Stacked power supply HELP!

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
You can build a stacked supply out of + and - full wave rectifiers and a center tap. It doesn't matter where ground is, only where the return connection from the load is (e.g. OPT primary winding signal current return).

All electrons do not originate from some pool in the ground... Think current loops... Here is a good place to start:
ETF Presentation

Your stacked design has 2 separate current loops also, with a common point between them (the point at which your 2 power supplies stack).

Here is a stacked supply I built that just happens to have "ground" in the center but the current loops are the same as if ground had been connected to the negative-most point or even at the positive-most point. Just don't exceed the breakdown voltage to chassis on any of your components.
 

Attachments

  • solstice-ps-schem.png
    solstice-ps-schem.png
    12.3 KB · Views: 277
Last edited:
DannyD, if the seller tells you something encouraging or you decide to take the bells off one of those transformers I'd like to see what's inside, if you can be bothered.

I bought a pair of those output transformers used from another forum member. I have not taken them apart, and I wouldn't recommend it since there is nothing keeping the gap constant except for the 4 tiny screws that hold the end bells on.

There is no magic in OPT's. The size and weight will determine the real power rating. These OPT's weigh about 6 pounds which makes them a 10 watt or so OPT. Yes you can crank 50 watts through them at 1 KHz, but if you go down to 30 Hz they will saturate at 10 watts. I got them for experiments and they don't sound half bad, but don't expect to run 50 watts through them.

I haven't gone beyond about 350 volts on them so I can't say how much voltage they will take before frying.
 
You can build a stacked supply out of + and - full wave rectifiers and a center tap. It doesn't matter where ground is, only where the return connection from the load is (e.g. OPT primary winding signal current return).

All electrons do not originate from some pool in the ground... Think current loops... Here is a good place to start:
ETF Presentation

Your stacked design has 2 separate current loops also, with a common point between them (the point at which your 2 power supplies stack).

Here is a stacked supply I built that just happens to have "ground" in the center but the current loops are the same as if ground had been connected to the negative-most point or even at the positive-most point. Just don't exceed the breakdown voltage to chassis on any of your components.


Hey Michael, thanks. I read the Lynn Olson article. I've been pondering this for the last 10 minutes or so.


So, even though the two DC supplies are derived from the same secondary on a single power transformer, having the top supply's ground referenced to the "top" of the bottom supply can work after all? :confused:


Your Solstice amp has an input transformer that isolates the input stage tube grids from ground. Did you do that because you had to (isolate the input stage from 0V ground) or because you wanted to have an input transformer for isolation, etc?


If you have two stacked DC supplies, and the bottom DC supply is the plate supply for the input stage, and that stage's input is to be DC-coupled, would the bottom supply's "ground" need to be referenced to 0V? That way the input signal can be referenced to the same ground as the input stage's DC supply, and a DC blocking capacitor won't be necessary on the input of that stage.

Then the DC supply on top of the stacked supply can have its "ground" referenced to the "top" of the bottom DC supply.

Is that all correct?

I believe I might be a bit confused...
--
 
Hey Michael, thanks. I read the Lynn Olson article. I've been pondering this for the last 10 minutes or so.


So, even though the two DC supplies are derived from the same secondary on a single power transformer, having the top supply's ground referenced to the "top" of the bottom supply can work after all? :confused:


Your Solstice amp has an input transformer that isolates the input stage tube grids from ground. Did you do that because you had to (isolate the input stage from 0V ground) or because you wanted to have an input transformer for isolation, etc?


If you have two stacked DC supplies, and the bottom DC supply is the plate supply for the input stage, and that stage's input is to be DC-coupled, would the bottom supply's "ground" need to be referenced to 0V? That way the input signal can be referenced to the same ground as the input stage's DC supply, and a DC blocking capacitor won't be necessary on the input of that stage.

Then the DC supply on top of the stacked supply can have its "ground" referenced to the "top" of the bottom DC supply.

Is that all correct?

I believe I might be a bit confused...
--

The center tap is in fact the common point where the (-) of the top supply connects to the (+) of the bottom supply. It's connected to chassis in my amp out of convenience. I had already decided on the input transformer because I use all balanced sources and I use these for PA gigs etc... OT but I rocked an outdoor festival last week with a pair of these 15W triode amps and Klipsch LaScalas!

I think your description of the stacking with the bottom supply grounded sounds right. Here is an example. I also have a third (75V) supply stacked to provide positive grid drive voltage, but the principle is still the same; separate current loops. The filament supply is even stacked on the same connection point. Definitely a separate current loop there!

Cheers,

Michael
 

Attachments

  • meteor-stacksupply.png
    meteor-stacksupply.png
    11.3 KB · Views: 259
If I use them with the Borbely design they'll get somewhere around 15-18W, my speakers are rated at 93db and my room is about 12x14, so hopefully I don't ever have to drive them very hard, at least not hard enough to cross that 10W threshold mentioned earlier.

Speaking of Borbely, does anyone have any comment on that design? It's sounding more appealing all the time given the fact that I already have all the parts for it, save the power transformer.
 
Sorry to have two topics going on in one topic. My thanks to the OP and responders for putting up with this.

The center tap is in fact the common point where the (-) of the top supply connects to the (+) of the bottom supply. It's connected to chassis in my amp out of convenience. I had already decided on the input transformer because I use all balanced sources and I use these for PA gigs etc... OT but I rocked an outdoor festival last week with a pair of these 15W triode amps and Klipsch LaScalas!

Cool!

I think your description of the stacking with the bottom supply grounded sounds right. Here is an example. I also have a third (75V) supply stacked to provide positive grid drive voltage, but the principle is still the same; separate current loops.

I think I get it. Here's the power supply I was thinking of. The idea is to use the lower (+370V) supply for two LTP driver pairs drawing 30mA for the two pairs (7.5mA per triode), and the upper (+350V) supply for two pairs of output tubes drawing about 200mA total for the two pairs (50mA per triode). The power transformer is rated for 275mA, and I'm staying under that by about 45mA.

I see what I think are the two current loops, each consisting of the diode bridge, choke and reservoir cap and on to load (where the current loop would continue and be closed). Right?
 

Attachments

  • Stacked_Pwr_Supplies_00.jpg
    Stacked_Pwr_Supplies_00.jpg
    62.4 KB · Views: 244
Sorry to have two topics going on in one topic. My thanks to the OP and responders for putting up with this.



Cool!



I think I get it. Here's the power supply I was thinking of. The idea is to use the lower (+370V) supply for two LTP driver pairs drawing 30mA for the two pairs (7.5mA per triode), and the upper (+350V) supply for two pairs of output tubes drawing about 200mA total for the two pairs (50mA per triode). The power transformer is rated for 275mA, and I'm staying under that by about 45mA.

I see what I think are the two current loops, each consisting of the diode bridge, choke and reservoir cap and on to load (where the current loop would continue and be closed). Right?

Not quite. There is a very short current loop through diodes, the choke, and the power transformer.

The bridge diodes do not create a center tap (as often mis-stated), they in fact create an "end tap" which creates a short circuit as shown in yellow when two opposite polarity "end taps" are connect together.

You need to use the transformer's center tap as the stacking point and use a (+) connected pair of diodes for the B+ and a (-) connected pair for the driver supply with the driver filter in the (-) branch to ground. Look at the Solstice supply and instead connect the 'B-' driver output to ground and connect the center tap to the 2 filter returns (which becomes the stacking point).

Don't worry about the filter choke in the line to ground. The whole transformer primary floats on the small amount of lower supply ripple and adds to the B+ ripple (series connection of the two power supply voltages + ripple).

Does it make sense?

Cheers,

Michael

PS I hope this side discussion is still helpful for people to understand the whole idea of stacked supplies. If we annoy anyone we can move to another thread.
 

Attachments

  • short-stack.JPG
    short-stack.JPG
    55.5 KB · Views: 219
Last edited:
That will do it. If you use a choke in the lower supply it can go between the rectifier or cap and ground.

Many thanks for coaching me through that!

So now, if I add the chokes as choke-input smoothing, and since the resulting output voltage will equal approximately VAC in * 0.9, I should get voltages as shown in the schematic below. (Ignoring DCR of the chokes and xfmr windings and other losses.)

Sec. 1: 415 * 0.9 = 373.5
Sec. 2: 395 * 0.9 = 355.5

Right?
 

Attachments

  • stacked_psu_single_xfmr_choke_input.jpg
    stacked_psu_single_xfmr_choke_input.jpg
    36.6 KB · Views: 192
Many thanks for coaching me through that!

So now, if I add the chokes as choke-input smoothing, and since the resulting output voltage will equal approximately VAC in * 0.9, I should get voltages as shown in the schematic below. (Ignoring DCR of the chokes and xfmr windings and other losses.)

Sec. 1: 415 * 0.9 = 373.5
Sec. 2: 395 * 0.9 = 355.5

Right?

:D
 
OK, I think I've got it now!

Last question, re: Additional smoothing of the two DC supplies by adding add'l RC sections...

Let's say I've got a 2-stage amp I want to DC-couple.

I'm using the 'stacking point' as the B+ for the first stage, 'top' supply is for the output stage.

I need to define ground for the first stage input, as I want that input to be DC-coupled (no input cap).

Do I now start adding RC filter stages to the center tap to use as the first stage supply, or is that causing a problem that I can't see?

schematic attached...

PS - I keep looking at that schematic... Dropping the voltage of the center tap feed would also drop the voltage of the top supply's lifted ground reference, no? It seems the two DC supplies would interact in all sorts of ways I can't decipher.

M. Jones' "Crystal Palace" amp has its LTP driver DC-coupled to its PP output stage. But the input stage output is RC-coupled to the driver stage's grids, which is why the input can be at ground in that 3-stage amp. A two-stage DC-coupled amp with the input at ground is looking like a challenge with this 'two stacked DC supplies from a single pwr xfmr' scheme...
 

Attachments

  • stacked_psu_single_xfmr_choke_input_+RC.jpg
    stacked_psu_single_xfmr_choke_input_+RC.jpg
    39.1 KB · Views: 193
Last edited:
Maybe I need to think of it as a dual-polarity supply, and the decoupling resistor needs to be in series with the choke, and then I define ground at the end of the chain...

PS - I'll call the previous schematic "take one," this one "take two." I'm pretty sure Take 2 is correct... ??
 

Attachments

  • stacked_psu_single_xfmr_choke_input_+RC_01.jpg
    stacked_psu_single_xfmr_choke_input_+RC_01.jpg
    39.2 KB · Views: 177
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.