EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: electronics technician on April 26, 2014, 05:24:44 pm
-
will this schematic work if i build it on a breadboard
-
It is typical audio bullshit, that is. Dont breadboard it, it needs to be air wired with gold plated silver wires...
-
thanks for the advice i knew it was too good to be true
if i have to correct the schematic what do i have to do
-
Audio circuit? Looks like a DC power supply circuit to me with no need for 'golden ear' components. ;)
-
Maybe NANDBlog shouldn't post under the influence of (whatever?). As retrolefty (welcome!) said, it is a rather old-school regulated power supply circuit. No reason it wouldn't work if you build it like that, but there may be better ways of accomplishing whatever you are trying to do (undisclosed?).
-
This is a power supply circuit... by the way it's designed it's aimed for audio idiots.
electronics technician, by the way you post your questions and the questions themselves, it looks like you're a beginner. This type of circuit is not something that a beginner should do, before having at least some basic understanding of the components and how they work.
Do you even know how expensive and large a 625VA transformer is? Do you know how big of a heatsink you'd need for those 2n3055 transistors? This isn't for you.
-
This is a power supply circuit... by the way it's designed it's aimed for audio idiots.
And you know this how?
-
There is nothing indicating that this power supply was intended for audio applications. You see this configuration in a lot of old test equipment from the time when NPN bipolar power transistors were much better than PNP transistors. This design also has the advantage of allowing the case (collector) of the power pass transistors to be connected to chassis ground so there is no need for isolated mounting on the heat sink. You could think of it as a floating low dropout negative voltage regulator using NPN pass transistors but since it is floating, the negative side can be grounded and the positive side used as the output. If you were to use a 7905 negative regulator to generate a positive supply, this is how you would do it.
If you want to see an old example of this configuration, check out the old Tektronix power supply designs used in their oscilloscopes.
As far as I can tell from a quick look, it will work fine. Keep in mind though that it requires a floating source which the transformer secondary happily provides.
-
I don't see anything obviously wrong with the circuit. Like others have pointed out it is using some old-school parts that don't seem to get a lot of respect (the 2n3055 and LM741, in particular.)
The schematic is also a little confusing. It refers to GND as -13.8V. I'm guessing that the implication is that if you ground the +13.8V terminal (to another supply) that the -13.8V terminal will have the voltage noted, which is the case. It just looks like it is a dual voltage supply, though, they way they drew the schematic. (It isn't.)
-
It also says "power supply", so I think it's supposed to be a power supply :)
If the schematic is correct, I couldn't say with certainty. At first glace it looks to be ok, but some things are a bit odd. I would guess at it being a psu for a charger or so, but why is a fuse used as an OFF-switch? Seems a bit wasteful.
And also, I'm not a fan of letting 20 amps run through a bread board. I have no idea what they are rated for, but I just don't feel comfortable with that.
Any more info on what you plan to use it for?
-
Ouch, poor 741 being abused near the input voltage range limit, and really good way to make a high power oscillator as well if you use modern 2N3055's with much higher gain and much higher transition frequency.
-
As others have said just an old school linear DC power supply. It's a floating output so one is free to ground either output terminal. Also of note is an old school 'crow bar' circuit that blows the fuse if you ever have a final pass transistor short and try and place full rectifier DC voltage to the output terminals.
-
That's a power supply for a car amp or a HAM. I'd go for a LM723 based circuit.
-
thanks for all the info guys
-
That's a power supply for a car amp or a HAM. I'd go for a LM723 based circuit.
As a Car amp it's going to have to need a rather long AC power cord, no? ;)
-
This is a power supply circuit... by the way it's designed it's aimed for audio idiots.
And you know this how?
Well, let's see ...
625VA toroidal transformer to produce 13.8v DC @ 20A .. so 625w transformer, to get 276 watts. Sounds very efficient, right?
toroidal transformer up to 18v AC ... rectified that's 25.4v, minus about 2v drop on rectifier and you get about 24v. At low currents, that transformer's going to output way more than 18Vac, so you may get more than 25v DC at the output.
How will those 60.000uF worth of capacitors rated for 25v are going to feel when you push more than 25v through them?
Why 25v DC at input, to output only 13.8v DC ? So that you can waste half of that in the heatsinks as heat?
C = Current / (2 x ac frequency x Vripple) .... 0.06F = 20a / 100Vr => Vr = 20 / ( 100*0.06) = ~ 3.3v
So you have enough capacitance there to keep the dc voltage between ~ 22-25v when you want to output 13.8v DC at 20A ... why? No need for such large capacitance. Large capacitors are more expensive than just choosing a better transformer with higher peak DC voltage, but I guess they're limited by the LM741 opamp (+/- 15v input voltage)
Luckily they added two NTC resistors in series to limit the inrush current caused by such big capacitance, otherwise such large toroidal would probably trip your house fuse and make those 3A/6A
fuses pointless.
Then you have the triac at the end, when you could just have a relay or something after the transformer that would cut the input voltage instead of shorting everything to (hopefully) trip the fuse.
Old, maybe a good design decades ago when those 2n3055 transistors were good and expensive but these days it's not worth spending time to make something like this.
-
It is a power supply for home-base powering mobile ham radio equipment (typically a transceiver).
You can find it on Mr. van Stralen's web page: http://www.qslnet.de/member/dk4dds/text_duits_engels/2009_dk4dds_down_load_gallery_dl_uk.htm (http://www.qslnet.de/member/dk4dds/text_duits_engels/2009_dk4dds_down_load_gallery_dl_uk.htm)
What remarkable twaddle is posted in these forums by dilettante "engineers".
-
PA1HFO is a HAM call sign. It is a high power mains power supply mostly for mobile (12v automotive battery powered) radio equipment. I don't see anything wrong with it. I have an old commercial unit of similar design.
The main feature of the circuit is the pass transistor collectors are ground so they require no isolation from the chassis. The commercial unit I have has the TO3 transistors mounted on the bottom of the aluminium chassis cans exposed. Internally there are some cheap aluminium sheets bent and riveted to the chassis forming heatsink fins.
-
As a Car amp it's going to have to need a rather long AC power cord, no? ;)
A car amp works also fine in your living room or lab :-)
-
As a Car amp it's going to have to need a rather long AC power cord, no? ;)
A car amp works also fine in your living room or lab :-)
Agreed, as long as you take it out of the car first. ;)
-
As others have said just an old school linear DC power supply. It's a floating output so one is free to ground either output terminal. Also of note is an old school 'crow bar' circuit that blows the fuse if you ever have a final pass transistor short and try and place full rectifier DC voltage to the output terminals.
The usefulness of crow bar circuits is under appreciated.
I would be really careful about grounding the positive output to make a negative power supply because then the chassis will be at positive ground unless isolation is added somewhere. I have some power supplies which can operate safely like this which isolate the outputs from the chassis ground under all conditions.
-
Ouch, poor 741 being abused near the input voltage range limit, and really good way to make a high power oscillator as well if you use modern 2N3055's with much higher gain and much higher transition frequency.
There is some concern about that but the emitter degeneration resistors used to enhance current sharing help and a little bit of frequency compensation could be added between the operational amplifier output and inverting input.
I should also point out that the output transient response would be greatly improved by adding a low value resistor between the base of the 2N3055 array and the negative supply at the input. Without this change, the design is going to be harder than it needs to be on low voltage RF power amplifiers.
-
It is not a bad supply, just a poor choice with more modern parts available to implement a better one.
Years ago I designed a power supply using what was at hand. 2N3773 transistors, plenty of them ( 6 in parallel on a massive piece of aluminium alloy extrusion) with another 2 as a darlington driving them, all driven from a 723 arranged as a 0-25V supply. I designed it for 60A but was limited by the input from the built in power supply of the building, which was supplied to the distribution panel as 28V at 100A, but which was then sent via a 1.5mm cable to the various workshop rooms ( I wanted to rewire but never did) so I really only got 15A into a short circuit. No insulation from power devices to heatsink, just a big Bakelite board use to insulate it inside the aluminium workbench so it did not touch. Ballast resistors were 6 equal length PTFE 18SWG wires soldered to the emitters, and one was used for both current sense and for metering with a single cheap analogue meter movement. I did manage to weld aluminium with it using a carbon rod from a D cell battery. That generated an incredible amount of RF noise, wiped out FM radio for around 100m. If I had been able to replace the cable and the breakers I would have made it a 200A supply, as the input was via a 100mm cable, and the 28V transformer rectifier unit was rated for a current of 10kA, and never ran at more than idle most of the time except when it was used to test start engines. Most of the time by us it ran a radio from the rail via a simple voltage regulator, typically a power transistor ( whatever NPN was to hand in a TO3 or TO220 case and with gain of more than 20) and a 13V 1W zener diode and a resistor. Some dispensed with the diode and used 2 resistors to simply divide the voltage down.
-
It is not a bad supply, just a poor choice with more modern parts available to implement a better one.
It is a great design though if you want to avoid isolating the power pass transistors from the chassis and want a negative chassis ground.
-
will this schematic work if i build it on a breadboard
Yes, the schematic will work but it won't run at the full power rating on breadboard, which is limited to low currents.
-
will this schematic work if i build it on a breadboard
It will, assuming that you build it correctly on a breadboard.
ua741 is a perfectly fine choice here, and it matches well with the rest of the design. Using a modern and fast part here may not be necessary and likely can cause stability issues. The basic topology is a LDO type, implemented with NPN output transistors.
-
The key thing about breadboards is to think of them for testing "electronic" circuits, with currents in the milliamps. They are really not intended or designed for "power" circuits.
If you want to "breadboard" a power circuit you should think in terms of terminal strips, point to point wiring, and other techniques. Though of course you could put the "control" part of the power supply on a breadboard--the op amp and surrounding components. Just don't try to feed the main power rails through it.
-
This is an old design for positive ground systems. Somebody just dusted it off. If memory serves me there should be a cap on the reference there or AC ripple will be introduced to the output. Its not a very clean regulation system, but for non audio stuff it is fine. There are fat better circuits out there.
-
This is an old design for positive ground systems.
Actually it was DESIGNED for negative-ground. That is why the collector of the series-pass transistors was hard-connected to ground.
-
If it helps, I made a version of the 13.8V >20A PSU in the link below way back in the 1980s to power a 100W HF ham radio.
http://warc.org.uk/?page_id=404 (http://warc.org.uk/?page_id=404)
The design has stood the test of time really well and I've even used it to start vehicles with a flat battery on numerous occasions as well as being very brutal to it in terms of short circuit tests when I first made it.
It's a nice design with a soft start, overvoltage trip and overcurrent trip. I put the current limit as an external control on my version. It has never ever developed a fault/failure in over 25 years service :)
-
If it helps, I made a version of the 13.8V >20A PSU in the link below way back in the 1980s to power a 100W HF ham radio.
http://warc.org.uk/?page_id=404 (http://warc.org.uk/?page_id=404)
The design has stood the test of time really well and I've even used it to start vehicles with a flat battery on numerous occasions as well as being very brutal to it in terms of short circuit tests when I first made it.
It's a nice design with a soft start, overvoltage trip and overcurrent trip. I put the current limit as an external control on my version. It has never ever developed a fault/failure in over 25 years service :)
Looks interesting as I have 4 24V@10A transformers I'd like to put to good use.
I'll take a closer look tomorrow. Thanks for the link. :-+
-
If it helps, I made a version of the 13.8V >20A PSU in the link below way back in the 1980s to power a 100W HF ham radio.
http://warc.org.uk/?page_id=404 (http://warc.org.uk/?page_id=404)
The design has stood the test of time really well and I've even used it to start vehicles with a flat battery on numerous occasions as well as being very brutal to it in terms of short circuit tests when I first made it.
It's a nice design with a soft start, overvoltage trip and overcurrent trip. I put the current limit as an external control on my version. It has never ever developed a fault/failure in over 25 years service :)
That's your standard LM723 circuit used in most linear power supplies today. Look closely at this other design at the output configuration and how it sinks current rather than sources it.
-
It is a power supply for home-base powering mobile ham radio equipment (typically a transceiver).
You can find it on Mr. van Stralen's web page: http://www.qslnet.de/member/dk4dds/text_duits_engels/2009_dk4dds_down_load_gallery_dl_uk.htm (http://www.qslnet.de/member/dk4dds/text_duits_engels/2009_dk4dds_down_load_gallery_dl_uk.htm)
What remarkable twaddle is posted in these forums by dilettante "engineers".
Indeed!
I was waiting for someone to throw a bucket of water on them! ;D
-
will this schematic work if i build it on a breadboard
Yes, the schematic will work but it won't run at the full power rating on breadboard, which is limited to low currents.
Not so much low currents -- the datasheets do usually say under 1A per contact, but you can put a few amps through, with a good sized pin (large, but not tight fitting -- don't break the board!). I've done as much as 5A at 500kHz, which did get the wires and pins rather hot!
But even with melty contacts, still not enough amperage for the project in question, of course.
What I'd be more worried about is... do you even have a breadboard big enough to fit all those 2N3055s, and if so... uh, how are you going to heatsink them? :P
Tim
-
There is nothing indicating that this power supply was intended for audio applications.
It is using 60.000uF capacitors on the input, nada on the output. And a crowbar. And TO2 transistors. No sane person would built this power supply in the 21 century, only if you already have the components. Or if you dont want to spoil your precious audio signals with "new" components.
-
There is nothing indicating that this power supply was intended for audio applications.
It is using 60.000uF capacitors on the input, nada on the output. And a crowbar. And TO2 transistors. No sane person would built this power supply in the 21 century, only if you already have the components. Or if you dont want to spoil your precious audio signals with "new" components.
Why wouldn't they? The TO-3 transistors support a high power dissipation and are more convenient to use when remote mounted on a large heat sink than TO-220 style packages. The crowbar protects the load from catastrophic power supply failure. The linear design is simple, cheap, and effective.
Astron and others sell lots of power supplies like this although they use the LM723 design with an emitter follower power pass element and separate bias supply. The floating regulator design shown here is more clever than that although this particular implementation has a couple minor problems.
-
- I'd draw a circle around the Darlington pair to indicate that it's a single Darlington-pair device and not two discrete transistors.
- I don't like the terminology of "-13.8V" on the ground rail. Just call it ground or 0V.
- The 1k LED resistor seems a bit small for a 13.8V rail... it's driving the LED pretty hard, perhaps increase it to 4.7k or so.
- The way the bases are drawn on the output transistor stack could be confusing for beginners to read, it might be better if it was re-drawn so it doesn't look like there's two base pins on each transistor.
- The use of a triac as the output crowbar seems unusual, normally you'd expect to see an SCR used.
Those are the only minor things that "don't look right" to me on the schematic, other than the obvious point (as discussed) that a linear regulator with such a big heatsink and inefficiency would be rare these days at power levels like this and you'd usually encounter a switchmode system.
-
- I'd draw a circle around the Darlington pair to indicate that it's a single Darlington-pair device and not two discrete transistors.
- I don't like the terminology of "-13.8V" on the ground rail. Just call it ground or 0V.
- The 1k LED resistor seems a bit small for a 13.8V rail... it's driving the LED pretty hard, perhaps increase it to 4.7k or so.
- The way the bases are drawn on the output transistor stack could be confusing for beginners to read, it might be better if it was re-drawn so it doesn't look like there's two base pins on each transistor.
- The use of a triac as the output crowbar seems unusual, normally you'd expect to see an SCR used.
Those are the only minor things that "don't look right" to me on the schematic, other than the obvious point (as discussed) that a linear regulator with such a big heatsink and inefficiency would be rare these days at power levels like this and you'd usually encounter a switchmode system.
All of that, plus as mentioned:
* The 25V C1 is underrated. Should be at least 30V, to allow for unloaded transformer overvoltage plus some margin.
* The 741 is being run very near it's Max for power rail voltage and input offsets.
Also:
* The current limit via Q2 works out to limit at around 24A. Which makes the 20A output fuse rather silly.
A trimmer to set the precise current limit might have been nice.
* The LM336 is a 3 pin device. Someone who doesn't know that, won't know from the schematic which pins to connect. It's a very poorly drawn schematic anyway, but to leave off pinout diagrams is inexcusable.
* Another mistake with the LM336 is that it should be labeled LM336-2.5 Not 25.0.
* But, worst of all, a WON'T WORK: the values of the voltage set divider R2, P1, R3 are wrong. The voltage at the midpoint of P1, relative to the high rail should be -2.5V to match the reference diode. But it works out to be -4.7V.
Did I make a mistake? 1.2mA in the divider, via 3K3 plus half of P1, ie 500R = 4.7V
I'm wondering if the designer arrived at those values by trial and error, and really the op-amp is operating out of permissible common mode range, and the wrong resistor values just fudged it to sort of work despite a distressed op-amp.
All these things, and especially specifying a triac instead of an SCR for the crowbar, makes me think the designer was something of an amateur. Ha ha, but then he IS apparently a HAM.
-
The use of a TRIAC instead of an SCR may have been intentional. With a TRIAC, if (somehow) a positive source is fed into the negative terminal, the crowbar will trip. This wouldn't be the case for an SCR.
-
* The current limit via Q2 works out to limit at around 24A. Which makes the 20A output fuse rather silly.
The 20A fuse won't blow straight away when there's more than 20A flowing through it. Maybe the 24A limit is to keep the current reasonable until the fuse can blow?
-
Those are the only minor things that "don't look right" to me on the schematic, other than the obvious point (as discussed) that a linear regulator with such a big heatsink and inefficiency would be rare these days at power levels like this and you'd usually encounter a switchmode system.
The popular large Astron linear power supplies make this one look small. Switching power supplies *are* nice at high power levels but linear supplies have the virtue of being inherently low noise (no RF birdies), easy to understand, and easy to repair.
* The current limit via Q2 works out to limit at around 24A. Which makes the 20A output fuse rather silly.
A trimmer to set the precise current limit might have been nice.
It is not there as a precision current limit. It is for protecting the pass transistors long enough for the fuse to blow.
* But, worst of all, a WON'T WORK: the values of the voltage set divider R2, P1, R3 are wrong. The voltage at the midpoint of P1, relative to the high rail should be -2.5V to match the reference diode. But it works out to be -4.7V.
Did I make a mistake? 1.2mA in the divider, via 3K3 plus half of P1, ie 500R = 4.7V
I'm wondering if the designer arrived at those values by trial and error, and really the op-amp is operating out of permissible common mode range, and the wrong resistor values just fudged it to sort of work despite a distressed op-amp.
You got the right results; the values in the schematic are wrong as shown. It looks to me like the LM336 is suppose to be an LM336-5.0 which solves both problems.
Interestingly enough, the even older LM301 has an input common mode range which includes the positive supply making it a better choice if the resistors were adjusted and the LM336-2.5 used. The LM301 is particularly useful as a high side current sense amplifier.
One major but easy to miss flaw is a lack of a low value base-emitter resistor across the output transistors (the darlington has this internally). Without it, when a high power load is released like a transmitter unkeying, the output voltage will surge more than it needs to. I have seen this mistake in commercial linear power supplies included in HF transceivers where their solution was to add a big transient voltage suppressor.
-
- I'd draw a circle around the Darlington pair to indicate that it's a single Darlington-pair device and not two discrete transistors.
- I don't like the terminology of "-13.8V" on the ground rail. Just call it ground or 0V.
- The 1k LED resistor seems a bit small for a 13.8V rail... it's driving the LED pretty hard, perhaps increase it to 4.7k or so.
- The way the bases are drawn on the output transistor stack could be confusing for beginners to read, it might be better if it was re-drawn so it doesn't look like there's two base pins on each transistor.
- The use of a triac as the output crowbar seems unusual, normally you'd expect to see an SCR used.
Those are the only minor things that "don't look right" to me on the schematic, other than the obvious point (as discussed) that a linear regulator with such a big heatsink and inefficiency would be rare these days at power levels like this and you'd usually encounter a switchmode system.
* The LM336 is a 3 pin device. Someone who doesn't know that, won't know from the schematic which pins to connect. It's a very poorly drawn schematic anyway, but to leave off pinout diagrams is inexcusable.
All these things, and especially specifying a triac instead of an SCR for the crowbar, makes me think the designer was something of an amateur. Ha ha, but then he IS apparently a HAM.
And Hams,unlike some Engineers, can use the brains they were born with to determine the active terminals of a 3 pin device without being "led by hand"! ;D
-
Well.. to be fair, I suppose a lot of HAMs just buy equipment and talk. No technical knowledge required (aside from passing the test, and whatever motivation is driving the interest). A wide gulf between that sort of player and a hard core techie such as yourself! :)
Overall, I've seen a lot of dubious-at-best designs from what "hi-fi" I'm familiar with, and I'm sure you've seen your share amongst HAMs. I've also seen (and had to fix..) my fair share of things from just general EE stock...
Tim
-
I've certainly seen plenty of schematics in handbooks for commercially made stuff where not showing pinouts were the least of their sins---especially amongst small,"niche" manufacturers.
.
Back when Australia had an Electronics Industry worth speaking of,the major manufacturers,were pretty good with diagrams,but even then,you'd run into few "dodgy" ones,among the more "fringe" firms.
Not many "hams: build stuff from scratch,nowadays,but a lot of them repair their own equipment.
Of course,their are still quite a number of EEs & Techs among the Amateur fraternity.
Most older commercially made Ham stuff has reasonable documentation,but one RF Linear Amplifier I worked on from the well respected "Henry" company had handbooks which were quite poor,with numerous typos & errors on schematics.
-
There is nothing indicating that this power supply was intended for audio applications.
It is using 60.000uF capacitors on the input, nada on the output. And a crowbar. And TO2 transistors. No sane person would built this power supply in the 21 century, only if you already have the components. Or if you dont want to spoil your precious audio signals with "new" components.
Why wouldn't they? The TO-3 transistors support a high power dissipation and are more convenient to use when remote mounted on a large heat sink than TO-220 style packages. The crowbar protects the load from catastrophic power supply failure. The linear design is simple, cheap, and effective.
Astron and others sell lots of power supplies like this although they use the LM723 design with an emitter follower power pass element and separate bias supply. The floating regulator design shown here is more clever than that although this particular implementation has a couple minor problems.
The TO2 has higher Rjc and higher Rch than many high power TO220. Crowbar is not the preferred method to protect a circuit, that is why we have hundreds of types of ICs to prevent over and undervoltage, and TVS diodes. I dont know about you, but I prefer automatic recovery over blown fuses.
Using 20A linear supply is nor simple, nor effective. This needs a heatsink which can dissipate a ton of power, weights a lot, you need cables in the system, and you need to train people to make the cabling, it is messy. Compared to this, a switching power supply coulf easily be a single board construction. And the claim that "switching is noisy" is not true anymore. Switching is noisy if it is not properly designed.
To be honest I really dont care about those companies which sell 20 year old technology in suits. I mean we also have legacy systems, they are inferior. Whenever I hear the term: this has been working for years, I only see people standing in the way of innovation.
-
Why wouldn't they? The TO-3 transistors support a high power dissipation and are more convenient to use when remote mounted on a large heat sink than TO-220 style packages. The crowbar protects the load from catastrophic power supply failure. The linear design is simple, cheap, and effective.
Astron and others sell lots of power supplies like this although they use the LM723 design with an emitter follower power pass element and separate bias supply. The floating regulator design shown here is more clever than that although this particular implementation has a couple minor problems.
The TO2 has higher Rjc and higher Rch than many high power TO220.
Motorola's famous application note 1040 indicates that TO-3 packages have from 1/2 to 1/10th of the Rcs of TO-220 packages. Even the old TO-66 package comes out better. TO-3 packages also support higher power through larger die sizes (although IRF and others are known for putting over sized dies in TO-220 packages) but that is less relevant since modern devices tend to eschew the TO-3 package for alternatives like the TO-218 and TO-247. Diamond flange packages are also easier to mount reliably because of their greater symmetry. The single fastener on the TO-220 tends to bend the area under the die upwards off of the heat sink surface.
Crowbar is not the preferred method to protect a circuit, that is why we have hundreds of types of ICs to prevent over and undervoltage, and TVS diodes. I dont know about you, but I prefer automatic recovery over blown fuses.
Thyristor crowbars have the advantage of protecting against collector-emitter shorts more reliably than TVS diodes. The later are limited to lower power levels although I have seen them short blowing the fuse anyway in which case they add to the parts which need to be replaced. Over voltage protection that relies on the transistor drive does not help at all in the event of transistor failure.
Using 20A linear supply is nor simple, nor effective. This needs a heatsink which can dissipate a ton of power, weights a lot, you need cables in the system, and you need to train people to make the cabling, it is messy. Compared to this, a switching power supply coulf easily be a single board construction.
Switching power supplies have a definite power density advantage.
And the claim that "switching is noisy" is not true anymore. Switching is noisy if it is not properly designed.
Even resonate mode switchers and low noise inverters are noisier than linear regulators. What you say is true insofar as most applications are concerned but in the context of amateur radio use, switching regulators are a problem for HF receivers. Either they result in birdies or they raise the general noise level. One trick to get around this is to phase lock the switching regulator's oscillator to the receiver to push switching harmonics out of the channel passband but that is hardly feasible with an external power supply. Broadcast band AM receivers may do this to keep their 455 kHz IF free of switching noise.
To be honest I really dont care about those companies which sell 20 year old technology in suits. I mean we also have legacy systems, they are inferior. Whenever I hear the term: this has been working for years, I only see people standing in the way of innovation.
The only innovation here is marketing. Switching regulators have real advantages but low noise and simplicity are not among them. They are not replacements for linear regulators in every application.
-
Whenever I hear the term: this has been working for years, I only see people standing in the way of innovation.
The OP asked about the quality of a homebrew linear PSU circuit. Clearly it isn't a quality design. I offered a proven alternative for a linear PSU design.
I don't know what the OP wants the PSU for, but the Linear PSU schematic in the first post is quite obviously aimed at ham radio use.
You imply that a modern switcher should be designed and built instead. But conducted and radiated emissions from a switching PSU can cause interference and can spoil the listening experience from a radio receiver. Sometimes the level of interference can depend on how the ham sets up the wiring and earthing.
The switching PSU also needs to be able to cope with high RFI that could be conducted or radiated from the ham transceiver. This also applies to linear PSU designs as the PSU can be tripped or the regulation can go loopy if RF is able to get inside the PSU.
The PSU needs to be able to cope with a rapidly changing load in the case of an SSB/CW transceiver.
It is possible to buy modern switch mode PSUs that can work really well. However, I'm not so sure it is going to be easy/straightforward to make a (successful) homebrew switcher for ham use. Some hams have had success when modifying switching power supplies taken from PCs but I would still recommend a decent homebrew linear PSU for ham use even if it is big and heavy and inefficient. Many will simply opt to buy a good quality switch mode PSU instead of trying to make their own.
-
We have zero information from the OP about what he is trying to do with this power supply.
I strongly suspect that he did not even ask and we are not responding to the right question.
This thread demonstrates why such a vague, open-ended query does not hold up as a generic question.
-
will this schematic work if i build it on a breadboard
Yes, the schematic will work but it won't run at the full power rating on breadboard, which is limited to low currents.
Not so much low currents -- the datasheets do usually say under 1A per contact, but you can put a few amps through, with a good sized pin (large, but not tight fitting -- don't break the board!). I've done as much as 5A at 500kHz, which did get the wires and pins rather hot!
But even with melty contacts, still not enough amperage for the project in question, of course.
What I'd be more worried about is... do you even have a breadboard big enough to fit all those 2N3055s, and if so... uh, how are you going to heatsink them? :P
Tim
I found another schematic that looks alot better than this one with short circuit protection current limiting temperature sensor with fan :D and more http://www.qsl.net/vu2upx/Projects/40apsu.htm (http://www.qsl.net/vu2upx/Projects/40apsu.htm)
-
i mount the 2n3055s on the heatsink :) with emitter resistors and long enough wires to reach the breadboard
-
Motorola's famous application note 1040 indicates that TO-3 packages have from 1/2 to 1/10th of the Rcs of TO-220 packages. Even the old TO-66 package comes out better. TO-3 packages also support higher power through larger die sizes (although IRF and others are known for putting over sized dies in TO-220 packages) but that is less relevant since modern devices tend to eschew the TO-3 package for alternatives like the TO-218 and TO-247. Diamond flange packages are also easier to mount reliably because of their greater symmetry. The single fastener on the TO-220 tends to bend the area under the die upwards off of the heat sink surface.
Instead of doing the famous strawman, I will point out only one point where you are wrong, and be done with it, as we are obviously looking at different angle at electronics (I never designed RF, you never did power elecronics), and I don't see the point arguing for no reason. I have an appnote also for you: AN-1012 from IRF. It states that with proper mounting 0,2K/W (case to heatsink) is reachable with TO220. There are transistors from IRF, ST, NXP and Infineon, which have 0,5 K/W junction to case.
If I look at the 2N3055 mentioned here (TO3), that alone has 1,5K/W junction to case thermal resistance, which is already the double. But even an old IRF640 can par with this.
-
Thermal resistance depends on the transistor, and more so on the manufacturer. Often a part that is multi sourced will have a varying thermal resistance die to case, depending on the attach method used inside the cover. TO cans at least will have the advantage in that, if made correctly, you can use a kovar heat spreader to reduce stress on the silicon die, so that it will not fail even with millions of thermal cycles from ambient to close to max die temperature, even with a steel case. Might not have as low a thermal resistance as a brazed die on a copper slug, but at least it will not fatigue off the tab with time.
i have decapped a few different "2N3055" devices and clones from assorted manufacturers, and the internal mounts are often vastly different, from a simple conductive epoxy with a single aluminium wire for each leadout, to some with an internal copper slug with a kovar interface brazed to the die, and multiple emitter wires of aluminium and a single base wire. Most are brazed to a spreader to spread the heat to the steel case, though there were a few Motorola devices with an aluminium cap, aluminium spreader and a base made from a hard aluminium alloy.
-
Motorola's famous application note 1040 indicates that TO-3 packages have from 1/2 to 1/10th of the Rcs of TO-220 packages. Even the old TO-66 package comes out better. TO-3 packages also support higher power through larger die sizes (although IRF and others are known for putting over sized dies in TO-220 packages) but that is less relevant since modern devices tend to eschew the TO-3 package for alternatives like the TO-218 and TO-247. Diamond flange packages are also easier to mount reliably because of their greater symmetry. The single fastener on the TO-220 tends to bend the area under the die upwards off of the heat sink surface.
Instead of doing the famous strawman, I will point out only one point where you are wrong, and be done with it, as we are obviously looking at different angle at electronics (I never designed RF, you never did power elecronics), and I don't see the point arguing for no reason. I have an appnote also for you: AN-1012 from IRF. It states that with proper mounting 0,2K/W (case to heatsink) is reachable with TO220. There are transistors from IRF, ST, NXP and Infineon, which have 0,5 K/W junction to case.
If I look at the 2N3055 mentioned here (TO3), that alone has 1,5K/W junction to case thermal resistance, which is already the double. But even an old IRF640 can par with this.
I think you may have posted the data for a different package than TO-220 there. I'd expect that a typical TO-220 with a screw mount to barely achieve 0.5degC/W c2h even if you got everything mounted correctly and used just the right amount of high quality thermal compound.
The nice thing about the 2N3055 is that it is cheap, fairly idiotproof to mount (because it has a larger footprint and two fixing points) and it can dissipate about 80W even with a case temp of about 75degC. So having four or more of them in a big linear supply like this is fairly bombproof.
It may have a 1.5degC/W j2c but it does have a huge footprint and this is good if you have to fit an insulator.
Like I said earlier, I've used my PSU with the current limit wound up to >30A and used it to start vehicles. I also spent a fair bit of time trying to break it by deliberately abusing it by shorting the output with the current limit way higher than it was designed for. I did this multiple times to test the design claim it could survive a short circuit. I also showed people how it could weld thin metal strips together. None of the 2N3055s failed. In fact nothing failed despite lots of angry and noisy sparks.
-
Motorola's famous application note 1040 indicates that TO-3 packages have from 1/2 to 1/10th of the Rcs of TO-220 packages. Even the old TO-66 package comes out better. TO-3 packages also support higher power through larger die sizes (although IRF and others are known for putting over sized dies in TO-220 packages) but that is less relevant since modern devices tend to eschew the TO-3 package for alternatives like the TO-218 and TO-247. Diamond flange packages are also easier to mount reliably because of their greater symmetry. The single fastener on the TO-220 tends to bend the area under the die upwards off of the heat sink surface.
Instead of doing the famous strawman, I will point out only one point where you are wrong, and be done with it, as we are obviously looking at different angle at electronics (I never designed RF, you never did power elecronics), and I don't see the point arguing for no reason.
I am sanguine that I only run the risk of considering alternatives and/or learning something.
I have done plenty of power and RF electronics and to the point where component failure modes include disappearance. I have a small drawer with literally blown up plastic packaged transistors and TO-3 packages cut up for inspection although except for the later, I have not added to it in a long time.
I have an appnote also for you: AN-1012 from IRF. It states that with proper mounting 0,2K/W (case to heatsink) is reachable with TO220. There are transistors from IRF, ST, NXP and Infineon, which have 0,5 K/W junction to case.
The IRF application note even mentions the mounting problem with TO-220 and similar tab packages and it does not cover TO-3 packages in comparison at all. These types of measurements are difficult to make and compare even without the confounding influence of different manufacturers using different methodologies. I picked that particular application note because of Motorola's recognized expertise, it is widely cited, and because it includes a wide variety of packages. The IRF application notes are great but I do know know of any which include measurements of TO-3 cases.
If I look at the 2N3055 mentioned here (TO3), that alone has 1,5K/W junction to case thermal resistance, which is already the double. But even an old IRF640 can par with this.
Your claim was about TO-3 packages in general and not the 2N3055 specifically. I even mentioned that IRF is known for extending the die size used in TO-220 packages. For a while there was a arms race to see who could put the largest die into the TO-220 package which reached diminishing returns for a while because of lead resistance until the drain bonding wire was replaced with a leaf. In a modern design I would avoid the TO-3 simply because newer packages like the TO-218 and TO-247 are less expensive for the same performance but that would not stop me from using the TO-3 if I had a bunch handy or they were easier to use in a specific application.
As a practical matter, mounting considerations are more problematic then case selection and it is easier to mount a TO-3 reliably than a tab style case at least if screws are used.
It is too bad that IRF includes case to heatsink thermal resistance in their IRF640 datasheet but not in their TO-3 datasheets which are older. I have a whole bag full of TO-3 IRF351s but my IRF databook is hidden away at the moment.
Here is a photo from a 2N5885 TO-3 I pulled apart a couple months ago for failure analysis which is from a linear power supply for a 200 watt HF amplifier. The design was all discrete but the parallel common emitter output pair lacked any emitter biasing so I suspect failure was caused by uneven current sharing and high temperature induced breakdown. Even worse in my opinion, the output transistors lacked base-emitter shunt resistance so the load regulation spike when the amplifier unkeyed required a TVS to protect the RF transistors. That may have resulted in the TVS shorting out thereby destroying the regulator. The current shunt resistor helpfully blew to protect the fuse after the transistors sacrificed themselves.
-
"AN-1012 from IRF. It states that with proper mounting 0,2K/W (case to heatsink) is reachable with TO220. "
I find that very hard to believe.
My rule of thumb for to220 is 10w.
-
will this schematic work if i build it on a breadboard
You want to build a 20A mains powered PSU on a breadboard? :o
That doesn't sound like a very good idea to me. This type of PSU will need a fairly large case, with a lots of airflow, heatsinking and a good, well laid out PCB for all the electronics.
-
will this schematic work if i build it on a breadboard
You want to build a 20A mains powered PSU on a breadboard? :o
That doesn't sound like a very good idea to me. This type of PSU will need a fairly large case, with a lots of airflow, heatsinking and a good, well laid out PCB for all the electronics.
no the heatsink with fan assembly will be separate from the breadbord but the rest of the circuit that controls the current and voltage will be on the breadboard but i agree with you about getting a well made pcb for all this thanks for the advice
-
Speaking of bodgy linear supplies with crappily drawn schematics, here's a little story and an entry for a 'worst of' competition.
Sometime in the 1980s I was living in a small place in Maroubra (one of the Sydney beachside suburbs.) I didn't have any kind of workshop, just some handtools and a few boxes of assorted electronics junk. One morning when trying to leave for work, my car battery was flat. So was the street, so hillstarting was out. I didn't have a charger, and I was poor enough that calling a taxi was not a good option. But, it wouldn't be a disaster if I was a few hours late for work.
And so, I designed and built a car battery charger in about an hour. It worked, got enough charge in the battery to start the car, and I made it to work about 2 hours late.
The supply operates in constant current mode, until the battery reaches a set voltage, then it operates in CV mode.
It's not exactly 'precision', but who cares? It works, and despite the rough build I kept using it for years.
Still have that charger, though it hasn't been used for, um... about 15 years. Just thought you guys might be amused to see pics of a really rushed and bodgy design and build. The schematic was a scribble, modified as I constructed the supply. The paper is actually yellow - it's on the back of a foolscap electoral flyer, that I had boxes of then due to them having a typo that made them unusable. It's been nibbled by insects, since the supply was sitting in an old garage for a long time.
Unfortunately in those days I didn't have the habit of writing the date on all notes and drawings I did. Really bad habit that, which I often kick myself for when looking at notes from my 20s.
-
Fairly sophisticated design,if not construction.
Most El Cheapo "trickle" chargers are just a transformer,a bridge, a meter,& two terminals,& will give you enough charge for one start in about 1.5 hrs.
Your one was capable of a lot better charge rate,though,as is evident by 1hour to build & 1 hour to charge!
-
crufty old heat producing design. get a meanwell switcher. itll do 13.2 volt 20 amp for years with 96 % efficiency.
-
Efficiency isn't a huge concern with Ham Radio use.
The duty cycle is so low that it won't be using a lot of power most of the time,as Hams do a lot more listening than talking.
The heating in a cold Ham shack would pretty much swamp it! ;D
-
TO-220s vary. Just as you should expect for their design...
I've tested new Fairchild MOSFETs -- some small things, which blew when driven just past the expected dissipation due to thermal resistance (something like 50W into RthJC = 1 C/W, case greased to heatsink with no insulator, temp and power slowly turned up until failure).
I've also tested bog standard IRF740s, which went easily past twice their rated dissipation. If I assume rated RthJC, the die was over 210C when it failed; or if I assume it failed at rated temperature, RthJC was more like 0.5 C/W!
The IRF740 die was something like 2.5 x 3.2mm, easily double the Fairchild device of similar rating (presumably, they never die-shrank the cheap old silicon, whereas the new device is more-or-less latest process.
As for thermal performance of '3055s... why even bother? It's not even the McDonalds of transistors. Why would you ever buy one? McDonalds sandwiches are literally specified better, and required to use better materials, than the ancient 2N3055 standard requires. You have no idea what you're going to get. Buy some beefy TO-247s from On Semi and forget about anything else. ;)
Tim
-
To be fair, when you've been a ham for more than a year or two you develop the uncanny ability to discern a callsign. You are REALLY good at it about 6 months after you get your ham license plates and nearly die a dozen times while looking for other ham plates.
So most people wouldn't think twice about seeing PA1HFO in gigantic huge bold print in the schematic title block.
-
crufty old heat producing design. get a meanwell switcher. itll do 13.2 volt 20 amp for years with 96 % efficiency.
The new Ham market targeted switching supplies are really nice. Lower noise and hash than they used to be.
-
"AN-1012 from IRF. It states that with proper mounting 0,2K/W (case to heatsink) is reachable with TO220. "
I find that very hard to believe.
My rule of thumb for to220 is 10w.
that is really far from the real capabilities of the TO220.
The design was all discrete but the parallel common emitter output pair lacked any emitter biasing so I suspect failure was caused by uneven current sharing and high temperature induced breakdown. Even worse in my opinion, the output transistors lacked base-emitter shunt resistance so the load regulation spike when the amplifier unkeyed required a TVS to protect the RF transistors. That may have resulted in the TVS shorting out thereby destroying the regulator. The current shunt resistor helpfully blew to protect the fuse after the transistors sacrificed themselves.
It is probalby thermal runaway. The transistor which got more heat became more conducting, generating even more heat, and so on. You can make this happen with uneven wiring, if you rely on the wiring resistance.
-
Fairly sophisticated design,if not construction.
Most El Cheapo "trickle" chargers are just a transformer,a bridge, a meter,& two terminals,& will give you enough charge for one start in about 1.5 hrs.
Your one was capable of a lot better charge rate,though,as is evident by 1hour to build & 1 hour to charge!
I built a cheapo version one day for the same reason TerraHertz did but just used a pair of big transformers each configured with a full wave rectified center tapped output and a variac on the input to adjust the output voltage. Charging time was irrelevant since the power supply could power the starter without the battery anyway although that would be hard on the car's charging system.
TO-220s vary. Just as you should expect for their design...
I've tested new Fairchild MOSFETs -- some small things, which blew when driven just past the expected dissipation due to thermal resistance (something like 50W into RthJC = 1 C/W, case greased to heatsink with no insulator, temp and power slowly turned up until failure).
I've also tested bog standard IRF740s, which went easily past twice their rated dissipation. If I assume rated RthJC, the die was over 210C when it failed; or if I assume it failed at rated temperature, RthJC was more like 0.5 C/W!
The IRF740 die was something like 2.5 x 3.2mm, easily double the Fairchild device of similar rating (presumably, they never die-shrank the cheap old silicon, whereas the new device is more-or-less latest process.
I ran across the same thing with power MOSFETs. Improved processes yielded higher voltage and current ratings for a given die size but power is mostly proportional to just the die size. Over time, the same voltage and current rating became available in a less expensive device (less expensive because the die was smaller) but the power rating went down with the smaller die. This was most apparent when looking for complementary pairs where the p-channel device with the same voltage and current rating as the n-channel device would have a power rating about 66% higher because of the larger die size needed. The larger device also had a larger gate charge requirement.
My conclusion (besides lateral MOSFETs being better for linear applications) was that complementary pairs of power MOSFETs should be matched by power rating and not current rating.
As for thermal performance of '3055s... why even bother? It's not even the McDonalds of transistors. Why would you ever buy one? McDonalds sandwiches are literally specified better, and required to use better materials, than the ancient 2N3055 standard requires. You have no idea what you're going to get. Buy some beefy TO-247s from On Semi and forget about anything else. ;)
Except for cost there is no reason. If you really want a TO-3 transistor because of mounting consideration then a 2N3773, MJ802, or 2N5885 is probably a better choice. Even without the cheap junk parts, there were several variations of the 2N3055 sold at the same time. My favorite example of this is Tektronix using 3 different 2N3055s in the same instrument simultaneously although they may have been grading to produce 2 of them.
For one off projects, I just look at what I have available in bulk. Besides a motley but large collection of various 2N3055s transistors, that includes IRF351, MJ11032, 7805, 7808, 2N5885, MJ802, MJ4502, 2N6545, 2N6053, 2N6055, and a tray of LM337s parts. The regulators make fine fault protected power pass elements in low current applications (like a cheap LM395) and the large package often allows forgoing a heat sink that would be required on a TO-220.
The design was all discrete but the parallel common emitter output pair lacked any emitter biasing so I suspect failure was caused by uneven current sharing and high temperature induced breakdown. Even worse in my opinion, the output transistors lacked base-emitter shunt resistance so the load regulation spike when the amplifier unkeyed required a TVS to protect the RF transistors. That may have resulted in the TVS shorting out thereby destroying the regulator. The current shunt resistor helpfully blew to protect the fuse after the transistors sacrificed themselves.
It is probalby thermal runaway. The transistor which got more heat became more conducting, generating even more heat, and so on. You can make this happen with uneven wiring, if you rely on the wiring resistance.
It was not entirely clear from the investigation what the initial failure was and I was just happy that the expensive power RF amplifier survived. The TVS shorting first would have caused immediate failure do to excessive current but thermal runaway under normal operating conditions could have shorted one transistor which shorted the TVS which shorted the other transistor. One transistor had the emitter bond wire blown off and the other had the base bond wire blown off which did not make a whole lot of sense given that they had base ballasting resistors which would have limited the base current.
My repairs included using a pair matched for Vbe and adding emitter ballast resistors.