The unfiltered drive of the PSU looks like this :

 

But the zener regulation along with the filter capacitors produces a rather good dc 12.8V-14V  positive and -5V3 negative.

Edited November 19, 20223 yr by BritishRacingGreen

Hello British RacingGreen. The best thing about the inverter is that it worked fine until I connected the ground and the PV input. So I know why he destroyed himself. I just don't understand why it doesn't work after replacing the faulty transistors and not finding another fault. In theory, the destruction could continue to the opticouplers or source for the igbt. Probably not anymore. It's weird for me to modify something that worked before. And despite the fact that I added the missing parts, the signal according to Coulomb is disturbing. If someone could measure the correct course of the signal on the igbt and perhaps also on other timing converters, it would help a lot in detecting the error.

26 minutes ago, nogers said:

Hello British RacingGreen. The best thing about the inverter is that it worked fine until I connected the ground and the PV input. So I know why he destroyed himself. I just don't understand why it doesn't work after replacing the faulty transistors and not finding another fault. In theory, the destruction could continue to the opticouplers or source for the igbt. Probably not anymore. It's weird for me to modify something that worked before. And despite the fact that I added the missing parts, the signal according to Coulomb is disturbing. If someone could measure the correct course of the signal on the igbt and perhaps also on other timing converters, it would help a lot in detecting the error.

Good day friend, fortunately i have been focussing on the igbt drives recently and willing to share  whatever i have gained. 

In the process i will do the required measurements for you, but only tommorrow. In the meanwhile i'll post the whole igbt drive schematic here, which we will use as a reference. 

@nogers please find attached the schematic for the IGBT drive section . This IGBT drive includes the DC-AC full bridge IGBT arrangement  as well as the BUCK converter IGBT . While  this schematic  is for the MAX 7.2 , the only difference there should be between your modul and this one is noted as follows :

1. The igbt opto-isolater chip is a 3-pin device , wheras on yours its an 8 pin T350 type . But interface / boundary functionality remains the same 

2. The MAx has two IGBT's in parralel whereas yours only one.

3. The IGBT transistor types differs .

There are three isolated IGBT power supplies that I have labelled PSU  A, PSU B and PSU C. 

4. PSU A drives one IGBT transistor in the full bridge 

5. PSU B drives one IGBT transistor in the full bridge 

6 PSU C drives two IGBT transistors in the full bridge  , as well as the BUCK IGBT. The reason for this is because all three IGBT's are reference to BUS-.

7. there is a little context diagram in the bottom left hand corner .

 

 

 

Edited November 19, 20223 yr by BritishRacingGreen
Misspelling

I'm glad to find people on this forum so welcoming and helpful.  Thanks.  I also thought of measuring the consumption of sources for the igbt with an external source of approx. 20V through a 1k ohm resistor and two have 2mA at a voltage of 17.5V, and the third connected to the BUS- takes 4.3mA and 14.5V.  Sources and static subscriptions seem to be fine.

5 hours ago, nogers said:

I'm glad to find people on this forum so welcoming and helpful.  Thanks.  I also thought of measuring the consumption of sources for the igbt with an external source of approx. 20V through a 1k ohm resistor and two have 2mA at a voltage of 17.5V, and the third connected to the BUS- takes 4.3mA and 14.5V.  Sources and static subscriptions seem to be fine.

Thank you .

So  the first thing I do is to power up my main board with  just about nothing attached to it except a current limited 60V power supply. So I remove all connections and all other modules including the control board (DSP) . The battery , mains and PV must not be connected to the bus . The bus voltage should be zero , if not bleed it off with a resistor. Make sure you have no short circuit on BUS+ and BUS- by multi-meter resistance measurement.

I am not too sure if my MAX  main power supply connection is the same as yours. The MAX has an external SPS module which receives mains as an input , and produces about 60VDC as an output . This is fed to the main board via a 4 pin Molex connector to power the main PSU . This is where you connect a  40V variable bench PSU and limit it to about 2 amps. If you don't have that high voltage , use the existing SPS module and manually feed it with mains. Anything lower than 40V and the main PSU will not bootstrap. Once the  main board is powered , you should measure the main system rail voltages , i.e.. 12V ,-12V and 5V . Please take care and make sure your external supply polarities are correctly matched to the main board.

Because there is no control module , a soft start will not be initiated and thus the BUS will remain at 0V . Also the DC-DC converter Mosfets will be switch off because their PWM controller is in shutdown state. The AC side IGBT's will be negated and there should be about -5.3V on the gates of the the 4 IGBT's as well as the Buck converter IGBT. So there is no control signal from the DSP , which allows you to measure the static dc levels of the IGBT drives . 

 

Following is list of DC (static)  measurements :

Remove your external power source , and test with a multi-meter that you have adequate isolation ( insulation) between the 4 sets of windings on transformer TX7 .  This does not prove breakdown at high voltages , but it will for sure expose any hard isolation fault.

 Turn on your power supply , and for  IGBT PSU A ,measure with your scope to inspect the quality of dc  over ZD7 (5.3V) and ZD8 (12.8-14.5V) . It should be smooth dc with no ripple to report. Repeat the same test for PSU B and PSU C after identifying the relevant Zener  diodes.

If this test passes , check that the negative voltage (the -5V3) is routed out of the opto-coupled driver to the gates of the transistor/s.

Now , for each IGBT drive , turn the IGBT On by shorting its control input to GND . Example to turn the gate voltage on for IGBT QB2 , you will have to short the CN11-1  pin to system ground (GND) . Now check your gate  voltage level to be at least +12.5VDC . If so you have turned the gate 'ON'.

Repeat this for all five transistor , making sure you do not experiencing sagging voltages and ripple. Now turn on all 5 transistors , and make sure your gate signal levels are  still ok.

If successful, you have completed the DC character test of the IGBT drives. However I suspect you will already experience problems here.

 

Looking forward to your response .

 

EDIT : your Hantek Scopemeter is just about your biggest asset in all of the IGBT tests, so this is a great positive 👌

EDIT : you may find that some of my descriptions is bleedingly obvious , and I also know in your posts that you are qualified and experience. Just bare with me , as these walkthroughs  will eventually be compiled for the purpose of other people that may be absolute beginners.

 

 

 

 

 

 

 

 

 

 

Edited November 19, 20223 yr by BritishRacingGreen

Hi. I did all the measurements again exactly as per your instructions. The only difference is the voltage level of zd8 (12.8-14.5v), in my case it is only 8.3V and on igbt 7-7.3v when all igbt transistors are switched on. The voltage without igbt switching on zd8 is 8.49V. All voltage is generated directly from the tx9 transformer. I have already measured these pulses and according to Coulomb everything is fine up to the level of positive voltage. No ripples or other deviations were detected. I used the original source for all measurements and it has an output voltage of 48.5v. My power supply is only 32v 10A.

On 2022/11/20 at 10:31 PM, nogers said:

Hi. I did all the measurements again exactly as per your instructions. The only difference is the voltage level of zd8 (12.8-14.5v), in my case it is only 8.3V and on igbt 7-7.3v when all igbt transistors are switched on. The voltage without igbt switching on zd8 is 8.49V. All voltage is generated directly from the tx9 transformer. I have already measured these pulses and according to Coulomb everything is fine up to the level of positive voltage. No ripples or other deviations were detected. I used the original source for all measurements and it has an output voltage of 48.5v. My power supply is only 32v 10A.

Hi @nogers, apologies , i never got a notification of a message on this thread . Next time if you can please make reference to @BritishRacingGreen  in your post .

Ok , thats interesting , i would have thought the problematic section to be PSU C because its driving 3 circuits. ZD8 is associated with PSU A which only drives one IGBT .

So how does your supply over ZD8 looks like on the oscilloscope  when you drive the transistor on , and also when you turn it off ?

The fact that you only getting 8V could be from the transformer secondary not supplying enough , or the circuit is loaded via the opto-coupler U3 or the transistor itself.

 

 

Hello @BritishRacingGreen. Such a small voltage is not only on zd8, but on all zd for igbt. I also measured the load of the individual Igbt sources and measured 2mA on two and 4mA on the third. This eliminated the possibility of a short circuit, even when all igbts were switched on. Sorry for the inaccurate description. I think the overall low voltage comes from tx9 and d20 (in my plan and on my d40 board) tx9 is a +12,+5,-12V source and these spikes are all fine. Therefore, it is not clear to me why the Igbt voltage is not correct.

10 minutes ago, nogers said:

Hello @BritishRacingGreen. Such a small voltage is not only on zd8, but on all zd for igbt. I also measured the load of the individual Igbt sources and measured 2mA on two and 4mA on the third. This eliminated the possibility of a short circuit, even when all igbts were switched on. Sorry for the inaccurate description. I think the overall low voltage comes from tx9 and d20 (in my plan and on my d40 board) tx9 is a +12,+5,-12V source and these spikes are all fine. Therefore, it is not clear to me why the Igbt voltage is not correct.

ok , on Friday I will send you waveforms  for the primary and the three secondaries of  tx7  so you can check if the primary voltage is within spec. You may even have a faulty tx7 transformer . I actually had to rewound my transformer and i know of others that also had them faulty.

Hi @BritishRacingGreen My converter is only a few months old and it doesn't look like any of the components around the igbt are damaged. It is also clear that the short PV to ground caused the destruction of the igbt. But if the transformer tx7 or tx9 should be damaged and the surrounding components are not damaged, it does not seem logical to me. I'm starting to wonder what could be causing it. Thanks.

Hi @nogers following are waveforms captured on the IGBT power supply  :

1. Below is the waveform on the primary of Tx7 , Notice the peak to peak swing of about 36V .

2. This is the Tx7 output across seconadry windings of PSU A . Note the transformar has a voltage gain , the swing is about 55V peak to peak :

 

 

 

3. This is the IGBT positive supply across ZD8 . Its about 14.4VDC and is rather smooth DC :

 

4. And here is the Negative IGBT supply across ZD7 . About 5v6  and also smooth DC :

 

 

 

 

Edited November 27, 20223 yr by BritishRacingGreen
Correction

hi @BritishRacingGreen. So everything absolutely agrees. As for the voltage level, it is about 25V at the input and your measurement shows 36V, that's the only difference. Maybe it's meant to be, and the problem is somewhere else entirely. Basically, I also have a borderline input voltage of 48V. and you said that the min is 40V and the max is about 60. This voltage would theoretically raise the voltage overall even on tx7.

3 hours ago, nogers said:

As for the voltage level, it is about 25V at the input and your measurement shows 36V, that's the only difference. Maybe it's meant to be,

I don't think it's meant to be that low. It should be fairly close to the 48 V of your power supply. I suspect that something is weak along the chain supplying the power supply: perhaps a diode or choke in series with the 48 V supply, the MOSFET itself, its source resistor, or the gate drive might be low. The source resistor seems like the most likely culprit. If somewhat high resistance, it will limit the current drive to the transformer, which would affect output voltage and possibly cause the voltage sag you also see.

It will be quite low resistance, less than one ohm, so it will be non-trivial to measure accurately. Zero your multimeter for this measurement. Consider using a current limited power supply and measure the current and voltage drop to calculate the resistance.

Edited November 27, 20223 yr by Coulomb

So I measured almost everything possible and did not find the error again. R215 has 149m ohm. Measured with a Grundig RLC200 four-point RLC meter. The power board has 48v written on its output and its output is about 47v. That's okay too. I also measured the excitation voltage of transistor Q36 about 10V and the input voltage of tx9. See picture. I think everything is fine.

5 hours ago, nogers said:

I think everything is fine.

if I'm reading correctly, your timebase is 2 milliseconds, and BritishRacingGreen's is 2 microseconds. Your oscillator appears to be running at about 350 Hz; it should be about 100 kHz. That's about 280x too slow. That would explain the sagging gate drive for the IGBTs.

Check the timing resistor (R208 on mine) and capacitor (C67); they should be about 3kΩ and 39 nF. Although RC is then 3 x 10³ x 39 x 10⁻⁹ ≅ 120 x 10⁻⁶, which seems 10x too long. So maybe the capacitor should be 3.9 nF, or I'm misreading BRG's DSO photos. Certainly, the transformer will not work very well at 350 Hz. If those are good, it may be the oscillator chip, U10. What does Vref read with respect to local ground, i.e. pin 8 to pin 5?

Edit: though now that I look again at my spare board, it says 89B for R208, which is 8.25 kΩ (and it measures close to that), so perhaps it's meant to be 8.25 kΩ and 3.9 nF, for a frequency of 1.72/RC = 1.72 / (32.2 x 10⁻⁶) = 53 kHz. At ≈50% duty cycle, the frequency may be 2x that, about right. I note that the datasheet says never to use a timing resistor value less than 5 kΩ, so the 3 kΩ value on my traced schematic must be wrong.

Edited November 29, 20223 yr by Coulomb

@nogers, @Coulomb  yes !!!!! , Coulomb picked up an important difference here , indeed the SMPS base frequency is around 100khz (10us).  Below is one of my early captures when the input voltage (SPS+) is about 56V , no control board , the battery side mosfet drivers disabled .At that input and load the pulse widt is about 800nS.

@nogers check your U10 (UC3845) that you have at least 11VDC on pin 7 and 5.0VDC on pin 8(vref) . Note of course your reference (0V)  is BAT- (not GND).

 

EDIT : when working with the SMPS , you can get some nasty and unforgiving failure of the rail supplies .  connect at the least a 18V transzorb (or do better ) on the +12V dc rail. The reason for this is the +12V rail is the one that's regulated by feedback control. All the other rails are just a derivative of this . (hence the reason why -12V and +5V rails has secondary linear regulators).  These failure modes can really destroy the rest of the board .

EDIT : at 56V input the pulse width is only 800ns in order to regulate the +12V rail at 12.3V . When the circuit fails and  U10 deliver its max of 5uS pulse width (50% of 10us) , you can imagine how the output rails will shoot up !

Edited November 29, 20223 yr by BritishRacingGreen

Hello @BritishRacingGreen. So after a long time I finally got around to measuring. pin 5-8 4.95V and 5-7 12.28V so absolutely fine. Resistor and capacitor also fine (3.9nf and 8.25ohm). The input to the transistor was measured and on the range of 2 microseconds it looks the same as your measurement. I don't see a problem here. I don't know how I got the 2ms measurement range. Any other idea what it could be?

All I can think of relates to this image:

I assumed that this was really bad ringing, but now I'm thinking that this is high frequency signals getting sampled. It might be that this is pure artifact, you just happen to sample at times that makes this pattern, but given how clean it looks at high frequency, I doubt that. My guess is that the power supply has ≈350 Hz ripple on it. Can you check what the power supply pin looks like with 2 ms timebase, please? It might be that the power supply has a huge sawtooth on it, due to some large capacitor not doing its job.

Edit: Or maybe the power to the chip is OK and protected against the sawtooth by the diode(s), but the power supply to the MOSFET and "transformer" primary may look like that.

Edited December 5, 20223 yr by Coulomb

Hi coulombe. So it's not a source either. The supply voltage of the source is smooth and stable. It should be noted that the converter is only a few months old and has been in operation for about a month. It was fully functional until I connected ground and the PV input. This caused the destruction of the Igbt and a short circuit. I believe it could destroy something on the output of the inverter and that is now destroying the new igbt again. The problem is that I can't even find any errors in the output. For example, a faulty capacitor or relay. Low switching voltage of Igbt should not destroy the inverter without load. Do you think this idea is correct?

5 hours ago, nogers said:

Low switching voltage of Igbt should not destroy the inverter without load. Do you think this idea is correct?

Not necessarily. The low gate voltage might cause the IGBT to switch more slowly, which could possibly cause shoot-through (both transistors conducting, shorting the DC bus). But your -5 V gate signal seems to be high enough, and that's what turns the transistors off, so you are probably OK with this assumption.