Siglent SPD1305X lab power supply repair

[mrmp17]

Long story short, I was testing a BLDC controller with regen functionality on my lab power supply (at around 25V, near the upper limit of the power supply) and decreased throttle command to quickly (powerfull regen). This must have generated a voltage surge well over 30V and back-fed into the power supply, damaging something. 
Now, sometimes the power supply works, but displays a few mA of output current with no load attached (which it didn't before), and sometimes it outputs around 0.8V (I measure the same value at the output posts) at 0.8A (again, no load attached) regardless of the voltage setting (current limit set higher than 0.8A). In this fault mode, I have located a TO-220 part, which is getting hot - a ST micro BTA08-600C triac near the output side, which connects to the positive output rail via a shunt resistor - R007 (see attached pictures).  Pin 1 of the triac seems to be connected to ground. Also, this described fault mode seems to happen at higher output voltages. I think I can also trigger it by slightly heating up that section of the board with a heat gun on low setting.

Does anybody know what is this triac circuit doing here in the power supply, near the output terminals? Is it some output protection circuit? Is it safe to assume that the triac is blown and replacing it should fix the problem?

[2N3055]

It is probably crowbar circuit. You can try desoldering triac, and check what is happening. Make sure NOTHING is connected to PSU, because regulator circuits might be damaged too.

If you were lucky, maybe crowbar circuit was only thing damaged.

[mrmp17]

Regulation circuit otherwise seems to be working correctly, both constant voltage and constant current (when not in 0.8V/0.8A fault mode).

There are two other active elements in this circuit (see picture), a SOT-23 (unknown yet) and a SOT-92 which is a TL431. This fits the crowbar circuit theory. Should I be looking just at the triac, or other elements in the crowbar circuit too? It seems kind of unlikely that the triac rated at 600V blew up from regeneration voltage of a hoverboard BLDC motor stopping quickly.

If I measure the output of the TL431, the voltage should probably be somewhere above 30V, as this is power supply's max output voltage. Would this confirm that only the triac in the crowbar circuit is blown?

[2N3055]

Crowbar circuit is parallel to output. Remove triac and verify complete work of PSU. It should work perfectly without it.
If it does then you need to repair crowbar circuit.
Crowbar might be fixed voltage (to just protect PSU from what happened to you), or might have some circuit that follows set OVP and trigger when you go over some set limit. It might even be made so it has some fixed threshold over set voltage where is engages.
I don't know I don't have that PSU.

But start with elimination process. Disconnect crowbar and verify if PSU is otherwise OK..


EDIT: from manual:

 Built-in overvoltage protection; When the actual output voltage of
the SPD1168X is greater than 22 ± 2 V or the actual output voltage of
the SPD1305X is greater than 36 ± 2 V, the output will automatically
short-circuit, and limit the voltage output. If this occurs, please
re-engage the output enable switch to resume normal output.

so it's a fixed limit...

[tautech]

I'll chase up tech support for a service manual tomorrow.
They have them for their other PSU's but not the SPD1000X models yet so I guess it's not been translated yet.

[2N3055]

Too lazy to draw circuit, I found it on the net...

http://axotron.se/index_en.php?page=26#:~:text=Precision%20crowbar%20circuit&text=A%20low%2Dcost%20adjustable%20shunt,of%20the%20TL431%20is%20used

The "Improved TL431-based thyristor crowbar circuit" should be what you have here.. With divider on TL431 set to trigger on needed voltage.
If I were you, I would order ALL active parts (TL431, transistor and triac) because they're so cheap. Also, you don't know how stressed they were... Since you're repairing it better make sure.
Also need to check passives when actives are out.

R189 is not a shunt but to limit discharge current of output capacitor..

Thinking about it, I would remove Q8 first, not the triac.

[mrmp17]

Thank you both, @tautech let us know if you get the service manual.

I have removed all three active elements, will of course replace them all. The power supply now seems to be working fine, but shows a few mA of current (open terminals), increasing with voltage (which I think it didn't before). If I measure the resistance between positive and negative output terminals I get around 0.95kohm, even with R189 removed (my bad, not a shunt, but there seems to be some current sensing going on). I have no idea if this is normal, but the resistance fits with around 3mA current showing at 30V.

There are not a lot of components that could be damaged and sinking those few mA of current, just the two output caps and a reverse protection diode (D4). Could one of these be at fault?

I have attache some photos of the circuit with components removed and the display with current showing.

[2N3055]

Thank you both, @tautech let us know if you get the service manual.

I have removed all three active elements, will of course replace them all. The power supply now seems to be working fine, but shows a few mA of current (open terminals), increasing with voltage (which I think it didn't before). If I measure the resistance between positive and negative output terminals I get around 0.95kohm, even with R189 removed (my bad, not a shunt, but there seems to be some current sensing going on). I have no idea if this is normal, but the resistance fits with around 3mA current showing at 30V.

There are not a lot of components that could be damaged and sinking those few mA of current, just the two output caps and a reverse protection diode (D4). Could one of these be at fault?

I have attache some photos of the circuit with components removed and the display with current showing.

That is 10-12 kOhm load.. Is it linear with voltage?.

That D4 is reverse protection diode directly on output, together with some capacitors, and voltage sensing circuit. Some of those might have been damaged so they pull some current.
Also, there is a chance that current measurement circuit was degraded somehow.

If you have capability to measure current precisely enough, you should verify output voltage and current compared to set and read by PSU.

So I would verify caps and diode on output (not very likely, that's a big diode but still) and trace a bit of voltage sensing circuit.

Let's hope Tautech will have more details..

[mrmp17]

Of course, 10k, not 1k, I might be an idiot  . I have tested it again and is not linear, I got a reading of 3mA across most of the range, but it does seem so sometimes switch to 4mA at voltages above 20V. I also tried heating parts of the power supply with a heat gun. Got no change in current when heating the reverse protection diode and caps, but when I heated the analog stuff under the transformer, it quickly increased to around 10mA, again fairly constant across the whole range of output voltages.

According to my multimeter (Extech EX520), voltage is spot on, but current seems to be a bit on the low side and I think it correlates with current showed at no load. For a set current of 100mA, I got a reading of 97mA (showing 3mA at no load connected) and after heating up the analog part of the circuit, I got 93mA (after disconnecting the multimeter, showed 7mA).

I think this confirm that there is something wrong with the current sensing circuitry, sadly. Will take a look around and hope it's just some protection element that is damaged.

[mrmp17]

I took a look at the analog stuff under the transformer. Selectively heating up components, I only got a change in current reading (open terminals) when heating up U19, an AD8512A op-amp which is probably the main current shunt amplifier before the ADC. Looks like this op-amp needs replacing too.

[2N3055]

Is current shunt low side or high side?

How much did you heat it?

That is precision JFET opamp. It should have 2-12 uV/ °C input offset voltage drift. You could look how it's connected, and try to calculate amplification factor. That should give you estimate how much it can change just because of opamp spec. Looking at the board it looks like some kind of difference amplifier..
Also take a look at input offset current and input bias current in datasheet. They change exponentially with temp, so if you heat it too much they might show up on output if resistors on input are large enough.

If you heated it 60°C over ambient, I would expect something to be visible even on a good part..

If opamp is OK and you change it, it's not a problem, but you will probably have to recalibrate PSU...
Chances that new one will have exactly same offset and drift are zero..

And if you do change it, put a AD8512B grade in if price is not an issue. Better offset and drift specs..

[mrmp17]

I have heated it up to about 50°C, so about 25°C rise.

The current shunt is 10mOhm, so I can compare the opamp offset to the voltage drop over the shunt resistor. Taking 4uV/ °C times 25°C rise gives about 100uV offset. 100uV drop over 10mOhm would be a current of 10mA (if it was a real current, not an offset voltage). So if my logic here is correct, I am getting offset of 4uV/ °C, which is more than typical, but still less than maximum given in the datasheet.

So it might be that I just don't remember power supply displaying a few mA (technically still in spec of the power supply) of current with open terminals, and this rise with temperature is normal behaviour? Could someone with the same power supply confirm?

[2N3055]

That confirms my hunch, it is probably OK.

Is PSU open for this testing? This slight offset might go away with closed case in thermal equilibrium...
If tautech can help with info, maybe recalibration of current range ?
For that you might need at least 5.5 digit multimeter, preferably 6.5 digit to do it right..

My advice would be to repair crowbar circuit, and verify if PSU works ok. Put it together and use it a bit to make sure it is OK.
After some time, if offset is still there you can recalibrate it.

But make note this 3 mA offset is in well in spec of : set and readback Current: ±(0. 3% of reading+10 mA), so probably nothing is wrong.

[tautech]

Asked for guidance as to what components might be blown from regenerative reverse voltage but none was offered, sorry. 
However the service manual was provided and attached.

[mrmp17]

@tautech thanks for the service manual. After some further testing with closed case, the power supply starts at 0 or 1mA offset at no load, and creeps up to 4mA after a few minutes. So, still in spec but not great.
Just for curiosity, how could an over-voltage event like this introduce an offset in the current shunt amplifier? Can an over-voltage event somehow change the offset voltage of an opamp, but not damage it otherwise? I would expect that the crowbar circuit dissipated most of the voltage surge, despite it getting damaged (it failed closed, not open). Is it expected for precision instruments to require re-calibration after overload events, even if not damaged?

[2N3055]

Opamp input can for sure sustain partial damage.. You can try replacing opamp as discussed, but likely will need current callibration afterwards.

[mrmp17]

Yes, I have ordered the opamp too. I'm just trying to understand what exactly happened.

[mrmp17]

I have now replaced the crowbar components (works ok) and the mentioned opamp (with the same A grade component). Temperature stability is now much better, I am getting only 1mA change while heating up that part of the circuit by the same 25+ degC as before. Looks like the opamp was indeed damaged. Calibration was of course required after opamp swap (done it with my DMM for now which looks to be kinda precise enough, but will do it with something with more digits when I get the chance).
So for anyone dealing with overvoltage damage, it's probably worth looking at this opamp, designated U19 on the board.

[2N3055]

Good to hear!!
Take care!

[tautech]

I have now replaced the crowbar components (works ok) and the mentioned opamp (with the same A grade component). Temperature stability is now much better, I am getting only 1mA change while heating up that part of the circuit by the same 25+ degC as before. Looks like the opamp was indeed damaged. Calibration was of course required after opamp swap (done it with my DMM for now which looks to be kinda precise enough, but will do it with something with more digits when I get the chance).
So for anyone dealing with overvoltage damage, it's probably worth looking at this opamp, designated U19 on the board.

If you could make a list of all the parts replaced it would be much appreciated by other owners that suffer from the same mistake you made with excessive reverse voltage.
Congrats for fixing it. 
TIA.

[mrmp17]

TLDR:
Damage: BLDC driver regenerating (slowing the motor) into the power supply (overvoltage, negative current).
Symptoms:
 - Crowbar circuit locking up a few minutes after power-up, showing around 0.8V/0.8A with no load attached.
 - Current reading developed an offset, showing 4mA with no load attached, offset drifting with temperature.
Repair:
- Replaced all active components in crowbar circuit (Q8, Q9, U22).
- Replaced current shunt amplifier opamp  (U19)
- Current set/read re-calibration needed after replacing the opamp (procedure described in service manual).
Needed components:
- Q9: BTA08-600C triac
- Q8: SOT-23 labeled "2A", MMBT3906 pnp transistor
- U22: TL431 reference
- U19: AD8512A JFET opamp