SMPS Troubleshooting

[eddie_ruff_]

Hi,

I'm trying to recreate the Studer MicValve Switch Mode Power Supply. It features 24VDC input and several different outputs (+-15VDC, elevated DC heaters, -300VDC, etc). The two main ICs are L4962 and L4960 (obsolete but still able to purchase for now).

I've built and tested the attached schematic and everything works except the +-15VDC rail. I've narrowed down the issue to L11. I initially saw this as a common mode choke, but I'm now thinking it's likely a 1:1 flyback transformer.
The service manual mentions that L11 is a 2*250uH storage inductor. But with 12 pins and a similar footprint as T7...I think it's acting more like a flyback transformer.



The issue I am having is a bit complex, when I first tested the PCB, I did so unloaded and I think I might have damaged the +15VDC rail. After I added a 10K resistor between +15 and ground, I was still getting a dead short through L11. So for some reason, I get DC on pin 7 of IC12 that is then shorted through L11 to ground. Shouldn't this be a PWM signal going into the primary like T7? Maybe my L4962 is fried since it isn't switching on the output...

Here are the two parts I've been thinking about using. I went with the second one for the PCB. Should I be looking for a 1:1 flyback instead?

CYSTF0805TL-251Y BC(Bao Cheng Elec) | C261394 - LCSC Electronics

CYSCM9085TL-162 BC(Bao Cheng Elec) | C261384 - LCSC Electronics

 

[Tubetec]

From a quick look it seems like L11 is just a standard common mode choke , the way you connect the windings does matter though , if its wrong the noise wont cancel as it should .
Im no expert on smps's but Im sure by morning others will be around to help out .
'L' denotes an inductor , 'T' stands for transformer .

 

[JohnRoberts]

You say the +/- 15V rails don't work. How about the +48v? If the +48V is good and no +/-15v then the L11 could be suspect.

JR

 

[ruffrecords]

Looks to me like the bottom half of L11 acts as the inductor for the -15V supply. The secondary of L11 is referenced to the 0V of the -15V so if it acts as a transformer it will create +15V referenced to this 0V which is what you want. I suspect you are right that it is really acting as a flyback transformer.

Cheers

Ian

 

[ccaudle]

The service manual mentions that L11 is a 2*250uH storage inductor.

The operation of a flyback transformer is technically considered a coupled inductor, because the energy delivered to the load is not on the same part of the cycle as the energy delivered to the primary, the energy is stored in the primary, then released through the secondary later. In a traditional transformer the energy couples directly from the primary to the secondary and there is no energy storage.

Maybe too much information, but that is a bit long winded way of explaining yes, you are probably right, that is a flyback transformer, which would likely have different design parameters than a common mode choke.

 

[eddie_ruff_]

The operation of a flyback transformer is technically considered a coupled inductor, because the energy delivered to the load is not on the same part of the cycle as the energy delivered to the primary, the energy is stored in the primary, then released through the secondary later. In a traditional transformer the energy couples directly from the primary to the secondary and there is no energy storage.

Maybe too much information, but that is a bit long winded way of explaining yes, you are probably right, that is a flyback transformer, which would likely have different design parameters than a common mode choke.

Thanks! That makes sense, and the footprint size/pinout makes it seem like a custom flyback. I'll start looking for something like a 1:1 flyback and test it out in circuit.

 

[eddie_ruff_]

You say the +/- 15V rails don't work. How about the +48v? If the +48V is good and no +/-15v then the L11 could be suspect.

JR

Yes, on first power up, the 48V output measured 47V and the -15V output was around -14V, but the +15V rail was stuck at <5V or so, and there was high current draw. So, it was trying to work but L11 seems suspicious.

 

[eddie_ruff_]

Looks to me like the bottom half of L11 acts as the inductor for the -15V supply. The secondary of L11 is referenced to the 0V of the -15V so if it acts as a transformer it will create +15V referenced to this 0V which is what you want. I suspect you are right that it is really acting as a flyback transformer.

Cheers

Ian

Thanks! This makes sense

 

[JohnRoberts]

Yes, on first power up, the 48V output measured 47V and the -15V output was around -14V, but the +15V rail was stuck at <5V or so, and there was high current draw. So, it was trying to work but L11 seems suspicious.

that tells me the switcher IC is switching and mostly working.... It doesn't look like the +15v is regulated separately so expecting a symmetrical loading on +/- 15V rails. Is anything connected to the +15V supply getting hot?

JR

 

[murrayatuptown]

L11 initially made me think common-mode choke, mainly because of two windings strongly indicated as being on one core, it seems like an odd use of one with one side going to ground, in a bipolar PS sub-circuit.

If it is not a CMC, then my confusion is justifiable. If it IS, I would put it further down the list of suspects as it's probably there for an EMI/EMC function. I could be wrong, but they don't usually go bad without provocation.

The interpretation of L12 (I think, and am having enough trouble typing ony phone with ads popping up, that I don't want to go reverify)...if you look at L9, L10, L12, L14 & L16, they are all 72 uH. That's 10 terminals. Did I miss another? The footprint comparison to T7 (12 terminals) made me count inductor terminals. Are all 5 of those inductors on the same core/bobbin? I've seen old (OK, maybe 'classic') SMPS designs with so-called 'coupled inductors' that look like they are working in multiple voltage outputs in some way I found mind-boggling. They DO, in combination look like a transformer, and by physics, surely must exhibit some of those characteristics...but I never grasped why and didn't need to pursue why there would be coupling desirable. But being named 'coupled inductor' proudly (no secret) announces the coupling serves some purpose...but in the ripple filtering?

There was another inductor, I think 250 uH?, I only noted one of...is it on the same inductor assembly? I don't think coupled inductors have to be the same value, especially since the power supplies are probably only sharing a common ripple frequency, if not load currents.

I think they are filter-function inductors, not converter types. But this is the first time I've looked at this schematic.

Some of the circuits I am remembering type had only a single voltage regulated off the transformer, and relied on turns ratio of the switching transformer windings to provide regulation 'by association'. Subsequent active regulation sometimes followed. There was often feedback from that single voltage, usually a logic voltage like 5 or 3.30thay provided voltage feedback to the PWM chip.

In troubleshooting, techs warned new guys lifting individual diodes off each secondary for isolation, but NEVER the one regulation feedback was derived from...because losing the regulated voltage feedback would result in all the less-regulated outputs rising because of the PWM trying to recover the missing FB voltage.

I should stop before I get even farther out in the weeds. I'll stop typing & go look at the schematic again.

Murray

 

[murrayatuptown]

L11 does look like the 'lower' winding is the switching inductor for that particular regulator. The lone 250 uH elsewhere is also (obviously?) The converter inductance.

The L496x's all have their own voltage feedback so ignore my rambling about...whatever needs ignoring.

 

[murrayatuptown]

L11 looks like pretty creative use... The regulator feeding the lower winding seems to use same as the required inductance for the L4962, and the D19 is the (-) HWR for -15V.

L11 is probably like a CMC in having identical windings, and having equal/opposite voltages output goes along with that thought. D18 is fed from L11 terminal 12 as (+) HWR. So each polarity HWR is followed by one of the 'half-windings' of L11.

Assuming CMC's have high coupling, a 'magnetic fault' is unlikely if one polarity of the +/-15 works properly, or at least reasonably.

C83/R102 = snubber at 2.26 MHz?


Are both the MBR460's OK?

 

[eddie_ruff_]

that tells me the switcher IC is switching and mostly working.... It doesn't look like the +15v is regulated separately so expecting a symmetrical loading on +/- 15V rails. Is anything connected to the +15V supply getting hot?

JR

Nothing hot on the +15V rail that I know of.

There was no load on the +15V rail when I first powered it up, and I'm wondering if that was part of the issue. This is my first time duplicating a SMPS and I remember reading somewhere that certain designs could be damaged if left unloaded. I initially tested this PSU PCB by itself under no load. I guess there are some paths to ground in the schematic, for example, the -15V rail feeds back through R159 and R160. Similarly, R308 and R344 provide a path to ground on the -300V and 6V3 rails. That means the only unloaded switched rail is +15V (I didn't build the VCC stuff, as I'm not interested in recreating the digital circuitry).

I now have a 100K resistor from +15V to GND, but no dice. I'll need to get a couple of spare L4962 and some flybacks to try out. I might've damaged those components already.

 

[eddie_ruff_]

UPDATE:
Good news! The issue seems to be resolved now. I was able to source a 1:1 flyback transformer for L11 and so far, everything seems to be nominal. I added some temporary load resistors on every output so I can test the PSU by itself before connecting it to another PCB. Here is the revised PSU PCB.



Here is the scope screenshot that shows +48V phantom, +-15V IC rails, and -300V plates.



And here are the DC heaters. These are currently set to 12VDC but all outputs have an adjustable trim pot now as well.


Here's a close-up of the dummy resistive loads on the plates and heaters. For example, R4L is 200K on the plates. So at -300VDC, there is 1.5mA of current or about 0.45W of power dissipation.

Here's a thermal shot to show what it looks like running "unloaded". 24VDC input voltage with 167mA of current draw or about 4W.


And another thermal when I connect the DC heaters to 4 12AX7 tubes. 24VDC input voltage with 666mA current draw or about 16W.


thanks again for the help, everyone! Next step is testing it with the full circuit and seeing how the noise and ripple do

 

[danparaschivescu2016]

For sales?

 

[eddie_ruff_]

While it did work, the heat sinks were quite large and got very hot. I think this was the best Studer could do at the time. I have since created a more modern version that is all SMD, no heat sinks, lower noise and smaller footprint. I might sell it on my website at some point, but I need to finish this project first!

SMPS YouTube

 

[danparaschivescu2016]

I look forward!