Connecting 2 power supplies in series for 2 sided +/- voltage

[bmdaly]

I've recently acquired a Tenma 72-10480, which I'm very happy with, but would like to add a second supply for 2-sided +/- voltage supply. This is the supply - http://ie.farnell.com/tenma/72-10480/power-supply-1ch-30v-3a-adjustable/dp/2251946
Are there any potential issues with simply connecting 2 of these supplies in series, or any precautions I can take to ensure neither supply is damaged by sinking current from the other? As they are independent supplies, they will not be powered on at exactly the same time and may be set to different voltages. I have seen a suggestion that a reversed diode be connected between the positive and negative terminals of each power supply to prevent any significant  reverse voltage being applied across the terminals.
Any advice appreciated.

[Performa01]

No worries!

Lab bench power supplies provide a floating output (not ground referenced), so they can be connected in series for either increasing the total output voltage or getting a dual polarity supply.
The reverse polarity protection diodes are certainly a good idea, but not actually required, as any halfway decent power supply will have these already built in. Just to be absolutely sure, I’ve just checked six of my bench power supplies, one of them identical to the one you’ve provided a link for, and sure enough they all have the protection diode fitted.

[bmdaly]

No worries!

Lab bench power supplies provide a floating output (not ground referenced), so they can be connected in series for either increasing the total output voltage or getting a dual polarity supply.
The reverse polarity protection diodes are certainly a good idea, but not actually required, as any halfway decent power supply will have these already built in. Just to be absolutely sure, I’ve just checked six of my bench power supplies, one of them identical to the one you’ve provided a link for, and sure enough they all have the protection diode fitted.

Thanks for your response. I'm trying to understand how real power supplies differ from ideal voltage and current sources. I have 3 further questions I'd like clarified:

1. I assume that in a series arrangement with the load in series with both supplies, the same current must flow through both supplies. In this case, will limiting current in either supply suffice to limit the overall current, or is this over-simplifying things?

2. Alternatively, with a dual polarity arrangement with load connected to a virtual earth where positive of one supply is connected to negative of other supply, both supplies' voltage and current settings would essentially be independent of each other?

3. Finally, with the power supply model linked above, can 2 supplies by connected in parallel with same voltage set to provide increased current capacity, or would this not be recommended for independent supplies, due to risk of imbalance in their voltages?

Once again, any input appreciated.

[Performa01]

1. I assume that in a series arrangement with the load in series with both supplies, the same current must flow through both supplies. In this case, will limiting current in either supply suffice to limit the overall current, or is this over-simplifying things?

2. Alternatively, with a dual polarity arrangement with load connected to a virtual earth where positive of one supply is connected to negative of other supply, both supplies' voltage and current settings would essentially be independent of each other?

3. Finally, with the power supply model linked above, can 2 supplies by connected in parallel with same voltage set to provide increased current capacity, or would this not be recommended for independent supplies, due to risk of imbalance in their voltages?

ad 1) In short: no, it will not. Current limiting just means that the supply will reduce its output voltage if necessary in order to not exceed the set current limit. It will do so down to zero volts, but cannot go negative - just the before mentioned reverse protection diode will prevent it from doing that. As an example, if you have 2 x 30V and one supply set to 100mA and the other to 5A, the first one will go down to 0V as soon as 100mA is approached, but then the load will still see 30V. Only if the load current exceeds 5A, the total output voltage will drop below 30V.

ad 2) Yes, both polarities are independent. There are dedicated dual power supplies that have a series tracking mode in a master/slave arrangement, where the voltage and current limit of the master channel is effective on the slave channel as well. But with two independent PSUs, you have to set them both independently and manually.

ad 3) It is perfectly possible to connect two PSUs in parallel, as any halfway decent PSU will have a reverse current protection, so there is not a problem even if the two voltages are not exactly identical. Imagine two supplies connected in parallel, both set for a current limit of 1A, the first one set for 30V and the 2nd one for 32V. You'll get an output of 32V/1A. If the load draws more current, the 2nd supply will limit at 1A down to 30V, where the first PSU chimes in and delivers another 1A, thus resulting in a total of 30V/2A. If this current would be exceeded as well, then the voltage will go below 30V accordingly.

[coromonadalix]

i would agree with Performa

But i would put an identical model in serie or in parallel to have the same specs / response time etc ....  since it is an switchmode model, i would personally not put  EX : an linear model in parallel.

[bmdaly]

Thanks for taking the time to write that detailed explanation. I'm still, however, unclear on point 3. Does the supply have a series diode (or some logically similar protection), allowing current only flow out of the positive terminal for reverse current protection, as well as a reversed diode across the terminals?

[bmdaly]

i would agree with Performa

But i would put an identical model in serie or in parallel to have the same specs / response time etc ....  since it is an switchmode model, i would personally not put  EX : an linear model in parallel.

Hi coromonadalix,

The supply referenced above is a linear supply. Are there any specific concerns you have placing two of these in series?
Thanks.

[Performa01]

Thanks for taking the time to write that detailed explanation. I'm still, however, unclear on point 3. Does the supply have a series diode (or some logically similar protection), allowing current only flow out of the positive terminal for reverse current protection, as well as a reversed diode across the terminals?

Well, you are perfectly right in that paralleling two power supplies is a bit more complex.

The PSU does not have a series diode. It just has a series pass transistor that drops as much voltage as needed to get the desired output. If you feed an external voltage into the output, that is higher than the set voltage, then the PSU will sense its output being too high and switch off the series pass transistor in an attempt to let the output voltage drop. But it does not drop, because of the external voltage. So this is the final state then, the PSU output essentially shut off.

Of course there is one little catch – the input voltage of the series regulator in the PSU (the voltage before the series pass transistor) is finite. Once the external voltage exceeds this input voltage, the series pass transistor gets polarity reversed and might be damaged. So there is a reverse voltage protection diode across this transistor and any output voltage exceeding the input voltage will be routed back to the input of the PSU, thus increasing the input voltage as well. This way the pass transistor will never see more reverse voltage than one diode drop, which will save its life. Of course there are limits, because of the max. voltage the input of the PSU can handle. This is limited by the max. reverse voltage of the rectifier and the max. working voltage of the reservoir capacitors, but maybe also other semiconductors in the PSU that are somehow connected to the input voltage. So for example it's probably not a good idea to connect 40V to the output of a PSU that has a max. output of just 5V. But as long as you connect similar PSUs, like 30V and 40V, there should not be a problem.

Apart from that, it is clear that the PSU will start drawing current from the external voltage source as soon as the reverse voltage protection diode starts to conduct and charge up the input reservoir capacitor. It depends on the construction of the PSU, when this happens. A simple linear PSU might have just a single input voltage of say 40V to produce output voltages from 0 to 30V for instance. In this case, no current will flow into the PSU as long as the external voltage does not exceed 40V, no matter what output voltage is set on the PSU.

It might be a more refined linear power supply that uses various transformer taps to adapt the input voltage in a way that the difference to the output voltage remains low, which enables higher currents and vastly reduces the power dissipation in the series pass transistor for lower output voltages. In that case, even a 30V PSU might draw current if set to only 5V and an external voltage of e.g. 12V is connected. But even then, it will still not be a problem for external voltages up to at least 40V, since the circuitry has to be able to handle the full transformer voltage anyway, so it does not matter what transformer tap is currently used.

It should be quite similar with a SMPS (Switch Mode Power Supply), at least with a simple step-down converter, since there is just a switching transistor instead of the linear pass transistor (plus some additional circuitry for charge transfer/storage) and the input voltage would be constant and rather high. The output voltage is determined by pulse width or density control of the switching transistor and quite similar to a linear PSU, the switching transistor will just never turn on if the output voltage appears too high (because of the external voltage).

In any case there will not be a problem as long as you don’t connect vastly different PSUs in parallel, i.e. their max. output voltages should be roughly in the same ballpark. 30/40V would be fine, 12/40V maybe not.

[bmdaly]

Great explanation, Performa01. The constraints around dual power supply usage are a lot clearer to me now.
Thanks again for taking the time.
Brian

[alm]

One thing to keep in mind: two floating power supplies in series will generally be fine as long as you do not exceed their maximum common mode voltage relative to ground (don't go put two high voltage supplies in series, or several low voltage supplies in series, without checking this). But the device under test may not be happy with incorrect sequencing of power rails. For example the negative rail coming up before the positive rail, or the positive rail shutting down due to current limiting, while the negative rail does not. That's where two tracking voltage rails might be valuable. Some dual power supplies offer this, and some older power supplies also had rear terminal strips that allowed you two link two identical power supplies for tracking operation.

The cases where this would actually damage the device under test are probably fairly obscure, but keep in mind that there is more to this than just the power supplies.

[Performa01]

One thing to keep in mind: two floating power supplies in series will generally be fine as long as you do not exceed their maximum common mode voltage relative to ground (don't go put two high voltage supplies in series, or several low voltage supplies in series, without checking this). But the device under test may not be happy with incorrect sequencing of power rails. For example the negative rail coming up before the positive rail, or the positive rail shutting down due to current limiting, while the negative rail does not. That's where two tracking voltage rails might be valuable. Some dual power supplies offer this, and some older power supplies also had rear terminal strips that allowed you two link two identical power supplies for tracking operation.

The cases where this would actually damage the device under test are probably fairly obscure, but keep in mind that there is more to this than just the power supplies.

… which is going to show that even an apparently simple topic could easily fill a book if we want to cover all possible eventualities and use cases

I’d like to expand on this.

The max. permissible common mode voltage is essentially the isolation of the PSU circuitry from the case and consequently mains earth ground. It is surprisingly hard to find any specifications for that, at least for the cheaper power supplies.

I did find some more detailed data only for a way more expensive PSU like the Rohde & Schwarz NGE 102/103. So I’ll use this as an example here:

Maximum voltage to earth   150V DC
Maximum reverse voltage   33V
Maximum inverse voltage   0.4V
Maximum permitted current in case of inverse voltage   3A

I figured this is a good opportunity to discuss specifications like this.

The max. voltage to earth is what has been referred to as common mode voltage before. So if this PSU is connected in series with some other PSU for increased total output voltage, then it should be made sure that neither of the output terminals of this PSU gets elevated more than 150V above ground potential (which will be the same potential as the PSUs metal case). I guess it will be next to impossible to find a PSU with even less isolation, specified or not, so as a general rule it is certainly safe to connect several PSUs in series up to at least 100V.

The max. reverse voltage indicates the point where reverse current starts to flow when an external voltage is applied to the outputs of the PSU, according to my earlier post with regard to paralleling PSUs. This specification is just 1V higher than the max. output voltage, hinting on a very efficient design with the input voltage only slightly higher than the max. output. This is a fairly conservative rating and I’m quite positive it does not mean the PSU will instantly blow up as soon as the reverse voltage exceeds 33V, just R&S don’t want to guarantee for anything in this case. Even so, with a little care it shouldn’t be too hard to meet this requirement.

The inverse voltage specifications are for the inverse voltage protection diode (I think I have incorrectly called it “reverse voltage protection diode” in one of my earlier posts) connected in parallel with the output. It has a forward voltage drop of 0.4V (hinting on a Schottky diode) and can handle a max. current of 3A. So any imbalance in the current limiting of several PSUs connected in series should be less than 3A. As a general rule, I would keep them identical within 1A, which should be safe in any case, even if the reverse polarity protection diode happens to be just a 1N4004 or something similar.

Regarding the sequencing of power rails, it is quite obvious that we’re hardly able to turn them on/off at exactly the same time, when it’s two independent PSUs. Thankfully, I cannot think of a circuit requiring a symmetrical dual supply that would get damaged if one supply rail drops out. It just should not become reverse polarity, but that’s why we have the inverse polarity protection diodes in the PSU.

I think most dual output power supplies provide a tracking mode as described in an earlier post, and the master/slave principle does not only ensure identical settings on both channels, but also symmetrical output voltage in all operating modes, i.e. even when the current limit is active. So if one rail gets overloaded and reduces output voltage in order to maintain the current limit, the 2nd channel will reduce its output voltage accordingly.

There are indeed PSUs where several of them can be linked for building a tracking multi channel power supply. The Manson SSP-7080 SMPS is an example for this.

http://www.manson.com.hk/products/detail/178

I own only one of these, hence haven’t tested the Master & Slave / link feature. Apart from that, I like this tiny little PSU a lot and it has proven to be reliable and convenient to operate.