Repairing a ZENTRO power supply

[Apfelmus]

Hello,

Has anybody serviced these units and might help me tracing the fault and eventually can explain how the switching cards in these units work?

Here is the story:

I have two power supplies from Zentro Elektrik, a German company based in Pforzheim near Stuttgart. They are still making power supplies and current sinks, and their older equipment is readily available, at least in Germany.

Currently I'm repairing an LD50, a power supply giving 50V and 10A max. This supply is build around a regulator card, 3 so-called "switching cards" and 11 power cards. The power cards work in parallel, the "switching cards" do a pre-regulation in 4 steps (30-45-60-75 volts) of the voltage that is sent to the regulator cards.

I have some circuit diagrams but they do not correspond exactly to my unit. It helped me tracing a fault in the regulator card. But there is still a problem in one of the switching cards, where a resistor starts to smoke when the voltage goes from 30 to 45 volts.


cheers
Martin

[Apfelmus]

Hello all,

I did some reverse engineering and correlated this with the diagrams I have.

These supplies use a pair of SCRs to "connect" and rectify a transformer winding that adds its voltage to the unregulated voltage path.

There are 3 cards working in series, each connected to another transformer winding. The circuit that controls the SCRs contains the smoking resistor. But I cannot see why the circuit should be faulty. Each of the 3 boards contains a slightly different value of the resistor (56, 100 and 220 ohms). The one with 220 smokes. The 100 Ohm looks as if he had a rough life, and the 56 looks good. So maybe its a design flaw, eventually amplified by the mains voltage that went from 220V to 240V since then.

I'll replace the resistor with higher wattage type, that should do it.

cheers
Martin

[karesz*]

Hallo Martin,
would you share your circuits please?
Karl

[Apfelmus]

Hello Karl,

I would do it if they were more than a barely readable sketch. That would require quite some work.

The circuits I have do not exactly correspond to the unit, and they are very badly labeled or not at all, so its very difficult to get an overall view of the circuits and all those 15 cards connected partly in parallel, partly in series.

I noticed the units do not have a mains voltage selector. Given their age they must have been designed for 220V. Now we have 240V, thats 10% more, and maybe the reason why some components are stressed above their limits. In one of the two supplies I noticed the 220Ohm resistor has apparently been replaced by a 5Watt type, whereas the one that smokes is a 1Watt type.

On the regulator card there are two resistors that also get very hot. They have approx. 2 and 3 kOhm, and something like 50V to burn. They are only rated 1 Watt and consequently heat up to approx. 150-200 °C (but do not smoke, only smell)

Is that bad design?

cheers
Martin

[orbanp1]

Hi Martin,

Looking at the Zentro web-page, it looks like that PS is still being made.
You could contact Zentro and ask for a manual. Hopefully that has revisions included. I can't see them sending out units that would overheat...

It is possible that something is wrong with your PS.
The design would be pretty marginal if 10% mains increase would result in failure.
Again, I can't see them making products like that. Eventually no one would buy from them!

Good luck, Peter

[karesz*]

Hallo Peter,
As you can reed it on her side: "Zentro-Elektrik gestaltet Zukunft und dies schon seit mehr als 60 Jahren."
Itsn`t seldom that Eu & US firms dont have such support for older products as 10-20 years and we can belive, their didnt have some equipment had in production, and they had it_but only in history and we dont find practically nothing on net and in actuell catalogs etc...
Zentro was and is a professional supporter for industrial and lab customers, so a 10% design margin with some overheatings is in my opinion inpossible, it must have such component/value failures.
BTW; Serie LD & ELA are, I think, over 20+ years in production program-maybe are released, but only internaly...
Karl

[Apfelmus]

Originally Posted by orbanp1

Looking at the Zentro web-page, it looks like that PS is still being made...

Hi Peter,

I know that. But I think there must have been some revisions since then. The diagrams I got from them (by asking) are from the 70/80ies. And yes, I also think this is a professional company so they would not do such designs.

But I cannot help wondering by looking at what is actually built.

The circuit diagrams I got do not all belong together, and some of them do not correspond exactly to my units. I'll try to make a compilation of pictures and diagrams the next days, if I can spare some time.

cheers
Martin

[orbanp1]

Hi Martin,

Here is an app note by the late Jim Williams about high efficiency linear regulators:
http://www.linear.com/docs/4126
Among others, there is a circuit with an SCR-based pre-regulator.
This might be some help to figure out how the circuit in your PS works, if not in the actual circuit, but in the principle.

Regards, Peter

[Apfelmus]

Hello and good morning,

Here are some pictures and circuit diagrams of the power supply.

The first picture is a bottom view with the wiring between the modules. On the upper left is the "Ausgangsplatte". It has a 31 pin connector holding the regulator board that can be seen in picture 2. On the rugulator board I had to replace C3 and D5 (upper left), both had a short. Resistors R4 and R5 are the ones heating up to 150-200°C in normal operation.

The 3 modules on the upper right are the switching cards (Umschaltkarten) with the SCRs. The 3rd picture shows these cards from the side, with the aforementioned resistors - the one on the right heats up to more than 300°C when output voltage raises above 12V.

All other cards are identical power modules. The card on the lower left side is used as a driver for all other modules. This card did never get warm when all the others heated up under load. I first thought it was defect and was ready to take it out for inspections when I noticed the slightly different wiring...

The circuit diagrams are not that helpful for understanding of the interconnections and global functioning of the unit. They do not always correspond, in particular the switching card which is only half of the circuit depicted. Pin-numbering also does not correspond or is missing either in the drawing or on the unit.

Its interesting to see how Zentro rationalized manufacturing at the expense of service-friendliness, at a time when other big companies still made exceptionally well designed units (mechanical-wise) and very good documentation. The modules are clamped by the rails of the cabinet - if you need to take out one you have to unscrew the rails, which makes the whole block of cards falling out of the unit. The two wires on top are not connected to anything in the circuit, they seem to have only mechanical purposes.

have a nice Sunday!

cheers
Martin

[orbanp1]

Hi Martin,

Hmm, interesting design! Splitting up 10A load among 11 output cards!

Looks like the schematics are not really for the cards you have in your PS!
I guess you need to do some more reverse engineering, optimally drawing out all the cards and the interconnects!
That is only four different cards, and hopefully the schematics you have are not that far away from your cards!

From the switching card schematics (pg. 5) it looks like they do not use pre-regulation, but they turn on raw voltage sources in stages, in cascode connection.
Also, it is probably not the schematics of your card(s), as your PS seem to have individual cards for the three raw voltage switching.

Sure sounds like fun!

Regards, Peter

[Apfelmus]

Hi Peter,

the unregulated voltage is ramped up in 4 stages. It starts with roughly 35V (idle), then 50V, 65V and 80V. The total current output of 10 amps is spread among 10 power cards. The 11th card is used as driver card, with identical layout of the card itself.

One of the 10 power cards has the bridge installed (with the asterisk in the upper left part of schematic). This output is routed towards the first switching card (the one closest to the transformer).

The circuit diagram of the switching cards is correct, but there is only one stage instead of two on my cards. You must suppress the upper (or lower) half of the SCR switching mimic and its corresponding parts on the right side. The power card diagram is correct, the regulator card diagram looks right, too.

I also think that a complete schematic would be helpful. Is there a good drawing tool for circuit diagrams on Mac?

cheers
Martin

[orbanp1]

Hi Martin,

I am slowly piecing it together!

When you say the cards are close to the schematics, do the card numbers match the schematics numbers?

What is missing for me are the numbers on the power cards (pg. 4.)!
One can guess, but knowing it accurately should help. Those are the connections to the switching cards, where you have the problem with the overheating/burned resistors. If I am getting it correctly, those are resistors R7 and R8 on the switching card schematics. I mean functionally, as your switching card configuration is different, your switching card has only one circuit, but you have three of those cards.

The functional block diagram is also cryptic, e.g. it is not entirely clear where does the control voltage to the pass transistors comes from, matching up the connections with the control board and output board is not straightforward!

I guess these are the things that you can verify on the PS!
After that it should be a bit easier!

Regards, Peter

[Apfelmus]

Hi Peter,

here you can find my annotations to the drawings. I totally agree on your opinion about the schematics - thats why I'm here.

The numbers on the power modules (PM) corresponds to the sequence of notches on the board. Not all are connected, thats why some numbers are missing. The sketch is of the switching card (SC). And NO, the module numbers do not match those on the schematics.

About cabling:
- Pin8 on PM goes to Pin14/15 on all SCs.
- Pin1 on PM is the output of the 3rd SC (Pin29/30)
- Pin2 and 3 on PM are fed the first pair of windings of the power transformer.
- Pin4 on all 10 PMs is fed by Pin6 on the 11th PM (used as driver)
- on this driver card, Pin4 is connected to the regulator board
- Pin2/3 on SC is connected to Pin29/30 on the preceding SC
- Pin 17/18 on SC do not seem to be connected.
- Pin 8/9 and 27/28 is the 2nd, 3rd and 4th tansformer winding. The middle tap seems connected to the bottom of the diode. I'm not 100% sure about the position and polarity of the diode in the circuit, there are several wires going on the heatsink but impossible to see exactly where.

Hope this helps.

I did not picture the circuit layouts of the PM and SC. They are soldered into the unit in a way that its not much fun to take them out.

cheers
Martin

[orbanp1]

Hi Martin,

Did some simulation of the switching card.
The bottom line is that R1, the 220 Ohm resistor dissipates quite a bit of power, about 1.9W.
This with my assumptions.
I modeled just one stage, and used raw AC voltages of about 25VRMS. Same voltage for the stage above that provides the voltage for the SCR gate currents.
The SCR used there needs typically 2mA (max 10mA) gate current according to the data sheet.

What i did not see, or do not know where and how are those rectified stage voltages filtered? Is it filtered stage-by-stage, or is it filtered at the end, when all the stage voltages are added together?
After all, the output is a linear regulator and that needs filtering of the raw unregulated voltage.

This is where the problem could lie.
If the stages are not filtered individually then the voltage for the gate current is not filtered either, it is a rectified sinusoidal voltage.
To fire the thyristors early on you need the small resistor that you would get the gate current, and turn on the thyristor, early in the phase.
I assumed this unfiltered case, and the peak gate current was about 63mA!
That is what causing the power dissipation on that resistor, and it seems that resistor was sized for the unfiltered case.

You could probably clear it up for me how the filtering is done!

Not touching the main filtering schema, extra filtering can be added to that auxiliary voltage that provides the gate current, separated by a diode from the rest of the circuits. In that case the R1 resistor value can be increased, the dissipation reduced. It works in simulation.

In any event, this my analysis so far, but please, look at it with a critical eye!

Also, the thyristor model I used was for another model, I did not find a spice model for the BT151-500R.

Other questions I would have what are the AC-RMS voltages of the various stages from the transformer? From the differing R1 resistor values it seems those are not the same!

Peter

[Apfelmus]

Hi Peter,

fascinating. What tool are you using for the simulation?

Unfortunately I put together the whole unit, so I cannot make measurements right now. I remember:

• at rest, with 0V output, the unregulated Voltage is approx. 35V,
• when the first switching card fires (at approx 12V) the Voltage rises to approx. 50V, and so on (65, 80) when the others come in,
• filtering is entirely done on the power modules (i.e. at the end),
• the 220 Ohm resistor is a 1 Watt type, heating up to more than 300°C (measured with an IR camera). In another (identical) power supply I noticed it has been replaced by a 5W type.

I do not understand why these R1 resistors on the switching boards have different values, and why the one with 220Ohm is heating more than the others. Its dissipation does not seem to depend on whether the other cards have fired. The moment of firing seems to depend only on the value of the Zener diode.

cheers
Martin

[karesz*]

Hallo Martin,
your measured 300C is massiv to correct for emissions factor, then soldering tin isn`t molded...
Can be interessant to see what is to find out if a too hot resistors card is in another place?
Karl

[Apfelmus]

Hello Karl,

yes indeed, 300°C would melt the solder. But the resistor is mounted approx. 1cm away from the board, so the soldering point should not get this hot.

cheers
Martin

[orbanp1]

Hi Martin,

I have been using LTspice, I would be glad to send you the simulation files if you are interested.

The power dissipated on R1 depends if the stage is turned on (irrespective of at what output voltage), on the voltage of the stage above, and on the value of R1.

This is the added rectification to the switching circuits. I am showing it on the card with two circuits.
This could be added to all the stages.
You might need to tweak R7 (R7 in this schematics, it is what we called R1 above) depending on the voltage in the stage above, under full load, when the voltages would sag.
With the voltages as used in the simulation, R7 (1 kOhm) dissipates 0.9W.

Again, these are simulation results, you would want to fully test it in real life!
The beauty of the testing is that there is no need to cut any trace on the boards, the cap can be tacked on, the diode used as the "wire" between the boards.

Regards, Peter

[Apfelmus]

Hi Peter,

yes, I would like to see the simulation files. I used LTspice myself some time ago.

Looking at your design change I wonder why Zentro did not do something like that...

cheers
Martin

[orbanp1]

Hi Martin,

Send a PM with your e-mail address!

Regards, Peter