My HP 6632 Power Supply has got crazy (SOLVED)

My 6632A s/n 3145A-06821

 

Symptoms:

• power on is ok; no self test errors on the display;
• VSETing from the front panel to values greater then 5/6 Volts triggers OVP.

Checks:

• with these symptoms always double check the sensing terminals (S+ and S-): they must not be left floating. Be sure they are directly connected to output terminals on the rear terminal strip or to a remote load. For testing purposes mine were connected to the terminal terminals on the rear output strip;
• checked OVP circuit as per SM: voltages at U110 are ok; both R158 and R155 checked ok.

Further investigation:
    • 0.1 Volts < VSETs < 4/5 Volts work but the readback (front panel) is out of, say, 15% and slightly unstable; the actual output is about 4x VSET and highly unstable: it jumps randomly up and down few volts.

NOTE: from now on I  connected my Arduino GPIB controller to the unit so that I can quickly issue commands to it without pushing the control panel buttons.
      
This is the result of “VSET 2” GPIB command:

The first spike is the direct consequence of VSET 2; the second is just pure random noise. Please note the rms voltage is more then 4 Volts instead of 2. The peak reached by the first spike was 20.4V, barely not enough to trigger OVP.

VSETting to > 7/8Volts produces an output spike. Fortunately the working OVP circuit triggers @ 21 Volts:

 

 

this is the result of VSET 2 while the output was @ was 1 Volt:

A 12.8V spike, .. and then down to about 4 volt, then instability and random noise. WHAT A MESS!

 

Where this kind of behavior can come out from?
OVP is not the cause: it works as expected triggering as soon as the voltage overcomes the limit.
So, what else?
I suspected the "voltage monitor subsystem" (sometime called the “error amplifier”).
Please note that these power supplies have sensor terminals useful to monitor the load voltage directly at the load terminals.

 

The S+ and S- terminals are available for this purpose and they are high impedance inputs (very sensitive to noise). Instabilities on the monitoring circuit behind the S+ terminal have great effects on the output voltage stability.
So I started looking at those terminals on the main board and ...BINGO!
Some of the PCB traces and solder joints in that area shown signs of corrosion.

 

An electrolytic filter cap (C110 - 470uF 35V) is present in the affected area. It appears it could have leaked, but it is in very good physical shape (no bulgin, no fresh leak around it).
The cap, measured off circuit with my LCR meter, was found to be in excellent electrical conditions: 472uF and ESR=0,1ohm.
In any case, while working on the board, I replaced it with a new ELNA one.
I also cleaned the offended PCB area with IPA (isopropyl alcohol).

 

THESE OPERATIONS SOLVED THE PROBLEM.

 

Here is my 6632A following a bunch of VSETs without any problem.

Aftermath note:
an incorrect voltage calibration can trigger OVP at turn on. This is because the rig resets to "VSET 0" during the turn on routine, but if the zero setting is well below zero, it fails and goes up to MAX V, triggering OVP.

Hope this helps!

Cheers.

Graetz Joker-834 portable radio repair What a challenge!

 This radio is as old as me: born in 1959.

Initial conditions

Found it in very tough conditions: dirty, smelly, not working (dead).
Inside the PCB was dirty and strangely greasy in some area.
I have found the schematic in my personal old paper archive.
Here is the annotated version I used during the repair job (look at the end of the article for a high res pristine version):

Please note this rig uses PNP transistors so GND is positive.

 

Looking around I immediately noticed an odd component that I had never seen before. It was badly leaky and one of its terminals was disconnected from PCB because of corrosion.

BIAS Cells

This is how I discovered the BIAS Cell technology used at the time this set was designed: it is a 1.5Volts battery cell used to provide the bias voltage required by the entire transistors strip (but the AF power amp transistors that have their own bias network).
You can locate it in the diagram at the rightmost part, highlighted by a red rectangle, just close to the main battery pack.
Undoubtedly the radio cannot work without that bias voltage, so, to be able to continue the fault finding procedure before establishing a definitive solution, I used my HP 6114A precision power supply to provide the required 1.5Volt.
I eventually modified the circuit to get rid of that BIAS Cell. I do not understand they used it.
The mod is depicted in a schematic excerpt at the end of the article.

I then went ahead with a full recapping (electrolytics only).

The speaker started some buzzing and clipping.. the set was alive.
LW, MW and SW checked ok even if devoid of any significant sensitivity.

FM module

FM was only lightly blowing and cracking...
Long story short: OC171 aluminum whiskers problem! I knew about it but had never met it before. Tapping the OSC-MIXER transistor (OC171M) had the effect of producing cracks in sync with the taps. The treatment consisting of delivering a powerful pulse of current between the terminals (all shorted together) and the transistor's chassis had positive effects as I started listening to FM stations.. but it was a non lasting solution. Hiss and cracks soon started again, FM stations faded out and no further treatment produced any effect.

Again long story short, I've replaced both the OC171's in the FM module with AF329's. 

 

Kicking the oscillator off

The OSC didn't want to self trigger oscillations until I installed a 22pf cap as shown in the following schematic (the red one) useful, I suppose, to compensate for less collector capacitance of the AF239 compared to the OC171 (comments on this are welcome!).

 

 

 

 

 

 

   

 

 

Initial LO alignment

I then spent my time realigning the OSC trim cap (23) and coil (25) trying to get the received signal injected into the IF tunnel through the first IF transformer (27). It was only after having invoked all the saints in heaven that I understood the IF of this radio is not the usual 10.7MHz but 6.750MHz. 

 

 

 

 

Without my best friends  (A.T. 3361B and HP 8640B) I would probably be still there attempting to inject a 10MHz signal into a 6.7MHZ xformer....

 

 

 

WARNING! fragile tuning slugs ahead

Once I've got a decent FM reception, I tweaked the first IF xformer (27) attempting to tune it at its best. Doing so I cracked the top of the tuning slug (it was soooo fragile...), becoming impossible to further turning it in either directions.
I had to dismantle the radio further to get to the xformer and desolder it. Fortunately the slug has a drive notch on the opposite side also. So I flipped it and so regained the ability to turn it from the xformer top.
Tweaking that xformer has an impressive effect on the FM performance (as expected).

Audio circuit

After that the FM was fully recovered but the sound was ugly... like heavy clipped. In fact I've found one of the side connection of the input push-pull xformer (320) disconnected (it look like someone disconnected it on purpose...). Reconnecting it improved the sound a lot (obviously).
The volume pot (302) needed an hard treatment as it lost one terminal and was really filthy inside. I managed to recover it.

AF power amp thermal instability

Everything was looking good now, but.... after about 5 minutes of operation the audio signal started to fade out and at the same time the set started to draw an uncontrolled amount of current. It didn't do that before.. Fortunately I had set up my bench power supply to a max current of 0.4Amps! That salvaged my set.
One (and only one) of the AF power amp transistors was warming up rapidly. Needless to say it was just the one on the push-pull side of the xformer terminal that was disconnected and that I had just reconnected.
Warming up the NTC during the avalanche phenomenon by mean of my hot air gun, returned the circuit back to normal operation conditions (audio and drawn current back to normal).

 Looking at the circuit I noticed two things:

- the final BIAS voltages are controlled only by the NTC (322);
- the circuit lacks of the usual emitter resistors useful to compensate for the negative temperature coefficient of the transistors' base to emitter junction.

 

 

 

Moreover the NTC is physically installed far from the transistors it should detect the heat of; it is sooo far that there is no chance it will have any thermal control effect.

Desoldering the NTC to displace it in a more appropriate position was considered a nightmare because of inappropriate physical layout.. a layout really not intended for circuit repair.

 

Installing a couple of 10ohms resistor in series with the transistor's emitters was sufficient to thermally stabilize the circuit and everything now works seamlessly for hours. 

 

The mod is depicted in a schematic excerpt at the end of the article.

 

 

Repair finished!

Additional resources:

This is my Graetz Joker

- Pristine high res schematic:

- AF P.Amp emitter and BIAS Cell elimination mods:

Greyed out lines and components have been eliminated. Red components have been added.
The final BIAS voltage I set the bias trimmer to is 1.5V. It has been determined by max the noise floor listening to MW, assuming that the more sensitive the OC71 strip is the best performance you get. BIAS voltage has no effect on FM as far as you stay in the 1.3-1.7 Volts range.

PCB layout:   

        

 

Hard Disk catastrophic failure

Have you ever seen inside an hard disk that suffered a very strong catastrophic failure?

I don't: I've never seen something like this before....

The disk suddenly went in FAILURE status inside my server. 

Here is what I've found inside:

 

 

RAID saved me!

Ciao.
Emanuele.