[SOLVED] Problem with an old 10KV/3mA DC high voltage module

[D.A.(Tony)Stewart]

Hello all

I am using an old 10KV/3MA DC high voltage module and a few times I repaired it because parts are very old but this time I have stuck. When I don't connect any load (usually about 3.3 mega ohm 1.5W passive resistors) I have 10 KV at the output but when I connect the load, high voltage get killed. There is a pin (0-10V) shows what KV I have at the output. With no load I have 10V there but with load I only have 2V. I used to have 10V with load. I need help to diagnose and find the problem. Those who are familiar with HV circuits please advice. Schematic is attached.

Agoo

My analysis would start like this;

1. Rated Linear Load = 10kV/3mA = 3 MΩ
2. Assuming @10% load regulation of 10kV,
3. ESR = 1kV/3mA= 1/3 MΩ
4. Rated Power = 10kV * 3mA = 30W

Test Results

• 2kV @ 3.3MΩ = 0.6mA load means ESR is (10-2) => 8kV /0.6mA = 4/3MΩ or 4x normal ESR

So look for causes of high ESR or shunt leakage in primary and secondary paths.

1. test Diodes with DMM Ohm tester for reverse leakage
2. test Caps for leakage or loss tangent.
3. or test power off decay time constant
If the storage caps are equivalent to 1uF the decay rate with =4/3M leakage and no load will be 4/3 second
If 0.1uF then it will be 4/30 second.
It is hard to read schematic values perhaps 0.2uF @ 3kV

You can make a capacitance Hv Load divider 1000:1 using
four series 0.1uF caps 3kV rated caps to one 0.025mF cap rated at 10V for a 10kV rated 1000:1 V divider.

Then if you find the caps are no good use these to replace the ones you got using two per 0.2uF part. $4.86 each

If it is the diodes, then replace those.

I wonder which type of caps are used.

or

If your decay RC times constant is >5 seconds , then the HV section is good. This Means 60% drop for RC=T , no load from 10 to 4 in seconds

You can also connect a DVM to TP #7 to read V

Ensure all dust is removed and clean with isoprop. with output shorted (dont forget memory effect in caps)

 

[betwixt]

Tony, see post #4, the HV output section has been eliminated by substitution.

The odd thing is the supply voltage to the transformer significantly drops under load but so does the current drawn by the oscillator transistors. I would expect it to drop as the current increased, not the other way around. It appears to be a first stage voltage regulation problem although the voltages are all varying in sympathy with the load. It could just possibly be a false overload detection causing the bias to be pulled away from the oscillator, that would account for the current consumption dropping but as far as I can tell, the overcurrent indication isn't showing.

Brian.

 

[D.A.(Tony)Stewart]

OK thanks Brian, Then the problem is primary side.

I see now 70V from load test from the 95V supply seems excessive.

THe Primary HV switch uses some quasi Darlington's ganged.

THe 2N3442 is rated for Vce(sat)=5V @ 10A, with 2 in parallel this means Rce(sat)= 0.5 and Re=0.5 5W
so the total is 1 each switch or 0.5Ω with ganged switches.

Current gain is only 5 on these old switches. MOSFETs today will do this with easy 1000x better. This is a very old German design.

If the test shows 2kV @ 3.3MΩ = 1.2 Watts and the voltage gain is 2kV/70V=29 instead of 10kV/90 = 111,
then our converter gain is too low
And the switch is too lousy.

Tony Stewart

- - - Updated - - -

With a 30W secondary power , the primary needs to be at least 50W due to poor efficiency of this design.
Thus 95V must supply 0.5A and drop only 0.25V under full bias.

Thus the bias from the uA741 via 3.3k from 12V out to Q9 is failing.
Q9 is BF257

Poor old BF257 is hFE @ 10/30m (Vce / Ic ) = 25 min

except poor old uA741 typically only supplied 25mA short circuited.

I would replace this front end design with better designed parts

For starters, try any old Darlington instead of Q9 ( of course rated for > 100V , NPN)

Then regulated primary supply becomes 90V instead of 70V and secondary becomes 10kV under load. if you are lucky with no other faults.

 

[agoo]

pplus
Brian

Any current overload or voltage overload will cause relay K1.1 (above posted image) to deactivate (through Q17 and Q18) and cut the voltage at point "G" but this is not the case. We have voltage at point "G" both with and without load.

Brian
I saw in your above post you have mentioned "The odd thing is the supply voltage to the transformer significantly drops under load". This is wrong. It was my mistake in my first post. As you realized it I got voltages mixed up. Correct supply voltage to the transformer under load is about 72V and with no load it is about 42V.

Agoo

 

[D.A.(Tony)Stewart]

Did you replace Q9 with any 100V Darlington yet?

• 10mA is all that is needed.
• Looks like a design weakness to me.

 

[agoo]

Tony

Any part number do you suggest for replacement Darlington for Q9? We have BSR50 but voltage rating is low.

Agoo

 

[D.A.(Tony)Stewart]

https://www.digikey.com/product-sea...=1&stock=1&quantity=1&ptm=0&fid=0&pageSize=25

 

[D.A.(Tony)Stewart]

Most common part for this

https://www.digikey.com/product-detail/en/MPSA29/MPSA29-ND/1048908

 

[agoo]

Tony, I replaced Q9 with TIP122 darlington but still the same. I can see HV arc without load when I do screw driver test but no HV arc when I connect the load.

 

[D.A.(Tony)Stewart]

If you cannot get 90 V out of power transistors with 95V in and only get 70V something in between in wrong.

If 95V drops, then that is a problem. No need to look further. If power gain is insufficient , add another 10k in parallel off darlington collector. 1W min. For each.

 

[agoo]

I think I don't get 90V because voltage limiting circuit (R41) is acting to limit the output voltage bellow10Kv which means power transistors output needs to be lower than 70V. Am I right?

 

[D.A.(Tony)Stewart]

But you aren't getting 10kV which is the crux of your problem.

Yes & No
R41 is the threshold adjustment for regulating HV out level.
U5 & U4 are two stage filters for PID ( Proportional, Integrator, Derivative) feedback to match this threshold.
If U4-out is high then,
Q9-C-out is low then,
Q8-C-out is high then,
Q25-C-out is high then,
Q61//Q71 pair E-out is high and you get 90V
THis transistor bundle is an LDO-like regulator to remove ripple in 95V input with < 5V dropout when driven hard and under voltage out on HV,

If OVP and OCP and any other feedback is not limiting the output HV and I assume the threshold is set properly on R41 for 10kV, Then you should be getting 90V out in order to power the 10kV inverter under load.

But it is failing under load, so always start from DC source and trace the loop.

 

[agoo]

Brian
Tony

Parts inside orange boxes (see attached schematic) are all located in a separate module and the rest of the parts are located on another module. I replaced the module of orange parts with another working one from our second working HV unit and I saw problem went away. I had 10Kv with both load and without load. So the faulty part is located among the parts inside these orange boxes for sure. Now with this knowledge please give your advice how to get closer to the faulty part.

 

[betwixt]

Thanks Agoo, that is very helpful. I'm sorry it seems to be taking a long time to make a diagnosis but without the unit in front of us it's difficult to try things out and being in different parts of the World doesn't help either !

Can you try shorting out the base and emitter pins of Q15 to see what effect it has. It will disable the overload protection to see if it is mistakenly thinking there is something trying to draw too much current.

If that makes a difference, can you tell me the voltage measured across C39 with and without load please. I am trying to see if the bias regulator is responding to changes in load.


Brian.

 

[D.A.(Tony)Stewart]

Agoo either the LDO is overloaded from the right side or the left side is failing as I described. Can you measure these transistor to find the bad one from my explanation?

 

[pplus]

Check the voltage +95V with load and without it. The matter can be quite simple - a source of DC lost capacity of filter capacitors.

 

[betwixt]

Pplus, you have taken us full circle to where we started from!

What we know:
The incoming supply is stable. (power source is good)
The 10KV is present without load. (HV oscillator must be running)
10KV drops to 3KV under 3mA load.
Voltage regulator seems to be trying to adapt to the change in load.
By substitution, fault is isolated to the HV generator circuit but NOT the rectifiers or output capacitors.

The overload circuit works by shunting the oscillator bias current, I'm trying to get Agoo to temporarily disable the current sense to see if it is prematurely detecting too much load. I suspect if the fault is in the oscillator circuit itself, it wouldn't generate any HV at all.

Brian.

 

[pplus]

Voltage at base of Q9 is -4V no load and 350 mV with load.

Q9 is closed

I measured a few points on a borrowed similar HV unit. Results are as follows:

Voltage at collector of Q9 ia about 90V with and without load.
Voltage at Q9 base is 0.5V with and without load.

Open

Voltage at Collector of Q9 is 90V with load and 89V without load.
...
Voltage at base of Q8 is 90V no load and 91V with load.

IR33=0.1mA

Any clue?

Leak in C13?

 

[agoo]

Hello All

I had to split the faulty module to 2 sections (section 1 and section 2). Section 1 contains all suspicious parts at the left orange box and section 2 contains all suspicious parts at the right orange box. disconnecting all section 2 connections and connecting section 1 to the section 2 on the working HV module I have HV both with and without load. So it is obvious section 1 parts are all OK now. Now faulty parts are located at orange box in section 2 (right orange box) for sure. At this orange box also CR37 and CR38 should be excluded because I replaced them with new ones. Attached schematic shows updated suspicious parts left in orange dots. Please renew your comments accordingly.

Agoo

 

[pplus]

We have two facts. Post #12 https://www.edaboard.com/threads/335654/#post1433961
1. Case - no load.

Output voltage is 10KV.
Supply voltage is 42V.
Voltage at emitter of Q7 and emitter of Q6 both are 40V without load.
Voltage of point "G" is 42V without load.

Measurement is wrong. The voltage at the point "G" can not be higher then voltage at emitter of Q7 and Q6.

Voltage at base of Q9 is -4V without load.

It is also wrong? The output stage in this case must be closed, but we have 42V output.

2. Case - with load.

Output voltage is 3KV.
Supply voltage is 73V.
Voltage at emitter of Q7 and emitter of Q6 both are 74V.
Voltage drop across R34 and R35 is 1V. The output current is 2*(1/0.5)=4A.
DC power is 74*4=296W. ??
Load power is 3000*0.003=9W. The conversion efficiency is 0.03.

Voltage at base of Q9 is 350 mV with load.
It is also wrong?

Conclusion.
Converter works, but is not effective. BDY56 is old slow transistors. If the conversion frequency is changed to a higher, through currents may occur, causing a significant power consumption. Check with an oscilloscope conversion frequency and operating modes Q12, Q13. Also check the timing capacitor C22, C23?, C26.