UD 2.7c "sawtooth" output

[Felix M.]

Hello everyone,

For around 2 months now I've been building my first drsstc. Lats week my bridge showed its first life sign with Vbus=60V and OCD=200A (it never tripped).
However I can't conduct more power tests due to the strange looking output of my UD 2.7C, since it might cause IGBT damage a higher power.
I have hooked up a signal generator at my desired frequency to simulate feedback. I did some scoping but couldn't find the problem.
This shows the UD's output with a 10ohm load resistor:


The 24V regualtor does its job and the ripples are minimal so my guess would be the logic. Here is an other shot of the MOSFET driver IC.


The signal on the gate looks like this:


After going to 24-25V, it ramps down to around 18V. (There is a 5.1ohm gate resistor, a diode for faster discharge and some 33V zener diodes for protection)


I appreciate every peace of information

Cheers

[Weston]

What are the specifications on your gate drive transformer / what does it look like / what core did you use?

The waveforms look like the magnetizing inductance of your GDT is too low (too few turns / core with too low an Al value). A low magnetizing inductance causes a high magnetizing current. The IR drop due to the magnetizing current causes the waveforms to look sawtoothed.

[Felix M.]

I'm using a large N30 ferrite core with 15 windings for my fullbridge setup. It's driving two SKM200GB128D bricks.

I think the problem isn't actually the GDT because like the first picture shows, the output without the GDT is allready not a square wave.
The GDT without any gates connected spits out the same sawtooth wave it gets as an input.

[Late]

Try probing along the signal path and see if waveform in all places looks okay. Mainly check the input of the high speed comparator, both outputs of it, inputs to the AND gate and inputs and outputs of the gate drive ic. Send all those scope shots here.

I have hooked up a signal generator at my desired frequency to simulate feedback.

What signal are you feeding in the input? sine or squarewave or smth else?

[Felix M.]

I'll do the scoping tomorrow, it's already late.

The feedback is a 80kHz sine wave at around 6-8V

[Late]

I'll do the scoping tomorrow, it's already late.

Okay, sounds good. No rush

The feedback is a 80kHz sine wave at around 6-8V

That sounds correct

[Felix M.]

One thing I forgot to mention: I don‘t have the SMD but the THT version of the UD 2.7C. It probably doesn‘t matter anyways but I saw you have experience with this driver in particular so if there‘s anything you had problems with, please let me know.

[Late]

It probably doesn‘t matter anyways but I saw you have experience with this driver in particular

Yeah it doesn't matter, all the parts and circuit is the same just tht. And I've made the driver 2 times and in neither of them did I have issues that the output is sawtooth. Usually when the output is sawtooth it means that the GDT is saturating.

I don't know of you mentioned it before but how many turns did you have on the GDT and what wire did you use to wind it? Also what's the core size and what igbts are you using?
Maybe send a pic of the GDT

[davekni]

Here is an other shot of the MOSFET driver IC.

Looks like roughly 4V amplitude.  If I'm understanding correctly that this is output of UCC27423 chip, it should be 9V amplitude.  Suggests one of three possible issues:
     9V supply is low (4 to 5V).
     UCC27423 chips are bad.
     Something is loading down UCC27423 chip output(s) excessively.

Good luck with debug!

[Anders Mikkelsen]

The UD output stage is not designed to be loaded by a resistor, but by the gates which are mainly capacitive. A 10 ohm resistive load in series with the 2.2 µF GDT coupling capacitor gives a 22 µs time constant, which would explain the sloping of the gate driver output waveform. If you use a more representative load, maybe some ten nanofarads in series with a few ohms, the waveforms should be closer to reality.

[Felix M.]

Thanks for your advice. I'm gonna do some test and post some scope shots soon.

[AeraCura_]

I believe what you're seeing is expected. See https://www.loneoceans.com/labs/ud27/,

Specifically the section labeled "Above shows the secondary of the GDT across a small resistor at 50kHz and 500khz." which looks exactly like your image with the 5.1 Ohm resistor.

Further, if you look at this image. https://www.loneoceans.com/labs/ud27/ud27atopview2.jpg you can see that Lone Oceans is using the exact same 5.1 Ohm resistor on the GDT.

[Felix M.]

Further, if you look at this image. https://www.loneoceans.com/labs/ud27/ud27atopview2.jpg you can see that Lone Oceans is using the exact same 5.1 Ohm resistor on the GDT.

The unloadet output of the driver should be a square wave:


In my case there is no GDT connected and it's already a sawtooth:

[Late]

In my case there is no GDT connected and it's already a sawtooth:

What does the input and output of the driver IC looks like?

[Felix M.]

Driver IC input:


Driver IC output:

[Late]

This looks very wrong. At least looks like the driver IC and the output stage isn't the issue. Could you probe along the signal path? from the input up until the driver ic.

Also what are you using as your interrupter signal?

[Late]

Here i attached the UD2.7C THT schematic. Probe in all the places I've circled red. for the driver IC inputs and outputs probe both at the same time using both of your scopes channels.

[Felix M.]

Thanky Late, I'm off to the lab 

[Felix M.]

Feedback input:


Ultra fast comparator input:


Ultra fast comparator output:


74hc08 output:


UCC27243 input:


UCC27243 output:


It's getting more and more strange since the 74hc08 output is directly connected to the UCC27243 input so the signal should be the same but it's not. For some reason my trigger and othe functiones of the scope didn't do what they are supposed to do while testing. So lastly I measured the output of the driver with channel 2 an I got this:


The scope was likely the problem all the time because it measures those ramps in square waves. It's good because it means the driver works but it's also really bad because now I have to troubleshoot on my scope. I will runn some further tests to verify this but it makes sence to me.

[Late]

Comparator output and 74hc08 output looks very weird and not like it should be. Looks like there's a problem with your ultra fast comparator.
Looks like one of the inputs and outputs for the driver ic looks okay

Have you connected the phase setting jumpers correctly? and are both of them installed? Also could you send a pic of the driver board?

[Felix M.]

That's the driver. It's powered by 24VAC. For "dry tests" the ODC is removed and the feedback is hooked up to the function generator.


I'm not too sure which jumper do you mean. Is it J4 or J6? J4 is open and the middle and lower pins are connected together for J6.

[Late]

I'm not too sure which jumper do you mean. Is it J4 or J6? J4 is open and the middle and lower pins are connected together for J6.

On the jumper J6 pins 1, 3 and 2, 4 or 3, 5 and 4, 6 have to be connected together.

In the pic it doesn't look like they are connected together.

[Late]

That's the driver. It's powered by 24VAC. For "dry tests" the ODC is removed and the feedback is hooked up to the function generator.


I'm not too sure which jumper do you mean. Is it J4 or J6? J4 is open and the middle and lower pins are connected together for J6.

Also i don't see any interrupter signal being fed in. Are you feeding in any for your dry tests?

Looking at the previously sent waveforms i suspect there's something wrong with the ultra fast comparator or some connection near it. Do you have another comparator you could put in to check of that is the issue?

[Felix M.]

Yes, I forgot to mention it. The interrupter signal is fed in by a fiberoptic cable. I'm going to buy some new comparators soon to check if it makes any difference bun I'm quite sure it was the scope all the time.

[Late]

Yes, I forgot to mention it. The interrupter signal is fed in by a fiberoptic cable. I'm going to buy some new comparators soon to check if it makes any difference bun I'm quite sure it was the scope all the time.

How is the scope connected to the circuit and is the probe set to 10X or 1X?

[Felix M.]

The probe is set to x1. I'm not too sure if it's a good thing but the circuit is floating due to the ringcore transformer I use as a supply so I can scope everything without worrying to blow something due to wroung ground connections.

[Late]

The probe is set to x1.

You should always keep your probe in 10X mode unless you're measuring very low voltages at low frequencies. In 10X mode the probe will also have a less of an effect on circuit. in 1X mode scope probe isn't rated up to the full bandwidth and voltage.

And was the ground lead connected to GND for all measurements?

[Felix M.]

And was the ground lead connected to GND for all measurements?

no

[Late]

no

uhh it should have been connected. otherwise you have just one connection and u need both. for most measurements connect it to circuits negative terminal but about the output be careful. if its isolated then you should be able to connect the ground lead to one terminal and probe to other. just make sure the other ground lead isnt connected to anything

[Felix M.]

I‘ll kep it in mind for the future. The problem I have right now is a probably damaged channel 1 on my scope

[Late]

likely just a probe or maybe even nothing at all. Have you tried connecting it to the scopes cal port for test?

And possibly the measurements looked like that because the ground lead of probe wasn't attached

[Felix M.]

Both pobes don‘t work on ch1 but on ch2 non of them has any problems. I connected them to the callibration port to see in ch1 that the square wave goes up like it should be but then instantly ramps down back to 0V. Ch2 on the other hand works fine.

[Late]

Both pobes don‘t work on ch1 but on ch2 non of them has any problems. I connected them to the callibration port to see in ch1 that the square wave goes up like it should be but then instantly ramps down back to 0V. Ch2 on the other hand works fine.

Hmm does switching from 1x to 10x on probe change anything? Ofc not counting the total amplitude

If not then yea something is cooked inside scope. Could be bc u measured something that's highish voltage with probes in 1x mode

[Rafft]

power just UD circuit only.

[Felix M.]

Hmm does switching from 1x to 10x on probe change anything? Ofc not counting the total amplitude

If not then yea something is cooked inside scope. Could be bc u measured something that's highish voltage with probes in 1x mode

Switsching between x1 and x10 doesn‘t chandge the strange measurements. Strangely a sine wave is no problem at all but the amplitude and voltage measurements in general are off. There was a case in the past with a CT where there was a potential of overvoltage but if I shot it with ov, wouldn‘t it measure anything?

[Egg]

Both pobes don‘t work on ch1 but on ch2 non of them has any problems. I connected them to the callibration port to see in ch1 that the square wave goes up like it should be but then instantly ramps down back to 0V. Ch2 on the other hand works fine.

I have the same problem with my oscilloscope. Both channels work but first channel reads -140/-150v rms at all times making measuring low voltages almost impossible. Calibrating does not help either.

[Felix M.]

Did you also have a situation where the scope might have seen too much voltage?

[Late]

Switsching between x1 and x10 doesn‘t chandge the strange measurements. Strangely a sine wave is no problem at all but the amplitude and voltage measurements in general are off. There was a case in the past with a CT where there was a potential of overvoltage but if I shot it with ov, wouldn‘t it measure anything?

Yea seems like you have cooked something in the scope. For the future remember when you measure voltages in a CT always put a resistor in paralel with output. Something like a 10-100ohm resistor.

If you repeat all measurements of the circuit with the second channel in 10x mode do you still get a sawtooth wave on output?



edit - Also always keep your scope probe in 10X mode. Your scope is only rated to the max voltage and bandwidth in 10X mode and you're less likely to cook something with it.

[Felix M.]

I didn't had the time yet to remeasure everything but the output of the driver looks good. It has some spikes but it's a squarewave. The shot was done with no load whatsoever.


I talked to a friend of mine who has a lot of experience in the whole field and he mentioned that the "sawtooth" might look like a capacitor's discharge curve:


This would explain the strange signal that accures when the probe is hooked up to the calibratin port at the scope which puts out a 1kHz squarewave. There the discharge curves are even more visible.

[Late]

I didn't had the time yet to remeasure everything but the output of the driver looks good. It has some spikes but it's a squarewave. The shot was done with no load whatsoever.

Yup the output looks good. It looks like its supposed to

[Late]

I talked to a friend of mine who has a lot of experience in the whole field and he mentioned that the "sawtooth" might look like a capacitor's discharge curve:


This would explain the strange signal that accures when the probe is hooked up to the calibratin port at the scope which puts out a 1kHz squarewave. There the discharge curves are even more visible.

Yeah sawtooth can be a capacitor discharge. Tho it certantly isnt what the signal from cal port of scope should look like. Something is deffinately cooked inside the scope

[Felix M.]

Yeah sawtooth can be a capacitor discharge. Tho it certantly isnt what the signal from cal port of scope should look like. Something is deffinately cooked inside the scope

This shot is just the UD2.7 output because I don't have a shot of the calport signal on hand but it also shows the problem.

[Anders Mikkelsen]

It looks like your measurement is AC coupled with a pretty short time constant. How are you grounding your scope to the UD? Do you have a picture or an illustration of your setup, showing how and where you make the measurement?

[Felix M.]

I don't have pictures of my measurement setup but like I said before, my UD is floating due to the ring core transformer I use a sa power supply. Scoping the output is just a matter of connecting the probe head to one and the ground to the other output terminal.

[Late]

Yea some pictures of measuring setup would be very helpful. Also do you have set AC or DC coupling for the scope? For this measurement it should be DC

[Felix M.]

The whole sawtooth problem is due to the fried channel 1 on my scope!

It doesn't matter how you turn or change it, it just saw too much voltage and died. Channel 2 confirms that the driver works fine and produces a squarewave. There are some further problems with the bridge but that doesn't belong to the "sawtooth" section.

[Late]

It doesn't matter how you turn or change it, it just saw too much voltage and died. Channel 2 confirms that the driver works fine and produces a squarewave. There are some further problems with the bridge but that doesn't belong to the "sawtooth" section.

A okay. Yea the waveforms on ch2 looked to be as they should be