2N3055 series pass questions.

[Diabolical Artificer]

, it’s good to have a sense of what 40 degrees feels like to a finger, or 60 degrees.

Indeed Al, with tranny's and tfmrs I go on , can I put my finger on it comfortably and leave it there and it's like that after an hour at full whack. Doesn't work with valves though : )

Not very technical but needs must when you have 3 tins full of old HS's and no temp device. Thanks Mark & Resto73, I will see if I can get a temp device in the future.

Bit of a puzzle why your original attempt at current sharing with emitter resistors didn't work ...

I think I had the Re's configured wrong. I had them between load and ground, I think they should have been between base and emitter. I looked online but could only find examples of parallel Q's in amps where Re goes to 0v or ground. No examples of current sharing in series pass config.

What puzzles me is how different devices in the same package can differ in current when they're all made more or less the same. What's even more mind boggling is the thiness of the junction wires going to a die, how the hell are they supposed to handle 25A in some cases, they're thinner than a fuse element. I asked about this once and it was explained by the length of bonding wires, that may be, just looks wrong. I also get that some Q's have bigger die's and thicker metal bases, still it's bl**dy amazing what Q's can do for their size.

I will definitely look at using SMPSU's and or buck convertors in the future Peter, my only reservation is their complexity. If a linear PSU develops a fault it's easy to fix, not so with a switching device. most folk seem to treat them as a single part, if they stop working, bin them and get another. This presents two problems - 1) if I fit one as part of an amp and it stops working after a year the chances of me finding the same SMPSU again is slim. 2) I hate this approach, as Aldous Huxley said "mending is better than ending"

One last observation, a PSU supplying heaters has a high inrush current,for this application I measured a peak of 5A +. I fitted a thermistor at one point, but the one I used, the only type I had got so hot it unsoldered itself. Am I right in thinking a linear supply can handle this better than a SMPSU? Does it present any significant problem to a PSU using a regulator IC on it's own, I'm thinking in terms of a reg IC's protection circuit.

Andy.

[Ed_Dinning]

Hi Andy, emitter resistors go between each emitter and load on series pass devices. Typically 0.1R
Also there were quite often base sharing resistors from the base drive to each individual base, perhaps 10R or so

Ed

[radiomobile]

A word of caution when choosing switch mode controllers. Many have short circuit protection which shuts down the output if there is an excessive current demand, and require the equipment to be switched off and back on again to re-start. Now a valve heater has a very low resistance when cold and draws a surge current many times its normal operating current when power is first applied causing the aforementioned shut-down. I found this to my cost when building my valve tester. The solution is simply to choose a smps which has adjustable current limiting as well as variable voltage output. They are surprisingly cheap at around £3

[dave cox]

Hi Andy, Ed's right, 0.1R should be fine in your situation. For more general cases you want the voltage drop across the emitter resistor to swamp variation in Vbe in the transistors. Usually a few hundred mV.

As an approximation, the Vbe voltage varies ...
about -2mV per C, so going down as the temperature goes up (-ve coefficient)
about +60mV per decade (x10) of current increase

The pretty horrific consequence is a +30C change in temperature would be equivalent to 10x the current. A series pair of transistors (no current sharing resistors) can result in 1 being 30C hotter and carrying 10x the current of the other! (then 60C hotter and 100x the current). Smoke will result

dc

[mhennessy]

Originally Posted by Diabolical Artificer

I think I had the Re's configured wrong. I had them between load and ground, I think they should have been between base and emitter. I looked online but could only find examples of parallel Q's in amps where Re goes to 0v or ground. No examples of current sharing in series pass config.

See attachment. It's from page 18 of that power supply PDF I sent you a few years back.

The second attachment shows the single pass transistor in the Thurlby PL310 PSU, and the 4 pass transistors in the 2A PL320. The current sharing resistors are 1 ohm each. They're arguably a bit higher than needed, but only dropping half a volt each, so no harm done...

Originally Posted by Diabolical Artificer

What puzzles me is how different devices in the same package can differ in current when they're all made more or less the same.

It's all about the die

The die in a BC548 - a 100mA small-signal transistor - is 0.35mm square, whereas a 2N3055 die measures 2.7mm square. And that's fairly modest for a power transistor - the MJ15003 is much bigger again, as the photos on Rod Elliot's article show. Obviously, you need a certain amount of area to handle the current, and the bigger the area, the lower the thermal resistance between the die and case.

I'm not an expert here, but honestly wouldn't be surprised if I was told that in some cases the die area was dictated by the thermal requirements as much as - if not more than - the current-handling requirements.

Originally Posted by Diabolical Artificer

I will definitely look at using SMPSU's and or buck convertors in the future Peter, my only reservation is their complexity. If a linear PSU develops a fault it's easy to fix, not so with a switching device. most folk seem to treat them as a single part, if they stop working, bin them and get another. This presents two problems - 1) if I fit one as part of an amp and it stops working after a year the chances of me finding the same SMPSU again is slim. 2) I hate this approach, as Aldous Huxley said "mending is better than ending"

Off-line SMPSUs are reasonably complex, but a DC-DC converter module isn't really - it's essentially just a chip, a diode and inductor, plus some caps and other passives. These parts can be changed if required, although given their cost, it does make sense to just buy a few extra spare modules if you're buying the Chinese ones from the usual places

Last time I needed a step-up boost converter I started looking at buying some ICs to do the job, but admit that I did just buy some of those modules. IIRC, I bought 5 for less than just one IC would have cost - which is crazy, but the reality at the time. Obviously for home projects that's fine - not really practical for production.

Originally Posted by Diabolical Artificer

One last observation, a PSU supplying heaters has a high inrush current,for this application I measured a peak of 5A +. I fitted a thermistor at one point, but the one I used, the only type I had got so hot it unsoldered itself. Am I right in thinking a linear supply can handle this better than a SMPSU? Does it present any significant problem to a PSU using a regulator IC on it's own, I'm thinking in terms of a reg IC's protection circuit.

Good questions. It depends on the specifics, but many DC-DC converters have soft-start built in. Though if you know what the inrush is, you might be able to use a converter that is beefy enough to cope with the increase. The datasheets should give details about soft-start and short-term current overload behaviour.

Personally, I've never been comfortable about relying on the over-current or over-temperature protection built in to linear regulator ICs. They are generally very reliable, but the failures I've encountered have been those that have run too hot or have been overloaded. It seems that some designers are happy to trust what the datasheets say, but I've always preferred to be more cautious.

Earlier I mentioned that the regulator IC dies can't run as hot as a simple transistor - only 125 degrees C for many. Contrast that with 200 degrees for a 2N3055. Also, the thermal resistance from junction to case is a massive 4 degrees per watt (though some sources say 2.3C/W) for the LM317K. Contrast that with the 2N3055 at 1.52C/W. For all of these reasons, I would always prefer a transistor over a beefy (and expensive) regulator IC.

Hope this is of interest,

Mark

[jamesperrett]

Originally Posted by Diabolical Artificer

If a linear PSU develops a fault it's easy to fix, not so with a switching device. most folk seem to treat them as a single part, if they stop working, bin them and get another. This presents two problems - 1) if I fit one as part of an amp and it stops working after a year the chances of me finding the same SMPSU again is slim.

I started designing gear with encapsulated DC-DC convertors in the mid 90's. While the exact part may no longer be available, I can still buy a pin compatible improved version that will be a drop in replacement for the original design.

Parts that I used for a design over 10 years ago are still available. The vast majority of the DC-DC convertors that I used were made by Traco with a few made by Vicor - not the cheapest but they seemed very reliable.

It may also be worth saying that some of the Vicor supplies are designed to be used in current limiting mode. They reduce their output voltage if the current goes above a certain threshold and restore the voltage when the current reduces.

[John_BS]

Originally Posted by merlinmaxwell

I bet there is some clever way of putting the cap after the rectifier so it doesn't need to be unpolarised!

Everyone missed your joke?
John

Everyone missed your joke?

Even me, please elucidate.

[John_BS]

Well, the problem with 'yer traditional capacitor, she don't pass DC.

And them four diodes, they's only pass DC!

J

[Herald1360]

They can pass a fair amount of AC too- the ripple current in the smoothing cap isn't DC.....

[Diabolical Artificer]

Thanks Mark, I didn't think to look there.

Andy.

[John_BS]

They can pass a fair amount of AC too- the ripple current in the smoothing cap isn't DC.....

True, but the term AC does imply a reversal of current. The smoothing cap sees AC as it's charged by the diodes and then has charge removed by the load. But the diodes are just supplying unipolar (but obviously varying) amounts of DC current.