Maker Pro
Maker Pro

Design of a transistor curve tracer

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have now switched to a N-channel JFET (BF245A) which is my favorite by the way, the plot looks extremely linear but it has to do with the fact that I now only run 0,5Vpp as Us (which is the same as Ug in this case), as you may remember I changed this to make my BJT work but now my JFET needs a couple of Volts to work so, sadly enough, I need to change the blue LEDs to some Zeners of some 5V so that I can run 0->3V at least even though this still will not cutoff the JFET, I would need some -5V for that.

Best regards, Roger

DSCN3921.JPG DSCN3922.JPG DSCN3923.JPG
 
Last edited by a moderator:

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have now managed to trim my BF245A-plot.

I switched back to 5V as step voltage (Us) and while BF245A is cutoff at some -2V I could, with my Us, fully cutoff the transistor, actually it sufficed with only three steps and thus some -1,5V.

I am very pleased with this new graph of mine, it looks almost exactly as in the datasheet.

I have even (roughly) hit Idss at some 4 milliamps.

My next step will be to realign the Us for BJTs because it has to be lower there (due to the current generating thransistor that needs a few volts to work properly).

While I use inverters it is simple to attenuate from 0dB to -20dB by changing the input resistor from in my case 10k to 100k.

When I have done this I will probably also change the current setting resistors because right now I have 0,1mA totally and thus 10uA/step (ten steps), this times some 200 of hfe gives only 20mA of collector current, I wish to increase this to 100mA which means I will have to lower the resistor with an magnitude of five, this means to 1k.

I also have a high current option (for TO3s) and the current "generating" resistor is 10 Ohms, 0,5V over 10 Ohms means 50mA as total base current BUT 50mA times a low Hfe like 10 is only 500mA which is far from the collector current a TO3 gan give. So here I also wish to lower the current setting resistor to some 1 Ohms which would mean som maximum base curret of 500 milliamps (thus some 5A of collector current).

Best regards, Roger
PS
Features I miss in your site is the possibility to write text next to pictures, a quating button that makes it possible to just copy a sentence, hit the quating button än place the copied part inbetween the tags, I also dislike the way bold and such stuff is generated, this should of course be done using tags and not bolding the post, I also miss the feature of hitting a button when you wish to share an article number of som supplier, finally I dislike preview being an "More options", because how can that be? I hit preview all the time and I do this to check my spelling, my gramatics, my choice of words and the exactness ot the technical elaborations, or do you not wish people to the that thorough? Finally, I dislike that you are not capable of editing your post after a certain time, why is that? If I feel that I have written someting wrong should I not have the opportunity to change that? I'm sorry but that is just plain stupid, why not do as elektronikforumet.com where you can edit your post as many times you like, all that happens is a row that tells people that an edit has been made.

DSCN3926.JPG
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Features I miss in your site is the possibility to write text next to pictures

I've honestly never tried to do this. I normally just post the thumbnail of the images with interspersed text leaving it up to the reader to get the full image if they're interested. As someone who tends to read this site on a phone more often than on a PC, text next to the images would probably not work for me anyway (although I'm sure it would be great on a larger screen).

a quating button that makes it possible to just copy a sentence, hit the quating button än place the copied part inbetween the tags,

If you highlight some text in a post, a little button will appear allowing you to quote just that. It can be a word, a sentence, a paragraph, whatever. I use this a lot. You may have seen it done in this thread (I did it here to just quote your postscript). I think it also works with the main quote button...

I also dislike the way bold and such stuff is generated, this should of course be done using tags and not bolding the post,

It can be done with tags. By default you're in the WYSIWYG (what you see is what you get) mode. There is a button near the top right corner of the tools pane (it is a page with a spanner) that will allow you to see the tags. Regardless of the mode, you can enter tags manually it you desire (I did it then -- which is a real pain on a phone). The easiest approach is to highlight the text and press bold (So much easier for this poor phone user, and probably easier for a man and his mouse, but the person who wants to keep their hands on the keyboard can do it the way I think you want)

I also miss the feature of hitting a button when you wish to share an article number of som supplier,

I normally embed links using the link button in the toolbar. Again, you can do it with manual tags. And if you just enter a link to a YouTube video, the software renders it as an embedded video.

finally I dislike preview being an "More options", because how can that be? I hit preview all the time and I do this to check my spelling, my gramatics, my choice of words and the exactness ot the technical elaborations, or do you not wish people to the that thorough?

I think this is influenced by the fact that WYSIWYG mode is the default. I read my post before I hit the "post reply" option. For very long posts with complex formatting (see the LED resource) I used the option to view tags and the preview option *lots*, but I use it far less often for more normal posts like this.

Finally, I dislike that you are not capable of editing your post after a certain time, why is that?

There are a number of reasons:
  1. We had problems with spammers who would go back and edit their posts to contain links to dubious sites. Edits get less attention from our members and the moderators, but no less attention from search engines. It was a problem we wanted to hit on the head.
  2. We had a a problem with some newer members who insisted on editing their earlier posts in a way that made the rest of the thread meaningless. We try to ensure that thread are useful, not only for the original poster, but for someone who has a similar problem in the future. Preventing continued edits resolved this.
  3. On a similar note, we had people edit their posts to remove their question and all further information once their problem was solved, or on other occasions when someone has pointed out an error in their posts which embarrassed them.

If I feel that I have written someting wrong should I not have the opportunity to change that?

Sure. That is why you can edit posts. If you find a good reason to edit them later, please PM a moderator. We're usually more than happy to do the edit for you.

I'm sorry but that is just plain stupid, why not do as elektronikforumet.com where you can edit your post as many times you like, all that happens is a row that tells people that an edit has been made.

The decision to limit access to edits was made after discussion and has worked well for a long time. As moderators, we decided to take in the work of doing edits for others (in what has turned out to be very rare cases) in exchange for less spam and more easily read threads.

You may have noticed that I have made some edits to your posts because of a corrective post you made. I decided that the thread would read far better if I went and applied a few changes to your posts. I won't detail these here, but feel free to PM me if you want a more detailed understanding on why I did it and the motivation for it.
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Interesting, please teach me some basic opamp theory.

Positive feedback works differently to negative feedback. Whereas negative feedback tends to reduce the output swing of a op-amp, positive feedback increases it.

So, if, with negative feedback, the voltages on the inverting and non-inverting terminals differ, a change in the output brings them back to equal. This limits the output voltage swing. In many cases this results something like a fixed gain determined by the ratio of a pair of resistors.

But with positive feedback, a slight difference in voltage at the inverting and non-inverting inputs causes a change in the output which further increases the difference. In a practical op-amp the output can't go further than the supply rails (many can't even go this far) so there is a limit at which the output cannot go any further, and so the difference between the inputs also can't continue to increase.

This page (which assumes you know what negative feedback does) shows the difference between no feedback at all (when the op-amp acts as a comparator) and positive feedback (when the circuit has hysteresis).
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
The power MOS are made in Japan by the company Hitachi, 2SK175 is their name. Sadly these are extremely hard to find but the datasheet is awesome, I/V-diagrams included.

Why not find something more modern as a replacement. The datasheet does have the typical curves, but nothing out of the ordinary.

I have now switched to a N-channel JFET (BF245A) which is my favorite by the way

Now, I'm not sure. Are you talking about what you're using in your circuit, or the devices you're drawing curves for?
 

Attachments

  • 2SK175.pdf
    167.6 KB · Views: 39

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
Why not find something more modern as a replacement. The datasheet does have the typical curves, but nothing out of the ordinary.
Do you perhaps know any TO3 N-MOS?

The once I find are TO220 N-MOS.

Is there perhaps some bright way of serching for such?

But sure, any capsule works so then there are lots.

Now, I'm not sure. Are you talking about what you're using in your circuit, or the devices you're drawing curves for?
The last curves are for BF245A.

Best regards, Roger
PS
I was about to involve my JFET-project here also but I think it deserves a thread of its own, also it will scramble things up.
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I now wish to show you the plan for my whole Transistor Tester (I do however use the same name for the whole system as for one single PCB in the system, KTT)

It involves as many as 6.5 PCB boards where the .5 board actually is no board but a Power diode Device called KPD built on a couple of "sockerbit" which diode-sums/selects two supply voltages of 7-18V where 7V is my estimation of finished 9V battery and 18V is a supplied by my KTD (Transformed Device) which only lacks transformer in the picture (it will be mounted in the lid).

The other units are called KSB (simple blink-circuit made out of only two transistors), KSU (a variable supply unit), KTS (a Transistor tester Supply, which boosts my tiny OP27 with regard to current), KTI (which is the Transistor tester Interface that may be called the heart of it all which produces the stimulis with regard to the Uc/Ud-ramp called Ur and the Ib/Ug-stepsignal called Us) and finally we have the KTT (Transistor Tester which is the brain of it all).

I have produced a wooden (plywood) cabinet that will hold all these boards and I have even gone and bought a special tool to make slits in the plywood so that you may use a sliding 2mm Aluminum panel on both front and back, among others my plan is to assemble all TO220 transistors on the rear panel just to cool them, I don't know if this is neccesary for the KTT transistors but I am planning to use up to 500mA and with 18V of supply it feels like this is neccesary (Uce is somewhere around 15V but the current form is not DC so let's say 250mA which would mean a collector loss of some 4W and while a naked TO220 can stand somewhere around 2W I probably have to cool them).

Best regards, Roger
PS
I have a couple of very nice TO220 coolers that you just push on the capsule which makes it withstand some 4W but the problem is that there is no room for them on the board. Now I wonder about the problems I may get when I draw cables from the PCBs to the transistors that are up to 1ft long.

I see now that there is no room for the four KTT transistors on the back panel, do I really need to cool them?

By the way, the non asssembled PCB is the KTT which I am triming right now, the next step is to set Us for BJTs to 0,5Vpp then I will probably also change the base current determining resistors to get the ranges I want that is some 0-100mA Ic for small signal BJTs and 0-5A Ic for power BJTs.


KTT_Layout_BCE_3_final_4_Batt_KPD_2_KTS_KSB_PCB_lek_corr_3_värde_Power_Trissor_mät.PNG DSCN3722.JPG
 
Last edited:

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
Do you perhaps know any TO3 N-MOS?

The once I find are TO220 N-MOS.

Is there perhaps some bright way of serching for such?

I don't know if any in the TO-3 package off the top of my head.

What I would do is go to the digikey web site and just search for "mosfet". It should give you a list of options, one will be for "single mosfets" or something similar. Importantly, this option will have tends of thousands in that category. Once you select that you'll get the first page of a list of every mosfet digikey know about. Above this list is a series of criteria you can use to reduce the scope of the list. In your case you would first probably select N channel, and possibly some other characteristics like the party being in a through hole package and maybe a range of Vds(max). Then you update the list, and continue to refine your choice. If you want to purchase the device from digikey then you probably want to subset the list to those that are in stock, but you might not do that if you want to search other sites for the device -- digikey can be a great search even if you're buying elsewhere, I've not found a better parametric search.

As for the package, TO3P is another option. It's like a TO-220 except the package is wider and the pins are more widely spaced. The package has better thermal conductivity, much like the TO-3, but easier to wire up. Also look at TO-247. Note that some of these packages can be named several ways, TO-3P-3 just means a TO-3P package with 3 pins, similarly there are several variations of TO-220.

Oh and digikey normally provided links to the datasheets, so their search can be very helpful during design.
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
In a practical op-amp the output can't go further than the supply rails (many can't even go this far) so there is a limit at which the output cannot go any further, and so the difference between the inputs also can't continue to increase.
As I've said before I don't understand so much about electronics but listening to your nice repaly on my rude statements I have now used the feature of highlightning and hitting the emerging button (reply), I thus think that the last bold part might be wrong because the voltage difference between plus and minus is only low (or insignificant, even) when the OP works linearly, but when it is "overdriven" quite many volts can exist there.

I am probably wrong in this statement but that is what I believe so feel free to correct me.

Bestt regards, Roger
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have done some calculations on the bus today and come to the concluusion that the visible part of the traces is some 20mApp of Ic (depending on actual hfe), the bias due to the diffrence i reference DC and sweep voltage DC is some 0,7V, (which I measured yesterday) this over the Ib generating resistor of 5k means an Ib(0) of some 0,14mA which times estimated hfe of 200 means 28mA Ic(0).

So the lowest trace is biased at some 28mADC Ic (let's say 30mA) and all the above "real" traces covers 20mA, this should mean that zero Ic is distanced quite far below from 30mA such that the lowest trace (30mA) should rest about in the middle of the screen, but I get the "sweep-dot without signal" to rest quite near the leftmost corner of the oscillloscope grid and yet the first trace starts about there.

There is obviously something wrong here, I do not understand what it is but I am prepared to do some measureing, I will temporarily solder an 1M resistor over the Ib generating resistor (5k right now) and measure the real effect of the stepsignal (Us), I am expecting some 0,7V offset but otherwise all the steps should constitude a nice 10-step ladder.

I am also planning on trimming the offset from 0,7V to some 100mV but not less, this while I am running the system on 12V and it can be as low as 7V which makes my LED-refrences drop some in voltage.

100mV would then mean some 100mV/5k=20uA of residual base current which we may multiply with 200 of estimated hfe for BC546B to get 4mA of residual collector current, 4mA compared to 20mA is only 20% so this may be accepted (right now I however have some 30mA and thus 150% which is not acceptable).

I'm not sure if this last fix will help but it sounds better, moreover I have already tried this not once but twice and did not get any good results but I suspect there was some short circuit because my test-pin sleeves aren't the best when it comes to 100mil distance...

Anyway, I will try this.

Best regads, Roger
PS
Hope you can follow what I mean even though my english isn't the best. The current determining resistor is by the way R21 (5k now).
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have now tried to do this but it turned out that it was more difficult than I thought.

My KTT PCB is crowded to say the least so when I tried to add my 1M test resistor one of the 10k//10k went loose because I was too fast with my soldering iron.

At that point I thought I could as well change it as I have planned so I changed the LC 5k Ib "generating" resitor to my later preferred 1k and in that same procedure, where I had stripped the board of many TP wires anyway, I went away and changed the HC resistors also, so now I have a Low Current (LC) resistor of 1k and a High Current (HC) resistor of 1 Ohms.

Let's look at this.

There are tree magnitudes between the HC/LC values, is this good?

Well for low current measureing we now have 0,5Vpp/1k=0,5mA of base current and this times 200 is 100mA, theoretically perfect.

For high current measureing we have 0,5V/1=500mA of base current and this times 10 is 5A, which also is theoretically perfect.

Now, in my rebuild I also managed to add a 1M over the Us bias resistor, this while the bias seems a little low, my goal now is to increase the bias somewhat to meet the 3,5V a blue LED gives (within 100mV).

But as I tried to repair my KTT many wires got removed but what's worse is that while trying to reassemble it all I wind up with a left over wire, I actually don't know between which testpoints (TP) it was connected :D

So it seems like magic if this will work again.

Best regards, Roger
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have now gotten it to work again :)

I had to build myself a new IC adaptor, this time I used a component adaptor, soldered the correct IC legs directly to it and bonded plus/minus-inputs, the adaptor got rigid.

Then I tried it out connecting all three OPs as voltage followers and injected a voltage of half Vcc, they all seemed to work.

After some modifications on the board, where I wanted to check the bias over the base current generating resistor and even made it possible to adjust it, I pushed the adaptor in place.

Then I did some measurements and I grounded the Us so that the voltage over the current generating resistor would be steadily measureable (ideally zero with no Us signal).

Then I mounted a trimmer over the resistor that controls the bias "adding" and by tuning that trimmer I got down from 0,8V offset to 0,1V offset (I left it at that because my LED refrences are not ideal when it comes to a supply voltage variation from 7V to 18V, which is my plan).

Remember that a 0,8V offset means an offset base current bias of 0,8/1k=0,8mA and while each step of the Us is 0,5V/10=50mV and thus 50mV/1k=50uA of base current, 0,8V offset is a lot!

So I trimmed this and got some fascinating results!

I am however sceptical that this is true but it seems like the BC546B actually compresses signal when it is driven around its maximum Ic current (that is, the hfe seems to be less than at midpoint Ic-wise).

I don't know if this is true but I hope so beacause otherwise my KTT just stinks :)

Can it really be that this is correct?

Some saturation feature when Ic-current gets too high?

But eaven the Earley Voltage looks strange now so I guess I am doing something wrong.

My two oscilloscope plots shows that when perfecly trimmed, exactly all eleven traces are visible but the high current performance is questionable, in the other plot I have "bueried" some traces in "zero" so that not all are visible but it looks better, I have made this plot by "faulty" trimmig like above.

Best regards, Roger
PS
If I'm not mistaken, I seem to remember that BJTs has the highest hfe at a point somewhere inbetween their "lowest" and highest working collector current.

I have now uploaded a datasheet on BC546B and ACTUALLY hfe dives sharply after some 30mA, the highest trace in my plot being some 80mA.


DSCN3929.JPG DSCN3930.JPG DSCN3928.JPG
 

Attachments

  • BC546B.pdf
    72.8 KB · Views: 37
Last edited:

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have now switched the DUT to a N-MOS called 2N7000 which I recently bought a hundred of for less than a SEK per peace, extremely cheap and it turns out that this MOS transistor actually is extremely linear.

I have one thing left to do with my KTT and that is to test a P-channel, MOS/JFET-wise this can be considered done because only an inverter is being invoked or not and I have tested an N-JFET and now an N-MOS.

BJT-wise an PNP-unit is however required to be tested also, actually it is more simple to test all that can be tested before you assemle it all, my KTT has lots of testpoints (TP) and if it turns out that the unit does not work it will be not so simple a task to disconnect it, repair it and put it back.

So I need to know that my unit has the capability of working before I assemble it all.

Next step is thus to test an PNP unit, I will try a BC556B.

Best regards, Roger
PS
I can't get my LaTex to work. Picture text: Sensitivity is 0,5V per step, the number of traces is always 11 but only 9 are visible which means that 2X0,5V as Ug is required to turn the MOS on, Y-scale or Id is some 20mA/DIV. The kinky loadline has to do with the buffer transistor being a tiny BC546B, but this is only what I think (event though maximum continous current of BC546B is only 100mA).

DSCN3932.JPG
 
Last edited:

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have now managed to test a BJT complementary pair (BC546B/BC556B) and my system works, if it wasn't for the fact that the traces are swept a bit different because sweeping NPN gives "hysteris" around the knees while sweeping PNP gives hysteris almost over the whole Uc range.

The difference is not that big but it is kind of irrittating because as you see right above a MOS transistor can be swept perfectly, it is only when base current is generated as I get problems.

The MOS (2N7000) graph above shows that voltage gate sweeping works very well while base current sweeping does not work so well, this is an enigma to me.

Anyway I am attaching them all.

Best regards, Roger
PS
Picture text:
1) 2N7000 N-MOS, Ug=0.5V/step, Y=20mA/DIV, X=1V/DIV, first visible drain current sweep is for 1,5V
2) BC556B PNP, Ib=50uA/step, Y=10mA/DIV, X=1V/DIV
3) BC546B NPN, same scale as for BC556B.

When it comes to the unlinear loadline for the N-MOS I think this is due to me buffering a TL071 with BC546B as lone emitter follower at some 160mA while its maximum continous collectror current is 100mA, but I'm not sure that this is the reason, what do you think?

DSCN3932.JPG
DSCN3934.JPG
DSCN3929.JPG
 
Last edited:

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have started to make a cabinet for my Transistor Tester (KTT).

Actually I am on the way of making three cabinets (for KTT, KAP and KGA) but I am kind of focusing on KTT.

Yesterday I sawed a front panel out of 2mm Aluminum, it was kind of hard work because my NEW saw wasn't so sharp (took about half an hour to saw 50cm) and the result also did not fully fit length-wise (my fault).

I have come to the conclusion that I am not competent enough to make this work so today's new plan is to let my friend in China do it for me, it is not only the exactness of the sawing or the exactness of the drilling it is also the kind of complicated way of fitting nice texts on the panel (I do however not have a competent program for that, it strikes me that I need to have that so my Chinese friend can make this work, or may MS Paint suffice? :) )

Actually, my requirements are not that high because I always aim at "Good Enough" but it is a sure thing that my Chinese friend (who has made both toroidal transformers of my design and PCBs of my design flawlessly for me) will do it better, if he can that is (but if he can fix as widely different type of electronics as transformers and PCBs I actually think he can, so I will contact him tonight and ask, to go easy on the guy I will begin to ask for a simple drilling of a tube amp copper chassis).

Best regards, Roger
DSCN3943.JPG
DSCN3944.JPG
DSCN3949.JPG
DSCN3950.JPG
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
I have decided to paint my KTT in screaming red.

A few days ago I bought a white foundation paint, now I will paint all three of my box projects including a sidetracked project which involves one of my tube amp boxes but this time for food-accessories only.

Please excuse my bad english :)

It is kind of hard to explain when I suck at detailed english but maybe this will clearify stuff, the tube amp box (KPA) will be tested with regard to looks by painting the exact same type of box but this time for kitchen knives and such.

The good thing with this is that if I don't like the result, I still have a kitchen-box to use.

So it's win-win situation :)

On the other hand, I am talking to my Chinese friend and he is going to make a true copper KPA chassis for me and maybe he can make a panel for my KTT also, the future will tell.

Best regards, Roger
 

(*steve*)

¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd
Moderator
Jan 21, 2010
25,510
Joined
Jan 21, 2010
Messages
25,510
We look forward to pictures of your screaming red transistor tester (SRTT?)
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
Hi Steve!

I thank you for your encouraging words and I am attaching two pictures that shows how my foundation-painted kitchen-box is turning out, my other box-projects that is KTT, KAP and KGA will get to look the same.

Tonight I will foundation-paint the bottom side of my KTT.

Due to lack of space in my small but cozy apartment I will have to do all the other tree box-projects bottom side plus top side.

Expected duration before I am finished is thus 6 days.

Best regards, Roger
PS
I spilled some paint on my trousers yesterday, for a moment there I kind of panicked BUT most paint today is water-based so I just used some water and soap and today there is absolutely no stain left :)

Could you please deroutate the picture?

DSCN3980.JPG
DSCN3981.JPG
 

rogerk8

Jul 28, 2011
179
Joined
Jul 28, 2011
Messages
179
Attaching my panel plan :)

I am very unsure about these measures and how to make it work.

The plan with this drawing is to determine the position of both all holes and all text, I have tested all positions using mm-paper and they seem to fit.

I have used text out of selfglueing OH-film before but the result is not so good.

Now I want to have serious text spray painted via a schablon/mask but I have difficulties specifying where the text should be located on the panel.

My plan is to specify in one single drawing the positions of all mechanical parts and their hole sizes as well as all text and then perhaps give it to someone competent to produce it for me, bacause I know that if I do it it will never be that good, the problem might however be the price.

Best regards, Roger

KTT_Layout_final_outsource_baraT&H_2.PNG
 
Top