28-05-2011, 10:56 PM
Hi Al,
Thanks for the schematics - most interesting. Very neatly drawn, and a level of detail that can only come from someone who has spent a lifetime working with professional gear from the likes of HP/Tek...
You invited constructive comments, so here's a few minor points that leapt out at me:
I'd be inclined to sprinkle a few diodes around - across the regulators (between in and out) to prevent damage to them when an external DC voltage is applied to the unit when it is unpowered. I know that you're unlikely to do this, but it's cheap insurance. By way of an example, a friend of mine was using a certain (well-regarded) Farnell bench PSU to charge some batteries. The mains failed (he lives out in the sticks), and the batteries damaged the transistor that drove the 2N3055 pass transistor. When the mains came back on, full unregulated input (circa 40 volts) was applied to the batteries. Which made a horrible mess :@
I would also put a diode across the output terminals to protect against external reverse-polarity. This source of reverse-polarity can come from unexpected sources - especially with dual rail units like yours. For example, an external DUT that has a short between +ve and -ve. Bob Pease wrote extensively (and very entertainingly, as usual) some years back. Incidentally, Bob's book "Troubleshooting Analog (sic) Circuits" is well worth a read, and seems to be freely available on Google Books...
Turning to the discrete regulator circuits (although the diode comments apply to these as well), it's really nice to see some "proper" circuitry - I think that discrete design is a dying art. I must admit that I often use discrete transistors in preference to ICs because I enjoy the process of design and debugging. A couple of small things I noticed:
1. Nice to see a current source supplying the operating current for the pass transistors - which reduces the ripple, of course. I note that the current source is biased with a 5V6 Zener, which increases the "drop-out voltage" needlessly. If you notice 100Hz ripple abruptly appearing at a certain output current, this might well be a source (as the ripple voltage "trough" dips below Vout + Vdropout). Of course, there is a million ways to build a current source, but perhaps the easiest (and prettiest) alternative might be a red LED instead of the 5V6 Zener. Then, you'd have 1V across R4 instead of 5V, so drop it to ~200R to maintain the current. Incidentally, I speak from direct experience here
2. I was surprised at C2 at the junction of the Darlington pair - not seen that elsewhere... What does that do - reduce ripple, improve transient response? I would expect to see a resistor (circa 100 ohms perhaps) between the base and emitter of Q3, and I'd normally add ~10R between the emitter of Q2 and Q3 base as a "stopper" to prevent HF oscillation in Q3. There is a risk - probably academic - that C2 could cause Q3 a problem in the event of a direct s/c across the output, as it is able to dump (relatively) significant energy into the BE junction of Q3.
I keep looking at it, thinking "I could add a half-wave rectifier to generate a -ve rail, then use 3 diodes to create a -1.8V reference, then return the diode string (Q5 emitter) to that, then the output control might get down to zero volts. But even if that's electronically feasible, it's probably not very easy to add to the completed assembly... Perhaps it's just me; ever since I made my first PSU - based on an LM317T when I was 14 (still have it!) - I've always endeavoured to build them to include 0V :D
It's great to see so much activity in the DIY section of this forum, although I'm getting frustrated now! In my last house I used to build loads of stuff, but here, the workshop isn't really set up for that. One day I'll have somewhere better than the damp cellar, but I will try to produce and document some small projects sometime soon. All my recent projects have been DIY and woodwork stuff, although I've done a lot of radio repairs and restorations over the last year.
And sorry again for the delay in commenting, but wanted to wait to get home so that I could devote some "quality time" to the task
All the best,
Mark
Thanks for the schematics - most interesting. Very neatly drawn, and a level of detail that can only come from someone who has spent a lifetime working with professional gear from the likes of HP/Tek...
You invited constructive comments, so here's a few minor points that leapt out at me:
I'd be inclined to sprinkle a few diodes around - across the regulators (between in and out) to prevent damage to them when an external DC voltage is applied to the unit when it is unpowered. I know that you're unlikely to do this, but it's cheap insurance. By way of an example, a friend of mine was using a certain (well-regarded) Farnell bench PSU to charge some batteries. The mains failed (he lives out in the sticks), and the batteries damaged the transistor that drove the 2N3055 pass transistor. When the mains came back on, full unregulated input (circa 40 volts) was applied to the batteries. Which made a horrible mess :@
I would also put a diode across the output terminals to protect against external reverse-polarity. This source of reverse-polarity can come from unexpected sources - especially with dual rail units like yours. For example, an external DUT that has a short between +ve and -ve. Bob Pease wrote extensively (and very entertainingly, as usual) some years back. Incidentally, Bob's book "Troubleshooting Analog (sic) Circuits" is well worth a read, and seems to be freely available on Google Books...
Turning to the discrete regulator circuits (although the diode comments apply to these as well), it's really nice to see some "proper" circuitry - I think that discrete design is a dying art. I must admit that I often use discrete transistors in preference to ICs because I enjoy the process of design and debugging. A couple of small things I noticed:
1. Nice to see a current source supplying the operating current for the pass transistors - which reduces the ripple, of course. I note that the current source is biased with a 5V6 Zener, which increases the "drop-out voltage" needlessly. If you notice 100Hz ripple abruptly appearing at a certain output current, this might well be a source (as the ripple voltage "trough" dips below Vout + Vdropout). Of course, there is a million ways to build a current source, but perhaps the easiest (and prettiest) alternative might be a red LED instead of the 5V6 Zener. Then, you'd have 1V across R4 instead of 5V, so drop it to ~200R to maintain the current. Incidentally, I speak from direct experience here
2. I was surprised at C2 at the junction of the Darlington pair - not seen that elsewhere... What does that do - reduce ripple, improve transient response? I would expect to see a resistor (circa 100 ohms perhaps) between the base and emitter of Q3, and I'd normally add ~10R between the emitter of Q2 and Q3 base as a "stopper" to prevent HF oscillation in Q3. There is a risk - probably academic - that C2 could cause Q3 a problem in the event of a direct s/c across the output, as it is able to dump (relatively) significant energy into the BE junction of Q3.
I keep looking at it, thinking "I could add a half-wave rectifier to generate a -ve rail, then use 3 diodes to create a -1.8V reference, then return the diode string (Q5 emitter) to that, then the output control might get down to zero volts. But even if that's electronically feasible, it's probably not very easy to add to the completed assembly... Perhaps it's just me; ever since I made my first PSU - based on an LM317T when I was 14 (still have it!) - I've always endeavoured to build them to include 0V :D
It's great to see so much activity in the DIY section of this forum, although I'm getting frustrated now! In my last house I used to build loads of stuff, but here, the workshop isn't really set up for that. One day I'll have somewhere better than the damp cellar, but I will try to produce and document some small projects sometime soon. All my recent projects have been DIY and woodwork stuff, although I've done a lot of radio repairs and restorations over the last year.
And sorry again for the delay in commenting, but wanted to wait to get home so that I could devote some "quality time" to the task
All the best,
Mark
