When I first started this hobby last year I had no bench power supply to reach the B+ voltages I needed to experiment. But I did have a lot of DIN rail SMPS supplies laying around from work.
At that time I entertained the thought of maybe stacking these regulated SMPS DIN supplies in series, as a quick and dirty B+ supply. Now I've come across this blog on a fellow members web site (Wauwatosa Tube Factory). Entertaining this idea too.
Stacked SMPS PSU for tubes – wauwatosa tube factory
The series stacked supplies will sum to the desired voltage, but along with that they also sum up the ripple, no? So 6 supplies is 6 times the ripple of one supply, no?
Is it feasible to go ahead and stack the SMPS supplies, but then end the chain with a capacitor multiplier circuit to actively kick the ripple back down? Or not a good approach?
Since the supplies are regulated voltage should be stable. My DIN supplies have a high isolation rating for series stacking, well, and they are free to me (discarded by my work after being upgraded).
At that time I entertained the thought of maybe stacking these regulated SMPS DIN supplies in series, as a quick and dirty B+ supply. Now I've come across this blog on a fellow members web site (Wauwatosa Tube Factory). Entertaining this idea too.
Stacked SMPS PSU for tubes – wauwatosa tube factory
The series stacked supplies will sum to the desired voltage, but along with that they also sum up the ripple, no? So 6 supplies is 6 times the ripple of one supply, no?
Is it feasible to go ahead and stack the SMPS supplies, but then end the chain with a capacitor multiplier circuit to actively kick the ripple back down? Or not a good approach?
Since the supplies are regulated voltage should be stable. My DIN supplies have a high isolation rating for series stacking, well, and they are free to me (discarded by my work after being upgraded).
Last edited:
Any ripple is of a high frequency and minimal. Stacking them will only increase any ripple if they are all perfectly in phase, which they cannot be.
If ripple shape and strength are identical for all power supplies, the resultant ripple has a tendency to be cancelled out. This is a consequence of statistics. For instance, drawing a number from a bag containing the numbers from 1 to 10 for several times and finding the average of the drawn numbers, will result in 5.
Thanks,
I didnt think of the unique phase aspect of SMPS. The phases would all be random, so any ripple would neither cancel or reinforce. The stacked ripple should be pretty close to the worst single supply in the chain?
This is different than totem polling linear supplies, with that they would likely all be in phase?
I didnt think of the unique phase aspect of SMPS. The phases would all be random, so any ripple would neither cancel or reinforce. The stacked ripple should be pretty close to the worst single supply in the chain?
This is different than totem polling linear supplies, with that they would likely all be in phase?
I've thought about this as well. I've stacked two 48V switchers for 96V in some tube circuits but that got me wondering if I could just keep stacking and whether it would be easy to filter the noise after they are all stacked like that.
I've had good luck adding small LC post-filters to these cheap switchers and getting ripple practically down to nothing.
I've had good luck adding small LC post-filters to these cheap switchers and getting ripple practically down to nothing.
I would talk to whoever is making those SMPS units to see how many you can stack. Some of them will have X/Y caps from one end of the supply output to the chassis of the SMPS which is supposed to be earthed, so you'd want to respect the limits of at least that component.
The ones I have used were plastic bricks (no chassis gnd) made by Meanwell and something like 3kV I/O isolation test listed in the datasheet. So far, I've only tried stacking two, both upward from GND and downward (to make +96V supplies or -96V supplies).
These are what I'm using for my Corona breadboard amp mosfet grid driver rails.
These are what I'm using for my Corona breadboard amp mosfet grid driver rails.
The Greenvalve GM70 amp uses multiple 48V smps units. The website does not appear to be available now but you can see the schematic at
826 -> GM70 AMP Design
I tried it with my GK71 amp and it worked fine. Make sure you do not connect the grounds for each smps. I have also used the stack for experimenting with other breadboard builds.
I think the Meanwell units would be a better solution than the surplus 48V units I used. The biggest problem is making sure all the DC wires are hooked up with the correct polarity which can be tricky with multiple stacked units.
Audio ratbag: The heart attack special - 6S17K-V powerdrive GK71 SE amplifier
The stack is very useful it your amp requires multiple voltages including negative.
I have moved recently and do not have access to my stuff but in my next amp I will try using small toroid inductors and small capacitors as an LC filter for each of the various output voltages.
ray
826 -> GM70 AMP Design
I tried it with my GK71 amp and it worked fine. Make sure you do not connect the grounds for each smps. I have also used the stack for experimenting with other breadboard builds.
I think the Meanwell units would be a better solution than the surplus 48V units I used. The biggest problem is making sure all the DC wires are hooked up with the correct polarity which can be tricky with multiple stacked units.
Audio ratbag: The heart attack special - 6S17K-V powerdrive GK71 SE amplifier
The stack is very useful it your amp requires multiple voltages including negative.
I have moved recently and do not have access to my stuff but in my next amp I will try using small toroid inductors and small capacitors as an LC filter for each of the various output voltages.
ray
Stacking the supplies to get a higher voltage could stress the dielectric capabilities of the isolating components, like power transformers, when B+ level voltages are involved. Be sure to pay attention to that.
Could you go into more detail about that? My understanding is that the transformer in each unit only sees 110 or 240 volts on the AC side and 48 volts on the DC side. Where the smps unit sits in the chain of smps units is immaterial. Of course, the grounds on the AC side need to be disconnected.
ray
ray
Here is a good article for managed sensing supplies. For plain old supplies maybe just some protection diodes in case the supplies don't all rise together? At work we use DIN supplies all over the plant, regularly hook them in series or parallel, but never to B+ levels, they have sense wires. The max isolation figure pertains to the output not the AC side. That the output can sustain the final sum. Every few years we routinely replace old supplies ahead of failure because power supplies are still the most common failures we have (truck manufacturer).
Learn to connect power supplies in series for higher voltage output. | Keysight
Connecting Power Supplies in Series | Acopian Power Supplies
Schematic for plain unmanaged supplies in series, note the voltages can be mixed for even more versatility.
TDK-Lambda Americas Blog: Operating Power Supplies in Series
I'd probably want to check the voltage rating of the filter cap to ground in the supply too, and change those if needed.
Last edited:
Each successive SMPS is floating on top of the voltages of the preceding units.
So if you had 3 in series, the first is ground referenced, the second floats atop the first, and so on. Take 48V supply, the first has 48V across it relative to ground, the second 96V relative to ground at the point of highest stress and the 3rd 144V. Stack enough of them and eventually the internal insulation will fail.
So if you had 3 in series, the first is ground referenced, the second floats atop the first, and so on. Take 48V supply, the first has 48V across it relative to ground, the second 96V relative to ground at the point of highest stress and the 3rd 144V. Stack enough of them and eventually the internal insulation will fail.
there are now inverter type dc to ac switching bricks capable up to +-450 volts and 150 watts, since these are high frequency switchers, ripple frequency will be very high....
this is in my bucket list of things to do...
a 12 volt high current isolation traffo, a bridge rectifier, why you can even use a high power 12vdc to 40A smps to supply filaments and the hv switcher....
i can imagine that emi/rfi can be an issue with the 12v to +-450 vdc converters, but the thing is, since frequency s going to be high and outside the audio band, then mitigating measures may not be so expensive...
i see this as the way forward....no more back breaking power traffos....and cheap to implement also...
this is in my bucket list of things to do...
a 12 volt high current isolation traffo, a bridge rectifier, why you can even use a high power 12vdc to 40A smps to supply filaments and the hv switcher....
i can imagine that emi/rfi can be an issue with the 12v to +-450 vdc converters, but the thing is, since frequency s going to be high and outside the audio band, then mitigating measures may not be so expensive...
i see this as the way forward....no more back breaking power traffos....and cheap to implement also...
why stack them if you can find a psu that had the right voltage to begin with?
but if you stack them, you can megger the bricks on top of the the ground refferenced module, bear in mind that you should megger with the bricks not yet connected in series, and meggering is from chassis ground to the negative terminal of the bricks going on top, do not megger between positive and negative terminal or you will destroy the psu, a 1000dvc test voltage is plenty...
and for safety, the metal enclosures need to be earthed to the chassis too for all the bricks in series...
but if you stack them, you can megger the bricks on top of the the ground refferenced module, bear in mind that you should megger with the bricks not yet connected in series, and meggering is from chassis ground to the negative terminal of the bricks going on top, do not megger between positive and negative terminal or you will destroy the psu, a 1000dvc test voltage is plenty...
and for safety, the metal enclosures need to be earthed to the chassis too for all the bricks in series...
This unit is literally everywhere on both Amazon and ebay, overwhelming most others on the market. This $5 model has a single +/- output, for a split +/0/- output model the price doubles to $10. Many different sellers are carrying this unit, I dont know if that means its reliable or just in demand. It runs at 75kHz. 200ma max output current, probably should be derated in use to 125ma or so? Commentators said it can be hooked in series for 780V.
OK, its cute, cheap and it makes high voltage DC at the ma rating of any bench tube supply. But at what noise cost? And can that noise be tamed well enough with a LC post filter stage or maybe a cap multiplier MOSFET stage?
DC-DC 8~32V to 45~390V High Voltage Boost Converter Step-up Booster. J7E1 U1P1 | eBay
OK, its cute, cheap and it makes high voltage DC at the ma rating of any bench tube supply. But at what noise cost? And can that noise be tamed well enough with a LC post filter stage or maybe a cap multiplier MOSFET stage?
DC-DC 8~32V to 45~390V High Voltage Boost Converter Step-up Booster. J7E1 U1P1 | eBay
you are limited to 400vdc and overvoltage trip sets in at 410 volts....
those are so cheap that you can buy many of them, 10 pcs and still come up cheaper...than if buying an EI power traffo say....
that is a free running multi vibrator ckt, square waves, probably rich in harmonics, but that a ferrite choke mounted outside of a few microhenries should be able to filter out much of the noise, a common mode choke even, there are plenty of those that can be harvested from discarded tv monitors....
i have been eyeing that unit too, there is one rated for 450 watts, and yes, plenty of derating will increase reliability....
those are so cheap that you can buy many of them, 10 pcs and still come up cheaper...than if buying an EI power traffo say....
that is a free running multi vibrator ckt, square waves, probably rich in harmonics, but that a ferrite choke mounted outside of a few microhenries should be able to filter out much of the noise, a common mode choke even, there are plenty of those that can be harvested from discarded tv monitors....
i have been eyeing that unit too, there is one rated for 450 watts, and yes, plenty of derating will increase reliability....
In my experience that bottom dollar stuff should be operated at 50% of its supposed ratings. I certainly wouldn't leave it powered up when leaving the house.