I find if the diodes are connected as an old style full wave rectifier a simple 10 nF between the two anodes is doing some good.
When I was 17 in 1975 I was taught all of this and seem to have run on autopilot since then. The 2 diode PSU seems the better choice for what I am doing this time . At 1.5 A load ( 2 x 12 V 50 VA ) . I get about 16.1 V using nice conventional 6A diodes and 16.4 V SR506 ( 60 V is a bit close for comfort ). 28 V AC in ( 242 VAC ). There does seem to be about 3 dB more noise at - 130 dB referencing 12 VDC when Schotky. The noise seems general but more so above 4 MHz. I was a little surprised that the diodes change the DC so little. Enough to pass the 207 V test, 210 V when the standard type . One general purpose diode was very noisy at 16 MHz. Simple low current diode tests say 0.43 , 0.27 , 0.18 V . 6A20G , SR16150 , SR506 ( 560 ) . A custom transformer solves this. Nice not to bother. Oddly a higher grade will not help me as it is the poor regulation of this 50 VA that helps ( 14 % and next one up 11% ).
Has anyone a definitive layout for LM317? I have read too much and now baffled. I have just a gound plane for 0V. The others to the top surface. It seems OK but my guess is I could do better. I get > - 70 dB rejection. What I get too much of is ripple harmonics. There is some 50 Hz which sould be much lower.
Andrew T said about ideal 0 V connection somewhere else. Alas no PCB to look at.
When I was 17 in 1975 I was taught all of this and seem to have run on autopilot since then. The 2 diode PSU seems the better choice for what I am doing this time . At 1.5 A load ( 2 x 12 V 50 VA ) . I get about 16.1 V using nice conventional 6A diodes and 16.4 V SR506 ( 60 V is a bit close for comfort ). 28 V AC in ( 242 VAC ). There does seem to be about 3 dB more noise at - 130 dB referencing 12 VDC when Schotky. The noise seems general but more so above 4 MHz. I was a little surprised that the diodes change the DC so little. Enough to pass the 207 V test, 210 V when the standard type . One general purpose diode was very noisy at 16 MHz. Simple low current diode tests say 0.43 , 0.27 , 0.18 V . 6A20G , SR16150 , SR506 ( 560 ) . A custom transformer solves this. Nice not to bother. Oddly a higher grade will not help me as it is the poor regulation of this 50 VA that helps ( 14 % and next one up 11% ).
Has anyone a definitive layout for LM317? I have read too much and now baffled. I have just a gound plane for 0V. The others to the top surface. It seems OK but my guess is I could do better. I get > - 70 dB rejection. What I get too much of is ripple harmonics. There is some 50 Hz which sould be much lower.
Andrew T said about ideal 0 V connection somewhere else. Alas no PCB to look at.
There is a good paper describing what is needed and shows where the traces need to connect.
I can't remember the author, but a few years ago another Member gave a link to that article and I went back to re-read it to make sure I was remembering correctly.
And since the application would be likely in very high power systems, possibly the TO3 shape is more amenable to airflows from cooling fans.
Perhaps also in the interests of reliability some manufacturers would use quasi-complementary designs since NPN transistors are less apt to fail in use than PNP. I am thinking about amplifiers in theatres and stadiums which are put to hard use.
In using only one power transistor type the numbers are doubled and for production purposes that might better suit the price break structure of their supplier. There is always the curse of company accountants. One could speculate the market for common the well established devices is likely to be pretty competitive.
Some amplifiers using plastic packages have a thick aluminum (U.S. spelling) bar spanning across the tops of the packages to clamp them to the finned heat-sink. If this is strategy is to deal with some dissipation or heat sharing problem it would cause an impediment to fan cooling with a long chain of multiple output devices.
The plastic packages no doubt are cheaper to manufacture - and would be adequate for the levels of output for common audio systems.
Last edited:
Andrew and everyone I sorted it out and gained the missing 22 dB. I said to myself the LM317 is not very different to a power amp and the old ones at that. In the past following Douglas Self and Ben Duncan has not made a dramatic difference to hum levels or distortion. Out of respect to commonsense I followed the sage advice. This time I took Mr self's advice and made a T from caps to ground plate. This was not as easy as the design is centre tapped with 2 diodes. It is anybodies guess where to best solder them. Half way up the T to start. The strange thing is with non adjustable regulators with a simple 1oz copper plate it was good enough. I must now get the insulation tape out and do me a PCB ( no handy pen alas ). I have a hunch half deadbug as this is will not be beaten. Just to help others. Everything deadbug to a ground plate .The caps using > 1.2mm dia wire formed in a T by bending . The upright to the ground plate 1/3 up 3 cmm. I suspect I can get 3 dB more by trying harder. I now only have 100 Hz as anything else is below noise which is very low.
Andrew it was what you said somewhere else that prompted me to try harder, thanks. Reading the LM317 spec sheet seemed OK until I realised I had by bootstrapped the setting resistor to ground via a big cap eliminating the gain degredation. As it is 10.8 V I had lost > 20 dB somewhere. Regardless of a very low resistance there were horrible problems.
I have sweet spotted the transformer. I am using about 30VA out of 50 VA. The ripple is 400mV peak to peak. It is a 0-12 0-12. If 80VA I would need a 0-13 0-13.
I have to say I am convinced the old valve radio PSU as here is rather nice. The Schottky's do seem to introduce a bit of noise. It can be dealt with. Against all of what is said 10nF Mylar ( the very cheap green ones) seems to work, perhaps a ceramic would be worse? Between the diode anodes seems fine. I was exspecting to use a snubber. The results go from -120 dB to - 132 dB 4 MHz bandwidth reference 12VDC. I had to make up a BNC cable and solder it on to get reliable results, the clip probes were no good. There was more noise switched off if not. Interestingly the Schottky noise seems to crop up in the audio band if left unfiltered. Not drastically so. We are talking 3 dB well below -120dB.
I have to say I am convinced the old valve radio PSU as here is rather nice. The Schottky's do seem to introduce a bit of noise. It can be dealt with. Against all of what is said 10nF Mylar ( the very cheap green ones) seems to work, perhaps a ceramic would be worse? Between the diode anodes seems fine. I was exspecting to use a snubber. The results go from -120 dB to - 132 dB 4 MHz bandwidth reference 12VDC. I had to make up a BNC cable and solder it on to get reliable results, the clip probes were no good. There was more noise switched off if not. Interestingly the Schottky noise seems to crop up in the audio band if left unfiltered. Not drastically so. We are talking 3 dB well below -120dB.
My question is whether it would be better to have a bypass to earth - through 100n capacitors between each secondary winding and the earth connected centre tap?
If the secondary windings feed the bridge through series resistors of a few ohms, more tuning, that should put some restraint on the recharging current pulses and have some low pass filtering effect.
Last edited:
Mjona. When I do things often it is not by the book, it is my book I follow. I only change one thing at a time no matter what others say. I even annoy myself by doing it, call it my religion if you like. The 10nF was from the bridge rectifier that went before. The transformer is in two sections so this was a minor modification. I simply stopped using two of the bridge diodes ( nice negative supply if needed there , I did for a while). The 10 nF is usually enough to pass regulations. Some of my friends insist 4 x 10 nF sounds worse to the point of calling me an idiot to have fitted them. The arguement they use is the 10 nF simply reduces the frequency of the problem and that makes it more audible.To calculate a snubber is dependant on the transformer inductance, doing it empirically got me nowhere ( using an analyser ). This is why I was overjoyed not to bother. The reason I said it at all is if others had stumbled upon the same thing? I have a hunch ceramic will not be as good ironically as it is too good. 1nF COG comes to mind.
My friend John accepted the analogy with FET's and Schottky when discussing Schottky noise. His statement was they at his workplace had to be very careful when a low noise application and often would use conventional diodes. His point being although Schottky store energy differently to a PN junction they release it very fast. Some would think they do not store energy by things I read. The forward voltage tends to say diffferently. I must say when I used some in a class A power amplifier I was very dissapointed. Perhaps I should have tried harder?
One thing to say. My friend has 227V. The PSU was just about possible at 227V and 220 V 1.2A. By careful component choice and the same transformer I almost have 207V. 206.8 V is where it has lost 10 dB again and is at the drop out limit, it is a LM317 generic with 1.3V drop out . The problem is I lost my way on the hum and refused to accept a large piece of copper is no substitute for doing it the way described way. My friend said by February would be great. With luck Christmas.
Some of the very fancy LDO devices can reduce the drop out by >1 V. This is mostly because they use a Super Alpha PNP pass transistor. The one I am using is still NPN. I found it to be steadfast in stability. The other device exspensive and so many warnings as to stability. The killer was 450uV of noise which drops to 260uV with 33uF Cout. LM317 is about 27uV. I have to say LM317 is a very nice device.
If you need more current try what I did . LM317 or LDO version. TIP2955 pass transistor as shown in many application notes. Set up your pass device at >1.5A and give it a simple fuse. The very large footprint of the TO247/TO3P is vastly better than TO220. The LM317 can still monitor the output temperature and assist the protection if on the same heatrsink. If my braiin is in gear it is very simple as collector and Vout on LM317 share the same die potential which means very few precautions need to be taken. You should find the loss in the pass device almost is of no importance as the LM317 will loose less than typical depending on how you set things up.
If you want to exceed a LDO device you can use an LM317 as the reference device and TIP3055 as the workhorse.This gives 0.6 to 1 V typical loss. Add a PP3 battery to raise the LM317 up enough. I bought a box of PP3 at $1=3. They make excellent stand in devices when an idea comes. Then the battle starts to equal them. That version eat this one for breakfast until last night. I like the 3055/2955 more than many devices, they often win. They are cart horse and race horse in one. 3 MHz for <$1 is great. One can even have posh ones for that money. I bought Indian 2N3055 for $1=4. They are great! I suspect they are something better of TO220 slice in TO3. 3055 often sounds better than many modern devices because we know better how to use it is my guess? Amcron DC 300 is a wonderful amp. The same John has one. It is so much better than the super amp he owns which is still wonderful. They both are wrong in that the modern one is slightly harsh and Amcron ( Crown in fact ) dull. As far as I know DC 300 is 3055? Reading Ben Duncan again he was saying how 80 kHz compares with 200 kHz when looking at slewing. For Hypex 50 kHz is about the limit. And guess what, it sounds lovely. Different needs but similar outcomes. I am very anti class D. Hypex is very OK. In fact slight upper midband brightness is what I hear. That must be the reisdual noise I think ? Without knowning and careful listening one would guess Hypex to be > 200 kHz . In a subtle way it is.
12.04 + 0.4 ripple peak to peak + 1.3V = 13.74. I have 13.5V so 0.24 V into the drop out zone at 207V. At 1.2 A which is the load my friend needs 200V. Reading theroy 24 bit does not need low noise. It can't hurt can it? My new computer is 64 Bit. I have absolutely no idea what that means in PSU terms. John yet again told me how digital averaging works and how a disgusting PSU would be fine. I suspect the PSU it has is digusting so that boox is ticked. The Hypex SMPS is mildly disgusting. John says they are the best in the industry at sane prices. I did by hours of work beat it using a conventional PSU. You need a transformer 4 times larger than you would imagine. The Hypex sure has some reserve power. If you use it's aux outputs build some LC filters. It will be a big deal and super fun getting there. I was using it for MC PU.
At the start of this some weeks ago I used a complimentary feedback pair and TL431. It was very good. The problem was it sometimes gave oscillation which I am fairly certain was the TL431. If the stability cap is taken to the upper limits it works fine. I had great doubts about that as something said not even a cart horse. Reducing the stability cap to 1nF seemed OK. Maybe I was a coward. Lets be clear it was a real benchmark set up. It's drop out very small. A Darlington was no more stable both homemade and readymade. Readymade was excellent. If homemade copy the feedforward resitors as they help a fraction. If using BD 139 TIP2955 100R seems about right to set up the pair. If BD139 TIP3055 try making the BD139 do much of the work. You almost win back more than you looose.
My friend John accepted the analogy with FET's and Schottky when discussing Schottky noise. His statement was they at his workplace had to be very careful when a low noise application and often would use conventional diodes. His point being although Schottky store energy differently to a PN junction they release it very fast. Some would think they do not store energy by things I read. The forward voltage tends to say diffferently. I must say when I used some in a class A power amplifier I was very dissapointed. Perhaps I should have tried harder?
One thing to say. My friend has 227V. The PSU was just about possible at 227V and 220 V 1.2A. By careful component choice and the same transformer I almost have 207V. 206.8 V is where it has lost 10 dB again and is at the drop out limit, it is a LM317 generic with 1.3V drop out . The problem is I lost my way on the hum and refused to accept a large piece of copper is no substitute for doing it the way described way. My friend said by February would be great. With luck Christmas.
Some of the very fancy LDO devices can reduce the drop out by >1 V. This is mostly because they use a Super Alpha PNP pass transistor. The one I am using is still NPN. I found it to be steadfast in stability. The other device exspensive and so many warnings as to stability. The killer was 450uV of noise which drops to 260uV with 33uF Cout. LM317 is about 27uV. I have to say LM317 is a very nice device.
If you need more current try what I did . LM317 or LDO version. TIP2955 pass transistor as shown in many application notes. Set up your pass device at >1.5A and give it a simple fuse. The very large footprint of the TO247/TO3P is vastly better than TO220. The LM317 can still monitor the output temperature and assist the protection if on the same heatrsink. If my braiin is in gear it is very simple as collector and Vout on LM317 share the same die potential which means very few precautions need to be taken. You should find the loss in the pass device almost is of no importance as the LM317 will loose less than typical depending on how you set things up.
If you want to exceed a LDO device you can use an LM317 as the reference device and TIP3055 as the workhorse.This gives 0.6 to 1 V typical loss. Add a PP3 battery to raise the LM317 up enough. I bought a box of PP3 at $1=3. They make excellent stand in devices when an idea comes. Then the battle starts to equal them. That version eat this one for breakfast until last night. I like the 3055/2955 more than many devices, they often win. They are cart horse and race horse in one. 3 MHz for <$1 is great. One can even have posh ones for that money. I bought Indian 2N3055 for $1=4. They are great! I suspect they are something better of TO220 slice in TO3. 3055 often sounds better than many modern devices because we know better how to use it is my guess? Amcron DC 300 is a wonderful amp. The same John has one. It is so much better than the super amp he owns which is still wonderful. They both are wrong in that the modern one is slightly harsh and Amcron ( Crown in fact ) dull. As far as I know DC 300 is 3055? Reading Ben Duncan again he was saying how 80 kHz compares with 200 kHz when looking at slewing. For Hypex 50 kHz is about the limit. And guess what, it sounds lovely. Different needs but similar outcomes. I am very anti class D. Hypex is very OK. In fact slight upper midband brightness is what I hear. That must be the reisdual noise I think ? Without knowning and careful listening one would guess Hypex to be > 200 kHz . In a subtle way it is.
12.04 + 0.4 ripple peak to peak + 1.3V = 13.74. I have 13.5V so 0.24 V into the drop out zone at 207V. At 1.2 A which is the load my friend needs 200V. Reading theroy 24 bit does not need low noise. It can't hurt can it? My new computer is 64 Bit. I have absolutely no idea what that means in PSU terms. John yet again told me how digital averaging works and how a disgusting PSU would be fine. I suspect the PSU it has is digusting so that boox is ticked. The Hypex SMPS is mildly disgusting. John says they are the best in the industry at sane prices. I did by hours of work beat it using a conventional PSU. You need a transformer 4 times larger than you would imagine. The Hypex sure has some reserve power. If you use it's aux outputs build some LC filters. It will be a big deal and super fun getting there. I was using it for MC PU.
At the start of this some weeks ago I used a complimentary feedback pair and TL431. It was very good. The problem was it sometimes gave oscillation which I am fairly certain was the TL431. If the stability cap is taken to the upper limits it works fine. I had great doubts about that as something said not even a cart horse. Reducing the stability cap to 1nF seemed OK. Maybe I was a coward. Lets be clear it was a real benchmark set up. It's drop out very small. A Darlington was no more stable both homemade and readymade. Readymade was excellent. If homemade copy the feedforward resitors as they help a fraction. If using BD 139 TIP2955 100R seems about right to set up the pair. If BD139 TIP3055 try making the BD139 do much of the work. You almost win back more than you looose.
I have used LM317 and LM337 regulators for years, including a supply with boosted with TO3 transistors along the lines of National Semiconductor Application Notes. My original inspiration came from Ben Duncan's writings in Hi-Fi News. I tried some of the tricks such as putting metal oxide varistors of suitable voltages across power supply lines but could not live with the resultant sound. I had a friend who thought outside the square about modifications these were simple things which lent to your idea (mine too) of doing things one at a time. The maxim he used repeatedly was "less is more".
Earlier this year I removed the regulator board from my pre-amp and replaced it with a simple circuit using an audio quality dual op.amp in harness with a couple of transistors. I did this as an experiment but I am still living with it. The preamp unit feed - by an external supply continues as built with LM317 and LM337 regulators.
Last edited:
Luckily there are some interesting people designing audio gear, and this, SoundStage! Global | SoundStageGlobal.com | Phantom Sound: Three First Impressions, makes all the "right noises" on a first reading - eventually the current lethargy in audio thinking will be broken, and this looks to be a healthy step in the right direction ...
Less is more. That's my song sheet.
I was very tempted to build an op amp based solution until I realised that the noise is mostly the voltage reference. Reports that zeners might be the best seem a bit exagerated when tested. A CCS feeding a resistor wasn't tried. Seems it should be very low noise. TL431 is exceptional. It is said ON semiconductors are the best. There are better things. Not to the extent I would pay the money. TL431 data sheet is a joy. The graphs show stability is either a very small or very large cap. Be very careful at 2.5 V. It is then at it's most difficult. Some say the > 10 kHz gain is not wonderful. To be honest that is true of most devices. TL431 ia binary device. It works or it doesn't. That is helpful. The cheapest scope will tell you.
I must say a couple of 6.2 V zeners, TIP3055, filter cap and plenty of zener current seems almost as good as it gets. I looked hard for problems and found none . I doubt any double shunt circuit would be as easy going. In 1974 I was told this is the circuit to never bother with. Take TIP3055 to be gain of 50 at 1.5 V. I would guess it could be very good sounding. Should give just under 12V. BD139+TIP2955 as a feedback pair might offer something.That feedback loop is ideal and I never saw stability problems in a PSU appliaction. A single PNP pass device more so when wrapped in LM317 feedback loop.
In a valve amp I built a NFET capitance multiplier ( TO220FP ). I tried every fancy trick to protect it without sucess. I then did some sums and realised that it's maximum current could be kept safe by lossing 5 V out of 450 V via a resistor. I did feel stupid to take so long to do the obvious. The sucess was both hum and sound. I doubt any choke PSU could better it. Hum was - 85dB reference 1 watt 8R. For a SE valve amp, that's not bad. The PSU was 2 x 680 uF giving 340 uF at 470 V. The pentode input valve had it's own nice supply taken from the 450V using class X2 1 uF as cheap good caps. Pentodes are very nice devices if the PSU is quiet.
I did get my 3 dB today. In fact at 1.5 amps 12 V Ethernet via mains on will hide the hum.
I was very tempted to build an op amp based solution until I realised that the noise is mostly the voltage reference. Reports that zeners might be the best seem a bit exagerated when tested. A CCS feeding a resistor wasn't tried. Seems it should be very low noise. TL431 is exceptional. It is said ON semiconductors are the best. There are better things. Not to the extent I would pay the money. TL431 data sheet is a joy. The graphs show stability is either a very small or very large cap. Be very careful at 2.5 V. It is then at it's most difficult. Some say the > 10 kHz gain is not wonderful. To be honest that is true of most devices. TL431 ia binary device. It works or it doesn't. That is helpful. The cheapest scope will tell you.
I must say a couple of 6.2 V zeners, TIP3055, filter cap and plenty of zener current seems almost as good as it gets. I looked hard for problems and found none . I doubt any double shunt circuit would be as easy going. In 1974 I was told this is the circuit to never bother with. Take TIP3055 to be gain of 50 at 1.5 V. I would guess it could be very good sounding. Should give just under 12V. BD139+TIP2955 as a feedback pair might offer something.That feedback loop is ideal and I never saw stability problems in a PSU appliaction. A single PNP pass device more so when wrapped in LM317 feedback loop.
In a valve amp I built a NFET capitance multiplier ( TO220FP ). I tried every fancy trick to protect it without sucess. I then did some sums and realised that it's maximum current could be kept safe by lossing 5 V out of 450 V via a resistor. I did feel stupid to take so long to do the obvious. The sucess was both hum and sound. I doubt any choke PSU could better it. Hum was - 85dB reference 1 watt 8R. For a SE valve amp, that's not bad. The PSU was 2 x 680 uF giving 340 uF at 470 V. The pentode input valve had it's own nice supply taken from the 450V using class X2 1 uF as cheap good caps. Pentodes are very nice devices if the PSU is quiet.
I did get my 3 dB today. In fact at 1.5 amps 12 V Ethernet via mains on will hide the hum.
You men should look over Esperado's site, he has some interesting ideas on power supplies.
Do not miss the TI Design Note AN 1849 on how to design power amp power supplies. It might as well have been written by Matti Otala, which shows that mosz of what we do has in fact been known for long time, but somehow fotgotten and is now being rediscovered.
On 317/337. I hve been told more than once that the 317 is MUCH better than the 337, and that in fact the 337 should be avoided. If memory serves, it has around 20 dB more noise than the 317, all else being equal.
Much in line with Nigel's views, I find that a simple zener diode assisted cap multiplier, if done right, is hard to beat. Its only caveat is that it doesn't handle currents above say 1 Amp very well. Whoever wants to push it, can reduce output impedance by adding a say BC 639/640, or MJE 06/56, as a driver for the MJE 15030/15031 transistors, creating discrete Darlingtons. Do throw in a gate stopper resistor to the driver, you probably won't need it, but just in case.
Do not miss the TI Design Note AN 1849 on how to design power amp power supplies. It might as well have been written by Matti Otala, which shows that mosz of what we do has in fact been known for long time, but somehow fotgotten and is now being rediscovered.
On 317/337. I hve been told more than once that the 317 is MUCH better than the 337, and that in fact the 337 should be avoided. If memory serves, it has around 20 dB more noise than the 317, all else being equal.
Much in line with Nigel's views, I find that a simple zener diode assisted cap multiplier, if done right, is hard to beat. Its only caveat is that it doesn't handle currents above say 1 Amp very well. Whoever wants to push it, can reduce output impedance by adding a say BC 639/640, or MJE 06/56, as a driver for the MJE 15030/15031 transistors, creating discrete Darlingtons. Do throw in a gate stopper resistor to the driver, you probably won't need it, but just in case.
Happy Christmas Dejan.
I measured a PP3 9V battery. It has more hiss than my PSU. That's no big deal. The big shock is more hum ! The high internal reistance of a PP3 zinc carbon is enough to pick up local hum. This explains the seeming strange 50 Hz I was getting on my tests. That needs very low impedance to avoid.
I have settled on LD1084. It is cheap and seems as good as LM317. It will tollerrate 5 amps and has 1.2 V better drop out. It is very stable and seems to loose heat nicely. My design will accept LD1085 eqaully well. For the moment the price is about the same. 1084V will do the same noise reducing tricks as LM317. That is if the gain resistor is boostrapped the noise created by the gain can be lost. I used 470 uF for interest, 10 uF is fine. 470 uF seems to give no problems and offers soft start. I dare say in transient terms better with none.
Anyone know of ideal ferrite beads to use for Vin on these TO220's ? You can't aways shift things to 0V. Best to say no entry.
Voltage regulators can not sink current I suppose? This makes the output cap important. I guess the 1N4007 reverese biased protection diode ( out to in ) helps? The load will sink some of that I guess? That being so should we protect the regulator with a small choke ? 47 uH + 1 uF ? My tests give me doubts.
I measured a PP3 9V battery. It has more hiss than my PSU. That's no big deal. The big shock is more hum ! The high internal reistance of a PP3 zinc carbon is enough to pick up local hum. This explains the seeming strange 50 Hz I was getting on my tests. That needs very low impedance to avoid.
I have settled on LD1084. It is cheap and seems as good as LM317. It will tollerrate 5 amps and has 1.2 V better drop out. It is very stable and seems to loose heat nicely. My design will accept LD1085 eqaully well. For the moment the price is about the same. 1084V will do the same noise reducing tricks as LM317. That is if the gain resistor is boostrapped the noise created by the gain can be lost. I used 470 uF for interest, 10 uF is fine. 470 uF seems to give no problems and offers soft start. I dare say in transient terms better with none.
Anyone know of ideal ferrite beads to use for Vin on these TO220's ? You can't aways shift things to 0V. Best to say no entry.
Voltage regulators can not sink current I suppose? This makes the output cap important. I guess the 1N4007 reverese biased protection diode ( out to in ) helps? The load will sink some of that I guess? That being so should we protect the regulator with a small choke ? 47 uH + 1 uF ? My tests give me doubts.
At first thing: "we wish you a Happy Christmas, we wish you a Happy Christmass...
"If after a well decoupled zener diode with a LP filter feeding a BC317, could we use a high gain Darlington such a TIP 142/7... to still have a low impedance on a large bandwith and then into a 0.5 ohm/1W resistor or a smd self feeding a 10 000 uf cap and only a little decoupling cap near the load ?
Here the reg would just works to feed the output cap with a constant voltage and low impedance, but this is not anymore the cap before the shunt reg which is feeding the load but this 10k uF output cap... we choose it with the lower esr possible then still using a low esr cap for the local decoupling of the load if the 10 k uF doesn't feet (no place) near it ?
Does a cap "alone" (the output cap) is no still faster than a cap which share its work with a zener, any shunt, then transistors then a load ?
two cents
You probably could Cristophe, but you should ask yourself why would I bother with making a discrete Darlington when so many are readily avilable?
The answer is bandwidth. No Darlington known to me can match MJE's 60 MHz bandwidth.
But, I do hear the gears working in your head, and that's good, that's what makes new ideas come to light.
The answer is bandwidth. No Darlington known to me can match MJE's 60 MHz bandwidth.
But, I do hear the gears working in your head, and that's good, that's what makes new ideas come to light.
www.ti.com/lit/an/snaa057b/snaa057b.pdf
Inrush Current Control
Power Up/Down Mute Control
Constant Brightness LED Circuit
- Status
- Not open for further replies.