Topics

Uncooperative 7834

 

Well darn it - the power supply works just fine with a dummy load. If I
connect it to the scope the familiar ticking sound rears its ugly head.

Measuring the resistance to ground at the test points listed at the bottom
right of page 4-8 of the manual gives me results that aren't far out of line
from the specified values given that I'm using a modern DMM rather than an
Simpson VOM.

Specifed Measured
+130V 6.6k 6.8k
+50V 1.8k 1.75k
+15V 70R 52R
+5V 9R 8.4R
-15V 100R 60R
-50V 500R 2.5k

Apart from scattergun replacing every bead tantalum I can see, does anyone
have any thoughts on how to pin down the fault?

Measuring across the tants doesn't show any obvious issues.

Thanks
Dave

Craig Sawyers
 

Measuring the resistance to ground at the test points listed at the bottom right of page 4-8 of
the
manual gives me results that aren't far out of line from the specified values given that I'm using
a
modern DMM rather than an Simpson VOM.

Specifed Measured
+130V 6.6k 6.8k
+50V 1.8k 1.75k
+15V 70R 52R
+5V 9R 8.4R
-15V 100R 60R
-50V 500R 2.5k

Apart from scattergun replacing every bead tantalum I can see, does anyone have any thoughts on
how to pin down the fault?
Something frigging with the shutdown network on the SMPS control chip - it gets an input from
various points in the mainframe?

Craig

Jim Cotton
 

I've used an HP 3478A on low ohms setting looking for the best short across the tantalum capacitors, followed by clipping one leadand applying power (the second shorted tantalum then lost the magic smoke...)  I found one once with the finger on top of each tantalum capacitor until you say "ouch" method.
Jim On ‎Friday‎, ‎March‎ ‎1‎, ‎2019‎ ‎01‎:‎12‎:‎13‎ ‎PM‎ ‎EST, David C. Partridge <@perdrix> wrote:

Well darn it - the power supply works just fine with a dummy load.    If I
connect it to the scope the familiar ticking sound rears its ugly head.

Measuring the resistance to ground at the test points listed at the bottom
right of page 4-8 of the manual gives me results that aren't far out of line
from the specified values given that I'm using a modern DMM rather than an
Simpson VOM.

    Specifed    Measured   
+130V    6.6k        6.8k
+50V    1.8k        1.75k
+15V    70R        52R
+5V    9R        8.4R
-15V    100R        60R
-50V    500R        2.5k

Apart from scattergun replacing every bead tantalum I can see, does anyone
have any thoughts on how to pin down the fault?

Measuring across the tants doesn't show any obvious issues.

Thanks
Dave

fiftythreebuick
 

Hi Dave-

My favorite way to troubleshoot a scope that is in "tick mode" is to use a storage scope to make measurements of the power rails and/or the shutdown circuits during the ticks, with the storage scope in single sweep mode. During each tick, you can make measurements of the scope as it tries to start and run. So far, each time I have used this method it has led me to the failed component. Something fast like a 466 or 7834 is nice but I've also used a 549.

The other procedure that has paid off for me in the past is the use of an HP-3469B on the 1 Ohm scale to find the lowest resistance on the bad power rail, which should be very close to the shorted component.

Those two methods oughta get you there....

Best of luck!

Tom

" David C. Partridge
Mar 1 #154830

Well darn it - the power supply works just fine with a dummy load. If I
connect it to the scope the familiar ticking sound rears its ugly head."

Harvey White
 

On Sat, 02 Mar 2019 10:46:27 -0800, you wrote:

Hi Dave-

My favorite way to troubleshoot a scope that is in "tick mode" is to use a storage scope to make measurements of the power rails and/or the shutdown circuits during the ticks, with the storage scope in single sweep mode. During each tick, you can make measurements of the scope as it tries to start and run. So far, each time I have used this method it has led me to the failed component. Something fast like a 466 or 7834 is nice but I've also used a 549.

The other procedure that has paid off for me in the past is the use of an HP-3469B on the 1 Ohm scale to find the lowest resistance on the bad power rail, which should be very close to the shorted component.

Those two methods oughta get you there....
There's another way, that's to take a current limited supply and pulse
it on and off. For a supply of X amps, I'd limit it to x/10 amps.
Now, if you have an HP current probe, the one that's made for shorted
nodes, you can see where the bright lights go, and that shows you the
track where the short is.

You might be able to make a probe that does that.

With this same setup, you can also take a probe and with a high
sensitivity range on the scope, look for the lowest amplitude of
pulse. You'd like to measure across the component itself, if at all
possible.

This is a different implementation of the second mode.

Harvey



Best of luck!

Tom

" David C. Partridge
Mar 1 #154830

Well darn it - the power supply works just fine with a dummy load. If I
connect it to the scope the familiar ticking sound rears its ugly head."


 

On Fri, Mar 1, 2019 at 07:12 PM, David C. Partridge wrote:


Apart from scattergun replacing every bead tantalum I can see, does anyone
have any thoughts on how to pin down the fault?
Did you check / disconnect the HV Multiplier ?

/Håkan

 

Disconnecting the supply to the HT circuit (i.e. no EHT, no 130V), changes the nature of the ticking, but doesn't stop it.

David

-----Original Message-----
From: TekScopes@groups.io [mailto:TekScopes@groups.io] On Behalf Of zenith5106
Sent: 03 March 2019 14:47
To: TekScopes@groups.io
Subject: Re: [TekScopes] Uncooperative 7834

Did you check / disconnect the HV Multiplier ?
/Håkan

 

On Sun, Mar 3, 2019 at 05:59 PM, David C. Partridge wrote:


Disconnecting the supply to the HT circuit (i.e. no EHT, no 130V), changes the
nature of the ticking, but doesn't stop it.
I guess you mean disconnecting the 25kHz from the power supply.
In my manual there is a note I made many years ago to myself which says:
"If disconnected the power supply will go into burst mode" with arrows pointing at the 25 kHz connectors.
So I'd disconnect the multiplier to start with.

/Håkan

fiftythreebuick
 

Definitely good ideas, Harvey! I particularly like the idea of pulsing a current limited supply into the rail and checking around along that rail for the lowest amplitude pulse. That should get you close to the offending component! Have to remember that one....



"Harvey White wrote:

There's another way, that's to take a current limited supply and pulse
it on and off. For a supply of X amps, I'd limit it to x/10 amps.
Now, if you have an HP current probe, the one that's made for shorted
nodes, you can see where the bright lights go, and that shows you the
track where the short is.

You might be able to make a probe that does that.

With this same setup, you can also take a probe and with a high
sensitivity range on the scope, look for the lowest amplitude of
pulse. You'd like to measure across the component itself, if at all
possible.

This is a different implementation of the second mode.

Harvey"

Majdi S. Abbas
 

On Fri, Mar 01, 2019 at 06:11:50PM -0000, David C. Partridge wrote:
Specifed Measured
+130V 6.6k 6.8k
+50V 1.8k 1.75k
+15V 70R 52R
+5V 9R 8.4R
-15V 100R 60R
The -15V worries me a bit -- I've seen a shorted tantalum
around 40 ohms on this rail.

The character generator board has 1 uF tantalums directly
across the -15V and -50V rails; try disconnecting it and see
what happens with your -15V measurement.

--msa

 

On Mon, Mar 4, 2019 at 12:04 AM, fiftythreebuick wrote:


There's another way, that's to take a current limited supply and pulse
it on and off. For a supply of X amps, I'd limit it to x/10 amps.
Now, if you have an HP current probe, the one that's made for shorted
nodes, you can see where the bright lights go, and that shows you the
track where the short is.
Decoupling caps and tants and conducting PN junctions and regular power supply current would (partially) short out your pulse, thereby hiding the current info you're looking for, *unless* you use a low level, very low frequency, low slew rate signal.
A few months ago, I successfully used a low-level sinewave (about 0.4 Vpp) signal of about 10 Hz across the power supply pins with an HP 547A current tracer on a spectrum analyzer board, containing rows and rows of digital logic, 0.1" apart with dozens of interspersed decoupling caps. To my own amazement, it took about five minutes to find a shorted cap!

Raymond

 

On Mon, Mar 4, 2019 at 12:27 AM, Raymond Domp Frank wrote:


low slew rate
Correcting myself: That snippet was left over editing away the erroneous mentioning of using a square wave (pulse). Please ignore.

Apparently, you don't have a full short. Still, traces of lower impedance readily come up by probing. Don't forget that the HP 547A is direction sensitive, so rotating it around an axis perpendicular to the trace indicates the direction of the current (modulo180 degrees)!

Raymond

 

So far I'm getting nowhere fast. I'm tempted to remove Q1254 (which will prevent the inverter stop signal from shutting the supply down), and using a 200W dim-bulb on the mains input to limit the current draw (supply is rated at 215W).

I think this will force the supply to come up and blow out the component in the mainframe that's stopping the supply from running.

What do the mavens here think of this somewhat brutal approach?

Thanks
David

Göran Krusell
 

No, I don´t think so. I got the impression that the problem occurs when and only when the mainframe is connected and not the dummy load. Try to lift one voltage to the mainframe at a time if this is doable.
Göran

Harvey White
 

On Mon, 4 Mar 2019 16:28:59 -0000, you wrote:

So far I'm getting nowhere fast. I'm tempted to remove Q1254 (which will prevent the inverter stop signal from shutting the supply down), and using a 200W dim-bulb on the mains input to limit the current draw (supply is rated at 215W).
ok, limiting is your friend here.


I think this will force the supply to come up and blow out the component in the mainframe that's stopping the supply from running.
That's mean. It'll work, but not sure about the collateral damage.

Since you have storage scopes, have you any idea of which supplies are
causing the problem? Can some of the supplies be run independently?
Do you, for instance, have the ability to disconnect the 5, and say
+/- 15 volt supplies and just use the scope's 50 volt supplies? (and
anything higher). A lot of this would be dependent on the design of
the motherboard stuff.

I take it you disconnected all the boards, and then connected them
back up one by one?

First thing I'd do is disconnect the high voltage section. Then
vertical, horizontal and sweep sections. At this point, you shouldn't
be able to damage the CRT, and might be able to figure out what's
going on.

You've probably thought of this already, though.

Harvey




What do the mavens here think of this somewhat brutal approach?

Thanks
David



Ed Breya
 

I can totally understand the temptation to get drastic, but recommend not defeating the built-in protection of the PS - it may work, but there's a good chance that the bad part in the mainframe will win out over something in the PS.

I have a similar situation, with three 7904s out of commission, all tickers. I have done all sorts of swapping and unhooking, and testing combinations, and know that in one it's the PS itself, and in others it's something definite, but not readily found, in the mainframe. In these cases, there is no obvious near or dead short, which would be preferable - I can see all the supplies come up to nearly the right voltage, but then it shuts down. I have fixed a lot of 7Ks over the years - sometimes it's easy, and sometimes very, very not. It can be very frustrating.

This recent predicament has motivated me to finally get going on building my super-duper 7K maintenance box, a project I envisioned years ago, but didn't pursue. A few weeks ago, I started piecing it together and gathering up the parts, and figuring out the design requirements and features. It's a work in progress, and not nearly ready for prime time, but I can share some of the design concepts developed so far. It's nothing that can't be done with the right assortment of other gear, but this unit consolidates the functions needed to test the PS unit, and the mainframe, separately, or combined, to view all of the supplies, including the high voltage CRT stuff.

Here are the major components, and the associated functions:

Scope AC powering and monitoring. A 700 VA variac, 0-100% V line, with PT and CT signals via BNC outputs for monitoring line input waveforms, and DC outs to metering circuit for rough V line and I line measurement.

Switchable test load resistor bank to test or run the PS low voltage supplies while unhooked from mainframe, and measure their voltages.

Forcing/running supply. A modified 7603 linear PS that provides the nominal 7K low voltages: +130, +50, +15, +5, -15, -50. The circuits are to be modified so that each supply can be faulted independently, without affecting the others. Supplies the voltages (except the +130) to front panel binding posts for experimenting and developing plug-in circuits. Also provides various DC for internal control circuitry.

Metering section, two 3-1/2 digit DPMs and signal switching, for selecting and measuring voltage and current on all LV supplies and line, and CRT HV section anode, cathode, and heater voltages. Also routes the selected supply voltage signal to a BNC for monitoring with a scope.

High voltage interface box. This function was originally going to be built in, but the practicalities of the HV interconnections warranted making it an external unit. This box contains the HV dividers for Va and Vk, and load R and HF current transformer for monitoring Vh. The results are at low voltage, sent via cable to the main unit.

I/O interface, two DB-25 connectors, one inputs from the PSUT, the other outputs to the MFUT. Corresponding DB-25 connectors will be rigged up as adapters to route the various PS voltages in and out, via Tek ribbon cables and harmonica connectors, according to the particular mainframe type.

The basic overall operation:

For PS unit testing, the adapter goes to the various connections on the PS, then to the test load, and the supply voltages can be measured if it's operational.

If the mainframe has a fault, an output adapter can connect the PS to the mainframe via the metering circuit (loop-through mode), and the load currents can be measured. The test load resistors are turned off.

If necessary, the force mode can be used, running the mainframe from the built in linear supply. The voltages and currents can all be monitored, and there will be no difficulties as with transient ticking. The power situation will be constant, and clear, and the parts (usually Ta caps) that don't necessarily show bad when not powered, will have the full, normal voltage applied, or have to take whatever (measurable) fault current is needed to lug down the supply.

If the low voltage supplies all seem OK, the HV box may then be used for investigation into possible HV faults.

If the PS unit has internal problems on its primary side, the variac and line signals can be used. Problems like bad line rectifiers or main caps, or startup circuit issues, can be detected by looking at the waveforms as voltage is applied and increased.

That's all for now.

Ed

Daveolla
 

It may not be used too often, but I would sure hate to be without my hp-547 Current probe.
And for those that have the 547A style probes; 545A logic, 546A pulser, that have missing tip protectors, I have found that those large Alligator StyleClips are a perfect fit and look good. If they are a little too snug A slight heat with a heat gun and the boot can be shoved onto a tapered dowel to stretch it a bit.I also think these rubbery boots are actually safer for the probe in that it absorbs more shock.

Dave

At 05:27 PM 3/03/2019, you wrote:
On Mon, Mar 4, 2019 at 12:04 AM, fiftythreebuick wrote:


There's another way, that's to take a current limited supply and pulse
it on and off. For a supply of X amps, I'd limit it to x/10 amps.
Now, if you have an HP current probe, the one that's made for shorted
nodes, you can see where the bright lights go, and that shows you the
track where the short is.
Decoupling caps and tants and conducting PN junctions and regular power supply current would (partially) short out your pulse, thereby hiding the current info you're looking for, *unless* you use a low level, very low frequency, low slew rate signal.
A few months ago, I successfully used a low-level sinewave (about 0.4 Vpp) signal of about 10 Hz across the power supply pins with an HP 547A current tracer on a spectrum analyzer board, containing rows and rows of digital logic, 0.1" apart with dozens of interspersed decoupling caps. To my own amazement, it took about five minutes to find a shorted cap!

Raymond

 

When the EHT transformer is connected to the 25kHz HV drive, then pin 13 on the control IC hits 0V pretty much immediately after the inverter starts causing a shutdown after 10mS. Anything below +50mV on this pin indicates a fault condition (inverter current > 5A).

I tried disconnecting the HV Multiplier, but that wasn't the problem. My suspicion is shorted turns in the EHT transformer.

With the EHT transformer disconnected the supply also ticks, but not because of over-current, but because Bal-Sense (pin 2) is just over +200mV when the 10mS timer expires. From what others have said this is pretty much normal behaviour.

Ho-Hum

Who needs some 7834 spares (including a working PSU)?

Dave

Ed Breya
 

Before totally giving up on it, you may want to check the HV filter caps and rectifiers in the cathode circuit. I don't recall if these use the old ceramic type, or plastic caps - ones typically rated for a few kV. I've had a few of the ceramic ones go bad here and there over the years. To properly test them, besides measuring the C, you should also juice them up to a fairly high voltage, comparable to where they operate, and make sure they don't break down or leak excessively. Usually the 1500 V peak, in DC-mode, on a 576 or 577 is enough to show problems with them - if they can't handle that, then they won't make it at twice that. For testing them without doubt, they should have the full cathode voltage applied, if you have a convenient HV source.

Ed