or not that path is part of the gain stages.
Interesting point to ponder is that the signals in our audio system are not referenced to the earths ground. This of course has a bearing on where you locate the safety ground since it does bring the earth into the picture. I myself have danced with the grim reaper for years and not used any safety ground, but that's a whole other discussion. I just don't see the point in bringing its effects into the sonic picture.
Disclaimer: I would never suggest anyone follow my dangerous approach to electronic construction techniques. Not using a safety ground will probably kill you. I've just been really lucky.
Regards, Mike.
Safety Grounds
Mike,
Interesting that you mention the safety ground. Of course, a lot of equipment these days only has a two-wire power cord anyway. Nevertheless, for the equipment with a safety ground included in the line cord, I agree with you that it is a potential source of unwanted ground currents and possible sonic degradation.
Here is what I did on a vacuum tube power amplifier that I recently gutted and re-built. Let me know what you think. I decided that I wanted some resistance in the safety ground path so as to break ground currents created by ground potential differences in the range of a few hundred millivolts and less. However, I did not want to forego the safety afforded by the safety ground. In other words, if, say, part of the primary of the power transformer developed a short to the metal chassis, I did not want the chassis carrying lots of line voltage.
I implemented a parallel network of a 4.7 ohm resistor and two large power rectifier diodes, connected oppositely. I connected the safety ground from the line cord to the chassis through this network. For small ground differentials, the 4.7 ohms pretty much breaks the ground loop and greatly reduces circulating ground currents. But if there is a fault, the power diodes kick in and blow a fuse or trip a GFCI. I know this probably would not get UL approval, but I think it does the job. Comments?
Bob
Mike,
Interesting that you mention the safety ground. Of course, a lot of equipment these days only has a two-wire power cord anyway. Nevertheless, for the equipment with a safety ground included in the line cord, I agree with you that it is a potential source of unwanted ground currents and possible sonic degradation.
Here is what I did on a vacuum tube power amplifier that I recently gutted and re-built. Let me know what you think. I decided that I wanted some resistance in the safety ground path so as to break ground currents created by ground potential differences in the range of a few hundred millivolts and less. However, I did not want to forego the safety afforded by the safety ground. In other words, if, say, part of the primary of the power transformer developed a short to the metal chassis, I did not want the chassis carrying lots of line voltage.
I implemented a parallel network of a 4.7 ohm resistor and two large power rectifier diodes, connected oppositely. I connected the safety ground from the line cord to the chassis through this network. For small ground differentials, the 4.7 ohms pretty much breaks the ground loop and greatly reduces circulating ground currents. But if there is a fault, the power diodes kick in and blow a fuse or trip a GFCI. I know this probably would not get UL approval, but I think it does the job. Comments?
Bob
Grounding
A good discussion can be found in the Rane site, "Grounding and shielding audio devices" for example.
They advocate direct signal and chassis (protection) ground connection.
Any contradicting experience regarding this?
Rodolfo
A good discussion can be found in the Rane site, "Grounding and shielding audio devices" for example.
They advocate direct signal and chassis (protection) ground connection.
Any contradicting experience regarding this?
Rodolfo
Re: Safety Grounds
Are you sure that in case of short circuit diodes won't protect the fuse?
I have the same network (two 1n5408 and 11R) but between chassis/heatsink and circuit's power ground. Metal parts, enclosures and so on should be in my opinion earthed as dura lex sed lex.
Bob Cordell said:
Are you sure that in case of short circuit diodes won't protect the fuse?
I have the same network (two 1n5408 and 11R) but between chassis/heatsink and circuit's power ground. Metal parts, enclosures and so on should be in my opinion earthed as dura lex sed lex.