1300VDC at the output of a PFC! Is it possible?

[delwin]

I have a strange experience last night.
I'm building a PFC stage. 230VAC/50Hz input, 390-400VDC output.
I'm using the NCP1654 pfc controller, and everything is by the reference design. I use the XLS file to calculate the needed values for the Rs and Cs.
I triple check it and still I'm getting one and the same strange behavior.
I have connected in series two 75W/230V tungsteen wire bulbs as load.
I'm using a DT-838 and DT-830B digital multimeters.
There is 1A fuse, the power switch is a N-MOS SPP20N60S5 (600V/20A/Rdson=0,21R), the diode is 15A/600V UltraFast.
Multimeter is connected across the + and - outputs of the PFC,and is switched to 1000VDC range.
After powering the damn thing, the bulbs get normally bright, as every one of it is connected to a regular 230VAC. Nothing blows, no smoke, but...
the multimeter says 1300VDC!!!!
I have tested it with the two multimeters, the same result.
What do you thing of this?

I'm puzzled, as if there were really 1300VDC the N-MOS and DIODE and the BULBs would be a history already.
Is the problem with the metering tools, or there really could be a 1300VDC?
The PFC is calculated for 1200W load nominal, but I'm testing it at about 75W just to see whether it works or not.

Some comments?
Thank you

The scheme the device is build around could be find at the end of the datasheet
https://www.google.bg/url?sa=t&rct=...sg=AFQjCNF9DmXgBO2gfrzs6PcYvqb_TGyiYg&cad=rja

 

[mister_rf]

Yes, that’s possible to read such a large value on the digital multimeters, as log we are measuring pulses on a DC scale. In this case you need to connect also an electrolytic capacitor on the PFC output.

 

[RF_Jim]

I have a strange experience last night.
I'm building a PFC stage. 230VAC/50Hz input, 390-400VDC output.
I'm using the NCP1654 pfc controller, and everything is by the reference design. I use the XLS file to calculate the needed values for the Rs and Cs.
I triple check it and still I'm getting one and the same strange behavior.
I have connected in series two 75W/230V tungsteen wire bulbs as load.
I'm using a DT-838 and DT-830B digital multimeters.
There is 1A fuse, the power switch is a N-MOS SPP20N60S5 (600V/20A/Rdson=0,21R), the diode is 15A/600V UltraFast.
Multimeter is connected across the + and - outputs of the PFC,and is switched to 1000VDC range.
After powering the damn thing, the bulbs get normally bright, as every one of it is connected to a regular 230VAC. Nothing blows, no smoke, but...
the multimeter says 1300VDC!!!!
I have tested it with the two multimeters, the same result.
What do you thing of this?

I'm puzzled, as if there were really 1300VDC the N-MOS and DIODE and the BULBs would be a history already.
Is the problem with the metering tools, or there really could be a 1300VDC?
The PFC is calculated for 1200W load nominal, but I'm testing it at about 75W just to see whether it works or not.

Some comments?
Thank you

The scheme the device is build around could be find at the end of the datasheet
https://www.google.bg/url?sa=t&rct=...sg=AFQjCNF9DmXgBO2gfrzs6PcYvqb_TGyiYg&cad=rja

Do you have access to an oscilloscope and an isolation transformer?

I think it would be prudent to do a quick inspect for 'spikes' which a non-RMS reading meter will read as 'peak' rather than rendering a more normal and expected reading.

Be very careful with hooking up the O-scope to your circuit; the scope chassis is not usually isolated from the 'ground' lead of the scope probe, so depending on your circuit you could have full line voltage on the scope frame!

Labs doing much work like this usually build enclosures in clear Plexiglas cases for the scope to be housed in, for safety sake, and personnel using it must be fully informed as to what can be done and what cannot about touching the instrument controls! We used to use short 'wands' made of clear acrylic plastic to 'touch' the controls of the scope through openings cut in the front of the protective Plexiglas case.

Jim

 

[delwin]

Yes, that’s possible to read such a large value on the digital multimeters, as log we are measuring pulses on a DC scale. In this case you need to connect also an electrolytic capacitor on the PFC output.

I have a elecrolytic capacitor at the output. 3x(200V/330uF) connected in series =330uF/600V, across the + - of the output.

Unfortnately I have no O-scope. Any other suggestions beside the scope, how to see the RMS or filter out the spikes you mention?

---------- Post added at 09:23 ---------- Previous post was at 08:42 ----------

Some images of the thing

 

[mister_rf]

Those high pulse sequences need to be smoothened by a intermediate DC capacitor 0.1uF/630V. Try to modify the PCB in order to accommodate both electrolytic capacitors. Use a 150k-220k parallel resistor for each capacitor. See the attached diagrams for details.

 

[delwin]

Those high pulse sequences need to be smoothened by a intermediate DC capacitor 0.1uF/630V. Try to modify the PCB in order to accommodate both electrolytic capacitors. Use a 150k-220k parallel resistor for each capacitor. See the attached diagrams for details.

I would definetely try this.
Are those hipotetical spikes usual in such a design. I'm really not far from the reference, even from the PCB they(ON semiconductors) originaly built.
besides the fast cap (0,1uF/630V) that could filter the spikes, I'll add resistors across each of the 330uF/200V caps.
Thank you

 

[mtwieg]

Well I think it's safe to assume you're not getting 1300V DC at the output, otherwise you would have had a catastrophic failure immediately. Even brief pulses at that voltage would probably destroy the diode and FET as well. So I think you must have some measurement error somewhere. Maybe your multimeters are not rated for the 390-400V you expect to see? Try putting a divider on the output and measure that.

 

[delwin]

Well I think it's safe to assume you're not getting 1300V DC at the output, otherwise you would have had a catastrophic failure immediately. Even brief pulses at that voltage would probably destroy the diode and FET as well. So I think you must have some measurement error somewhere. Maybe your multimeters are not rated for the 390-400V you expect to see? Try putting a divider on the output and measure that.

I agree. Infact I'm planning to do a mesurement on the lower resistor of the divider with an analogue volt meter. Eventhough I don't think a 400VDC are something a cheap chinese multimeter could not meter correctly.

 

[RF_Jim]

I have a elecrolytic capacitor at the output. 3x(200V/330uF) connected in series =330uF/600V, across the + - of the output.

Unfortnately I have no O-scope. Any other suggestions beside the scope, how to see the RMS or filter out the spikes you mention?

---------- Post added at 09:23 ---------- Previous post was at 08:42 ----------


Some images of the thing

The other suggestion might be:

The Fluke meters used to have a low-pass filter as part of the 'input conditioning' on their (now old) 8020A and 8024A series DVMs such that any pulses or high-frequency energy would not be read by the meter ... if you were to use say, use a 1k R in series with the voltmeter leads and then shunt the meter with like a .047 uF cap that might 'snub' the spike enough ('low passing' the DC) to get meaningful readings from the inexpensive Chinese meters ...

I have seen the venerable Triplett 630 VOM (simple meter with a diode to rectify the AC into a cap) 'read high' on an AC Line when a noisy RADAR Transmitter/Modulator was fired up (and created spikes on the AC Line) whereas the Fluke 8020A read true (and note: the 8020A is not an RMS instrument, it's that the LP filt made that much differeence) ...

With a 10 Meg Ohm input R DVM just placing a 1 Meg R in series with the meter leads and given the internal meter Capacitance this might be enough to dampen the spikes affecting the meter.

Jim

 

[delwin]

Hi there,
I have updates
last night I connected 5 x 100kR in series, across the output of the PFC. Then I meter the voltage across just one of them. The readings were correct 75VDC, not very stable though (70-75VDC) this would gives us 350-375VDC. The stability is an issue by itself, as I think +-3,3% is not that good.

However, another strange thing happens.
As I turn the power on, the output gets to about 275VDC (the controller do not start imediately), in some seconds (6-7) it goes on and the output gets to 375VDC.

Any suggestions? The power to the IC is through a 15V zener and an external battery (18V), the limiting resistor is 180R (it should supply 16mA to the output cap 100uF). The battery is connected to the PCB before the power is applied, so the IC is on, at the moment the Line goes high. So the supply of the IC is not the problem, as I see it.

Anyone?

Thanks again, and good day

 

[mtwieg]

Hi there,
I have updates
last night I connected 5 x 100kR in series, across the output of the PFC. Then I meter the voltage across just one of them. The readings were correct 75VDC, not very stable though (70-75VDC) this would gives us 350-375VDC. The stability is an issue by itself, as I think +-3,3% is not that good.

The ripple voltage on a PFC output can be pretty high, depending on the size of the output capacitor bank and the output power (Output ripple=Pout/(2*pi*120*Co*Vo). But given your numbers, you shouldn't be seeing that much variation at 75W.

However, another strange thing happens.
As I turn the power on, the output gets to about 275VDC (the controller do not start imediately), in some seconds (6-7) it goes on and the output gets to 375VDC.

Any suggestions? The power to the IC is through a 15V zener and an external battery (18V), the limiting resistor is 180R (it should supply 16mA to the output cap 100uF). The battery is connected to the PCB before the power is applied, so the IC is on, at the moment the Line goes high. So the supply of the IC is not the problem, as I see it.

You're really going to want an oscilloscope to debug the controller itself...

 

[delwin]

Yes, indeed.
The night before that, the IC fires instantly, and the output goes "1300VDC" (without the 500kR), I'll try it again this evening with no divider.