last nights attempt at M51 with 1100D

[mindburner]

HI

Well here goes, I hope I have got it right

I tested the PSU with a load of a 12v/35W halogen downlight bulb

12v selected on the pot

I uncommoned all the bases of the transistors and measured them one by one to the emitter of the BD139

The only measurement I don't have is the output from each 0.1r resistor as they are commoned and a bit of mess to split, due to my hasty construction. But I can do this if necessary.

Interesting base output on Q4&5

Also V in from the rectifier is 33v not 24

24 was the AC V before the bridger rectifier

----------------------------------------------------------------------------

BD139

VQ1-C=33.00

VQ1-B=14.1

VQ1-E=13.40

2N3055's COLLECTOR

VQ2-C=33.0

VQ3-C=33.0

VQ4-C=32.9

VQ5-C=33.0

BASE

VQ2-B=13.91

VQ3-B=13.92

VQ4-B=13.83 TO 13.72 FALLS & GETS HOT

VQ5-B=SAME AS ABOVE

EMITTER

VQ2-E=12.95

VQ3-E=12.94

VQ4-E=12.96

VQ5-E=12.97

0R1 RESISTORS INPUT

R1=12.93

R2=12.93

R3=12.96

R4=12.95

47R RESISTORS

INPUT TO RESISTORS

VR5=12.87

VR6=12.87

OUTPUT FROM RESISTORS

VR5=12.20

VR6=12.19

FINAL OUTPUT

VOUT=12.12

[DrRobin]

Hi,

Thanks for all the measurements. Based on these all of the transistors are working and supplying current as follows:

Q2 = 0.6A

Q3 = 0.6A

Q4 = 0.9A

Q5 = 0.8A

Total = 2.9A Your 35W load is 3.19A, but that is within the tolerance of the load and the measurements/resistor values.

The power dissipated in each transistor is

Q2 = 12W

Q3 = 12W

Q4 = 18W

Q5 = 16W

So not perfect load balance, but you will probably never achieve that any way with this type of circuit.

I suspect the Q4 & Q5 are getting hot as the thermal efficiency of the transistor case to heatsink is not as good as it should be and then they are dissipating quite a bit more power than the other two.

Any chance you can post a picture of the back of the unit, from the outside, whilst I work out the heat dissipation please?

[DrRobin]

Hi,

The base voltages are fine drifting like that. Transistors change their characteristics when they get hot. Actually most tend to start drawing more current, which makes the hotter, which makes them draw more. It is called 'thermal run away'.

Usually it ends in a bang and sometimes the offending part comes flying out of the circuit white hot and shouting 'I am the faulty part'.

[Gina]

When making power supplies don't forget that a full wave rectifier and reservoir capacitor will give you around 1.4 times the RMS AC voltage.

[mindburner]

When making power supplies don't forget that a full wave rectifier and reservoir capacitor will give you around 1.4 times the RMS AC voltage.

Yes While making this PSU I have found out a lot of stuff and remembered things I had filed away deep in by brain somewhere

[mindburner]

hi

thanks for the info on the circuits. Is it just a case that this design of PSU is basically liable to burn up unles a huge heatsink is used?

here's some pics of the rear:

The 2 leftmost 2n's are the ones that get very toasty

[DrRobin]

Hi,

I can't believe it, I typed a long reply about temperature and heatsinks but it hasn't posted it. I will have a look on my desktop PC incase it cached a copy.

Basically, your four transistors are working correctly, but ther heatsink is not up the job. I reckon it is about 5 C/W and to make it work you will need 1 C/W for each 2N3055.

If I can't find my full reply, I will have to re-enter it.

[DrRobin]

Hi,

I don't have a cache of the posting, even though I copied it to the clipboard first.

The good news is that your circuit is working correctly and all four transistors are supply current, although not even, in terms of Q2 to Q5, they are 0.6A, 0.6A, 0.9A, 0.8A.

At 33V input and 12v output with 3A you are dissipating around 60W in the four transistors, i have subtracted the dissipation in the resistors. The thermal resistance of the 2n3055 is 1.5 C/W, the mica washer about 0.5 C/W and your heatsink is about 5 C/W, giving a total of 7 C/W.

At 60W dissipation your junction temperature is going to get up to 420 degrees over ambient, but the transistor burns out at 200 degrees.

The power dissipated in the transistors is higher for high currents and higher for lower voltage outputs as it is dependent on how much current flows through them and the voltage drop across them.

By the way, I would move your mains cable, it is PVC covered and that is only rated for 70 degrees C. If it melts your case will become live followed by you doing a high jump and I can't have that, not when I am looking to you to tell me how to take good pictures of the Messier objects!

You have one of a number of options:

1) if you don't need more than 12V output, disconnect one of the transformer outputs, this will reduce the head voltage from 33V to around 16V. You will still be able to get the same current and around 13 - 14 volts out, but dissipate far less power in the transistors.

With the lower head volts at 12V 10A you will dissipate 40W in the transistors, still too much for your heatsink but a lot closer, you could get 8A max, if your transistors were balanced perfectly.

At 5V you will need to reduce the load to around 3A to be in spec with your current heatsink.

Or you could change your heatsink. There isn't much you can do about the junction to case or case to heatsink thermal resistance so to make it work you need a heatsink that offers 1 C/W or less per transistor. At that with 33V input and 12V output you could get 2A per device, so 8A overall.

Unless you need more than 14V, I think I would do both, reduce the head voltage by removing one output coil on the transformer and fit 1 C/W heatsinks to each 2N3055.

You could of course look at force cooling, but the calculations are quite difficult, so I would just buy a heatsink and fan with a quoted thermal resistance (C/W). Sometimes thermal resistance is quoted in K/W, that is Kelvin and for the sake of this analysis is the same thing.

I haven't selected any part numbers for heatsinks as I don't know who you buy from, but if you paste some links here, I can check them out for you. Incidentally, a linear voltage regulator like an LM317 would have exactly the same problem. The only way to get round this is to use a switcher or a class D (PWM) output.

I once design a PWM PSU, worked very well and with the amount of smoothing caps you have, would probably work well for you too. Mine was built from discrete logic and op-amps, but there is probably a chip to do it now.

Sorry, I have lost the full explanation, it had quite a few calculations in it, but I didn't save it anywhere else and it's just going to take me too long to try and remember and retype it all.

[DrRobin]

Hi,

If you do only need 12V output, you can parallel up the output coils on the transformer to give double the current. If you parallel them straight out of the transformer you need to get the phase right, should be in phase, otherwise it shorts the transformer out and goes bang very quickly.

If you pu a full wave bridge rectifier on the output of each coil, then you can just parallel the two + and - outputs together. They will load share in a similar way to your transistors, not perfectly, but reasonably near.

[DrRobin]

Hi,

I don't have a cache of the posting, even though I copied it to the clipboard first.

The good news is that your circuit is working correctly and all four transistors are supply current, although not even, in terms of Q2 to Q5, they are 0.6A, 0.6A, 0.9A, 0.8A.

At 33V input and 12v output with 3A you are dissipating around 60W in the four transistors, i have subtracted the dissipation in the resistors. The thermal resistance of the 2n3055 is 1.5 C/W, the mica washer about 0.5 C/W and your heatsink is about 5 C/W, giving a total of 7 C/W.

At 60W dissipation your junction temperature is going to get up to 420 degrees over ambient, but the transistor burns out at 200 degrees.

The power dissipated in the transistors is higher for high currents and higher for lower voltage outputs as it is dependent on how much current flows through them and the voltage drop across them.

By the way, I would move your mains cable, it is PVC covered and that is only rated for 70 degrees C. If it melts your case will become live followed by you doing a high jump and I can't have that, not when I am looking to you to tell me how to take good pictures of the Messier objects!

You have one of a number of options:

1) if you don't need more than 12V output, disconnect one of the transformer outputs, this will reduce the head voltage from 33V to around 16V. You will still be able to get the same current and around 13 - 14 volts out, but dissipate far less power in the transistors.

With the lower head volts at 12V 10A you will dissipate 40W in the transistors, still too much for your heatsink but a lot closer, you could get 8A max, if your transistors were balanced perfectly.

At 5V you will need to reduce the load to around 3A to be in spec with your current heatsink.

Or you could change your heatsink. There isn't much you can do about the junction to case or case to heatsink thermal resistance so to make it work you need a heatsink that offers 1 C/W or less per transistor. At that with 33V input and 12V output you could get 2A per device, so 8A overall.

Unless you need more than 14V, I think I would do both, reduce the head voltage by removing one output coil on the transformer and fit 1 C/W heatsinks to each 2N3055.

You could of course look at force cooling, but the calculations are quite difficult, so I would just buy a heatsink and fan with a quoted thermal resistance (C/W). Sometimes thermal resistance is quoted in K/W, that is Kelvin and for the sake of this analysis is the same thing.

I haven't selected any part numbers for heatsinks as I don't know who you buy from, but if you paste some links here, I can check them out for you. Incidentally, a linear voltage regulator like an LM317 would have exactly the same problem. The only way to get round this is to use a switcher or a class D (PWM) output.

I once design a PWM PSU, worked very well and with the amount of smoothing caps you have, would probably work well for you too. Mine was built from discrete logic and op-amps, but there is probably a chip to do it now.

Sorry, I have lost the full explanation, it had quite a few calculations in it, but I didn't save it anywhere else and it's just going to take me too long to try and remember and retype it all.

[DrRobin]

Hi,

Somehow, my last post appears twice, something weired going on with my collection of PCs.

Anyway back to topic, I discovered last night how to use Image Analyzer correctly, most importantly how to stretch the 'shadow' and 'mid-tones' without over exposing the bright stars, so I did a quick post process of M51 and improved it again, I think. Now I can see more centre detail whilst still been able to see the detail in the spiral arms. New picture on the left, older one on the right.

I found out how to use this whilst messing around with an LED panel to produce a flat frame. Looks like I am set up now, scope and PHD sorted, Bahthinov mask made, flat light box ready (just needs mounting method), dew shield almost made, post processing sorted, clear skies (oh no, they are missing!).

[Gina]

Hi,

I can't believe it, I typed a long reply about temperature and heatsinks but it hasn't posted it. I will have a look on my desktop PC incase it cached a copy.

Basically, your four transistors are working correctly, but ther heatsink is not up the job. I reckon it is about 5 C/W and to make it work you will need 1 C/W for each 2N3055.

If I can't find my full reply, I will have to re-enter it.

I have that problem at times. Sometimes I save a copy as I'm writing the post but I always seem to forget when I write a really long one - drives me up the wall!!

[Gina]

Regarding power supplies, I'm using a switch mode supply in the form of an ex-PC PSU for several loads. The one I'm using gives 15A at 12v, 30A at 5v and several other voltages. For my mount I'm using a linear mode 13.8v 7A bench supply from Maplin. I'm planning to add an over-voltage crowbar and fuse to my mount to prevent blowing up the mount electronics in the rather unlikely event of the Maplin supply failing disastrously.

[DrRobin]

Hi,

If it was me, I would still float charge a decent sized battery, say about 35Ahr. That combination will never suddenly go off, go over voltage and will be super smooth.

You should of course fit a fuse in the mount power cable and possibly a main battery fuse as well, to protect the battery against short circuits.

[mindburner]

hi

Thats great info that the power supply is working ok. I think I will check RS for a decent heatsink. I do have a few lying around here for single To-3 transistors.

I will for sure change the mains input. It is just through the heatsink as I ran out of space in the build. It's not worth it though as my wife says my life insurance is not that good at the moment.

But I can take it through the side of the case and have a decent gromet and strain relief on order, plus the side is plastic so no shorts.

I think the option of taking off one of the 12v taps is a very good idea, plus upping the heatsink. I rarely use over 12v anyway and I think I was just being greedy having the 0-30+ volts, sort of just in case it was needed.

Nursing a bad white wine hangover today and going to see Prometheus movie tonight, so will make some changes to the power supply tomorrow just in case.

I will post up how I get on and again I owe you one for all the help you have given me in this build.

The pictures you have posted are superb. That M51 is excellent. Very nice colour and detail. The dust lanes in the 'tail' of m51 are some of the most detailed I have seen. Do you find you have to have tracking really accurate for the longer subs?

best regards

Alan

Hi,

I don't have a cache of the posting, even though I copied it to the clipboard first.

The good news is that your circuit is working correctly and all four transistors are supply current, although not even, in terms of Q2 to Q5, they are 0.6A, 0.6A, 0.9A, 0.8A.

At 33V input and 12v output with 3A you are dissipating around 60W in the four transistors, i have subtracted the dissipation in the resistors. The thermal resistance of the 2n3055 is 1.5 C/W, the mica washer about 0.5 C/W and your heatsink is about 5 C/W, giving a total of 7 C/W.

At 60W dissipation your junction temperature is going to get up to 420 degrees over ambient, but the transistor burns out at 200 degrees.

The power dissipated in the transistors is higher for high currents and higher for lower voltage outputs as it is dependent on how much current flows through them and the voltage drop across them.

By the way, I would move your mains cable, it is PVC covered and that is only rated for 70 degrees C. If it melts your case will become live followed by you doing a high jump and I can't have that, not when I am looking to you to tell me how to take good pictures of the Messier objects!

You have one of a number of options:

1) if you don't need more than 12V output, disconnect one of the transformer outputs, this will reduce the head voltage from 33V to around 16V. You will still be able to get the same current and around 13 - 14 volts out, but dissipate far less power in the transistors.

With the lower head volts at 12V 10A you will dissipate 40W in the transistors, still too much for your heatsink but a lot closer, you could get 8A max, if your transistors were balanced perfectly.

At 5V you will need to reduce the load to around 3A to be in spec with your current heatsink.

Or you could change your heatsink. There isn't much you can do about the junction to case or case to heatsink thermal resistance so to make it work you need a heatsink that offers 1 C/W or less per transistor. At that with 33V input and 12V output you could get 2A per device, so 8A overall.

Unless you need more than 14V, I think I would do both, reduce the head voltage by removing one output coil on the transformer and fit 1 C/W heatsinks to each 2N3055.

You could of course look at force cooling, but the calculations are quite difficult, so I would just buy a heatsink and fan with a quoted thermal resistance (C/W). Sometimes thermal resistance is quoted in K/W, that is Kelvin and for the sake of this analysis is the same thing.

I haven't selected any part numbers for heatsinks as I don't know who you buy from, but if you paste some links here, I can check them out for you. Incidentally, a linear voltage regulator like an LM317 would have exactly the same problem. The only way to get round this is to use a switcher or a class D (PWM) output.

I once design a PWM PSU, worked very well and with the amount of smoothing caps you have, would probably work well for you too. Mine was built from discrete logic and op-amps, but there is probably a chip to do it now.

Sorry, I have lost the full explanation, it had quite a few calculations in it, but I didn't save it anywhere else and it's just going to take me too long to try and remember and retype it all.

[mindburner]

I have that problem at times. Sometimes I save a copy as I'm writing the post but I always seem to forget when I write a really long one - drives me up the wall!!

same here. I am a CTRL A, CTRL C junkie now

just in case

every few lines

[DrRobin]

Hi Alan,

No problem with the help, you gave me enough hints and tips over M51 and the 250/300 dobs, so only too happy to help out. RS eh, we use them work and Farnell/Rapid Electronics.

If you are going for one heatsink then it will need to be 0.2 C/W or somewhere around that. RS stock heatsinks around this value, but they are quite expensive, so I would probably go with 1 C/W or less and fit one per.

I found this

http://www.ebay.co.uk/itm/Artic-Freezer-7-Pro-CPU-Cooler-Heatsink-Fan-AM3-AM2-939-754-1366-1156-775-/110889941522?pt=Computing_ComputerComponents_Fans_Heatsinks_SR&hash=item19d18e3a12

might not be the easiest thing to attach and might be a bit big for the back of your PSU, but nice all the same and might give you some ideas. It is certainly a bit cheaper.

Regarding your two secondary outputs, you could connect them through a big switch (or relay) so that you can use one coil for 14V or less and two coils for up to 28V. You will still get the same current out, regardless of which setting you are on, just watch out for low voltages and high currents.

Regarding that last picture of M51, it was the first time I have used PHD in anger. I couldn't get it to stay locked when I first tried and then it just seemed to work so I didn't fiddle with it.

I use an ST4 pulse guider (home made) so I have my Syncscan handset left plugged in. The Syncscan was left on tracking, I am guessing I should have turned tracking off and let PHD do all the work? I wonder if the two were fighting against each other?

Still a bit of a novice with all of this, so I think my M51 photo was more down to luck than design.

[KeithFarmer]

My most recent effort at M51.

SkyWatcher 200PDS Newtonian

HEQ5 Pro Synscan

Pentax K-x DSLR (12Mp)

10 subs + 2 darks 30s@ISO800

Processed in DeepSkyStacker and Photoshop PS5

Hateful isn't it.

[DrRobin]

Hi Keith,

That is nice for only 10 frames at 30s. Your stars are much sharper than mine, guess you have the tracking better sorted? When you enlarge it there is a lot of detail in the spiral arms.

[KeithFarmer]

Hi robin, by the time you zoom right into the galaxy it begins to show the slightly soft focus the optics have, even with the fine control and the bahtinov it just never reaches pinsharp focus.

[mindburner]

hi Robin

just an update on the PSU. I managed to find 2 large heatsinks and 2n4348 transistors already mounted to them. I decided to give them a try with the power supply. I have the 2 running with dual output from the transformer.

I was dawing about 4 amps and the heatsinks get very warm, they do not get really scalding. I recon with some active cooling, like a small quiet PC type fan or fans, they should be stable. This is just an experiment and unfortunately I do not know the heatsink spec. Theyare quite meaty. There is only 2 transistors now sharing the load. I may go to just 1 transformer output depending on the overall heat and if I can get cooling under control. I probably won't be drawing more than a couple of amps most of the time.

I also re routed the mains cable so that it comes in at the side through the plastic via a double grommet. Much safer and more secure.

Below is a picture of the heatsinks:

I will need to get them mounted on the rear of the PSU

[Gina]

Ah, that's much better Don't forget to allow an air gap at the bottom. For my 10A battery charger I used an aluminium diecast box (6"x4"x4" approx) with a number of strips of ali channel about 1/2"x1/2" and 1/2" apart on the sides. Gets a bit warm when giving nearly 10A charging a tractor battery but not too hot to keep hands on. You may find those heat sinks adequate without a fan.

[DrRobin]

Hi,

Looking good. I will gestimate the heatsink size and rerun the calcs later this morning. Your transistors will get hot with a temperature gradient across the heatsink. The important thing is that it doesn't get too hot to burn if touched (or shield it) and the junction temperature doesn't get too hot for the transistor.

You could mount the heat sinks on the top of your PSU, heat rises so that would work reasonably well.

[DrRobin]

Hi,

I would estimate that your heat sinks are about 1.2C/W, give or take a little bit. 2N4348 have the following

120W

10A

200C

1.46 C/W J-C

So pretty similar to a 2N3055.

Your test above 12V (?) at 4A, and 33V input, is 40W power dissipation per transistor. Your thermal resistance is about 2.2C/W, so your transistor junction would have got to about 90C over ambient if both transistors were load sharing correctly, well within spec. The heat sinks would have about 45C over ambient, probably too hot to leave your hand on it for too long, but not hot enough to burn?

Did you notice if both heat sinks felt about the same temperature? That would be a good indication that both were dissipating the same amount of power?

I reckon if you remove one of the secondaries from the transformer, you would be okay at 5V up to around 8A and 12V at 10A. Just monitor heat sink temperature, if it goes over 100C you are probably going to burn out the transistor, which will then cause the other one to go very quickly and probably the driver transistor.

[mindburner]

hi Robin & Gina

Yes it does seem to work quite well. I mounted the heatsinks on the rear panel just to see how they performed. I have a very nice typhoon PC fan that sports low RPM and noise. It keeps the heatsinks warm but not roasting. Even without, the heatsinks seem to do a good job of dissipating the heat. I can keep my finger on the transistors indefinitely.

Both heastsinks get equally warm, so the load seems to be shared pretty well.

Current limiting also works. YEA!

I have 2 LED's which borrowed from another circuit. These basically just get brighter the more voltage is applied. I must read up on what they actually do and if I have them wired correctly.

The digital V meter is powered from an isolated supply and calibrated using a small pot.

Thanks again for all the help. It;s been good to build the circuit and I have learnt quite a bit along the way.