UCC28060: VREF Vs Temperature

Hi Miguel,

The datasheet specifies Vref to be between 5.82V and 6.18V from -40° to 125°C when there is no load on Vref. I will need to check if the data you requested is available for UCC28060. Is there any circuit connected to Vref that might be loading it down? This might explain the additional voltage difference you are seeing on Vref.

Best Regards,
Ben Lough

Hi Benjjamin,

thanks for your reply. VREF sees only about 0.13mA constant in our design, not sure if this is load is enough to cause additional drift? Additionally, the input voltage to the 28060 is also regulated, so there should be no line regulation issue either.

Thanks!

Miguel Valdez

Hello Miguel,

As Benjamin indicated, the VREF voltage should stay within +/-180mV of nominal over the entire rated temperature range of the IC.
I contacted the designer of the UCC28063 and he had made some minor upgrades to the VREF block of the original UCC28060/61.
But he said there were not intended to correct a 270mV shift in voltage over temperature. The reference of the UCC28060 should not have such a large shift and should conform to the specification table range, and typically perform much better.

How are you measuring this shift? Does it happen on other ICs as well, or only one?
If only one unit, perhaps there may have been some limited ESD damage or overstress during a probing mishap that weakened the circuit.
Please check these possibilities.

Regards,
Ulrich

Hello Ulrich
the thermal drift we are seeing is across all our units, not only one. I'm measuring VREF indirectly via VSENSE. Perhaps there will be a small difference but Id don't expect to be that large. If the newer ICs have an improvement in this regards I would like to try the -063, *-063A or even the -064A. What version would be the most compatible to the -060, or at least the one that will require less circuit changes?

thanks!

Hello Miguel,

The voltage at VSENSE should follow the voltage at VREF since it uses an integrating error amplifier.
For a typical 390V output, a 270mV shift at VSENSE should scale to ~17.6V shift at the PFC output voltage. You didn't say which direction (up or down), but it should be easily measurable.

If the PFC output does not closely track the VSENSE (and/or VREF) voltage shift, then I would suspect the VSENSE measurement instrument being effected somehow by the temperature change. Can you check this please?

The UCC28063 and -063A are more similar to the -060 (and -061) and mainly require component value changes to adapt.
The -063A is almost identical to the -063 except the "A" does NOT have open-pin and shorted-pin protection on the TSET and CS inputs.
The -064A has made some pin function changes that require some different circuitry. Here is a document that lists the differences between the -064A and the -063, in section 5: www.ti.com/.../slua878a.pdf

I could not find a comprehensive comparison guide for all these parts, so I'll have to summarize changes from -060 to -063 here:
VINAC = 060 has gain change for high and low AC input; 061 has no gain change, 063 same as 061
Brownout = 060 detects brownout after start-up; 061 detects brownout before start-up; 063 same as 061, hyst current from 7uA to 2uA
HVSEN = 060 2nd-level OVP stops PWM; 061 same as 060; 063 same as 061 except hyst current down from 36uA to 12uA and OVP triggers full soft-start
VSENSE OVP = 060 OV stops PWM; 061 same as 060; 063 LOW_OV at +8% pulls COMP down, HIGH_OV at +11% pulls COMP and stops PWM
Burst mode = 060 enters Burst at COMP < 150mV; 061 NO burst mode; 063 NO burst mode; 064A new Burst function
Phase Fail Detect = 060 disabled when PHB <.8V; 061 same as 060 or COMP < 222mV; 063 same as 061, CS threshold changed, too

There are a number of other changes in the UCC28063/A to reduce bias power and improve controllability.
Further improvements in the UCC28064A help to reduce light load losses. Please see the respective datasheets for the details.

Regards,
Ulrich

Hello Miguel,

Having received no further communication from you in the past week, I will conclude that the issue as been resolved and I will close out this thread.

Regards,
Ulrich

Hi Ulrich, thanks for your inputs and I'm sorry for the delay. I decided to order some samples of the 28063 for testing. I will comment on this later. Answering your questions: Our regulated Vout is 385V, and the output voltage is drifting up with temperature. Also, your estimations about 17.6V drift is about right for that drift on Vsense. I installed the UCC28063 in place of the 28060 and this time I measured Vref directly. The voltage on Vref is pretty much stable with temperature but Vsense does drift with temperature as you can see below. I put a thermocouple above the IC to get an estimate of the temperature.
I also used 53K resistor in series with my test probe on Vsense to minimize noise. Our low side resistor going from Vsense to gnd is a 10K with a 1000pF in parallel. I'm using a Fluke45 on ea. Vref and Vsense points.

Temp Vref Vsense Vout
35C 5.992 5.978 384.7
40C 5.992 5.982 385
45C 5.992 5.990 385.5
50C 5.993 6.006 386.5
55C 5.992 6.024 387.7
60C 5.992 6.060 389.6
65C 5.992 6.110 392.7
70C 5.992 6.160 395.4
75C 5.992 6.175 396.9
80C 5.992 6.189 397.6

How can Vsense be drifting with temperature if it is supposed to follow VREF as you mentioned before?

Thanks!
Miguel V.

Hello Miguel,

I'm sorry for my own delay. I managed to find a UCC28060EVM and ran a simple heat gun test on it.
I loaded it up with 10K (about 15W output). I didn't take meticulous data but got some rough numbers.

At lab ambient ~24C, Vout = 386.1Vdc.

At heated air stream 85C~100C, Vout = 388.8Vdc.

I held the heat gun (narrow nozzle directed at the IC at 7~8cm above) in place for several minutes and the 388.8V stayed solid.
Thermocouple in the air ~1cm from the IC provided the temp reading. I did not measure VREF.
All parts of this family -060/-061/-063/-064 should behave the same way in this regard.

I wonder if the method that you are using (53K inserted in series with VSENSE) is somehow allowing a leakage current increasing with temperature to shift the regulation point. I do believe that VSENSE = VREF at the error amp input, but with any series R, the "remote" VSENSE' may not track VREF as closely if any leakage currents can get in to that node. Also, can you please verify that the 53K is 53K and not 530K which would "amplify" the effects of a small leak current?

If you re-measure your system with normal connection to VSENSE (no additional measurement apparatus attached to it) and raise the temperature, does the Vout still increase by 12V+ or only a couple of volts?

Regards,
Ulrich

Hi Ulrich

thanks for your reply. Answering to your questions, I have verified that series resistor used was indeed 53K. I've also done measurements without series resistor and Vsense was a little higher. I've also checked our system without any measurement device attached to either Vref or Vsense and the output voltage stills drifts at that high level for instance at 50C ambient (I've measured around 20C-25C Trise to where the 28060 is located inside our unit).

I also used a heat gun to apply heat to the IC but didn't use a narrow nozzle, I'm going to experiment with that to see if there is any change.

Regards

Miguel V.

Hi Miguel,

I must admit, I am at a loss to explain this. I haven't heard of this issue happening before with other customers.
(It sounds like a cliché, but we have dealt with plenty of other issues, just not significantly rising VOUT with rising temp.)

I am making assumptions about your design, such that the VSENSE input network is simply the usual divider with R to GND and several series R's to VOUT and nothing else (no stand-by transistor or on/off switches, etc). If you have any deviation from this divider, please let me know it. If you can post your PFC schematic, or select portions of it, it could help.

One possibility for an unwanted parasitic component is uncleaned flux residue on the board after soldering.
Sometimes "no-clean" flux can be hygroscopic and form a leakage path from anywhere to anywhere else, especially where high voltages are present. Or the additional heat may change the residue's conductivity. Anyway, it's a long shot, but please check to make sure your board is clean at least along the VSENSE network.

Regards,
Ulrich

Hi Ulrich

thanks for your respond. Unfortunately our design is not as straight forward as you mentioned above or as on your evaluation board. We do use some transistors and other devices to maintain regulation at high line. I have repeated the heating test with nozzle and can see that heating the device alone doesn't cause that much Vout drift. So it may be something external that can be causing that. Unfortunately due to company policy I'm not allowed to share even portions of the schematic. I'm still concerned about seeing Vsense not following Vref though...which I this point I haven't find yet why. I'm currently looking at noise levels at Vsens and Hvsense inputs as we go high with temperature.

Thanks for your help.

Hi Miguel,

I understand about protecting proprietary information.

In general, then, if there are any PN junctions or MOSFET Vgs thresholds involved with the VSENSE signal path, then they could have a significant impact on the voltage at VOUT, based on their temperature coefficient. Delta Vbe or Vgs is "gained up" by the divider ratio.

As I mentioned before, I believe that VSENSE tracks VREF at the error amplifier inputs within the IC, which should also be the same at the IC pins. Perhaps there is an impedance between where you are measuring "VSENSE" and the actual VSENSE input pin. I would not be able to explain why VSENSE (at the pin) would not closely follow VREF (at the pin), unless your measurement gear is also affected by the temperature. Without knowing what your circuit looks like, I can't say how... I just can't explain it any other way.

If this helps you to resolve your issue, please let me know, and if I may close out this thread.

Regards,
Ulrich