Connect Power Supplies in Series -- Practical Tips?

[flapdash]

I am connecting three switched power supplies in series. Each supply is 30VDC, to give 90VDC output. The current rating on each 30VDC module is 50A. I've verified that series operation is supported by the manufacturer (Astec/Emerson). The load is a motor, and I plan to install shunt regulation and capacitance near the motor to help with this inductive load and guard against harmful regeneration.

Someone recommended putting diodes between each module, as shown in the picture attached to this post (D1, D2, D3...see picture). Why? Something about one module's positive terminal going lower than the adjacent module's negative terminal and therefore these diodes would guard against unwanted current flow in the "wrong" direction?

Can anybody provide an explanation of why D1, D2, and D3 might be necessary?

Thank you!

 

[VEBill]

I would have put the three 'mystery diodes' across each power supply, with the anode of these diodes connected to each power supply's negative terminal (not all three anodes to the common).

The explanation is as simple as imagining suddenly turning off the middle power supply while things are running. Avoids forcing (conventional) current into the negative terminal.

I'm not exactly sure that the method shown accomplishes the same thing or not.

 

[itsmoked]

I agree BLL. Across the supplies not from each supply to ground.

Why this bizarre setup anyway?

Keith Cress
kcress -

http://www.flaminsystems.com

 

[waross]

What is "shunt regulation"?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[VEBill]

Shunt regulation. Conceptually, a huge zener diode to prevent the voltage from exceeding some value.

But I'm not sure if that addresses any actual system requirement in this case. Unless the test setup is actually an electric car rolling down a very steep hill (presumably with three very long extension cords).

 

[waross]

Would that be the electric car with the large trailer with a huge heat sink?
Am I getting the picture?

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[VEBill]

As drawn, the three diodes scheme doesn't make sense. For example, imagine that the top-most power supply is "accidentally" turned off while things are running. The negative terminal of that power supply would be at +60 and the positive terminal would suddenly be at zero (via the load, neglecting any additional issues with the load being a motor spinning at speed).

The diode drawn next to that power supply will be at zero and zero (no help). The top most power supply will thus be back-fed with reversed voltage and things inside, such as electrolytic capacitors, will be damaged.

I put quotes around the word "accidentally" because sooner or later you'll need to turn things off.

All the above are concerns even if the load were resistive.

The next thing to think about is the spinning motor...

 

[MacGyverS2000]

Don't be so short-sighted, VE... those extension cords don't need to be long if the car is driving in a circle. ;-)

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

It'd be a spiral, not a circle.

 

[biff44]

My experience with lower current output supplies is that they can do some really weird things on startup, or with varying loads, if you try to connect them in series.

That said, I would add some capacitors across each supply, and a fixed load resistor (maybe drawing 200 mA) at 90 V just to give the supplies something to drive if the motor is acting funny.

I think the three diodes are recommended because the supplies can only "supply" positive voltage. If for any transient reason any one of them is asked to supply a negative voltage, all heck breaks loose. The diodes might help in that regard?

http://www.MaguffinMicrowave.com

Maguffin Microwave wireless design consulting

 

[biff44]

Actually, a more accurate way to say it is that the power supplies can usually only supply positive current. If one of them is asked to "sink" current, even briefly, it will probably flip out. If one supply flips out, they all will as they are referenced to each other.

http://www.MaguffinMicrowave.com

Maguffin Microwave wireless design consulting

 

[itsmoked]

Yep, totally agree.

The OP asks for "practical tips". The practical tip response would be "don't do it".

Keith Cress
kcress -

http://www.flaminsystems.com

 

[flapdash]

Hello,

Thanks, everybody, for your input.

VE1BLL's comments make the most sense to me, and VE1BLL provides some good reasoning behind them. The post about the top supply failing and getting current back-fed into it makes sense (with the original topology that I posted).

So…I’ve attached the modified diagram to this reply/post.

Let’s narrow down the failure case just to make it a more tractable problem: let’s consider only the case where any *one* of the supplies/modules fails.

Perhaps there are three “failure modes” for any one of the 30V modules. When a module fails it could look like: 1) an open-circuit, 2) a short-circuit, or 3) a non-zero and non-infinite impedance.

Considering VE1BLL’s suggestion, and putting the diodes across each supply (anode on neg terminal, cathode on pos terminal), perhaps this would be the behavior in each of the above three cases:

1)
If any of the supplies gets turned off, resulting in an effective OPEN-circuit for that supply, then current would be bypassed through that supply's diode. Thus, we would see 60V on the output. That sounds like a fairly graceful failure to me. If I was to go with my original diagram, when either of the bottom two modules fail, there might be an opportunity for current flow from DCCOM through the diode associated with the failed module (since the cathode of that diode is tied to the negative terminal of the above/adjacent module). I think this is enough of an argument to go with VE1BLL’s suggestion.

2)
If any of the supplies fails in an effective SHORT-circuit for that supply, then all the current would just flow through the effective short-circuit, bypassing even the bypass diode. Eventually, I would anticipate that the failed module in this case would have some thermal problems and would degrade further to an open-circuit condition. But failure modes can be hypothetical a lot of the times.

3)
If any of the supplies gets turned off, resulting in some non-zero and non-infinite impedance, then current would be bypassed through that supply's diode. And again, if any one of the supplies fails, then we would see 60V on the output.

In response to user “itsmoked”, who had two comments:
>>“Why this bizarre setup anyway?”
– Well, that’s why I posted this question. If I knew the answer, I wouldn’t have posted. The devil’s in the details, and it’s very interesting and helpful to see everybody’s input on this!
>>“The practical tip response would be "don't do it"
--Well, that’s not very practical at all since it doesn’t solve the problem. I have used Astec supplies in series before (two 48VDC/25A modules in series) with no problem. Using three 30VDC modules is the only way I can get my power requirement of 4500W for this given product line.

In response to user “waross”:
Shunt regulation is common in motion control systems where a motor is present and the possibility of motor-generated back-EMF exists. For example, we develop “gravity-assist” machines that can be moved up or down. When the operator is moving the machine down, the motor is behaving like a generator and generating a voltage on the DC bus. Also, back-EMF can be present during motor decelerations as a part of normal, coordinated movements (not just in “gravity-assist” situations). The simplest form is a Zener along with a power resistor to regulate the voltage if it rises above a certain value and to shed excess power. More sophisticated versions might use a comparator and some active loads (power FETs), or a tiny microcontroller (e.g. PIC) that monitors DC bus voltage and then switches the line through a power resistor. For example, Maxon Motor makes an off-the-shelf shunt regulator:

http://shop.maxonmotor.com/ishop/article/article/309687.xml

In response to user biff44:
The capacitor recommendation is a good one…very interesting. How would you connect three caps: each positive lead of each cap connected to each module’s positive lead, and then the negative leads of all the caps to DCCOM? Or connect both leads of the caps across just the module terminals for each supply (that is, in parallel with the diodes when connected as VE1BLL suggests and as shown in the newest diagram).

Overall, I like VE1BLL’s suggestion, and it also provides me with some idea of what kind of ratings those diodes should have (50A bare minimum…probably 100A to be safe).

I do have a follow-on question:
Do I put a diode on the bottom-most supply (V3)? If that bottom-most supply fails and opens, then it’s diode (D3) would have anode at DCCOM and cathode at the negative terminal of the middle supply…would D3 then be conducting? Would that be OK? I think so….?

Thanks again!

 

[ScottyUK]

Why not rent or buy a proper regulated power supply with the usual features like programmable current limit etc? Sorenson do a nice line in big supplies - something like their PRO 110-90T would be a suitable choice for your load - and HP/Agilent have a range up to at least 20kW. There are quite a few on the surplus market and at a fraction of their list price.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[flapdash]

ScottyUK:

Thanks for the suggestion on other power supply manufacturers.

This is kind of a tangent to the original post (practical tips for properly wire switched mode supplies in serial), but...

I had a relatively difficult time finding data sheets on Sorensen supplies (a.k.a. Xantrex, Ametek?)...the only suppliers of Sorensen I could find were test equipment rental houses. Even a Google search had difficulty finding the original Sorensen web site where I could easily access data sheets. So I've decided that if it's difficult to find data sheets on Sorensen, I have to move on.

So I checked into Agilent as you suggested. Their N87xxx family has the power I need. They offer the N8759A, at 100V and 50A (5kW)...so at 90V I could expect 55.5A. That indeed does seem like a good fit. The only issues I have with the Agilent include: 1) it's a lab/benchtop solution which translates into a bigger physical package, and 2) the price. The Agilent is 17.4" x 16.7" x 3.5" and costs at least $6,112 (that's probably the entry model from this family, so I suspect my desired model is a bit more).

As comparison, the Astec/Emerson unit I am using is 11.5" x 8.0" x 5.0" and costs $2,200. Since we are building multiple units with this supply, we need moderate quantities (~20) so cost is (unfortunately!) a consideration.

Otherwise, your suggestion for Agilent, in a lab environment, is very good.

Thanks again.

 

[ScottyUK]

The old HP6475C is a monster of a power supply, 110V @ 100A. There are a number on ebay for under $2k, and plenty of them from various other sources. Of course it helps when you have a specific model number to pursue - sorry for not posting it last time.

Sorensen are hidden among the Elgar group's products alongside Ametek and Elgar's own products.


----------------------------------

If we learn from our mistakes I'm getting a great education!

 

[waross]

Every time I read this thread, I remember your first post where you stated

I've verified that series operation is supported by the manufacturer (Astec/Emerson).

If series operation is supported by the manufacturer, I suggest following their recommendations for series operation.

Bill
--------------------
"Why not the best?"
Jimmy Carter

 

[Skogsgurra]

Two questions:

1 Is a DC motor your only choice? Or can you use an induction motor? VFDs can usually be fed from a DC source, if you have a battery and if you have AC or three phase available, most any VFD can be used unmodified.

2 Have you considered a DC chopper or a thyristor controller? They are light-weight and can be had in a wide power range. From FHP to 1000+ kW. Certainly where your needs are.


Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[itsmoked]

AC motor/VFD, excellent point Gunnar. More efficient too.

Keith Cress
kcress -

http://www.flaminsystems.com