When the chinese can cheat, they cheat. Yes good parts and manufacturing can be had at a price, and when you keep a QC thumb on them and select the best manufacturers, which means you need to buy from a reputable company even then, if they QC their products and their suppliers.
Remember the dog melamine food deaths from a reputable US manufacturer! ? Just one of a litany of many.
Watch this
https://www.youtube.com/watch?v=tVXubFu7GZM
So yes I agree with you 100%, you pay for provanance! But where do you go when a reputable firm no longer sells that part and its only available in surplus bins? A hard task thats easy to prey upon those in need.
Not only the eye watering prices, each line item now must be at least $100, and the total order is now $250, not DIYers friendly at all .....
So cheap? Rochester sell to military and aerospace suppliers who won't choke at paying 1000s of dollars for some old part, because its the only way to keep the equipment qualification valid. The end user has to accept the large bill. This is the 250USD hammer market
A while ago I bought two pairs of 2N5684 and 2N5686 from ebay, a chinese seller at a too good to be true price. When I got the parts I immediately noticed that the PNP transistor had much thinner leads and was on the skinny side overall compared to the NPN which had thick nearly unbendable pins and nice and crisp lettering.
Fast forward a year and I finally used those devices and PNP went pop pretty soon after. I ground the top off (along with some other transistors I that had failed in past) to see what was inside and here we go :
Left one is legit Russian KT818GM (100W) and right is supposed 2N5684 (300W) :
Close up of the fake :
And the KT818GM for comparison :
That fake part's die is less than quarter size of the 100W part...
I kinda hoped to see two tiny dies in parallel, I would have got a nice laugh out of it then at least hahaha
I'll have to replace the PNP in the other amp channel too, and I'll grind open one of the NPNs that may be fake...
Fast forward a year and I finally used those devices and PNP went pop pretty soon after. I ground the top off (along with some other transistors I that had failed in past) to see what was inside and here we go :
Left one is legit Russian KT818GM (100W) and right is supposed 2N5684 (300W) :
Close up of the fake :
And the KT818GM for comparison :
That fake part's die is less than quarter size of the 100W part...
I kinda hoped to see two tiny dies in parallel, I would have got a nice laugh out of it then at least hahaha
I'll have to replace the PNP in the other amp channel too, and I'll grind open one of the NPNs that may be fake...
Hi There
I have been busy with X on a circlotron, however the amp do have a uge temp run away on the IRFP250N mosfets, the strange thing is it run out in just a few seconds to a idle current of some amps and blowing itselfs, like what happens with X
In my opinion, a mosfet temp runaway who go so fast the tracking transistor did not track it means the silicon get hot very fast before the mosfet housing do>?.
I do cool the amp with a fan, it stays quite stable on 500 mA but when I do take the fan away it do drop to 120 mA who do prove the transistor do work and is overcompesated, but it do not track on time, the irfp heats up so quickly inside that it is already in the amps region before the outside of transistopr body gets hot, very strange, It looks like a igbt I did have in past who has the same behavior and was not a solution for, the elektor amp.
Here pics of the mosfets, maybe she are not genuin but then very good made.
thanks.
kees
I have been busy with X on a circlotron, however the amp do have a uge temp run away on the IRFP250N mosfets, the strange thing is it run out in just a few seconds to a idle current of some amps and blowing itselfs, like what happens with X
In my opinion, a mosfet temp runaway who go so fast the tracking transistor did not track it means the silicon get hot very fast before the mosfet housing do>?.
I do cool the amp with a fan, it stays quite stable on 500 mA but when I do take the fan away it do drop to 120 mA who do prove the transistor do work and is overcompesated, but it do not track on time, the irfp heats up so quickly inside that it is already in the amps region before the outside of transistopr body gets hot, very strange, It looks like a igbt I did have in past who has the same behavior and was not a solution for, the elektor amp.
Here pics of the mosfets, maybe she are not genuin but then very good made.
thanks.
kees
Attachments
the case and heatsink heat up slowly as it has a large thermal mass. I saw large differences between plastic top temperature and case temp. so maybe better to glue the temp sensor directly on the inscription side of the fet. the sensor should have a low themal mass not a large metaltab type transistor.
I have a bd139 on the bold, glue it is maybe not such good idea because when it looses amp blow up.
Then there is thermal resistance between the case and the silicon , for 30 amps mosfets these need not to heat up that quickly causing very fast thermal runaway.
I however did als discover the J310 casoded current source in lpt do warm up giving a run when cold. so I need some work to do.
thanks for advice.
the BD139 "op het boutje" makes it slow. a smaller transistor with true thermal toothpaste on the marking makes it much quicker. have to adjust for VBE though. I tested an audio amp with lateral mosfets (they luckily have no runaway) but the temperature rise of the plastic top was almost immediate, mainly driven by the poor junction to caste resistance of 1.25K/W. I have an optris lasersight thermal probe that has a spot size of 1mm, that helps a lot. you can find the die with that thermally..
Hmm op het boutje, hier een nederlander die lekker in spanje vertoeft, ik was zelf van plan dat te gaan doen, heb kontakt met een spaanse
heb namelijk artrose en met warmte geen of weinig pijn..
Oke, the boutje has also contakt with the heatsink so it draw temperature away and get a slow response.
I have also discover that the cascoded current source with j310 jfets on 10 mA do cause a uge drift, did lower the dissipation by lower voltage and it is much more stable, so I need a temperature stable one or jfet who do lower current when get warm, not I get more and the circlotron do run, yesterday I can adjust 500 mA and now I did get 850 mA without much run after lower voltage on ccs to 8.2 volts.
And now back on topic because I am busy here with my design. http://www.diyaudio.com/forums/solid-state/291767-allfet-circlotron-new-post.html
regards
heb namelijk artrose en met warmte geen of weinig pijn..
Oke, the boutje has also contakt with the heatsink so it draw temperature away and get a slow response.
I have also discover that the cascoded current source with j310 jfets on 10 mA do cause a uge drift, did lower the dissipation by lower voltage and it is much more stable, so I need a temperature stable one or jfet who do lower current when get warm, not I get more and the circlotron do run, yesterday I can adjust 500 mA and now I did get 850 mA without much run after lower voltage on ccs to 8.2 volts.
And now back on topic because I am busy here with my design. http://www.diyaudio.com/forums/solid-state/291767-allfet-circlotron-new-post.html
regards
Try the Caribbean with 25C-28C night and 28C-32C day all year round.
3 years and counting till my name pops up as permanent resident at Kranshi in Willemstad.
(kontakt met Spaanse ? Boutje ? I don't want to know)
I have removed the irfp250N transistors and put in irfp240N transistors, now I have not a problem with my design, it is quite stable on 1.21 amp class A current as X want, I go lower.
When start cold it has 720 mA and steps up in 50 seconds to 1.20 amps and stay there happenly.
These transistors are or fake irfp250N or these are not happy in a amp, I think it are fakes, I think it are or faulty ones or igbt,s who has the same behavior as I have experienced in the past.
The silicon can be bad set on the case so cooling it down do not work well causing it to heat up fast and so it go jump high in seconds.
What I did experineced with the irfp250 was that it just sometimes jumps from 420 mA to max current of supply and was 1,3 amps, in just a second, then it jumps back again etc.
regards
When start cold it has 720 mA and steps up in 50 seconds to 1.20 amps and stay there happenly.
These transistors are or fake irfp250N or these are not happy in a amp, I think it are fakes, I think it are or faulty ones or igbt,s who has the same behavior as I have experienced in the past.
The silicon can be bad set on the case so cooling it down do not work well causing it to heat up fast and so it go jump high in seconds.
What I did experineced with the irfp250 was that it just sometimes jumps from 420 mA to max current of supply and was 1,3 amps, in just a second, then it jumps back again etc.
regards
Attachments
Boutje is a bold, and spanisch is far away and I am 60 over 2 months.
So See what I go to do.
He he the amp has a fake mosfet or reprinted igbt, sadly, I have not originals and these do fine.
Various makes of the same part and even newer design generations from the same make can behave very differently in threshold behaviour, especially temperature coefficient.
Current suddenly jumping could be due to RF oscillation.
Don't assume all strange behaviour is due to fakes
Hi davidsrsb,
Right you are. When I see current jump non-destructively, RF oscillation is the first thing that comes to mind. The very next thought is to kill the power.
That is why it`s so important to use known good parts when you are creating something that you can`t say will work for certain. Unless you know your parts are good, any problem you find could easily be either a bad part(s), or a design that needs tweaking. You can build a kit, but unless you have everything identical to the working model, you are creating a new device. That means the same PCB, chassis, heat sink -everything-, you don`t know otherwise. Building from a schematic only means you are definitely creating something for the very first time.
The one huge factor is, lead dress. That`s how you route every wire in that design of yours. Even rebuilding a commercial product that works will turn into a monster you can`t tame if you don`t place each component and wire where they were before you attacked it. They may not, but assume they will.
-Chris
Right you are. When I see current jump non-destructively, RF oscillation is the first thing that comes to mind. The very next thought is to kill the power.
That is why it`s so important to use known good parts when you are creating something that you can`t say will work for certain. Unless you know your parts are good, any problem you find could easily be either a bad part(s), or a design that needs tweaking. You can build a kit, but unless you have everything identical to the working model, you are creating a new device. That means the same PCB, chassis, heat sink -everything-, you don`t know otherwise. Building from a schematic only means you are definitely creating something for the very first time.
The one huge factor is, lead dress. That`s how you route every wire in that design of yours. Even rebuilding a commercial product that works will turn into a monster you can`t tame if you don`t place each component and wire where they were before you attacked it. They may not, but assume they will.
-Chris
I have now put in irfp240 these behave very good, when cold idle go up and get regulated after, and stays on 350 milliamps, however the tem sens resistor is overcompensate.
The irfp250 did go up so fast the temp transistor can not regulate on time, I have no oscillations, did test this on the scope. Maybe the 250 is not suitable for the amp, I did have such problem in past with a igbt amp, did not get stable with idle.
Here a square of 20 Khz, and have some overshoot but that is easly get rid of with compensations.
The irfp250 did go up so fast the temp transistor can not regulate on time, I have no oscillations, did test this on the scope. Maybe the 250 is not suitable for the amp, I did have such problem in past with a igbt amp, did not get stable with idle.
Here a square of 20 Khz, and have some overshoot but that is easly get rid of with compensations.
