Any reason why this wouldnt work?

[3nity]

I'm trying to put up with this diy circuit...
Using a 48VSPS to put out 24V to power some 24V preamps.
just before i etch it!!

I based this schem on this circuit...

thanks...

 

[Rochey]

this is basically a regulator, right?
if so, your going to dissipate 24V (voltage left in the transistors) x current.
Get ready for a heatsink folks.

That is... if the circuit works. My transistor theory is terrible.

I would be tempted to use a switch, followed by an LDO.
So, switch to 27V or something, then clean it up with an LDO.
http://power.ti.com suggests a TPS54160

/R

 

[3nity]

the idea is to make it as compact as possible...and easy to build...easy to etch and easy to get components...
So yeah basically a regulator...R1 its a divider 10K, R2 could be a 10k trimmer.
All caps are 47uf/63V.

Yeah Heatsinks are unavoidable there!

Any more comments?

 

[sahib]

Go for  good ole' 2N3055, it will run cooler.

Regulate the base drive voltage with a zener. Cut the connection between node 1 and 2. Connect the wiper (node 2) to the base. Connect a zener between the ground and node 1. Still a very crude circuit but better. Breadboard before you etch.

 

[PRR]

> to power some 24V preamps.

Does it have to be 24V exact?

Is your "48V" known to be a steady 48V, or is it unragulated?

What value is R1?

Why is R2 variable to zero? Do you ever want zero V out?

Is there some objection to a 25V Zener?

The way I see it:

The "48V" probably is regulated.

"some 24V preamps" is -probably- under 100mA. Therefore over 240 ohms.

The emitter impedance should be much-much less than 240 ohms. Pencil 24 ohms.

The emitter impedance is the impedance at the base divided by current gain.

Assume Darlington gain is over 1,000. Pencil 24K base resistance.

The approximate voltage ratio is 2:1, so two equal resistors of 48K each will work fairly well.

These resistors draw 24V/48K or 0.5mA, dissipate 0.012 Watts. There is no objection to taking more power. This will also reduce output impedance. Pencil two 10K resistors.

> R1 ... 10K

OK, you are ahead of me.

Effective base impedance is 5K, effective emiter impedance is <5 ohms, which is reasonably small for ~~240 ohm loads. A 0.1A load will cause <0.5V sag, which is good for a simple buffered divider.

The actual output voltage is 1.2V less due to Darlington Vbe drop. Two diodes would give excellent compensation. Or observing that 1.2V is 5% of 24V, you could select R2 10% higher than R1, plus a bit for current sag: here is where you trim. R1=10K, R2=10K fix plus 2K trimmer. That gives you roughly 25V-23V range, assuming perfect 48V.

Now short the output. (Don't tell me this never happens.) What current flows? While C2 holds, current is limited ONLY by Q1 self resistance, possibly 50 AMPS, and near-certain destruction. If it gets past that, the <5 ohm output impedance suggests >10A current, at 48V, is 500 WATTS dissipation, which will distress Q1 to death.

There are a bazillion ways to limit current.

But you also have (48V-24V)*0.1A = several Watts heat in Q1. You can heat-sink it easily.

But imagine a 200 ohm resistor in series with Q1 collector. 0.1A will force 20V across the resistor. Now there is 4V across Q1, it dissipates 0.4W, and needs no sink.

The resistor is dissipating 2 Watts. A 10W part is inexpensive, and perhaps easier to mount than a heatsink.

AND the maximum current to a short is 48V/200= 0.24A, well within Q1's rating. And while that is happening, Q1 voltage is nearly zero, dissipation is maybe 0.3W, still smoke-proof. The resistor is 11 Watts... a "10W" part will sit there and stink for many minutes. No instant-death, and if you ignore it, it will burn-open in minutes or days.

If you -know- your load, figure this resistor to sop-up most of the excess voltage. If uncertain, figure to sop-up some of the excess, to allow some leeway, yet limit "infinite" disasters to some slow-smoke situation.

> use a switch, followed by an LDO.

Good for your company. Possibly cool and clean. But "some 24V preamps" implies an era when power was pretty crude, yet sound was good. Maybe they sound better when powered better. Or maybe the onboard caps do their job: make power quality moot. And he's not concerned with cleanliness nor efficiency. I'd use 1970 power techniques with 1970 classic-amps.

Anyway, doesn't TI still make/market/license the "TIP" parts? That's a penny for TI.

> good ole' 2N3055, it will run cooler.

How? It still dissipates V*I. TIP122 is a reasonably large part, not as big as 2N3055, but plenty big for this chore. (Anyway I already eliminated the heatsink.)

And the '3055's much lower current gain will demand much more current in the base feed. The 10K penciled above, as a simple divider, has to be more like 500 ohms. A Zener won't need huge current for low impedance, true. But it still wants over 2mA divider current to ensure a '3055 can dump 100mA without starving the base.

 

[sahib]

> good ole' 2N3055, it will run cooler.

How? It still dissipates V*I. TIP122 is a reasonably large part, not as big as 2N3055, but plenty big for this chore. (Anyway I already eliminated the heatsink.)

And the '3055's much lower current gain will demand much more current in the base feed. The 10K penciled above, as a simple divider, has to be more like 500 ohms. A Zener won't need huge current for low impedance, true. But it still wants over 2mA divider current to ensure a '3055 can dump 100mA without starving the base.

Purely because of a mechanical point of view. It has more surface area for dissipating heat and a cheap part. You eliminated the heatsink at the expense of a large power resistor which is your choice but I am not a fan of big heating elements. And for a couple of hundred miliamps current draw it can still be mounted on the pcb. For slightly larger power dissipation; perhaps with a small heatsink made of a single sheet of aluminium, neatly folded right angle, which would not only provide a convenient heat transfer to the chassis but would also serve as a mounting bracket. Overall module would still be nice and dinky.

However, you are right on the lower current gain. I shouldn't have assumed that he could use a smaller transistor to drive it, I should have mentioned.

 

[3nity]

I breadboarded it last night and here are some observations:
Without heatsink the TIP gets stupidily hot it shuts down after 10 seconds...but thats with a load in there (1 channel of Fetboy)

WIth a normal TO-220 heatsink..it got very warm but never shutdown...after a 15 minutes trial it all was fine only the voltage drop around 23V instead of 24V.
But on the audio zone it performed very well and never heard oscillations with the psu.

@ PRR: the 48V its a DC switching power supply...sometimes goes to 49V buts its normal with all the psu's i've had..
And yeah i could use a 25V Zener instead of the trimmer no problem at all.
2 preamps drawning 80ma each ..so you sugest a 200R/10 Watts resistor in series before the collector? this will avoid the need for a heatsink?
thanks.

 

[JohnRoberts]

It is not very planet friendly to throw away as much power as you are using.

It might be an interesting learning exercise to hack into the switching regulator at tweak it to put out the voltage you want. Could be as simple as changing a couple resistor values.

JR

 

[tv]

http://www.national.com/mpf/LM/LM2575HV.html

???

 

[3nity]

As i already said i want something easy to build with easy to get parts...
the 48V at the input will serve as Phantom at the same time..

thanks

 

[Gus]

IMO you are doing it backwards.

What supply section has the highest current draw and what supply section has the highest power draw?

This is the drawing board and you should be using ohms law and calculating DC power used

It might be better to use a X2 on the 24VDC supply for the 48VDC

Rochey post is good idea if you decide to stay with the 48VDC down to 24VDC.

 

[Rochey]

i don't think part availability is an issue any more. Most TI parts are available from Mouser without any major costs.
PCB's are cheap enough to do in single unit runs (www.batchpcb.com)

You could knock this design out in a fortnight for under $20

 

[3nity]

Gus you mean using 24V Switching psu? and double the 24V to put out 48V?
Yeah you're right the 24V needs the highest current!

Yeah Rochey idea is just nice but all the parts you sugest as well as the national parts are SMD...

I'm not plungin into smd right now...

Thanks you made a point here..

Thanks

 

[3nity]

Coming back to the doubler for the 48V from a 24V.
I remember reading that its impossible to double a DC voltage with diodes..

Thanks.

 

[JohnRoberts]

Coming back to the doubler for the 48V from a 24V.
I remember reading that its impossible to double a DC voltage with diodes..

Thanks.

True.. but if you have a switcher, one trick that I have used successfully is to make the low current phantom doubler/tripler from the HF PS switching waveform instead of from mains frequency AC. I made a phantom supply with relatively small value SMT caps. IIRC I used one electrolytic cap for the final reservoir. 

Of course if you already have the 48V supply, that link to a switching stepdown regulator looked pretty simple, while you will need to manage noise so layout will matter.

JR

 

[Rochey]

Yeah Rochey idea is just nice but all the parts you sugest as well as the national parts are SMD...
I'm not plungin into smd right now...

Okay, use National. I wasn't trying to promote a TI solution. I was merely trying to help you get the solution you needed without burning too many watts, and needing a heatsink.

There's no need to feel unsure about SMD... Until february, I'd never worked with SMD myself (i'm a marketing guy... i can tell you what you could use, but wasn't really capable of doing it. 

I rolled my sleeves up after we got some pcb's back, and I found it actually faster to solder SMD's in many cases than it is to do thu-hole. 0805 resistors are easy peasy, even without magnifying goggles.

Regardless of what solution you choose, I wish you the best of luck... please post pics

Cheers

/R

 

[3nity]

Thanks Rochey i wasnt trying to beat up TI neither praising National..
I was a little scared on the SMD parts i can barely solder trough hole components..

Thanks and i;ll post some updates as soon as i get something done.

Cheers.

 

[tv]

That's why I linked to that Nat's part - it actually comes in 5-pin TO-220 and you can get a free sample of "ADJ" part ... LM2575HVT-ADJ ... scroll down

 

[PRR]

> you should be using ohms law and calculating DC power used

Is that legal? Moral ? ? ?

> 2 preamps drawning 80ma each ..so you sugest a 200R/10 Watts

How much is "2 ... 80mA"? 160mA or 0.16A.

What is 0.16A times 200 ohms? 32 Volts.

What is 48 Volts minus 32 Volts? I leave the bead-slinging to you, but I bet it is less than you want.

If you just can't math, try it. (Although beware the consequences of under-volting things. Your preamps will probably survive, but I once roasted a fan-motor with under voltage. Think before you smoke.)

> 2N3055 has more surface area for dissipating heat

"Somewatt". I'd put 1+W in a naked TO-3, 0.7W in a naked TO-220. I would not be happy about either one, even though the spec-sheets say these devices can run very hot. And a TO-3 is more work to mount. Either TO-3 or TO-220 can sink to PCB or scrap metal. And if the sink-to-air interface is small, the device-to-sink area hardly matters: a bad solid-solid joint is far better than any dissipation to air.

And looking at the specific job: asking ~~4W from a TO-3, naked or on-PCB, may be legal but it isn't moral to my thinking.

I'd call this very secondary to Gain and Availability. A bird in the hand beats a shipping charge.

> I am not a fan of big heating elements.

The plan is going to throw heat. Doesn't really matter what part the heat comes out of, still the same total heat. Resistors are made to do that, cheap. Transistors can, and the differential cost for DIY is nil, true.

But transistors have other chores, are built small, and very-very often fail SHORT. In this case that puts uncontrolled 48V on his 24V preamps.

Anyway the Short-Proof is worth the extra part. There are great ways to "short-proof" transistors, but none so simple, nor so reliable. (And at 3nity's apparent level of experience, simple and short-proof is best, IMHO.)

> available from Mouser ... PCB's are cheap enough ... You could knock this design out in a fortnight for under $20

As your own colleagues will tell you: it is tough to buck decades of momentum. I think I can still get 5K resistors and TIP-120 at Radio Shack, NOW. And I know I have that stuff in a box somewhere. And -I- would wire it dead-bug or term-strip... it's 4 parts, 9 legs, whadda I need a PCB for?

> It is not very planet friendly to throw away as much power as you are using.

Yes, agree.

But what is the waste? Around 4 Watts. A nightlight. I'm thrifty, stingy, but I got 87 Watts of lights on right now which I am not using. 4W is one of the smaller wastes of power-mad modern life.

If he runs this 100 hours a year for a decade, his future cost for the wasted energy is about a half a buck.

Shipping from Malaysia to Mouser to Canada, or driving 30 miles to the Radio Shack, is easily ten times that much energy.

And we really could ask HTH a mere preamp can be pulling 2 Watts.

I can do a preamp under 0.1W. One of my sweetest was just raw 36V filtered to 30V, which fed 4mA of chip and 4mA of Phantom ("most" good Phantom mikes are fine with 30V). And that is for stereo. <0.1W per preamp channel, and each mike took as much as the preamp.

Not that I have always been so thrifty. I had a 286 PC powered by a LINEAR 100W power supply. And a very generous job it was. 12"x5" finned heatsink, about 10 pounds of iron. Wasn't even 50% efficient (but it would tolerate huge wall-sag). Sitting in the open in a wooden trough it didn't make a bit of interference to radio or audio.

> impossible to double a DC voltage with diodes..

Anything is "possible". Convert DC to AC. Multiply AC. Convert AC to DC. Can't get there from here? Go somewhere else, you'll find a way.

John and Rochey are trying to tell you that stuff like this (general switcher techniques) is easy. It is... for them.

And John has a specific point: the "48V" has a reference divider, and this can be gimmicked for a different output. I diddle 3.3V switchers to 4.2V and 2.6V a lot. But finding the diddle can be a head-scratch, and even with parts-change there is some limit to how far it can go and stay happy. And you probably want 48V for something. Can't get two voltages at once from one output.

And Rochey's specific point is valid: the chip-makers make it "EASY" to enter your requirement (assuming you can add 80mA+80mA) and get back a picture and a parts-list you can drop at Mouser or DigiKey. My old 286 linear supply was linear because those computations were brain-busters in The Old Days. We live in wonderful times. And the several-part switcher will have far less waste heat, and therefore be smaller, which generally means cheaper. At least if you make more than one and don't have various spare parts on hand.

 

[Rochey]

PRR

A good reply as always. (And a pleasure to read)
Your point regarding the energy costs to transport the parts versus energy dissipated in use is an angle I hadn't previously considered.

It's a 'gud one

/R