Hi,
circuit B is a cheapskate version that has poor control of the constant current function on the right hand half.
D.Self did a similar thing and then wrote a chapter about his failure to equalise the slew rate in the positive and negative directions. He never published the result of doing the CCSs properly and may have avoided the need to alter his circuit as he ended up doing, although it did give him a chance to show dependance on a cap from base of control transistor to Vas collector, he called the effect feedthrough.
Many others copy circuit B to save a resistor. Can't understand why when they cost little board space and parts of a cent.
Circuit A gives equal priority to both CCSs and since the VAS usually has more Ic it deserves good control.
BTW. most designers recommend degenerating the emitters. You also need a collector load to drive the VAS.
circuit B is a cheapskate version that has poor control of the constant current function on the right hand half.
D.Self did a similar thing and then wrote a chapter about his failure to equalise the slew rate in the positive and negative directions. He never published the result of doing the CCSs properly and may have avoided the need to alter his circuit as he ended up doing, although it did give him a chance to show dependance on a cap from base of control transistor to Vas collector, he called the effect feedthrough.
Many others copy circuit B to save a resistor. Can't understand why when they cost little board space and parts of a cent.
Circuit A gives equal priority to both CCSs and since the VAS usually has more Ic it deserves good control.
BTW. most designers recommend degenerating the emitters. You also need a collector load to drive the VAS.
Circuit B is to be preferred.
The single resistor and LED produce the reference voltage, the independent output currents are trimmed by their respective emitter resistors.
The base resistor for the right hand current source prevents the reference voltage from collapsing on a negative clip (recovers from clipping cleaner).
The single resistor and LED produce the reference voltage, the independent output currents are trimmed by their respective emitter resistors.
The base resistor for the right hand current source prevents the reference voltage from collapsing on a negative clip (recovers from clipping cleaner).
They're all over the map in terms of parameters.
They modulate with light... admittedly not a big issue if you are a) in the dark or b) always in a closed chassis.
Spectral ran into this problem big time when they used them in their DMC-10 preamp, which was very pretty inside. They supplied it to their showrooms with a plexi cover to show off the guts. It took a while to figure out where the 60cycle hum was coming from... since none of the units back at the factory seemed to have that problem!
Ymmv.
_-_-bear
They modulate with light... admittedly not a big issue if you are a) in the dark or b) always in a closed chassis.
Spectral ran into this problem big time when they used them in their DMC-10 preamp, which was very pretty inside. They supplied it to their showrooms with a plexi cover to show off the guts. It took a while to figure out where the 60cycle hum was coming from... since none of the units back at the factory seemed to have that problem!
Ymmv.
_-_-bear
It depends on the circuit. For audio you're probably OK.
IIRC, Bob Pease of National Semiconductor wrote a heads up note in Electronic Design a few years back.
http://www.national.com/rap/
Dear Bob:
Roger Sheker's letter about the photosensitivity of input-protection diodes, which appeared in your recent Mailbox column (Electronic Design, Aug. 19, p. 66), reminded me of a trick that I started using many years ago when working on low-level circuits. In the days before SMT components, I always found it very frustrating that signal diodes, including low-leakage diodes, were inevitably in clear glass packages. (Not all are. Fairchild made many low-leakage diodes in black-painted packages. But when the paint gets scratched, it's almost as bad. /rap)
As such, the photocurrents often far exceeded the actual leakage of the diode, particularly when the board was out on the bench for calibration, etc. Putting black heatshrink tubing over the diode packages improved the problem, but was a pain in production. Then I realized that I could use an inexpensive small-signal transistor (2N3904, etc.), in a completely opaque TO-92 package, as a diode. The C-B junction is fairly low in leakage, has a high breakdown voltage, and the transistor may actually cost less than special low-leakage diodes of lesser performance.
But if you can tolerate the modest reverse-breakdown voltage (such as in back-to-back diode applications), the B-E junction makes an exceptionally low-leakage diode. Although transistor data sheets caution against breaking down the B-E junction, I have found that in protection or clamp applications this isn't particularly harmful if the current is limited to reasonable values. Hence, the B-E junction can make a useful low-leakage bidirectional input clamp, providing a forward drop in one direction, and a clamp of about 6 to 9 V in the other—without any problems due to photosensitivity.
By the way, amazingly, some small metal-can transistors DO show photosensitivity because the ceramic/glass header and epoxy seal at the bottom are translucent. Of course, today with SMT technology, you can finally buy signal diodes in opaque packages! (Yeah, but many of those junctions are still LOUSY in terms of conduction and leakage. /rap)
Eric Kinast
via e-mail
~snip~
The main subject I wanted to comment on was your mentioning of typical signal diodes' sensitivity to light. I had an amplifier in which I couldn't tame the 60-Hz noise. It had vent holes in the top lid, and an incandescent light was shining down on it in its usual place of use. (Incandescent lamps tend to be bright at dc, with only a small 120-Hz component of noise. That's because the large thermal mass of the filament doesn't let the light intensity fluctuate much. It's the fluorescent lamps that make a lot of 120-Hz hum! And at 360 Hz. Not much at 60 Hz. /rap)
When I moved it to the test bench, removed the cover, and focused my shop high-intensity light on it, it measured very high noise. (I haven't studied halogen lamps' emissions, nor noises, very much. /rap) I was lucky as the light generated enough heat to cause other problems. When I moved the light away from the unit, the noise stopped. This was when I remembered reading of this phenomenon in your article. Taking an ultra-black magic marker to the input limiting diodes solved the problem. Thanks again and please publish more practical engineering stuff. I love it!
Roger Sheker
via e-mail
Isn't it funny how removing a noisy source correlates well with the noise going down a lot?—RAP
IIRC, Bob Pease of National Semiconductor wrote a heads up note in Electronic Design a few years back.
http://www.national.com/rap/
Dear Bob:
Roger Sheker's letter about the photosensitivity of input-protection diodes, which appeared in your recent Mailbox column (Electronic Design, Aug. 19, p. 66), reminded me of a trick that I started using many years ago when working on low-level circuits. In the days before SMT components, I always found it very frustrating that signal diodes, including low-leakage diodes, were inevitably in clear glass packages. (Not all are. Fairchild made many low-leakage diodes in black-painted packages. But when the paint gets scratched, it's almost as bad. /rap)
As such, the photocurrents often far exceeded the actual leakage of the diode, particularly when the board was out on the bench for calibration, etc. Putting black heatshrink tubing over the diode packages improved the problem, but was a pain in production. Then I realized that I could use an inexpensive small-signal transistor (2N3904, etc.), in a completely opaque TO-92 package, as a diode. The C-B junction is fairly low in leakage, has a high breakdown voltage, and the transistor may actually cost less than special low-leakage diodes of lesser performance.
But if you can tolerate the modest reverse-breakdown voltage (such as in back-to-back diode applications), the B-E junction makes an exceptionally low-leakage diode. Although transistor data sheets caution against breaking down the B-E junction, I have found that in protection or clamp applications this isn't particularly harmful if the current is limited to reasonable values. Hence, the B-E junction can make a useful low-leakage bidirectional input clamp, providing a forward drop in one direction, and a clamp of about 6 to 9 V in the other—without any problems due to photosensitivity.
By the way, amazingly, some small metal-can transistors DO show photosensitivity because the ceramic/glass header and epoxy seal at the bottom are translucent. Of course, today with SMT technology, you can finally buy signal diodes in opaque packages! (Yeah, but many of those junctions are still LOUSY in terms of conduction and leakage. /rap)
Eric Kinast
via e-mail
~snip~
The main subject I wanted to comment on was your mentioning of typical signal diodes' sensitivity to light. I had an amplifier in which I couldn't tame the 60-Hz noise. It had vent holes in the top lid, and an incandescent light was shining down on it in its usual place of use. (Incandescent lamps tend to be bright at dc, with only a small 120-Hz component of noise. That's because the large thermal mass of the filament doesn't let the light intensity fluctuate much. It's the fluorescent lamps that make a lot of 120-Hz hum! And at 360 Hz. Not much at 60 Hz. /rap)
When I moved it to the test bench, removed the cover, and focused my shop high-intensity light on it, it measured very high noise. (I haven't studied halogen lamps' emissions, nor noises, very much. /rap) I was lucky as the light generated enough heat to cause other problems. When I moved the light away from the unit, the noise stopped. This was when I remembered reading of this phenomenon in your article. Taking an ultra-black magic marker to the input limiting diodes solved the problem. Thanks again and please publish more practical engineering stuff. I love it!
Roger Sheker
via e-mail
Isn't it funny how removing a noisy source correlates well with the noise going down a lot?—RAP
Hi,
I am sure glad my LEDs are not painted black, I would lie awake at night worrying that I might have scratched them.
I think we are really referring to other diodes being sugject to light modulation, not LEDs.
I have seen figures for LEDs of 50khrs and 100khrs. before performance drops off significantly.
I am sure glad my LEDs are not painted black, I would lie awake at night worrying that I might have scratched them.
I think we are really referring to other diodes being sugject to light modulation, not LEDs.
I have seen figures for LEDs of 50khrs and 100khrs. before performance drops off significantly.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.