开云体育

Copied in below is Ed's excellent message 76038 about the 22xx singing power supply problem from 2012.
tom jobe...


Ed Breya
Apr 29, 2012
It depends on the perception level. If a "piercing whine" means it's unbearable to be near it, then there is probably a circuit failure such as the cracked core problem or other components. If it's a "perceptible" or irritating whistling - like the sound of the horizontal frequency of the old TV sets - it's then a matter of degree, and it depends on the hearing response of the listener.

This "singing power supply problem" (that was the official name) arose just as the 2230 was going into production. I investigated it and found that the true cause is the beat between the running frequencies (and harmonics) of the pre-regulator and the chopper stages, which are asynchronous and highly variable with line voltage, load, and temperature.

The main "speaker" is the pre-regulator power choke - the pot core, which emits from the surface but mostly from the center hole in the core. I used the old mechanic's trick of poking around the circuit with a plastic tube held to the ear - you can do the same for locating specific noise sources.

It turned out that this was an inherent characteristic of the 2200 series power supply topology, but the 2230, which used basically the same stuff, aggravated the tendency because it took a lot more power. Of course, it never happened during the entire design project, but in production the true variability of components, and the huge number of units ramping up started to expose the problem.

The best solution technically would have been to modify the design to synchronize the sections, but the simplest and most expedient was to minimize the transducer efficiency by plugging the pot core hole with silicone goop, and on the occasional one that was still too loud (I think one of the line workers with good hearing was assigned to judge them), to change the transformer or "factory select" one of the oscillator timing parts to move the frequency a bit. I think that later that became the norm for all of the 2200 series.

So, if you have a 2200 scope that "sings," you can look at the pot core and see if the hole is plugged. If not, you can likely quiet it down a bit by filling it with electrical grade silicone goop or almost any kind of inert, nonflammable material. Do not use anything magnetic or conductive - it is a gapped core with an intense magnetic field in the center that will burn the material or possibly upset or damage the circuit.

Ed

Here's the link in the groups.io message base:

/g/TekScopes/messages?msgnum=76038

Bob.

Craig

Thanks for that info.
Yes, I have seen other RIFA snubber combo units (cracked) which I junked, and replaced with separate R+C - but that was without doing any brainwork.
Now I want to understand this concept properly.

Now, going back to the Tek A2A1 schematic, there is a R+C snubber R1018 30R + C1018 68nF , and also R1016 68R + C1016 68nF

I have seen R1016 burnt a few times, which I have replaced with a Ceramic Composite.

BTW, Metal Oxide (IMO) is unsuitable for these pulsed circuits.
A non-inductive Ceramic Composition resistor would be the best choice.
Any inductive resistor is going to cancel out the benefits of the snubber, and the ringing will simply continue.

Snubbers are usually applied to dissipate the energy in reactive transitions. If you use a capacitor
only,
it will tend to form a resonant circuit that can ring. It's good to have some resistance (loss) to
convert it
to heat, and quickly damp it out. The resistance also limits the peak currents in the loop.

For what it is worth, Quad (the audio company) used to use a RIFA 100nf/100ohm X2 snubber across the
switch output. In other words not across the switch. That was present in the 33 preamp and FM4 tuner
at least, and probably on more products too.

Now I would not recommend a RIFA part, because the epoxy casing splits and they emit massive amounts
of smoke (plenty in the archives about this), but there are other manufacture snubbers of the same
values out there.

Craig

Here's some background info on this:



If you simply change some of the parts, it may apparently solve the problem because the frequencies change a bit, but the tendency is still there.

Ed

Snubbers are usually applied to dissipate the energy in reactive transitions. If you use a capacitor only, it will tend to form a resonant circuit that can ring. It's good to have some resistance (loss) to convert it to heat, and quickly damp it out. The resistance also limits the peak currents in the loop.

Ed

Kevin

Thanks!
My goal is to reduce/eliminate arcing in a mechanical switch (or across relay contacts) - that's all.


1.
You've stated between 1-10nF is suitable - is there a definite method to calculate the optimum capacitance?

2.
What about the voltage rating of that optimum capacitor?
Should it be a DC-rated (up to a few thousand volts), OR an AC-rated Class X or Y capacitor?

I don't know why one would look to a snubber to protect a switch. Snubbers, RC, RLC, are typically employed to control dv/dt on semiconductors. I don't consider a capacitor alone as a snubber. It is pure reactance and therefore non dissipating. A snubber needs resistance to dissipate energy.
A switch should be able to sustain far more transient voltages than semiconductors. Switches in old equipment can be worn out in several ways. Spring fatigue, contact erosion or oxidation, lubricant contamination, time, all contribute to ware out.

Switch failure on closing a normal load, even a cap input load, would seem to indicate a warn out contact. This could be worn springs that allow excessive bounce an arcing.

Switch failure on opening a load can be a different issue. If the switch is overloaded, arcing will eventually erode the contacts and lead to some kind of failure or weakening of springs.
For normal switches in "typical" electronics, a 1000-10000pf cap across the switch should suppress most contact arcing. (~300-3000K). I'm sure I'll get some flaming on this, but mechanical components work different than semiconductors.

Kevin

"the PTC is in parallel with the resistor, not in series"
Just to confirm that it is definitely an NTC, not a PTC, in the 2465

"EMF is a high voltage phenomenon, not a high current phenomenon.
The current cannot be higher than the operating current of the supply.
The voltage can rise to thousands of volts."

To clarify:
If the voltage can rise to thousands of volts, then the capacitor which bypasses an on-off switch should rather be a DC capacitor rated to 5KV, and NOT an AC Class X1/Y1 unit.
Is my thinking correct here?

Would a metal oxide varistor be more suitable?

On Mon, 30 Jul 2018 15:52:29 -0400, you wrote:

The supply has to deal with (effectively) a ramp up of the input
voltage rather than an abrupt step. May confuse some chips that need
a minimum dv/dt, but I'd guess that depends on the design of the
regulator and the rest of the circuit.

Harvey

A snubber capacitor is a capacitor, used in conjunction
with a series resistor, to dampen, or eliminate the back EMF
transient caused when an inductor abruptly loses its
current source.

They are useless in capacitor input power supplies, like
used in the 2465 family of scopes, because they have no
appreciable inductance to make a back EMF transient.

However, capacitor input filter type power supplies have
their own switch eating problems, in that potentially,
they could draw extremely high inrush current when the
power is first applied to the discharged filter capacitor.

This inrush will quite nicely burn, or weld, the contacts
of an otherwise properly sized switch.

To cure this inrush problem, tektronix put a thermistor,
with a shunt resistor in series with the power switch.

The thermistor starts out at a high impedance, causing
much of the initial charging current to be limited by
its resistance. As the current passes through the
thermistor, I2R losses cause it to self-heat, which
because of its negative temperature coefficient (NTC),
will cause it to lower in resistance to a point where
it is near zero ohms. At that point, it self regulates,
never getting hotter, or cooler than is necessary to
supply the current to the scope.

Back to snubber capacitors: The snubber can be sized
such that it completely eats any back EMF transient.
To do this, it would be chosen to have a capacitive
reactance equal in absolute value to the inductive
reactance of the load it switches. Its resistance
would be chosen to equal to the DC resistance of the
inductive load.

This is rarely done, because, it is not necessary to
eliminate the back EMF peak voltage, only to dampen it
down to a manageable value that will not burn out the
switch. Usually 10 to 100x the load's resistance is
adequate. Similarly, the capacitance can be reduced
too... but either reducing the capacitance, or increasing
the series resistance will cause the EMF spike to rise.

-Chuck Harris


M Yachad wrote:

I have now seen two 2465 PCB blocks with a burnt-out on/off switch.

The damage is caused by the arcing across the now-separating terminals, when the machine is switched OFF.

Replacement is easy with NE182UEEP6AMP, but examining the circuit shows me that this failure could well be preventable with the addition of either a single ceramic snubber capacitor of between 3.3nF to 10nF (example Vishay’s VY1 or VY2 series), OR an R+C snubber network – I calculated a 120 ohm 2W ceramic comp in series with a 33nF X2 film cap, across the Tek switch’s on-off terminals.

What are the considerations for selecting EITHER a single ceramic cap, OR making up a R+C snubber?

I’m not looking at the apparent cost savings of installing a single cap vs constructing a R+C.
The objective is long-term reliability.

Also, how would one select the capacitance value of the single ceramic cap?

I’m thinking of applications on vintage stereos.
I noticed that a new Yamaha hi-power stereo uses a 10nF, and a 20-year old CD-player uses a 3.3nF.
Is the current draw a factor?

Over to the experts….

Menahem

On Mon, 30 Jul 2018 11:50:21 -0700, you wrote:

I have now seen two 2465 PCB blocks with a burnt-out on/off switch.

The damage is caused by the arcing across the now-separating terminals, when the machine is switched OFF.

Replacement is easy with NE182UEEP6AMP, but examining the circuit shows me that this failure could well be preventable with the addition of either a single ceramic snubber capacitor of between 3.3nF to 10nF (example Vishay’s VY1 or VY2 series), OR an R+C snubber network – I calculated a 120 ohm 2W ceramic comp in series with a 33nF X2 film cap, across the Tek switch’s on-off terminals.

What are the considerations for selecting EITHER a single ceramic cap, OR making up a R+C snubber?

I’m not looking at the apparent cost savings of installing a single cap vs constructing a R+C.
The objective is long-term reliability.

Also, how would one select the capacitance value of the single ceramic cap?

I’m thinking of applications on vintage stereos.
I noticed that a new Yamaha hi-power stereo uses a 10nF, and a 20-year old CD-player uses a 3.3nF.
Is the current draw a factor?

I'm far from an expert, but I'll guess that you're talking:
1) energy storage. You have to look at what happens when the primary
magnetic field collapses, that ought to give you a relative idea of
capacitance. Your example indicates more power = more capacitance.
2) I suspect that once you calculate the short circuit collapsing
current through the capacitor, the resistor is there to limit it at
the cost of more ringing.

Harvey

Over to the experts….

Menahem

Several people have mentioned their interest in schematics, others have offered to try and make image files or PDF's.
Being the original poster, I'll collect scans/images and if the package is more or less complete, clean up the files if and where necessary and post them. I guess that'll be at Didier's (K04BB). TekScopes' file area would not be my preference, since the instrument is SRS, not Tek...

Currently, it seems that the images will come from the SR780 manual, since AFAIK, no offers were made from SR760 owners.

Raymond

On Tue, Jul 31, 2018 at 12:43 AM, <johnasolecki@...> wrote:

I'm not at the computer that has the PDF's right now so I can't check.

John,
Yes, please have a look if you have PDF schematics, that would make life a lot easier.

Raymond

You could take the schematics to
almost any Office Depot and copy
them full size to a ThumbDrive.

Then post them in the files area.

73, Dick, W1KSZ

On Mon, Jul 30, 2018 at 3:54 PM, Raymond Domp Frank <hewpatek@...>
wrote:

On Tue, Jul 31, 2018 at 12:43 AM, <johnasolecki@...> wrote:

I should be able to photograph them though and forward the shots.

OK, John, thanks.

Raymond

Everyone - thank you very much for the pointers! I will attempt as described - a "spit and shine" cleaning, then checking rail voltages and ripple, then following the H?kan's article (special thanks), and, - if all else fails - the MOSFET and electrolytics.

Dmitri

How about our hero Hakan!
He comes through with the best information on the planet, time after time!
Also, a big 'Thank You' to everyone who contributes to all of the fine discussions that come across this message board.
The current 'snubber capacitor' conversation is going to turn into another 'winner' before it's done, with many secrets laid bare for all to see.
tom jobe...

On Sat, Jul 28, 2018 at 11:44 PM, Raymond Domp Frank wrote:

Ask Stanford directly - AFAIK they have always provided Service Manuals
with
schematics and still do.

Thanks for responding, Dave. As I said, I'm going to try at SRS but seeing
only SM's for their equipment that stop after the parts list, I wasn't
hopeful.
I'll report back what their response is.

I received an answer from SRS. They only supply schematics with paper manuals @ USD 110,-....

Raymond

On Tue, Jul 31, 2018 at 12:43 AM, <johnasolecki@...> wrote:

I should be able to photograph them though and forward the shots.

OK, John, thanks.

Raymond

I'm not at the computer that has the PDF's right now so I can't check. I know for certain I have the schematics in the printed manual but they're 11" x 17" which is too large for my scanner. I should be able to photograph them though and forward the shots. In the next day or so I'll send you the table of contents and you can tell me which specific ones you want. If I remember correctly there's 20-25 pages of schematics so, to start, please keep the list short!

John