Mackie Th15 SMPS low Vout

[madan1]

Hello,
I'm troubleshooting a PA speaker Mackie Th15 (schematics attached).
It is basically a smps with amp and some dsp.
The issue is that on the secondary side I read +/- ~23V instead of the expected +/- 38V. Indeed all secondary side and feedback/aux rails are like 1/2-3/5 of the expected values.
Q2, Q3, C35, C40, C33, C36, C41 and C42 were tested out of circuit with a component tester and read fine. Most of the other components on the primary were tested in circuit and seem to be fine. +HV to -HV on primary is ~330V.
By the way, the ferrite core of L10 is broken, but I do not believe this might have any affect on the smps.
Any ideas what might be wrong?

[momaka]

Yeah, the broken / cracked L10 common-mode filter choke shouldn't be the issue here.

Looking at the schematic of this SMPS, I see there's no feedback from the secondary side to the primary (no optocouplers), so I don't quite grasp here how stable voltage regulation on the output can be achieved (or if that was even part of the design in the first place.) According to the FAN7631 datasheet, if you look on page 16 (I'm looking at the OnSemi datasheet, found here), you can see that feedback from the secondary side is provided via OP1 optocoupler, by pulling pin 1 (CON) below 0.4V, which disables switching of the primary MOSFETs. But on this Mackie SMPS, pin 1 (CON) of the FAN7631 controller is tied to pin 2 (RT) via R57 (4.99k resistor), and there is no optocoupler to pull it low to achieve any kind of regulation. So to me, it looks like this power supply is simply left to run as an open-loop design (no secondary feedback). Looking at the Current Sense (CS) circuit connected to pin 9 (CS), I don't see how that can be used for any kind of feedback from the secondary by way of current sensing. So yeah, I'm thinking this is just an open-loop design... in which case, I don't think the output voltage can be a "solid" value.

That said, what happens to the output of the PSU if you disconnect it from the PA system amp and load it down with, say, an external (resistive) load? Does the voltage at the output sag a little? A lot? Or maybe it even increases?

[madan1]

That said, what happens to the output of the PSU if you disconnect it from the PA system amp and load it down with, say, an external (resistive) load? Does the voltage at the output sag a little? A lot? Or maybe it even increases?

Thanks for the reply.
Unfortunately the woofer amp (tda8954th) is soldered to the same board and on this stage I really do not want to desolder it, but what I can do is to remove the +/-38V rectifier diodes.
The HF amp (tda_something) is on a subboard and is disconnected. The DSP/control board is disconnected as well and I'm not even sure that it runs when connected. Indeed it reacts (power led blinks) as if the PSU is cycling, but no wonder, when none of the rails is reaching the needed/expected V.
I do not see how the secondary side could control the primary and that is why I've been looking on this as just a SMPS that should work as soon as it has power. I'm not familiar with what other ways are there to sense what is going on the secondary without optos.

Is it possible the Q2 and Q3 to be bad while still showing fine on dmm and component tester?
Also is there a way to check if the fets are switching properly. I have one of those DIY scopes which I do not think can be used with this kind of voltages..


P.S. Just did some more measurements
On the schematic are the voltages, which my dmm shows.
The other 3 photos are from my DIY scope hooked to the coil between ground and diode D19 (+DIG rail). So I guess the smps controller indeed is cycling and the DMM is measuring what the caps were able to collect.
Well, now I "just" have to find out why that FAN7631 controller is cycling.

[R_J]

What voltage do you have ACROSS C33 (47µf/35v) and ACROSS C36 (47µf/35v?
I would change C40 (1µf/50v) and check R49,50,51.

[madan1]

What voltage do you have ACROSS C33 (47µf/35v) and ACROSS C36 (47µf/35v?
I would change C40 (1µf/50v) and check R49,50,51.

Thanks for the reply.

across C33 --- 11V
across C36 --- 11.5:V
Each of R49, R50 and R51 individually measures around 0.6-0.8 MOhms in circuit.. there is something there that my dmm does not like and can't lock on a certain value.
I just measured them when powered.
"Ground" (on negative leg of C33) to:
-- HV+ --- 317V (same as HV+ to HV-)
-- R51xR50 --- 200V
-- R50xR49 --- 102V
-- R49/pin LS --- 5V

C40 was the first thing I desoldered to check. The original cap was something like 0.9uF ESR ~3ohms. Installed 1uF ESR ~2.5ohms. At the moment I do not have low ESR, checked the FAN datasheet, did not find anything special about that cap, so thought low ESR is not that necessary. Should I get a low ESR one?

[R_J]

There is no need for low esr for C40, Have you changed C33 and C36 as well? The voltage across C33 should be at least 20~25vdc, Q4 acts like a regulator who's base zener sets the voltage which is 15vdc, and Q4 emitter should be about 16vdc, This voltage is to maintain the vcc for the ic, if it is only 11 volts that may be why it is shutting down and restarting.
If you have not replaced C33 I would start, there. The Lvcc srtart voltage is min 11.2v and the turnoff threshold is around 10 volts If C33 is weak it will not charge to around 20v

[stj]

the schematic says the FAN chip runs at 100KHz - is it? or is it running slow?
also, check none of the output diodes are shorted and check the 22ohm snubber resistors and series caps across them.

[madan1]

Well guys, sorry for wasting your time.
By chance I noticed that there is a wire hanging in the breeze beneath the transformer. After further investigation and looots of desoldering, it turned out that part of the transformer chassis is broken and apart from the hanging wire, one of the wires of the aux/feedback coil is also broken. The hanging wire seems to be left like this by the manufacturer, it is not connected to anything so I left it as is. Soldered the broken wire, soldered back the transformer and now the PSU is working (at least is generating the expected power rails, still have not tested it as an amp).
Unfortunately I killed the NTC TH1 (while troubleshooting I installed wrong mosfets which blew immediately and apparantely took along the NTC ).
The original NTC is SCK-055 (5 ohms 5A). Replaced it with mf72 5d11 (5 ohms 4A, salvaged from old ATX PSU ). I guess 4A should be enough (450+100? watt amps < 4amps x 220Vac)? What do you think, should I get a 5A NTC?

[R_J]

It's not a waste of time, good to hear you found the problem, I was going to suggest checking the windings, but they usually don't give a problem, this broken wire would cause the problem you were having. the NTC should be ok

[momaka]

Unfortunately I killed the NTC TH1 (while troubleshooting I installed wrong mosfets which blew immediately and apparantely took along the NTC ).

Build a dim bulb tester if you haven't already, like this (scroll down to post # 70 for details):
https://www.badcaps.net/forum/showpo...4&postcount=70
This should limit current in the even something goes wrong.
Note that some PSUs may not work too well or at all with the dim bulb tester, especially with smaller wattage bulbs. So in such cases, it may be necessary to use 2x 100W parallel bulbs as the "dim bulb". Also the dim bulb tester doesn't usually allow the PSU under test to be loaded much or at all (typically only 10-20% of the power of the dim bulb, max.)

The original NTC is SCK-055 (5 ohms 5A). Replaced it with mf72 5d11 (5 ohms 4A, salvaged from old ATX PSU ). I guess 4A should be enough (450+100? watt amps < 4amps x 220Vac)? What do you think, should I get a 5A NTC?

Probably should be fine... though what power is the amp / PSU rated for?
If you think the amp will often be used near its maximum capabilities / output power, then the 4A NTC might start running too hot after long sessions, and possibly crack its solder joints eventually. Now if you don't mind making some improvements... basically if you have a 12V relay (and a diode), you can use that to bypass the NTC when the PSU is fully functional. The secondary side + or -15V rail will likely be a good source for that (may need to use a resistor in series with the relay's coil to drop the voltage a little so that it runs cooler.)

[madan1]

[QUOTE=momaka;n3575873]...[./QUOTE]

Well, I have a bulb current limiter, but this PSU is one of those that does not want to work (even when broken as in the beginning) with a limiter.
The blown mosfets are totally on me and from now on I will always check the part numbers before installing them. Unfortunately this will not help when the part label is wrong from the manufacturing.. real story, but another topic.

The woofer amp IC (TDA8954) is rated to 2x210-230W, the tweeter is something like 75-100w, add 30% heat, preamp, dsp, lights and I would give it 700W max.. and that is based on the amps datasheets and my guessing. If I'm looking at the drivers, I would say they will be fried way before even 500watts. Still, this is a "professional" loud noise maker so it is expected to run on max for long periods.
By the way, just checked the datasheet "300W + 100W peak".. well I guess Mackie stayed real and did not go with the "Chinese watts" .



One way or other I will have to visit the electronics shop in the near future for a suitable fuse (yes it blew with the NTC ) so will get and a proper NTC... I'm definitely not in upgrading or modifying it... at least not on the amp side, but have some thoughts about the sound making parts.




By the way, something more serious - how do you proceed when you have a broken transformer winding and there is no way to get the same transformer? I got lucky that mine was broken right next to the leg but what about if it was inside the transformer? Would you try to rewind it or macgyver something with multiple transformers, or just scrap the whole thing?

[momaka]

Well, I have a bulb current limiter, but this PSU is one of those that does not want to work (even when broken as in the beginning) with a limiter.

Yes, unfortunately, there are some PSUs like that... though it mostly comes down to the power rating of the bulb and how much load there is on the PSU output.
As a general rule of thumb, you can only have the PSU loaded with about 10-15% of the bulb's power rating... so for a 100 Watt incandescent bulb, 10-15W would be about the max load the PSU will likely be able to sustain before it starts "acting up" (as the line voltage it "gets" from the bulb will start to drop as the voltage across the bulb increases) - and that's just an observation I had that seems to apply mostly to older ATX PSUs. That said, I think LLC resonant PSUs should still be able to work OK with a "dim bulb tester" device. Only APFC PSUs tend to be more demanding. For those, I switch to 200 or 300 Watt halogen... and if that doesn't work, I switch to higher-power heating elements (that I have from scrap appliances): 500 W, 700W, 1000W, and 1400W. Obviously, the higher the power rating of the heating element / dim bulb, the higher the chance you can still fry something. Still, compared to straight plugging in things in the wall AC socket, even the higher power elements give you a bit more time to react if something isn't right and the damage isn't likely to be as bad, since both current and power are still limited.

But anyways, I suppose that boat has sailed when it comes to the relevance of this thread. Hope you might find it useful for future repairs, though.

The blown mosfets are totally on me and from now on I will always check the part numbers before installing them. Unfortunately this will not help when the part label is wrong from the manufacturing.. real story, but another topic.

A case of cheap no-name Chinese MOSFETs going ka-pow?
If so, you wouldn't be the first one to get burned by them.

By the way, something more serious - how do you proceed when you have a broken transformer winding and there is no way to get the same transformer? I got lucky that mine was broken right next to the leg but what about if it was inside the transformer? Would you try to rewind it or macgyver something with multiple transformers, or just scrap the whole thing?

If you can open the transformer without too much damage to the magnetic core part, you should be able to rewind it or at least get to the open/ripped winding to mend it back together.
Unlike regular (and typically smaller) line AC type transformers, SMPS transformers tend to have much fewer turns with relatively thick wire... so it's very very rare to have a winding burn out, especially on the secondary side.

In any case, there was a Romanian guy here by the name of Th3_uNiqu3 some years back, who used to make and showcase a lot of his home-brew PSUs made from cheap ATX PSUs. One of his projects, I remember, was an amp PSU using a standard ATX PSU transformer that he opened and rewound himself. So I believe it's doable (but probably time-consuming). I actually have an ATX PSU project that's been waiting for a few years now that needs a transformer rewind (5VSB 2-transistor circuit burned out and took out the primary winding on the 5VSB traffo with it.) I only got as far as opening the 5VSB traffo and getting down to the shorted winding... which ironically was the 1st layer, so I had to unwind everything from it. Just waiting for the day where I feel zen enough to sit down and do it (even though it's probably not worth it and I could probably replace it with another.)

But in the case of the PSU in this thread, it uses an LLC resonant topology, so having the same exact transformer with same exact number of windings is likely crucial to the operation of the PSU - after all, it is a carefully tuned resonant circuit. Using a different transformer with different number of windings & core (and thus inductance values) will likely not work.

[madan1]

A case of cheap no-name Chinese MOSFETs going ka-pow?
If so, you wouldn't be the first one to get burned by them.

I do not count anymore the fake transistors and ICs, always buy at least few more just to have a second try if the first one fails.
The thing I got in mind above was a batch of bridge rectifiers with reversed polarity. Somehow the die had been reversed in manufacturing and the polarity was opposite to the markings and the datasheet. Good thing those were low voltage ones and there was no bang.. just some very unhappy and hot caps.

About the transformers - I rewinded a toroidal transformer (another PA speaker), but the wires there were at least 0.5 square mm so no real danger of screwing it up if you have enough time, nerves and focus. From what I've watched about dismantling smps transformers, the wires there were tiny, but I guess those were all low power ones. So good to know that it is doable.


Another general question about SMPS - let's say a half-bridge one. What's the voltage on the fet gates? I guess it would be in reference to the floating negative, so no grounded testing equipment? Is there another way of testing for switching signal on the gate apart from a proper oscilloscope with the correct probes?

[momaka]

From what I've watched about dismantling smps transformers, the wires there were tiny, but I guess those were all low power ones. So good to know that it is doable.

Depends on the transformers, really.
Larger transformers typically have pretty thick wire turns. But the smaller stuff, like standby supplies, may have pretty thin wire, though not hair-thin. The one that I mentioned for the 5VSB of a PSU has about 120 turns of wire about the same thickness as an average 6" full-range audio speaker.

Another general question about SMPS - let's say a half-bridge one. What's the voltage on the fet gates? I guess it would be in reference to the floating negative, so no grounded testing equipment?

Generally, the most safe and meaningful way to measure Gate drive signal is to measure between the Gate and the Source of the MOSFET, since the Gate voltage relative to the Source is what determines whether the MOSFET is On or not (or in linear region.) So if the Source is primary-side "grounded", then correct, you can't really use any grounded (earthed) equipment.

Is there another way of testing for switching signal on the gate apart from a proper oscilloscope with the correct probes?

Yes. In case of grounded (earth-referenced) o-scope / measuring equipment, either you would have to have the DUT (device under testing) powered by an isolation transformer or your scope / measuring equipment connected to the isolation transformer.