R&S ZVR VNA power supply +15v wide ripple

[mrf245]

Hello all,
recently I got a bad R&S ZVR nice and clear, there are some problems in it.
At first I checked power supply, it was made by other manufacturer not R&S. I surprised found there are very wide and stable ripple in +15V output line, about 3VP-P, 244Hz, whether in empty load or normal. Please see picture below.
Another my friend has the same power supply and also have this problem, the same in 15V.
In ZVR service manual there is few words describe it, I can only understand 15V is output from secondary DCDC after main transformer.
Anyone can give me some advice? Thanks.

[PKTKS]

Impossible to guess without the schematic or even a clue
about what topology and rated power the PSU

Usually a long shot guess would say that the regulator just
have gone wild for a walk.

Even faulty output capacitors would kindly be compensated
by a good linear regulator working feedback loop - when unloaded

Things will go really bad when loading the rated power.

Put the regulator CHIP ID or datasheet to at least have
pins to do some checks

Paul

[mrf245]

Its difficult to describe details of schematic because I don't have that.
Only block diagram.
15V is output from DCDC, input is a common 28V, other DCDCs also supplied by it.
Output filter capacitor is 2200uF/25V, its OK, low ESR.
I check PCB find these DCDCs all use UC3845, LM358 and N-MOSFET.
By connect "failed" UC3845 all pins out to check it by oscilloscope, I found some stranges:
UC3845 pin2(feedback) is not continue waveform, its a continue pulse, frequency is the same as output ripple;
UC3845 pin6(driver output to MOSFET) is also pulse, it contain driver PWM waveform;
So it can explan why output still fluctuant.
But why fluctuant regular, no explan.

[PKTKS]

Several possible issues

Top list being a faulty IC - the 38xx family is widely used
in several topologies, looks like you a have a simple buck.

Remove that IC and do the OPEN LOOP test from the manufct.
If the oscillator and the internal VREF are still ok you can go
to the next items ...

Being that the feedback loop error parts (if the OPAMP is part of it...)
a bad error amp or current sense error will mess any SMPS

SMPS can go wacky for dozens of reasons.
The FB margin is narrow and if you have protection (OVP/OCP)
you should care about they are working as well

TEST pdf attached - 38xx family members differ in just some
 features being the test gig practically the same

[PKTKS]

I would change the note about being a simple BUCK...
Once these ICs are most widely used in forward converters....

**IF** your NMOS is driving a transformer... e.g. you have
a forward converter and the uc38xx works in classic
"CURRENT SENSE" forward mode

otherwise check how your mosfet is being driven e.g.
a buck would be a synchronous mosfet.

In forward mode your optocoupler should be checked
as if it fails you have no feedback loop at all

(if you have an optocoupler.. some cheap designs
ignore the fact that a forward should be isolated.)

2 very different scenarios - being the CURRENT MODE forward converter
the most widely used for these controllers

Paul

[PKTKS]

Typical LOW POWER BUCK for uc38xx series..

not a very common found  topology as the current sense
forward mode is the killer topology in these chips

2 cents to help decide how the OPAMP and MOSFET fit

[mrf245]

Thanks Paul,
As you mentioned feedback loop issue, I check waveform again by combine them together, that is answer......
See below.
Yellow is output voltage, blue is UC3845 pin2(feedback), purple is UC3845 pin6(PWM output), green is UC3845 pin1(compare output)
output voltage ripple is from 13V to 16.5V, when voltage rise up, feedback is no change along with output. But when PWM output stop, feedback start to rise up to as much as 6.5V........then drop down.
I geuss some wrong in feedback path, and other protect circuit force output off when detect voltage up to 16.5V, if output drop to 13V, it will work again...
So I need check feedback path. It a big project.

[PKTKS]

You may test the chip itself outside the loop...

If the chip failed there is no chance to succeed.

But even failing and putting a new in.. you may have other
culprits that caused the failure.

I would not skip the chip OPEN LOOP test in a simple breadboard.

These chips design date back from the 80s.

The "thing" with these series is the "CURRENT MODE" operation
In that mode you have stability direct proportional to the "inductor"
current which MUST BE in CC mode for proper operation

the other error amps just care about putting the loop in
proper output voltage being the "SENSE" pin the real
loop regulator.

That idea indeed works being a current driven regulator
the  inductor (e.g the transformer) should operate in linear
and CC mode all the time...

so that SENSE input working directly driven the MOSFET
is subject to some spikes.. that pin may indeed fail and the
regulator goes dead wacky

That is the thing with the uc38xx


Paul

[mrf245]

Paul,
Thanks for your guide, finally I finished repair working by find out 2 capacitors lost their value.
These are beside DCDC main inductor but not final output stage filter, there is a 2200uF capacitor with very good status.
A current feedback loop is connect beside main inductor so that if too much ripple there, it will lead to unstable loop detect so you can see waveform before.
After replace them, everything goes great again.
You are right, that's I forgot some unimpressive place.

[PKTKS]

BAD CAPS are common place in this SMPS things..

They just can not stand the heat inside or cumulative outside them

Take a closer look in the current "sense" pin of ucxx chips
because all regulation in these chips are based on that current sense

They (UCs) are good even being designed in the 80s ..
clones exist till today

Paul