Question on Power Supply Diodes

[ledtester]

I think it would be a good exercise to plot measurements of the transistor case temperature (at equilibrium) versus various output voltage and loads.

You can start small -- e.g. 16V at 100mA, 200mA, 300mA, 400mA and then move up to higher power dissipation, e.g. 12V and 8V. Record the transistor case temperature after it has plateaued -- this will probably take only a couple of minutes at the most.

The graph of temp vs. output current for each choice of output voltage should be a straight line and from the line you can determine the thermal resistance between the transistor case and ambient air -- i.e. the effectiveness of your heatsink.

[Electro Fan]

Thanks for all the analysis and info.

The temp varies somewhat depending on the specific spot being measured on the transistor or heat sink etc.  The numbers I gave are the hottest temps - generally (but not always) right on or neat the screw that holds the transistor to the heat sink.

[MK14]

The main issue (assuming you are below the transistor's maximum junction temperature), is the hotter the power transistor gets to, the shorter its life time. It is best to not run it too hot, otherwise, it will fail eventually.

As a rule of thumb, the transistor life, drops by a factor of 2, for every 10 deg C, junction temperature rise.

So, if a transistor had a life of (ultra approximate figure, as I don't know how long it would last, really, at 100 deg C) 10,000 hours at 100 deg C, it would be more like 5,000 hours at 110 deg C and 2,500 hours at 120 deg C. Etc. I'm referring to Junction (internal) temperatures.

They (power transistors) also tend to fail short circuit (but not always!), which can put the full output voltage onto your circuit, with potentially unpleasant results (or pleasant firework demonstration, for around 5 seconds, then unpleasantness  ).

[Electro Fan]

I think it would be a good exercise to plot measurements of the transistor case temperature (at equilibrium) versus various output voltage and loads.

You can start small -- e.g. 16V at 100mA, 200mA, 300mA, 400mA and then move up to higher power dissipation, e.g. 12V and 8V. Record the transistor case temperature after it has plateaued -- this will probably take only a couple of minutes at the most.

The graph of temp vs. output current for each choice of output voltage should be a straight line and from the line you can determine the thermal resistance between the transistor case and ambient air -- i.e. the effectiveness of your heatsink.

I’ve done some of this and so far the heat seems the greatest when DC voltage and current (into a 1 Ohm test load) reach a point that causes AC transformer current to rise past 2.5 Amps to 3.0 Amps and beyond.  Don’t know if it’s a straight line but based on what I’ve observed so far I can believe it might be.

One other observation is that as the transistor approaches the ~60 C max the Fluke IR temp shows more fluctuation than I noticed in the lower (40s C) temp (when DC was at 1.5 V and A into the 1 Ohm load).  I’m wondering if the several degree fluctuations around 60 C are reflecting the non-sinusoid AC ?  Just a wild guess.  Or maybe something else is starting to stress with some oscillation.

My interest in this is to learn more along the lines of what you guys describe/explain so well and to see if there is a way to raise the performance a step at a time by finding and upgrading the weakest link.

[Electro Fan]

The main issue (assuming you are below the transistor's maximum junction temperature), is the hotter the power transistor gets to, the shorter its life time. It is best to not run it too hot, otherwise, it will fail eventually.

As a rule of thumb, the transistor life, drops by a factor of 2, for every 10 deg C, junction temperature rise.

So, if a transistor had a life of (ultra approximate figure, as I don't know how long it would last, really, at 100 deg C) 10,000 hours at 100 deg C, it would be more like 5,000 hours at 110 deg C and 2,500 hours at 120 deg C. Etc. I'm referring to Junction (internal) temperatures.

They (power transistors) also tend to fail short circuit (but not always!), which can put the full output voltage onto your circuit, with potentially unpleasant results (or pleasant firework demonstration, for around 5 seconds, then unpleasantness  ).

Roger on heat and life span.  I’m trying to go easy on the PS while measuring and studying it.  It has run so well since it first turned on that I’m starting to develop an affinity for it - especially since this design had some skeptics - it’s kind of become an over performing under dog that you guys helped nurture and train up and I’d like to strengthen it further by finding and upgrading it’s weakest parts - so I would like to avoid fatiguing out it’s useful life.  It’s not my robot friend but it is kind of cute looking and sounding (with the fan). 

[Jwillis]

Congrats on your success .
One of the great things about this circuit is its really easy to switch out the power transistor to a higher power one without many problems. And because it uses dual rail power to the op amps you can change them to higher voltage ones directly with only moderate component changes in the circuit .
It's kind of fun to just experiment with and expand upon. Nice little learning kit.

[Electro Fan]

New finding, and another question.

A big part of the testing and learning process seems to be tracking down what causes heat.  Mostly I have been looking at temps in the PS at various voltage and current settings with various loads, but I've also been measuring load temps.

Along the way I noticed that when I use a 10 Ohm 100 Watt resistor as the load, at about 7 Volts and 700 mA DC (which is about 17.6V and 1.35 A AC), once everything warms up, the ~7 cm long resistor measures about 21C on one end of the resistor and 54 C on the other end of the resistor.  If I had thought about it I could potentially have imagined that the resistor would heat to ~50 C or so, but I don't think I would have guessed a ~33 C difference from one end of the resistor to the other end.

One question that comes to mind is should (for example) an 18 gauge solid copper wire heat to a lower temp than 18 gauge stranded wire for the same current put through both?   (I haven't found a table comparing solid vs stranded temps by AWG, or even a current carrying comparison table by AWG showing solid vs stranded.... in case anyone has a link.)  Thx

[ledtester]

but I don't think I would have guessed a ~33 C difference from one end of the resistor to the other end.

That suggests the resistive element is concentrated on one end rather than distributed evenly throughout.

I am very skeptical of the "100W" rating of those small power resistors. By comparison, here is a 120W 8 ohm power resistor made by Dale in the 70's:

[Jwillis]

One question that comes to mind is should (for example) an 18 gauge solid copper wire heat to a lower temp than 18 gauge stranded wire for the same current put through both?   (I haven't found a table comparing solid vs stranded temps by AWG, or even a current carrying comparison table by AWG showing solid vs stranded.... in case anyone has a link.)  Thx

https://www.engineeringtoolbox.com/wire-gauges-d_419.html

[Electro Fan]

One question that comes to mind is should (for example) an 18 gauge solid copper wire heat to a lower temp than 18 gauge stranded wire for the same current put through both?   (I haven't found a table comparing solid vs stranded temps by AWG, or even a current carrying comparison table by AWG showing solid vs stranded.... in case anyone has a link.)  Thx

https://www.engineeringtoolbox.com/wire-gauges-d_419.html

Thanks - that chart looks even better than what I was looking for....

I'm guessing that Single Core means solid wire and cores means "strands" of wire that add up to the AWG?

If that's the case the chart is pretty/very interesting.  For example, 16 AWG solid copper is rated for a load of 15 Amps but when it's 3 strands it drops to 10 Amps and to 7 Amps if it has 7-24 strands.  Do you think that's what it's saying, that cores = strands?

Either way, thanks for finding and posting the link.

[Electro Fan]

It looks like changing wires here or there could potentially help the power supply minimize some heat and improve current capacity but I'm thinking the bigger opportunity to improve performance might be in finding a better transformer to get more headroom with the AC current, and thereby enable the power supply to handle more DC current?  So far it seems that the best I can get on the DC side is about 1.5 Amps because if I turn up the DC voltage to get more DC current the AC current is hitting the 40 VAC transformer's spec of 2.5 Amps.

I'm not sure where the threshold is in the rest of the PS and I don't want to blow up the PS but maybe there is room to move upward toward 18-24 Volts and more current handling capacity, or maybe it might also be worth looking at a toroid transformer?

Any further suggestions/education on the transformer selection process?  Thx

Edit:  saw this link in another thread:

https://www.hammfg.com/electronics/transformers/power/266

Looks like pretty much everything listed that is north of 16V 2.5A (40 VA) is going to push the budget for this experiment, but I guess I'll have to deal with that after I figure out the preferred transformer spec and model.

[Jwillis]

If that's the case the chart is pretty/very interesting.  For example, 16 AWG solid copper is rated for a load of 15 Amps but when it's 3 strands it drops to 10 Amps and to 7 Amps if it has 7-24 strands.  Do you think that's what it's saying, that cores = strands?

Either way, thanks for finding and posting the link.

Yes. Strands = cores.  Those amperage ratings are for a 30^oC ambient , for PVC-insulated wire . I believe PVC softening point is around 85 to 100^oC . Other wire will have different  current ratings based on the insulation . The type of wire you use depends on application and design .  Datasheets for the type of wire you use will tell you the except able current ratings and temperature .

You take fewer core losses with a toroid than with other types of core . 5- 10%  where EI cores can be as high as 15% .

The current I in amps is equal to the apparent power S in volt-amps, divided by the voltage V in volts:
I_(A) = S_(VA) / V_(V)

For example if you need     3 amps at 40VAC on your secondary the core will need to be 120VA .  This also applies to the primary as well . A 120VA core at 230VAC will draw about 500mA  . Like volts current is proportional between primary and secondary .
 
Not sure where you live but here's another place you can look for toroid transformers :https://www.antekinc.com/transformers/
I have a few of these and they are built really solid .

[MK14]

I'm not sure where the threshold is in the rest of the PS and I don't want to blow up the PS but maybe there is room to move upward toward 18-24 Volts and more current handling capacity, or maybe it might also be worth looking at a toroid transformer?

Here is one a member did, which does 40 V at 4 amps.

https://www.eevblog.com/forum/beginners/success-prototype-40v-4amp-mc1466l-power-supply-thankyou-everyone-that-helped-!/

I'm pointing it out to you, to hopefully illustrate the work, and changes needed, to cope with higher capabilities.

Note well, that it has FOUR output transistors. That is because a single output transistor (like you have), can only handle a limited amount of power dissipation.

At a quick glance at the circuit, I just linked to, it seems reasonably straight forward (to make).

EDIT: Also, it seems to use 2 separate secondary windings/taps. Which allows for the much higher output voltages, without having to panic about a design which doesn't really want to handle too high an output voltage.

Analogy: You just bought a 5 watt output audio amplifier, and now want it to do 7 watts.
So, some people carefully help you to achieve this.

In 2 weeks , you now want it to go to 10 watts, output. Which turns out to be a very tricky modification and needs lots of calculations and changes.

What I am trying to say, is that you would be better off, buying (or getting the circuit for) a 50 or 100 watt power amplifier. Problem solved. Rather than trying to start with a 5 watt one, and increasing the voltage/current/power, watching what gets too hot, and trying to continually modify it, all the time.

What you have fundamentally, doesn't handle higher voltages, without a number of modifications. Also, the output current is also fundamentally limited, because it only has one output transistor.

Can you tweak it and get a bit more from it ?
Yes.

But really, I would suggest thinking about getting a kit/circuit or whatever, which can handle higher output voltages and currents, that you are happy with.

Opinions can vary. Some may think tweaking it, to get a bit more is worthwhile (theoretically, you could tweak it and tweak it, to very high voltages and currents). But, I don't really agree.
Better, to use a design which is designed for suitably high voltages and currents.

EDIT2: Or read through the various threads about your psu, and perform ALL the higher voltage modifications. There seem to be at least 2 or more. I suspect the higher voltage mods will rule out increasing the output current rating, as it will make the output transistor, significantly hotter.
The output transistor, is already getting rather hot, I'm not sure you would get much more from it, as it stands.

[Electro Fan]

If that's the case the chart is pretty/very interesting.  For example, 16 AWG solid copper is rated for a load of 15 Amps but when it's 3 strands it drops to 10 Amps and to 7 Amps if it has 7-24 strands.  Do you think that's what it's saying, that cores = strands?

Either way, thanks for finding and posting the link.

Yes. Strands = cores.  Those amperage ratings are for a 30^oC ambient , for PVC-insulated wire . I believe PVC softening point is around 85 to 100^oC . Other wire will have different  current ratings based on the insulation . The type of wire you use depends on application and design .  Datasheets for the type of wire you use will tell you the except able current ratings and temperature .

You take fewer core losses with a toroid than with other types of core . 5- 10%  where EI cores can be as high as 15% .

The current I in amps is equal to the apparent power S in volt-amps, divided by the voltage V in volts:
I_(A) = S_(VA) / V_(V)

For example if you need     3 amps at 40VAC on your secondary the core will need to be 120VA .  This also applies to the primary as well . A 120VA core at 230VAC will draw about 500mA  . Like volts current is proportional between primary and secondary .
 
Not sure where you live but here's another place you can look for toroid transformers :https://www.antekinc.com/transformers/
I have a few of these and they are built really solid .

Thanks for the confirmation on strands = cores, and thanks again for the chart.  All very informative and helpful.

Also, thanks for the transformer explanations with the math.

And thanks too for the antekinc link.  Those look like a very good selection of toroid transformers.

 

[Electro Fan]

I'm not sure where the threshold is in the rest of the PS and I don't want to blow up the PS but maybe there is room to move upward toward 18-24 Volts and more current handling capacity, or maybe it might also be worth looking at a toroid transformer?

Here is one a member did, which does 40 V at 4 amps.

https://www.eevblog.com/forum/beginners/success-prototype-40v-4amp-mc1466l-power-supply-thankyou-everyone-that-helped-!/

I'm pointing it out to you, to hopefully illustrate the work, and changes needed, to cope with higher capabilities.

Note well, that it has FOUR output transistors. That is because a single output transistor (like you have), can only handle a limited amount of power dissipation.

At a quick glance at the circuit, I just linked to, it seems reasonably straight forward (to make).

EDIT: Also, it seems to use 2 separate secondary windings/taps. Which allows for the much higher output voltages, without having to panic about a design which doesn't really want to handle too high an output voltage.

Analogy: You just bought a 5 watt output audio amplifier, and now want it to do 7 watts.
So, some people carefully help you to achieve this.

In 2 weeks , you now want it to go to 10 watts, output. Which turns out to be a very tricky modification and needs lots of calculations and changes.

What I am trying to say, is that you would be better off, buying (or getting the circuit for) a 50 or 100 watt power amplifier. Problem solved. Rather than trying to start with a 5 watt one, and increasing the voltage/current/power, watching what gets too hot, and trying to continually modify it, all the time.

What you have fundamentally, doesn't handle higher voltages, without a number of modifications. Also, the output current is also fundamentally limited, because it only has one output transistor.

Can you tweak it and get a bit more from it ?
Yes.

But really, I would suggest thinking about getting a kit/circuit or whatever, which can handle higher output voltages and currents, that you are happy with.

Opinions can vary. Some may think tweaking it, to get a bit more is worthwhile (theoretically, you could tweak it and tweak it, to very high voltages and currents). But, I don't really agree.
Better, to use a design which is designed for suitably high voltages and currents.

EDIT2: Or read through the various threads about your psu, and perform ALL the higher voltage modifications. There seem to be at least 2 or more. I suspect the higher voltage mods will rule out increasing the output current rating, as it will make the output transistor, significantly hotter.
The output transistor, is already getting rather hot, I'm not sure you would get much more from it, as it stands.

Hi MK14, as always you have the right info - just what I seem to be seeking over the next horizon is what you consistently identify. 

When you say "one member" if I followed the link correctly it looks like that might be a reference to non other than Jwillis.  The schematic he posted looks like it is indeed a next up from my present power supply project.  I probably couldn't have appreciated that design until I'd built and studied the present design but now I think I can see the beauty in the design he posted.

I think I'm going to twiddle a bit more with the current design - I'm sure there is more I can learn from it - and then I'm inclined to embark on a new PS project.

As long as I have two excellent power supply teachers here, what would you and Jwillis recommend for a circuit board for the design he posted?  I'm not ready to use a software tool to layout a PCB and send it out for manufacture, so what's the next best approach?

Thanks again!

Edit/Update:  looks like the lineage traces to here:
http://www.learningelectronics.net/VA3AVR/circ/ps4002/ps4002.html

[Jwillis]

As long as I have two excellent power supply teachers here, what would you and Jwillis recommend for a circuit board for the design he posted?  I'm not ready to use a software tool to layout a PCB and send it out for manufacture, so what's the next best approach?

Thanks again!

Edit/Update:  looks like the lineage traces to here:
http://www.learningelectronics.net/VA3AVR/circ/ps4002/ps4002.html

Unfortunately the MC1466L and MC1566L  has been discontinued and is getting very difficult to acquire. There are versions done with discrete components  both  through hole and SMD . You can find both versions with board designs online .

[Electro Fan]

These are not them?

https://www.ebay.com/sch/i.html?_nkw=MC1466L&_trksid=m5467.l1311

[MK14]

I think I'm going to twiddle a bit more with the current design - I'm sure there is more I can learn from it - and then I'm inclined to embark on a new PS project.

That sounds like a good idea, for sure. Learn all you can. 

By learning a lot more about linear power supplies. It has, and will more so (the more you tweak it), make you appreciate and understand what you really need/want, and why.

As regards PCBs. In some cases, the PCB design has already been made, and is available as files, to freely download. Which you can send to the PCB production houses, to make for you.
Some of the PCB production houses, even have designs, which you can just request, and they will make and send them to you.
But, those are things for later.

Although I agree the MC1466L's, are very difficult to get hold of, it can be solved, let me explain.

Many power supplies these days, have moved towards so called, switch mode power supplies (smps), rather than linear power supplies, like you have just built.

So, because linear power supplies are a lot less popular, than they use to be. I can understand why, many of the ICs, that made the job of making them, easier, are not made any more.

But, many popular chips, are no longer available. In fact many of the through hole components, especially ICs, are disappearing.

But you could kindly ask on the buy/sell forum here. Someone might have one or can tell you a reliable ebay seller, or something, who has one.

Warning: Re-the link, you just posted, while I made this post.
Many of the chips are fakes, and either won't work (fake), are poor quality copies, or salvaged second hand parts, that may not work properly.
Rarely, they are genuine/working parts. It is tricky, to get the working/genuine ones, but not impossible.
It is a long and complicated subject, how to solve that problem.

Or use another design, which has parts which are easier to obtain. Don't panic, anyway!

[Jwillis]

They have been discontinued since 2002 . If they look like they have brand new shiny plastic with really bright white labelling I would question authenticity . Also they came in a ceramic package . If they don't have a white band around the perimeter where the pins are then they are not authentic.
Not to say they won't work , just don't expect longevity or accuracy .
Also check for MC1566L

[MK14]

They have been discontinued since 2002 . If they look like they have brand new shiny plastic with really bright white labelling I would question authenticity . Also they came in a ceramic package . If they don't have a white band around the perimeter where the pins are then they are not authentic.
Not to say they won't work , just don't expect longevity or accuracy .
Also check for MC1566L

I've just had a quick look. The supplied photograph looks promising. But, sadly the supplied photograph, and the chips you actually receive, don't always match.

But, these seem to be in ceramic packages, and have the white line, you seemed to talk about.

https://www.ebay.co.uk/itm/1PCS-VOLTAGE-AND-CURRENT-REGULATOR-IC-MOTOROLA-CDIP-14-MC1566L/273461231717

EDIT: On reflection, the name in the photograph, doesn't match the ebay sellers name. So I would be very wary (probably best to avoid). Also some of the feedback, seems to talk about apparently fake chips, which is another bad sign.

[Electro Fan]

Ok, thanks for all the heads-up on clone and fake ICs.

Looks like a couple steps are finding the right regulator, and then finding a schematic that could be turned into a PCB.  Might be doable.

As for the regulator, it appears that the MC1446L and MC1566L are similar but the 15 handles a much larger temperature range and has an instantaneous input voltage max of 35V (vs 30V for the 14).  There are probably some other differences but I'm guessing that if either could be found the 15 would be preferable?

Apparently the white line around the regulator for both the 14 and the 15 is a requirement but not a guarantee for authenticity?  Not sure if what's in the photo is real or not.... but it looks like it has the white line.

Other questions on Power Supply Project 2:

It uses two transformers?

Also, in the article labeled "Precision Power Supply © 1986 Doug Bedrosian and 2010 Tony van Roon" it says the PS is good up to 40V and 2A.  But Jwillis' March 2018 PS hit 3.98A and in a more recent project in the same thread (just yesterday) another user (j2lr) apparently reached 5A using a MC1466L.  I'm guessing these are variants of the original 1986/2010 design? 

http://www.learningelectronics.net/VA3AVR/circ/ps4002/ps4002.html

https://www.eevblog.com/forum/beginners/success-prototype-40v-4amp-mc1466l-power-supply-thankyou-everyone-that-helped-!/

Back on the existing power supply (Power Supply Project 1) .... any of the Antek toroids trip your trigger the most?

https://www.antekinc.com/transformers/

[MK14]

I don't know how good a make they are (not experienced them), but assuming they are reasonably good.
This one catches my eye (don't worry, I'll be more careful with those sharp dangly wires next time    )

https://www.antekinc.com/as-2212-200va-12v-transformer/

For the following reasons:

The 100 watt ones look nice, but I can't see any 2 x 12 V ones (but there are circa 20V ones).

But at 200 watts, they do 2 x 12V.

A pair of 12 V secondary windings, means that you can switch between 12 V and 24 V, AC.
24V AC seems to be the limit, if you make the various higher voltage changes to your existing setup, so 2 x 12 V is good.

The switching means that you can halve (approx) the heat dissipation in the output transistor(s). The switching can either be through a switch, or eventually through an automated relay (harder to do). Which automatically switches between 1 or both secondaries, in series, making either 12 V or 24 V.

200 watts gives you plenty of current. Too much current capabilities is usually better than too little. It means you can consider multiple output transistors, for even higher output currents.

But I only had a quick look. Don't take that initial one as set in stone, there are other options.

Also, the reason the PSU design uses 2 transformers, is so that the voltage regulator control can use a lower voltage, without depending on the output voltage. Hence the output voltage stage (second transformer), can be very high, as there is no need for the regulator IC to handle the high voltage. E.g. 125 Volts. So called 'Floating regulator' (technique).

[MK14]

They do, do a 100 watt dual 12 V secondary. Their website, didn't easily find/show it, though.

Here:
https://www.antekinc.com/as-1212-100va-12v-transformer/

The choice 100 watt vs 200 watt, depends on your requirements with the eventual output currents.

Approximately, the 100 watt should give you around 2.5 Amps DC (maybe a little bit more), and hence 5 Amps for the 200 watt model. Both at 24 V AC (2 x 12 V).

In theory, you can actually do more complicated designs, whereby the pair of secondaries at 12 V, are actually paralleled up, at the lower voltage setting. Hence doubling the possible output current, compared to the high/full voltage range. But at this stage, it is probably best to not do that.
E.g. You'd have to double the current capabilities of lots of things, to handle so much current. So, it is not so easy or straight forward. The vast bulk of power supplies, DON'T bother to use this method of increasing the output current.

[Jwillis]

To work out a rough guess on transformer secondary voltage , take your desired output voltage of your PSU and divide by 1.414 . For example if you want 30V at the output  30/1.414 = 21.2  so I would choose a 22V transformer.  Same number works to get a estimate of the filtered voltage after the rectifier and smoothing caps .  Secondary voltage X 1.414 = Filtered Voltage .

Give some over head on the secondary current and voltage on your transformer . This would insure stability at full power.
Its been some time since I looked inside that project and remember now that the transformer I had was inadequate to get the full 4A . Just haven't got around to changing it . It was one I ripped out of an old UPS  . The second  transformer only needs to be 25VA to 30VA . 10VA might be to small.
Also when choosing the power transistor you need to work out the SOA (Safe Operating Area) based on the voltage at the collector and not the voltage at the emitter.  So if the voltage is 40V at the collector and your PSU is set for a maximum of 30V , the transistors SOA must be calculated at 40V not 30V . 

[mikerj]

One question that comes to mind is should (for example) an 18 gauge solid copper wire heat to a lower temp than 18 gauge stranded wire for the same current put through both?   (I haven't found a table comparing solid vs stranded temps by AWG, or even a current carrying comparison table by AWG showing solid vs stranded.... in case anyone has a link.)  Thx

https://www.engineeringtoolbox.com/wire-gauges-d_419.html

Please note that "multicore" in that web page does not mean "stranded".  It means multiple conductors within a cable e.g. like a three phase power cable may have four cores.  The more conductors you bunch together in a cable, the hotter they get for a given current which limits the maximum safe current.