A Test Set for Nondestructive Safe-Area Measurements

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Hi all,
To the newcomer:
This tread is set up with the aim to build a A Test Set for "Nondestructive Safe-Area Measurements under High-Voltage, High-Current Conditions" for transistors.
It is a well known fact that there are many counterfeit power transistors that do not live up to the code they carry.
Look here: http://sound.westhost.com/counterfeit.htm

It is obvious that such a batch, built in a new project or the repair of a beast of an amp can create havoc and cost a lot of money.
Not only that; you still have to find replacements that are genuine.

A search in to my memory brought up some Application Notes from RCA dating back to the early seventies.
See here:
http://www.diyaudio.com/forums/showthread.php?postid=1402235#post1402235


I finally found one of the few I was looking for.
Look here for the AN-6145:
http://www.diyaudio.com/forums/showthread.php?postid=1421066#post1421066

This will be the subject of this tread.
We will analyze the circuit (by the many great thinkers, better as I am) and build a modern instrument that will help our fight against fake transistors.

Now have a look at the AN-6145.
Start with proposing circuit changes so that actual components can be used.
Make parts of the instrument simpler.
Let's make it usable for not only NPN but also for PNP power transistors.
Is it possible to add a sort of curve tracer?
Controlled by a computer?
etc, etc.

Let's make something out of it.

Thanks and see ya on this tread....

Tarzan
 
Tarzan said:
...................
Start with proposing circuit changes so that actual components can be used.
Make parts of the instrument simpler.
Let's make it usable for not only NPN but also for PNP power transistors.
Is it possible to add a sort of curve tracer?
Controlled by a computer?
.................



Obviously, it would have to be made with components which are available now.

NPN and PNP would be better but the power supplies would need careful thought to avoid complication / duplication.

Curve tracer and computer control.... nice ideas but let's get the basic thing going first !!
 
Not much response I must say.

RCA AN 6145 One File.pdf
http://www.yuntaa.com/FileManager/Download.aspx?ContentID=45AA9C44149C0C0CE040A8C003025250

I tried to analyze the schematics and came up with a simplified block diagram.

Simplified.jpg
http://www.yuntaa.com/Photomanager/SnapView.aspx?ImageID=466D66CC8127CF6FE040A8C003027DCA

The whole test setup can be made up of 5 main blocks.

1. An adjustable Collector voltage source.
This can be a 0 .. 150 or 300V source / 5 .. 20 Amps

2. An adjustable load for the transistor under test. Must be able to withstand the full VCC at reasonable current (.. 20 Amps)

3. An adjustable Bias supply for the load under 2.

4. A fast flip Flop and Crowbar that removes the bias in the event of a failure of the DUT

5. A pulse source that can drive the Load circuit.

It would be nice to treat each part on his own and later on combine them to complete the test setup.

An exception could be the different power supplies.
They can all be brought together on one board.
Some voltage are not known because some blocks aren't worked out yet.
I would say we stuck with + 12 V for logic and relais to stay with commonly available parts.

Cheers and hints are welcome
Tarzan
 
Interesting....

With modern stuff like MOSFET, SMPS, microcontrollers, high speed gates and comparators, I think that a compact portable automatic tester for both NPN and PNP is not that hard to achieve. It may be funny as a future project ;)

I hadn't figured out that non-destructive testing could be achieved by removing power within a few hundred nanoseconds in case of overcurrent. With nowadays components I think that power can be removed within less than 100ns of fault detection.
 
Looking at your block diagram, it seems to me that Rload would be better in the collector lead of the Load Switch transistor, then the base drive is simplified. Seems that Load Switch should be built as a switched current source so that the test current can be easily set.

You might want to consider running the logic on 5V so that you can take advantage of the advanced high speed logic families, AC for example.

I would have the main collector supply be an external power supply so that a bench type unit can be used with precise voltage and current control. It offers more flexibility.

I've been thinking about building a SOA test system for some time now.

Pete B.
 
Yep PB2, there is indeed a negative supply towards the emittors of the load switching transistors.
I attached a new drawing with the changes.

I have a problem attaching it to this message. Find it here:
Simplified 080219-1.jpg
http://www.yuntaa.com/Photomanager/SnapView.aspx?ImageID=467F67910DE20514E040A8C0030215AE

RLoad is now connected to a -14V supply and not to the ground.

I do have a problem with understanding the use of the Vbb (0..20V) that goes to the puls generator.
Can some one explain me that?


Your comments PB2:
"Looking at your block diagram, it seems to me that Rload would be better in the collector lead of the Load Switch transistor, then the base drive is simplified. Seems that Load Switch should be built as a switched current source so that the test current can be easily set."
Can be looked at.
I recall the RCA rep telling me that he knew of an engineer in Belgium who built this unit and was very happy with it.
But that's more then 30 years ago...
So the original setup should work.

"You might want to consider running the logic on 5V so that you can take advantage of the advanced high speed logic families, AC for example."

That's a good proposal. The pulse generator and flip Flop-crowbar can be made of these.

"I would have the main collector supply be an external power supply so that a bench type unit can be used with precise voltage and current control. It offers more flexibility."

Yes can be. Allthough a test set-up that stands on his own is very handy to. I think the HV supply is the most problematic.
Some schematics show a simple HV supply. Not powerfull enough but that can be done.

Tonight I try to upload all I've found and leave a notice here.

Cheers,
Tarzan
 

Attachments

  • simplified 080219-1.pdf
    55 KB · Views: 162
Hi Tarzan

I wonder if a MOSFET would be better as the switch in the emitter of the test transistor?

Might be faster.... but there again, perhaps the switch transistor could be an MJL3281A which would be fine up to ~5A and two for 10A...

I think your blocks pretty much sum up the RCA circuit.

I guess the Vbb supply is just for the pulse circuit. Probably could run from the same psu as the negative load supply.

Need to protect the switch circuit against high voltage pulses ...in case ... maybe discrete instead of 5V logic ?



cheers
John
 
A Mosfet could do the job as I've seen Power Supply Loads equipped with a fet. But where?

Attached is a schematic available on the net of a HV psu. Adjustable from 0 to 300 plus volts. Can be fine tuned to give even more or less...

A low voltage supply is included but that should be no problem at all with the regulators available.

Now I'm stuck with the values of the pulses needed to trigger the load switching elements.
Also what voltage these pulses have to be.

Some details of the used HP 214A
Product Specs:
Amplitude Range (0.08 V to 100 V)
Source impedance: 50-ohms on 50 V and lower ranges; 1500-ohms on the 100 V range.
Internal Repetition rate from 10Hz to 1MHz (5 ranges)
External from DC to 1MHz, 15 ns transition time on the +50 V range.


Impressive and not so easy to build.
But I do not think it has to be so complicated.
A circuit with a couple of NE555's and a power amp (to get the desired output voltage) can do the trick.

As I can think in block schematics, details of the schematics are not my best parts.
Therefore more ingenious readers can come in to discuss the details of each circuit (or part of it)

I found some idea's on the pulse generator but if I know a few more details on the width, frequency and voltage I can see if it is use full.

Cheers,
Tarzan
 

Attachments

  • hv psu.pdf
    85.3 KB · Views: 102
The push button - test - triggers a one-shot m/v.

A similar continuous pulse generator could replace the HP source.

The pulses are from 0 - 20v and have width of 50mS (test) or less (HP)

The 0 - 20v supply provides drive to ensure the correct current pulse.

At least that is what I read.

Andy
 
hi,Tarzan

thanks for your infos.I have shared the RCA paper to my friend Locky_z which designed a curve tracer of transistor.

http://www.diyaudio.com/forums/showthread.php?s=&threadid=85076

after reading the AN6145,I suspicion that the schematic of AN6145 miss the important circuit of mercury wetted pulser(only show the coil driver),and the active test-time is very short like about less than 100ns.

you can find the detail of mercury wetted pulser in the tek 109 pluse generator sevice manaul.

ftp://bama.sbc.edu/downloads/tek/109/

X.G.
 
Hi X:G:
As the link provided in the tread looks dead (http://www.hifidiy.net/bbs/boke.asp...owtopic.92.html)
It looks nice and the results are great.
Details off the sw should come in handy for that project.
Can you get locky_z in to the discussion?

Yes it is not at all clear what the relay K1 in fig 5 is actually doing.
I also can not find a reference to any contact of that relay.
That stays a mistery.
But I assume that the relay is switching VBB to the base of the load switches (Q1..QN) at the pulse width set by the circuit of fig5.
l'lI have a look at the TEK 109 circuit.

Well, after some thought; why inventing the wheel if it's allready there.
The gating, flipflop and crowbar do work and excist already.
Just copying it and use modern components would do the trick and no hasle with "exotic" sensing chips that will make repair or change in the future quassie impossible.

But if the test setup will be alltered so that PNP transistors can be tested, then maybe another setup is needed.
Questions I ask myself and I have to get back on the ground.
Try (I) to build this one first (for NPN) and then think about modifying...

Tarzan
 
X.G. said:

I had read carefully all the circuits of AN6145.It is not show the connecting of relay contact which is the key of pluser.:bigeyes:

As I stated above, the coil for the relay is driven by Q1, as in diagram Fig.5 (which also show the relay).

Internal / external pulse selection is by SW2.

SW1 is the test button.

But I assume that the relay is switching VBB to the base of the load switches (Q1..QN) at the pulse width set by the circuit of fig5.
Correct.

The text makes only a passing reference and relies on the diagrams.

It is quite clear!!!:smash:

What part do you not understand ??


Andy
 
Tarzan said:
Hi X:G:
As the link provided in the tread looks dead (http://www.hifidiy.net/bbs/boke.asp...owtopic.92.html)
It looks nice and the results are great.
Can you get locky_z in to the discussion?

I had discussed with him for this SOA test set.we live in the same town, I can meet him in 10 minutes after contacting.I will tell him for your wish.

you can find more infos about his curve tracer.

his blog (but lost some important crticles)

http://blog.hifidiy.net/?176

those are his posts of the curve tracer.he haved developed this tracer for a few years.

http://bbs.hifidiy.net/viewthread.php?tid=63530&highlight=+LOCKY\_Z

http://bbs.hifidiy.net/viewthread.php?tid=22044&highlight=+LOCKY\_Z

http://bbs.hifidiy.net/viewthread.php?tid=66451&highlight=+LOCKY\_Z

http://bbs.hifidiy.net/viewthread.php?tid=76088&highlight=+LOCKY\_Z

http://bbs.hifidiy.net/viewthread.p...OCKY\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\_Z&page=1
 
QUOTE]Originally posted by Tarzan
Yes it is not at all clear what the relay K1 in fig 5 is actually doing.
I also can not find a reference to any contact of that relay.
That stays a mistery.
But I assume that the relay is switching VBB to the base of the load switches (Q1..QN) at the pulse width set by the circuit of fig5.
l'lI have a look at the TEK 109 circuit.

[/QUOTE]


yes.

you will know the theory of mercury wetter pluser and the missing of AN6145 after read the TEK 109 manaul.
 
I cannot access the manual.


However....................

the relay is employed as a fast closing switch. This bounceless switch has fast rise times (typically 10 ns or less) and low, stable on-state resistance.


So what is the problem.?

The switch operates the one-shot m/v.
The one-shot drives the transistor.
The transistor drives the mercury-wetted relay coils.
The relay closes and sends a voltage to the base of the current source transistors via the resistor.


The internal pulse generator in the original schematic is only a one-shot to act as a "test".



Andy
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.