Psu Multi Voltages And Doubt About

[r2d2]

Hi to all ,

checking an ac to dc psu of a friend ,
for caps replacement etc..

and there is a connection that I find for the first time ,
please check attached image,

the psu have 2 +18v sections
and 1 +48v for phantom

the "strange thing" is that the +18v out of the bottom section
is connected together with the :
negative / 0V / gnd , of the other +18v section in the middle ,
and the negative / 0V / gnd , of the 48v,

there is a specific reason about this connection or it is an error ?

any help-info about is welcome,

thanks.

 

[Ricardus]

Image?

 

[r2d2]

Image?

hi "Ricardus"
now attached
sorry : browser "drunk"....

 

[Rob Flinn]

This is a way to make a bipolar + & - 18v psu. It is 2x 18v supplies one stacked on top of the other. The point where the +18v of one psu is connected to the 0v of the other is considered to be the Audio common or 0v. The 0v of the psu with it's +18v connected to Audio common is now becomes -18v with respect to Audio Common & the other supplies +18v is positive 18v with respect to Audio common.

 

[r2d2]

This is a way to make a bipolar + & - 18v psu. It is 2x 18v supplies one stacked on top of the other. The point where the +18v of one psu is connected to the 0v of the other is considered to be the Audio common or 0v. The 0v of the psu with it's +18v connected to Audio common is now becomes -18v with respect to Audio Common & the other supplies +18v is positive 18v with respect to Audio common.

Hi Rob Flinn and thanks for the post ,
so this would be an "unconventional" way to obtain a negative voltage from a positive ?
is still reliable ?

more thanks.

 

[Brian Roth]

Hi Rob Flinn and thanks for the post ,
so this would be an "unconventional" way to obtain a negative voltage from a positive ?
is still reliable ?

more thanks.

It is a VERY common method used for eons. I've maintained some very large Amek desks that used multiple, single-ended power modules that were "stacked" to create bipolar voltages.

Bri

 

[r2d2]

It is a VERY common method used for eons. I've maintained some very large Amek desks that used multiple, single-ended power modules that were "stacked" to create bipolar voltages.

Bri

Hi Bri and thanks for the post,
so it is a "different" way to obtain a negative voltage ,

sorry for the insistence ,
but it is reliable as a dedicated psu negative section,
or if possible better to have the section dedicated ?
thanks.

 

[Rob Flinn]

It is very normal to do it this way. Some Amek consoles do the power this way for example the Einstein.

 

[Brian Roth]

I am unaware of any downsides.

Bri

 

[Matt Syson]

There are no particular downsides although you have to be careful not to 'reverse' the supply which can happen if ONE of the two 18 Volt rails comes up before the other which in some cases can prevent the opposite rail coming up and even possible damage to the supplies. This is usually prevented by having a decently powerfully rated diode connected 'across' each supply unit. This was a well known phenomenon when the LM7815/7915 fixed regulators (and all of their related families) were introduced and you will see in their data sheets the recommendation that 'reverse connected' diodes are installed to prevent this problem. Of course 'industrial' power supply module manufacturers are very aware of the likelihood of users stacking or even parallel connecting their modules and therefore instruct the end user to add diodes typically rated greater than the largest current rated unit in any 'stack'. They also point out that the total stacked voltage should not exceed (often) 500 Volts as it runs the risk of stressing the insulation of ALL the components in the supply with respect to the chassis of the supply. AMEK and many other mixing desk manufacturers used diodes on their DC input to desks either to prevent accidental reversal or to facilitate parallel supply operation (hot backup or sharing) duties.
Of course for any equipment using op amps and circuit deigns using op amp type topologies BOTH power rails MUST be present all the time as many op amps will 'crowbar' if one rail is ' absent'. Meaning the internal transistors of the op amp can all switch on together and effectively try to 'short' the positive supply pin to the negative supply pin until it catches fire.
Of course there is no defined 'ground' pin on usual standard type op amps so the non inverting input pin is always connected to a circuit node that humans tend to call 'ground'. This MUST be at a DC level tat is somewhere between the DC potential of the Negative power pin and the Positive power pin.
Earlier monolithic voltage regulators were most common in one 'polarity' for conceptual reasons giving a positive DC output such as the 78XX series and the 'negative' regulators appeared a good while afterwards and Negative variety are very rare above 4 Amp rating thus dual secondary transformers or interesting circuit design was necessary to obtain a dual output supply unit.
There were only a couple of 'dual rail' power supply chips ever made as far as I know and both of them can only supply a couple of hundred milliamps by themselves or then need extra 'pass transistors and other supporting components to get more power.

 

[ct.waveform]

Hi Bri and thanks for the post,
so it is a "different" way to obtain a negative voltage ,

sorry for the insistence ,
but it is reliable as a dedicated psu negative section,
or if possible better to have the section dedicated ?
thanks.

I'd say it's the standard way to connect a power supply with only one bridge rectifier. The 4 pins on the bridge are two marked ~ (for AC connection), one + ve and one -ve. So where would you put the main Ground? It goes on a nice thick bar (hopefully with a 'T' off it) between the two reservoir capacitors. I'd say it is a dedicated negative as I can't see any difference in its treatment to the positive. Each of the windings has 18V RMS across it (giving you 25.45V peak and, after the bridge, perhaps 24V on the rails) so if you connect them together you'd have 36V RMS. You're taking the halfway point and (maybe or maybe not) connecting it to chassis ground which shouldn't be more than a few volts away from actual Earth potential so, as is the way with transformers (which have no "Reference" voltage), each winding aligns itself to the reference that has been imposed on it. The negative side of the lower winding therefore moves to -18V. It has to be 18V below its other terminal.

Incidentally, there's no shame in not having this completely clear. Either we were not taught it properly, or it looked so simple in Physics classes that we felt we could always work it out from first principles when the time came, but in fact power supplies are incredibly thorny. And badly designed or laid-out ones introduce distortion through quite a few mechanisms. The worst I've seen is a Linn Amplifier where they thought it would be a good idea to have the reservoir caps on the amplifier board about 10 inches away from the transformer. Inevitably the switching, high current charging spikes sprayed RF everywhere a hundred times a second, and this showed up rather obviously on the output. One wonders if they ever measured it? But even discounting schoolboy errors like that, it is a very taxing area of amplifier design. At least I find it so.

 

[r2d2]

Really thanks guys for all your info posted !!!

So for it is a fairly reliable way of getting negative ,
even though it apparently looks more like a trick
to make up for the absence of the appropriate mode,

and the best thing still remains to have the appropriate section/rail & parts,

about the fact that it was used this way for get negative voltage
suggests that perhaps the most appropriate material was not available
or the time to build the appropriate section,
and then opt for this type of modification to one of the sections with +18v ?

As the Ac to Dc psu have 1 "783" Vreg for 48v phantom
and 2 "317" Vreg 1 for each +18v out,

any more info about as well ,
is welcome,
and more thanks too about !

 

[ccaudle]

and the best thing still remains to have the appropriate section/rail & parts,

You keep insisting while providing no evidence for that view, even though people with decades more experience have explained that there is really no downside to the way it was done (floating supplies in series).

What exactly is not "appropriate" about a design which has suitable performance and uses affordable off the shelf parts?
If you had provided measurements or even calculations showing that there was a limitation due to currents through parasitic capacitance, or an inconvenience due to the case being connected to one of the terminals, or a change in EMI performance depending on which terminal was connected to common there would be a legitimate complaint. Without that it seems you have a religious objection to the way the schematic is drawn, and religious discussion should go in the Brewery forum and not the technical forums.

 

[ct.waveform]

Yeah. I'm at a bit of a loss to know what @r2d2 would think is a usual or "normal" execution. He surely isn't expecting a second transformer dedicated to the negative? Besides, you've pretty much got that with the second winding, and aren't going to do much different after it. Though you get some advantages and additional flexibility, nor are two bridge rectifiers much different to this. So what is the "appropriate mode" (way) of doing this? I'm genuinely interested in finding out.

 

[SIXTYNINER]

Probably like the one on the attached image ?
(available here file:///Users/administrator/Downloads/Schematic_GC76.pdf)

 

[ct.waveform]

Probably like the one on the attached image ?
(available here file:///Users/administrator/Downloads/Schematic_GC76.pdf)

Isn't the left hand part of this (pre the LM317/337) exactly what he's looking at? Though here it's probably centre tapped rather than two windings.

 

[Rob Flinn]

Here is an example of a PSU I made usin this technique for a trident trimix I used to own that blew the psu & it used an expensive obscure regulator for the negative rail ( 79HGK). I subbed in this circuit & the +48v & 5v logic rail which were from the original trimix supply. The circuit was ripped off from an AMEK Einstein PSU. I built it in the original Trimix case so it could use the same lead to connect to the console. The only thing is the 2 halves of the PSU need to be fed from separate secondary windings on the mains transformer (no centre tapped secondaries so I couldn't use the original transformer for these rails). In this case I used 2 separate toroidal transformers which I bolted on the back of the case. I built the regulator circuit on a tag board (see pics) & was able to bolt the LM388 regulators to the original heatsinks in place of the obscure regulator. It worked really well and was very quiet.

 

[Newmarket]

Really thanks guys for all your info posted !!!

So for it is a fairly reliable way of getting negative ,
even though it apparently looks more like a trick
to make up for the absence of the appropriate mode,

Can you please take the hint that it's not a compromised solution. It's pretty much what you do when hooking up a lab bench power supply to provide dual rail supplies

 

[ct.waveform]

Here is an example of a PSU I made usin this technique for a trident trimix I used to own that blew the psu & it used an expensive obscure regulator for the negative rail ( 79HGK). I subbed in this circuit & the +48v & 5v logic rail which were from the original trimix supply. The circuit was ripped off from an AMEK Einstein PSU. I built it in the original Trimix case so it could use the same lead to connect to the console. The only thing is the 2 halves of the PSU need to be fed from separate secondary windings on the mains transformer (no centre tapped secondaries so I couldn't use the original transformer for these rails). In this case I used 2 separate toroidal transformers which I bolted on the back of the case. I built the regulator circuit on a tag board (see pics) & was able to bolt the LM388 regulators to the original heatsinks in place of the obscure regulator. It worked really well and was very quiet.

Interesting that you used LM338s. It's a very overlooked part, probably because in K form it was so expensive. But the TO220 version was a huge bargain for a time at something like 80p ea. Now it's not so cheap but still very useful and rugged, with pretty good performance out of the box. I've used them with 680uF of capacitance on their output without any problems. Mind you, you seem to be really putting them through their paces here with 40V going down to 18V. That really must be just about all they can manage.

Out of the dozens of possible permutations for a power supply it has never even crossed my mind that one could do a dual supply from two transformers. I'm usually thinking of how much easier it would be with more windings rather than fewer; and I've probably missed some bargain possibilities not having that possibility on my list. I'm also now a bit surprised not to have heard of some esoteric end of hifi saying it's a must-have if you want good sound. We do often go for two per stereo amp (and even 4 sometimes) but I don't think any of them are single winding.

 

[Rob Flinn]

Mind you, you seem to be really putting them through their paces here with 40V going down to 18V. That really must be just about all they can manage.

I never said I dropped from 40v to 18v so I'm not sure where you got that from. I can't remember exactly what the secondary voltage of the transformers was because I built that psu 15 or 20 years ago, but it definitely wasn't 40v.