Other trigger modes resulted in the same anomaly being present once warmed up.
Just tried it and on single capture (SGL) the anomaly is not present.
Just tried it and on single capture (SGL) the anomaly is not present.
Last edited:
What ground are you using as a reference for your scope?
Does it having the same problem in differential mode?
Does it do this with other amps?
Does it having the same problem in differential mode?
Does it do this with other amps?
I have not located another amp with an SG3525 yet, so I tested one with a TL494 PWM chip. It was clean and remained clean throughout my testing.
Still working on figuring out differential mode testing. And I need to open up more amps to see if I have another with the SG3525.
Still working on figuring out differential mode testing. And I need to open up more amps to see if I have another with the SG3525.
Differential mode is below. Another option may be to use pin 4 or pin 7 to drive the external trigger input of your scope. You'll have to select the external trigger as the source.
A differential input uses two inputs to produce a single waveform. The simplest way to get a differential input is to use a differential probe. A differential probe has two signal leads and a mixer amplifier built into it. It feeds the scope a normal signal (a composite of the two signals input into the differential probe). The problem with differential probes is that they're expensive.
The alternative is to use two scope probes and and both inputs of your oscilloscope. This is how you have to set up your scope:
Two probes
Both scope inputs used
Input set to add
Both channels set to DC coupling
Both channels set to 'cal'.
Both vertical amps set to the same voltage
Ch2 input set to invert
Bandwidth limited (works best for most measurements in car amps)
Trace aligned to the reference line on the scope's display
Ground leads for both probes connected together (not always necessary)
After setting up the scope, you need to confirm that it's working as it should. With the vertical amp set to 5v/div, touching the probe that's connected to Ch1 to the positive terminal of your 12v power supply should make the trace deflect about 2.5 divisions up from the reference (like it always does, seen below). Doing the same with the probe connected to Ch2 should make the trace deflect down about 2.5 divisions. Touching both probes to the positive terminal of the 12v power supply should cause no deflection. If it does, something isn't right.
I know that this may not be as simple as the isolated scope but if you take the time to learn it one time (even if it takes an hour or more of your time), you have that knowledge and this tool to use for the rest of the time you need to use a scope. Using the analog scope will give you much larger and cleaner waveforms.
A differential input uses two inputs to produce a single waveform. The simplest way to get a differential input is to use a differential probe. A differential probe has two signal leads and a mixer amplifier built into it. It feeds the scope a normal signal (a composite of the two signals input into the differential probe). The problem with differential probes is that they're expensive.
The alternative is to use two scope probes and and both inputs of your oscilloscope. This is how you have to set up your scope:
Two probes
Both scope inputs used
Input set to add
Both channels set to DC coupling
Both channels set to 'cal'.
Both vertical amps set to the same voltage
Ch2 input set to invert
Bandwidth limited (works best for most measurements in car amps)
Trace aligned to the reference line on the scope's display
Ground leads for both probes connected together (not always necessary)
After setting up the scope, you need to confirm that it's working as it should. With the vertical amp set to 5v/div, touching the probe that's connected to Ch1 to the positive terminal of your 12v power supply should make the trace deflect about 2.5 divisions up from the reference (like it always does, seen below). Doing the same with the probe connected to Ch2 should make the trace deflect down about 2.5 divisions. Touching both probes to the positive terminal of the 12v power supply should cause no deflection. If it does, something isn't right.
I know that this may not be as simple as the isolated scope but if you take the time to learn it one time (even if it takes an hour or more of your time), you have that knowledge and this tool to use for the rest of the time you need to use a scope. Using the analog scope will give you much larger and cleaner waveforms.
I believe I have the scope setup for differential measurements based on your response and the user manual. But I am not getting the expected results, when I place one probe on ground and one on 12V+ all I get is noise. I tried cycling trigger modes, volts/div and timebase with no luck. Probing the SG3525 output results in similar or the same noise.
Scope setup for differential, with channel 1 inverted based on my translated manual it says channel II invert is disabled in addition mode.
CH1 on 12V+ CH2 on ground. Same result if I swap probe positions and if I completely swap probe channels.
Same measurement points lower volts/div.
CH1 on output pin of SG3525 CH2 on main ground.
I tried it with and without the ground jumper between the probes, it looks the same.
The closest I got was trying some old Tektronix probes (mismatched). I did get close to a DC level when probing the 12V+ and ground, but the waveform was a bit wavy (not a straight line). Based on the fluke probes kinda working I am thinking maybe I should order two new probes.
Here is the translated manual page I found the add settings on.
Scope setup for differential, with channel 1 inverted based on my translated manual it says channel II invert is disabled in addition mode.
CH1 on 12V+ CH2 on ground. Same result if I swap probe positions and if I completely swap probe channels.
Same measurement points lower volts/div.
CH1 on output pin of SG3525 CH2 on main ground.
I tried it with and without the ground jumper between the probes, it looks the same.
The closest I got was trying some old Tektronix probes (mismatched). I did get close to a DC level when probing the 12V+ and ground, but the waveform was a bit wavy (not a straight line). Based on the fluke probes kinda working I am thinking maybe I should order two new probes.
Here is the translated manual page I found the add settings on.
Is the scope ground connected to the 12v power supply ground? If not, connect them.
If you use either probe alone, do you get a clean waveform? You'll have to choose the trigger mode for whichever probe/channel you're using.
If you use either probe alone, do you get a clean waveform? You'll have to choose the trigger mode for whichever probe/channel you're using.
The grounds were not connected to the 12V ground, I thought that would be correct for a differential measurement. Where channel 1 probe tip would measure the pin on the SG3525A and channel 2 probe tip would be connected to 12V ground. I did try it with the probe grounds connected together, it did not make a difference.
With the probes un-grounded the waveform displayed is noise, no resemblance to a square wave.
I just tried again with channel 1 (inverted, AC coupling) probe tip would measure the pin on the 12V ground and channel 2 (not inverted, AC coupling) probe tip connected to SG3525A. The result was the same noise, until I connected the ground (at which point I could remove probe 1 from the 12V ground with no change to the displayed wave form). But I feel like this is not a differential measurement.
Differential or not, the waveform behaved exactly the same, it starts out clean and then the a second out phase square wave appears. Cooling the chip makes it go away until it heats back up.
With the probes un-grounded the waveform displayed is noise, no resemblance to a square wave.
I just tried again with channel 1 (inverted, AC coupling) probe tip would measure the pin on the 12V ground and channel 2 (not inverted, AC coupling) probe tip connected to SG3525A. The result was the same noise, until I connected the ground (at which point I could remove probe 1 from the 12V ground with no change to the displayed wave form). But I feel like this is not a differential measurement.
Differential or not, the waveform behaved exactly the same, it starts out clean and then the a second out phase square wave appears. Cooling the chip makes it go away until it heats back up.
Use a square wave generator to see if you can get the differential mode to work.
Invert ch2 only.
Use DC coupling.
Invert ch2 only.
Use DC coupling.
I was able to get a clean signal in differential mode from a 20kHz square wave (I see now I was in AC coupling).
I was also able to create the "ghost" square wave by messing with the trigger level and trigger type (DC vs LF)
So I switched back over to a single probe like before and the "ghost" square wave I was seeing before is gone. I have had the amp powered on for an extended duration and it the signal is consistent and clean. I tried both probes, signal is now clean with both.
I am moving on, calling the output SG3525 good at this time. Gonna move to troubleshooting elsewhere.
I was also able to create the "ghost" square wave by messing with the trigger level and trigger type (DC vs LF)
So I switched back over to a single probe like before and the "ghost" square wave I was seeing before is gone. I have had the amp powered on for an extended duration and it the signal is consistent and clean. I tried both probes, signal is now clean with both.
I am moving on, calling the output SG3525 good at this time. Gonna move to troubleshooting elsewhere.
Attachments
I think I am making good progress now. All four of the power supply MOSFETs gate to drain are shorted. I saw this earlier and discounted it since they were still in circuit and I did not find that as a common failure mode for MOSFETs.
So I pulled them out today and confirmed that the gate to drain is shorted on each. They are Nexperia PSMN8R7-80PS. So I read a voltage spike could cause this, any opinions on slapping 4 new ones in?
So I pulled them out today and confirmed that the gate to drain is shorted on each. They are Nexperia PSMN8R7-80PS. So I read a voltage spike could cause this, any opinions on slapping 4 new ones in?
Define 'voltage spike'.
You need to check the drive signal. Do you have a scope?
PS FETs shorted gate to drain is the most common failure. They will short d-s after that happens but I've seen a lot of FETs shorted g-d and not shorted d-s.
You need to check the drive signal. Do you have a scope?
PS FETs shorted gate to drain is the most common failure. They will short d-s after that happens but I've seen a lot of FETs shorted g-d and not shorted d-s.
I have no insight on what damaged the MOSFETs as I bought the amp broken. I read a voltage spike may cause this type failure, I was asking to gain insight on replacing them or if I should dig deeper.
Gate drive signals show below.
Gate drive signals show below.
Did you check the output transistors to see if any were shorted?
I'd check the drive signal on the gate pad with the FET location loaded with a 0.01uF capacitor.
I'd check the drive signal on the gate pad with the FET location loaded with a 0.01uF capacitor.
Just checked the output transistors, none are shorted.
I soldered in a 0.01uF ceramic capacitor to one of the MOSFET location across the gate to source and measured the drive signal with the scope. Here is the result. 10x probe, 1V/div, 10uS timebase.
I soldered in a 0.01uF ceramic capacitor to one of the MOSFET location across the gate to source and measured the drive signal with the scope. Here is the result. 10x probe, 1V/div, 10uS timebase.
The scope needs to be on DC coupling (I think it is) and the trace aligned to the vertical reference before connecting the probe to the circuit.
If that's what you have, it looks OK.
If that's what you have, it looks OK.
Yup it is on DC coupling and the trace was centered before connecting to the circuit. I ordered the MOSFETs, should be here the middle of next week. I will get them installed and report back.