Switch mode PSUs question

[MrAl]

Very scholarly I'm sure, but of little relevance to the thread.

We're not discussing two separate power supplies, but rather two power supplies combined to make one split power supply, so they shouldn't be an option to switch only one of them on.

Obviously if you used two entirely separate supplies with separate mains leads then you're asking for problems, but that would be a very foolish thing to do.

Hello there,

I usually dont get as far off track as you seem to be suggesting

Try reading post #9, paragraph 3: "Why consider this case?"

 

[alec_t]

However, turn only ONE of those supplies on first and leave the second one off, and we could see -12v across the second supply when it should be +12v.

I still don't see how (unless assumptions are made about the supply internals). Nor can LTspice .

 

[Nigel Goodwin]

Re.#9..I assumed MrAl was talking about a momentary situation that might arise if the PSUs did not spark up at the same instant...Of course they would all be powered by a common mains switch in any scheme I would try...

'instant' wouldn't be a problem at all anyway, and identical supplies will perform pretty identically in any case.

 

[MrAl]

I still don't see how (unless assumptions are made about the supply internals). Nor can LTspice .
View attachment 113374

Hi,

You are assuming that the two supplies are ideal sources. Obviously that cant be true.

Power supplies are not ideal sources, and many of them have NPN pass transistors that have the output emitter connected to the positive output terminal (or nearly so). That means that a negative voltage (from the other power supply that starts up first) may force the NPN to conduct emitter to base, and so we may see a negative voltage at the base too. That may or may not work.
For power supplies with PNP output stages, it could mean the supply looks totally open. That would put a negative voltage across the output caps.

So what we have here is an indeterminate system where we cant be sure what will happen.

Just for the heck of it you can try to simulate two LM317 voltage regulators and tie them in series, then add a load, then start one up before the other, see what happens. You could also try one LM317 and the negative version of that IC in a split supply hookup with load from +12 to =12 (not to ground) and see what happens.
They may or may not work.
You can also try the reverse diodes, and see if they prevent latchup.

 

[JonSea]

So if what MrAL is saying is a worry and the power supplies connected to the same mains power switch for whatever reason don't power up at the same instant, there's a simple solution:

Have one switch to power up the supplies, and a second switch to connect the load after the supplies have a monent to stabilize. If the load switch is open, there's no way one of the supplies can reverse-polarize another.

 

[Nigel Goodwin]

Hi,

You are assuming that the two supplies are ideal sources. Obviously that cant be true.

And you're assuming linear analogue supplies, which almost certainly won't be the case. Why do you seem determined to mess the thread up by posting irrelevant rubbish?.

 

[MrAl]

So if what MrAL is saying is a worry and the power supplies connected to the same mains power switch for whatever reason don't power up at the same instant, there's a simple solution:

Have one switch to power up the supplies, and a second switch to connect the load after the supplies have a monent to stabilize. If the load switch is open, there's no way one of the supplies can reverse-polarize another.

Hi,

I actually suggested that previously

 

[MrAl]

And you're assuming linear analogue supplies, which almost certainly won't be the case. Why do you seem determined to mess the thread up by posting irrelevant rubbish?.

Hello,

Quite judgemental on your part which is not echoed by other postees.
I had to work in the power supply industry for many years so i've seen a number of different designs. They vary, and when we dont know the configuration we assume the worst. You are free to assume any config you wish, and i am free to assume the worst case. That's a large part of engineering, recognizing the worst case scenario and taking precautions.

 

[alec_t]

You are assuming that the two supplies are ideal sources. Obviously that cant be true.

That's why I was seeking clarification as to what assumptions you were making about the internals.

 

[MrAl]

That's why I was seeking clarification as to what assumptions you were making about the internals.

Hi,

Ok that's fine, no problem here. But i did want to make it clear that i think the most important thing about this question is that we dont know what power supply circuit we are dealing with so i assume the worst case.

If you like you can try a few simulations with different regulator circuits and see what turns up. You may find hat some can work ok and others can not. In one of my Mastech 30v 10amp supplies there is already a reverse diode across the output, in the other there is a fuse in series with that because i upgraded it with a fuse too so the fuse would blow. That kind of power supply will take a very brief reverse voltage because of that diode, but the diode is not rated for full current so if it was left too long like that it would blow the diode. Dont bother looking on the schematic for this diode either because it's not on the original schematic, and it is also hidden just behind the front panel which is hard to see without taking considerable stuff apart. I think i have a pic somewhere though of that diode and fuse mod too, i'll see if i can find it.

 

[unclejed613]

I'm putting together an amp for our band on the cheap and have bought a 100watt/2n3055/MJ2955 amp module off eBay.

for 100W into 8 ohms, power supply rails will need to be +/-60V@8A

 

[Nigel Goodwin]

for 100W into 8 ohms, power supply rails will need to be +/-60V@8A

It's almost certainly 100W to 4 ohms (not 8 ohms), but your supposed requirements are far too high even for 8 ohms - 120V at 8A is a PSU capacity of 960W, rather excessive for a 100W amplifier. 120V would be for 200W into 8 ohms, or 400W into 4 ohms. However, the +/-24V he suggests is too low for 100W - which was why I mentioned that the 24V modules I linked to are easily altered to give more output.

For 100W to 4 ohms you need about 60V, so a PSU 0f about 65-70V to allow a little loss, and around 2-2.5A.

 

[cornishlad]

The replies received have all been thought provoking and useful..One result is that I have spent a while reading up on SMPS - and DC>DC converters. It seems the latter are generally never isolated while the former cannot guanteed to be...The Xbox ones probably are, but while the galvanic isolation of a traditional mains transformer is usually assumed to be beyond reproach, maybe the tiny transformer in an SWPS may be specified differently..depending on the nature of the equipment it was originally intended to supply..ie perhaps largely an all plastic case with only manufacturer supplied plugins.
A music amp on the othe hand MUST be beyond reproach in that regard, with guitars and hand held mics all possibly being connected..
All that aside, I have decided to order two of the units linked to by Nigel...they are very cheap and I will use an RCD plug top just in case.
The more I thought about what I was proposing, the more Ithought it was a bit impractical to have a bunch of power bricks with all the wiring complications that would be involved. Some time I will find a different use for 6x 12a power bricks - perhaps parallel them and make a spot welder
As to the amp module itself....it will replace a similar looking job in a powered cab which is supposedly rated at 65watt...but the mains transformer looks too small to me..and the 15" loudspeaker init appears to be 4 ohm..( I'll see how that goes) The modules are specified as 100 w into 8 ohms and the designer/seller specified a 28-0-28 transformer and 4700uf per rail. noting on the diagram provided : 40V. This will be have poor stabisation of course. I too fail to see how 60-0-60 is arrived at..BUT althoughI didn't say, I'm only looking to get somewhere around 50 watts from it hence thinking of 24-024v as a reasonable substitute.

 

[Nigel Goodwin]

The replies received have all been thought provoking and useful..One result is that I have spent a while reading up on SMPS - and DC>DC converters. It seems the latter are generally never isolated while the former cannot guanteed to be...The Xbox ones probably are, but while the galvanic isolation of a traditional mains transformer is usually assumed to be beyond reproach, maybe the tiny transformer in an SWPS may be specified differently

I don't think reading up is helping you

SMPSU's are used in almost everything these days, and are almost always isolated - probably just as much so as conventional mains transformers (which are available isolated or not as well).

In either case (SMPSU or conventional), and for your specific application, you simply connect the mains ground to the chassis of the amplifier, and to the common (0V of the 24-0-24 output).

SM transformer are smaller because they operate at a higher frequency, not because they are less isolated - and both almost certainly have to meet the same legislation.

..depending on the nature of the equipment it was originally intended to supply..ie perhaps largely an all plastic case with only manufacturer supplied plugins.

Such items as that are 'double insulated', as most domestic electronics is these days, these means the actual PSU has to provide BETTER insulation than a class one (earthed) device, adding an earth to the modules I suggested gives double the protection for that reason - and certainly any guitar or PA amps MUST!!!!! be earthed.

As to the amp module itself....it will replace a similar looking job in a powered cab which is supposedly rated at 65watt...but the mains transformer looks too small to me..and the 15" loudspeaker init appears to be 4 ohm..( I'll see how that goes) The modules are specified as 100 w into 8 ohms and the designer/seller specified a 28-0-28 transformer and 4700uf per rail. noting on the diagram provided : 40V. This will be have poor stabisation of course. I too fail to see how 60-0-60 is arrived at..BUT althoughI didn't say, I'm only looking to get somewhere around 50 watts from it hence thinking of 24-024v as a reasonable substitute.

You mentioned that it used 2N3055's, these aren't capable of 100W to 8 ohms, only to 4 ohms - they have too low a voltage rating for 8 ohms.

Maximum power output is trivial to calculate from the power supply:

Maximum voltage swing can only be from rail to rail (or double that for bridged amps), so assuming your 28-0-28 transformer provides 30V+30V (60V) at maximum load then it can't provide more than 60V p-p.

Now convert that to RMS - 60 / 2.828 = 21.22V RMS - round that to 20 (I like simple!).

Now apply the formula to calculate power W = V*V/R - 20*20 is 400 - divide that by the speaker impedance, which gives 100W for 4 ohms, or 50W for 8 ohms (or 25W for 16 ohms).

Crude and nasty calculations, but they are 'in the ball park'.

You mention power stabilisation, that's not really a concern for an amplifier - and the PSU voltage dropping can even make lower voltage transistors survive, as it might drop low enough to save them under high drive. The SM modules I linked are of course stabilised - and if you want are easily modified to increase there voltage.

 

[unclejed613]

It's almost certainly 100W to 4 ohms (not 8 ohms), but your supposed requirements are far too high even for 8 ohms - 120V at 8A is a PSU capacity of 960W, rather excessive for a 100W amplifier. 120V would be for 200W into 8 ohms, or 400W into 4 ohms. However, the +/-24V he suggests is too low for 100W - which was why I mentioned that the 24V modules I linked to are easily altered to give more output.

For 100W to 4 ohms you need about 60V, so a PSU 0f about 65-70V to allow a little loss, and around 2-2.5A.

for 9 out of the last 10 years i was working at a service center for a large retailer, and most of the manufacturers of home theater receivers are running power supplies of +/-55-60V on amplifiers rated at 100W/ch. the majority of them usually read around 57 to 58V. keep in mind that the 100W/ch figure is with 2 or more channels driven, and there is going to be some sag in the rail voltages. these manufacturers are continuing to use "heavy iron" in their power supplies.

 

[cornishlad]

Do you ever get the feeling that you don't know whether you've (me) said too much or too little in a post !!

QUOTE="Nigel Goodwin, post: 1329648, member: 10131"]
You mentioned that it used 2N3055's, these aren't capable of 100W to 8 ohms, only to 4 ohms - they have too low a voltage rating for 8 ohms...........> power stabilisation, that's not really a concern for an amplifier - and the PSU voltage dropping can even make lower voltage transistors survive, as it might drop low enough to save them under high drive. [/QUOTE]
The amp module is listed on eBay with the title "100 watts RMS power audio amplifier module 2n3055/MJ2955 retro Maplin derivative " RMS stated...not even the dreaded "music power" thing.
Using the same trivial calcs...if the 28/0/28 transformer could hold up to 40/0/40 then the answer does come out as 100 watts into 8 ohms...just saying... The builder / designer/ seller would need to comment on his claims.
In all my thoughts on this I've been super cautious..it's my friends who will be using this, not me.
The earthing arrangements in random pubs and village halls can not neccessarily be relied on to be low impedance - or even there. Tripping the buildings RCD is also embarrasing.
The Xbox bricks are quite probably perfectly well enough isolated but they have no earth connection, so clearly were not used in equipment that had a earthed chassis. As I'm now not going to use them that won't be an issue either

SM transformer are smaller because they operate at a higher frequency, not because they are less isolated - and both almost certainly have to meet the same legislation.

I do understand that but there could be an issue here as to what/whose legislation or standards the C****** manufactures of products like this are working to. But I am going to get a couple anyway...


We have used commercial amps which of course now have SWMPS with an earth and not doubted their safety. ..(but our only eqipment failure at a gig was a SMPS...not that that's relevant in this context.)
unclejeg posted while I was writing...thanks

 

[Nigel Goodwin]

for 9 out of the last 10 years i was working at a service center for a large retailer, and most of the manufacturers of home theater receivers are running power supplies of +/-55-60V on amplifiers rated at 100W/ch. the majority of them usually read around 57 to 58V. keep in mind that the 100W/ch figure is with 2 or more channels driven, and there is going to be some sag in the rail voltages. these manufacturers are continuing to use "heavy iron" in their power supplies.

I've been working in a service centre for over 46 years, but not any more

The subject under discussion isn't home theatre systems (which will be specced for 8 ohms) but guitar/PA amps which are normally specced for 4 ohms - the HT system will also probably have a much greater supply margin as distortion needs to be kept low, and the supplies will sag a lot more as there's potentially 8 amps fed from it (7.1).