Re: PSU Ripple

On Sun, 10 Mar 2024 14:02:14 -0400, ehsjr <ehsjr@verizon.net> wrote:

On 3/10/2024 9:35 AM, Cursitor Doom wrote:

Returning to this Marconi signal generator which has a lot of ripple
on the (linear) PSU output, I managed to get probes through a barely
accessible crevice and get a couple of screen shots of the rectifier
outputs I'm seeing on the scope.
  So there's mains incoming which goes into a toroidal transformer and
thence to the rectifiers. There are two secondary windings on the
transformer and they each get their own bridge rectifier. This is the
waveform that's being applied to the storage caps of the PSU (which
I've disonnected for testing purposes).
The outputs of neither rectifier look at all correct to me. What does
the Panel make of them?
 
 
https://disk.yandex.com/i/CP8qRMy-QA-fCg
https://disk.yandex.com/i/ubNazf1pFhuNtg
 
(probes are on 10x and I did compensate them first)

>
If we assume that the probe is (one probe and 1 ground
only!) properly connected across the output of 1 bridge
rectifier ONLY and the load is not connected to that
bridge, you should see an almost perfect waveform.

Quite. And that *is* how I've measured it.

Do a google on bridge rectifier waveform to see what a
perfect waveform looks like.  That's a big assumption,
based on your description of the accessibility problem.

I know what a full-wave rectified signal should look like and what I'm
seeing doesn't much look like one! It does look like a diode's gone
down in each bridge, but it could conceivably be just one that's
failed in certain circumstances I guess.

Assuming all is well with the measurement technique,
those wave forms are wrong.  Both bridges look bad -
replace them.
>
With the above action in mind, you must disassemble things
enough to replace the bridges  --- SO that means you
must have unfetterd access to the bridges before removal
so that you can make measurements and KNOW you have your
probe and ground connected properly. With that much
disassembly you can use your VOM on diode test to check
the bridges.
>
To me, it really looks like you _MUST_ do the work, even
though it may be a PITA, to verify that those really
are the waveforms and not a measurements error.

I fully agree. It's just a nuisance not having two consecutive hours
to get stuck into this job. I'm lucking if I can find a spare hour
once a week to work on it. :(