A lot of DIYers receive great comfort when their "POS_OK" and "NEG_OK" LEDs (on the power supply PCB) glow brightly. It's a quick bit of reassurance that something is alive or partially alive. Am I even plugged into the wall socket? Yep, there is a glow.
Me, I'd put the LEDs downstream of the rail fuses. If fuse is popped, supply_OK LED is dead.
Me, I'd put the LEDs downstream of the rail fuses. If fuse is popped, supply_OK LED is dead.
Keeping it simple has two benefits for my needs. First, it's a great skill building exercise (learning to walk before you run) for me. And, the author (miro1360) of my intended project claims a very simple power supply will produce exceptional results. I'm building modular power supplies in KiCAD so they can be easily swapped out as my skill and ability progress.
Regards,
Dan
Regards,
Dan
That's why do this, a lot. helps not have a probe slip/fizz moment, esp. inside our 240Vac -powered equipment. 'remember it's live, you dummy...so you remain so...')
ETA: lesson recently learned - fixing a CD player for a friend : the problem came down to a cracked track (Vregs fixed to the case for heatsinking, the pcb could /just/ flex in transit - a hair. No - the crack wasn't at the Vreg pin solder connection, but about 1/2" away, for other constraint reasons.
If I'd had a small LED - at the pin supplied, not on the vreg.. flexing the pcb gently would have revealed the problem and trace the fix in about 2mins flat.
Might add that to miy diagnosis toolkit - a couple of pairs of cheap LEDs, antiparallel +4k7, with a few inches of wire to those nice tiny miniclips. Easy to apply to the VDD test pin in question, bright, and polarity-agnostic in use.
So much easier than trying to use a dvm, while wishing for a 3rd & 4th hand to manipulate ( -and still prone to error becasue the DVM input impedance is huge and easily mis-led.) Flash/flash go/no go, would be so much faster! next time...
Last edited:
I'm curious, if I do it right ...
Referring standard circuit from the first post .
From PDF adjust terminal current for TI 317 in range 0.05-0.1mA
so in overall R1 + R2 , can be increased to pass let say 0.5mA , and for 12V out let say it can be 2K + 16K =18K , 0.6mA
for 10uF and 2K , F would be around 8Hz, and startup time around 20 mS.
for 60 Hz, it would be an additional -20dB
Referring standard circuit from the first post .
From PDF adjust terminal current for TI 317 in range 0.05-0.1mA
so in overall R1 + R2 , can be increased to pass let say 0.5mA , and for 12V out let say it can be 2K + 16K =18K , 0.6mA
for 10uF and 2K , F would be around 8Hz, and startup time around 20 mS.
for 60 Hz, it would be an additional -20dB
Most engineers take a very conservative view of line 3 in Table 7.3 of the LM317 (datasheet) . Most of them ensure that R1 and R2 in their circuit design are low enough values, that the "min" value is met even when the load is disconnected from the regulator. That way the regulator operates within recommended conditions even with no load. So if a lazy repair technician measured the regulator output with load disconnected, s/he does not get a misleading or erroneous result because the regulator is operating outside Recommended conditions.
You'll get better transient response if you design the feedback network to draw the minimum load current required for stability too. For the LM317 that means the resistor from OUT to ADJ needs to be 120 Ω (or 121 Ω if you follow E96/192). For the LM317L you can increase that resistor to 240 Ω as the L version (100 mA, max) only requires 5 mA to regulate.
If you want lower quiescent current, go with a more modern regulator. The TPS7Axxx series from TI is excellent.
Tom
If you want lower quiescent current, go with a more modern regulator. The TPS7Axxx series from TI is excellent.
Tom
it clear enough , just some self education :
consider it only for feedback, and min load are separate and not counted,
does it make sense to decrease Freq. , as i see some design adding 200ohm 200uF, or not, as in overall noise floor basically achieved in few KHz spectrum ?
consider it only for feedback, and min load are separate and not counted,
does it make sense to decrease Freq. , as i see some design adding 200ohm 200uF, or not, as in overall noise floor basically achieved in few KHz spectrum ?
The title of the thread says LM317, though. This is from the TI data sheet:
The 3.5 mA is the "typical" number. 10 mA is worst case. If you want a robust design you design for the worst case.
You're correct that other flavours of the LM317, such as the TI LM117 and the ST LM217, require less current to regulate. 5-10 mA is still a lot to burn for "nothing". Something like the TPS7A470x runs on ~700 µA. And I do get that the LMx17 and the TPS7A470x is not an apples-to-apples comparison.
Tom
The 3.5 mA is the "typical" number. 10 mA is worst case. If you want a robust design you design for the worst case.
You're correct that other flavours of the LM317, such as the TI LM117 and the ST LM217, require less current to regulate. 5-10 mA is still a lot to burn for "nothing". Something like the TPS7A470x runs on ~700 µA. And I do get that the LMx17 and the TPS7A470x is not an apples-to-apples comparison.
Tom
..long ago I tested this, and found it was worthwhile burning up to 20-25mA for low outputZ if the device powered didn't demand anything like as much - obvs only useful for low-voltage supplies and low input-out voltage differential (3v min, set your own max governed by your tolerances for mains Vac variation on the transformer secondary, your tolerance for wasting energy, & heatsink space requirement)
The LM317 is a BJT pass device, and not that much internal gain... waste even more current and the notional Z-out benefit is lost to practical implementation issues for the most part. So - just place the output bypass cap right at the load served in preference (always!)- the LM317 Cealocanth does not require it at the regulator for stability.
The LM317 is a BJT pass device, and not that much internal gain... waste even more current and the notional Z-out benefit is lost to practical implementation issues for the most part. So - just place the output bypass cap right at the load served in preference (always!)- the LM317 Cealocanth does not require it at the regulator for stability.
Yes, correct, but that's all folks know about the generic part and not much about it's derivatives.
You know that LM117,217,237 are tested and selected for improved performance.
TPS7A4701RGWT is a wonderful part @$9.45 CDN, but a solution for DIYAudio with a VQFN-20 package?
If you need it, can justify it, handle using it, then go for it. No DIYAudio bean counter to justify your choice in parts 🙂
More word salad,
Coelacanths (order Coelacanthiformes) are an ancient group of lobe-finned fish (Sarcopterygii) in the class Actinistia
The species was believed to have gone extinct over 65 million years ago but was rediscovered off the coast of South Africa in 1938.
More knowledge 🙂
Last edited:
sorry about salad, trying multitasking,
my means about PSRR and how C2 affect it,
does it make sense increase C2; increase R1 as well, to increase PSRR below 100 Hz and above 10K (or speed of 317 wasn't sufficient for).
especially as a filter for switching PS.
my means about PSRR and how C2 affect it,
does it make sense increase C2; increase R1 as well, to increase PSRR below 100 Hz and above 10K (or speed of 317 wasn't sufficient for).
especially as a filter for switching PS.
Ever heard of solder paste stencils and toaster reflow ovens? 🙂 I've hand-soldered those. It's not super fun, but with a little patience it can certainly be done.
Increasing C2 makes sense. I wouldn't go crazy with it, though. Increasing R1 won't do anything for filtering. Within the regulation bandwidth of the LM317 the AC voltage should be close to 0 V across R1.
Tom
@tomchr
Yup, in 1985 I was servicing the smt line at Motorola com factory in Toronto. They also had a ceramic substrate line as well. George(tubelab) remembers because ironically he was doing the same thing in the Florida plant.
One nice summer day, while the production line was on lunch break, a green industrial Eng, decided to run a FR4 assembly through the ceramic line. It smoked out the whole factory, fire department etc, we got the rest of the day off while the air cleared. I still remember the guys name, Neil Bowman. 🙂
Another industrial eng, Eric D. was one of the 3 founding members of the smt centre, (smtc.com) now a multi-million $ operation. I serviced their excelon pick and place machine on the side in their startup industrial unit. Fun times.
Rick
Yup, in 1985 I was servicing the smt line at Motorola com factory in Toronto. They also had a ceramic substrate line as well. George(tubelab) remembers because ironically he was doing the same thing in the Florida plant.
One nice summer day, while the production line was on lunch break, a green industrial Eng, decided to run a FR4 assembly through the ceramic line. It smoked out the whole factory, fire department etc, we got the rest of the day off while the air cleared. I still remember the guys name, Neil Bowman. 🙂
Another industrial eng, Eric D. was one of the 3 founding members of the smt centre, (smtc.com) now a multi-million $ operation. I serviced their excelon pick and place machine on the side in their startup industrial unit. Fun times.
Rick
Last edited:
HA! That reminds me... When I was at National Semiconductor in Federal Way, WA (also known as Felony Way, but that's another story) one of our applications engineers decided to do a little DIY assembly of a sizeable part in a QFN package. I think it was the LMK04800. Tim was pretty good at soldering those parts using a temperature controlled hotplate. He'd use a piece of copper clad FR-4 as a heat spreader.
Anyway. I think he forgot to turn the hotplate off and just absolutely cooked that piece of FR-4. Filled the lab with smoke. The FR-4 delaminated completely and the lab stunk for a good two weeks after. Our lab manager was not happy. Shortly thereafter our site manager approved the purchase of a $20k METCAL soldering machine to solder BGA, QFN, etc.
There's a reason I get nervous tics when people talk about hotplate or frying pan reflow... 🙂
Tom