UCC39002 System Implementation

Hi David,

Is the circuit similar to what's shown on page 5 in the data sheet? Can you attach a schematic?

Jeff,

I'm attaching the TINA-TI version. This is a schematic only; not to be simulated. The one thing to note is that each DC/DC converter and load controller reside on different PCBs. As a result, from a top level view, the inputs and outputs are connected using copper wire and headers. Again, this goes to my original question about using these devices in a system that contains multiple PCBs. I'm interested how it would affect the load sharing function considering the grounds are separated by the layout.

-David

0451.UCC39002-Rev1.TSC

Jeff, Here are the measured values from one PCB:

I_load=11A, R_sense=0.005Ohm A_csa=68 V_cso=3.7Vrms, 2Vp-p ripple VDD=9.9Vrms, 1Vp-p ripple V_adj=9.3Vrms, 1.5Vp-p ripple V_eao=650mVrms, 800mVp-p ripple V_LS=3.7Vrms, 7Vp-p ripple DC BUS=27.5Vrms, 600mVp-p ripple

The concern stems from the voltage on the load share pin. The ripple is very high. Any reason why that may be the case?

-David

Thanks for the added info David. We'll look at this.......

David,

I looked over the schematic you sent and it looks like the EAO and ADJ pin connections are swapped compared to Fig. 4 (HV App.) in the Data Sheet. This could be causing the problems you are seeing.

Jeff,

I sent you the wrong version. Sorry, the pins were swtiched on the first one. This new version refflects the change. Here is some addtional testing information:

V_CE=17.2V,  It does not appear to change with load current.

When I have two boards in parallel, one of the UCC39002 on one board should become master (set automatically by the internal diode) and the other should become slave.  I am noticing that the power supply that is not sourcing any current, the V_CSO<100mV and the V_EAO=3.6V (makes sense).  However, with the power supply that is sourcing current, the voltage across the sense resistor is 65mV and V_CSO=4.3V, but V_EAO=3.6V!  I also measure about 6mA through the adjust resistor on both boards!  This suggest that both chips think they are the slave.  The LS bus=4V.  I don't see how the chip with high V_CSO can also have an error.  By definition, V_LZ=V_CSO=>no error!

-David

8468.UCC39002-Rev2.TSC

The module being paralled is SMV-28-500, made by Astrodyne.
http://www.astrodyne.com/Product/smv500.asp

From the Astrodyne site, Technical Support:
http://www.astrodyne.com/Main_Faqs.asp 

Parallel operation does not appear to be recommended.
The SMV-500 datasheet makes no mention of it.
Converters with SYNCHRONOUS RECTIFICATION on the outputs have a very difficult time being paralleled.
Large circulating current may result.

Since we don't know the exact nature of the output stage, and there is nothing explicitly stating these modules can be shared, my conclusion is that they should not be paralleled. I suggest contacting Astrodyne with this question.
The UCC39002 WILL WORK with modules that are designed to be shared.

Note: Astrodyne does mention using ORING diodes for paralleling for redundancy, and this would also work for load share I believe though not stated by Astrodyne.

From Astrodyne website:
Back to top

Parallel Operation for Increased Power Output
Unless the power supply is specifically designed for parallel operation, parallel operation is not recommended. Two output voltages from fixed output converters will not be exactly equal. The converter with the higher output will try to provide the full load, causing it to go into current limit. The second supply will then do the same. Even if an adjust function is used to set the outputs, any drifts with time and temperature will cause the same reaction. Astrodyne recommends that you refer to the detailed specification sheets provided in downloadable PDF format on this web site, or Contact our Technical Support department.

Parallel Operation for Redundancy
Connecting power converters in parallel for power redundancy is an acceptable practice. The converters must have their outputs connected in parallel through two diodes. Each converter must be capable of driving the full load. The diodes permit one output to fail without affecting the other. Contact Astrodyne's Technical Support department for further information.

Upon further review, if the manufacturer of the module states they can be load shared, I have the following suggestions.

1) Remove the connection to TRIM, let it float, the UCC39002 is intended to adjust the Remote Sense (S+) of a module.

2) Following figure 4, High Voltage Application in the data sheet, the collector of the NPN should be tied DIRECTLY to the S+ of the module.
    There should be no series resistance like your schematic shows.

If this does not allow proper function, or at least improves it, we will need more answers from the module manufacturer.