Hi guys, sorry but I don't think so. These transistors are produced on wafers that will hold 15000-20000 on each wafer. And each wafer will have a pretty similar current gain.
Yes, the bell curve still exists but for a population of 2-3 wafer lots, which would be 1.4M parts.
So, chances are all the 200 transistors in one bag do come from the same wafer, hence the distribution is much tighter. Unfortunately thats what we have to deal with here.
"Searching" the transistors is much more expensive than just buying more (if you need binning, that is). In industrial applications they just use the appropriate grade and design for the worst case; the manufacturer of the transistors then just has to do an end-of-production test to guarantee that the transistors are better than the spec, thats it.
so its highly unlikely that these are rejects from somebody else but are genuine. They just happen to be quite variable in parameters. As we have never seen a similar thing with our beloved JFETs
Yes, the bell curve still exists but for a population of 2-3 wafer lots, which would be 1.4M parts.
So, chances are all the 200 transistors in one bag do come from the same wafer, hence the distribution is much tighter. Unfortunately thats what we have to deal with here.
"Searching" the transistors is much more expensive than just buying more (if you need binning, that is). In industrial applications they just use the appropriate grade and design for the worst case; the manufacturer of the transistors then just has to do an end-of-production test to guarantee that the transistors are better than the spec, thats it.
so its highly unlikely that these are rejects from somebody else but are genuine. They just happen to be quite variable in parameters. As we have never seen a similar thing with our beloved JFETs
I’m blown away by the accusations here.
I purchased my own parts as I was too late for the parts group buy. It took HOURS to research parts for correct value, size, type etc. Even then I screwed up on the trimmers which were side adjust instead of top adjust. I had to re-order trimmers at a cost of $25.
For the transistors I bought Fairchild in ammo pack. Almost all the NPN’s measured between 500-550 Hfe. Almost all the PNP’s measured between 350-400 Hfe. I had to re-order more transistors myself and measured 1250 transistors to get enough NPN and PNP to match in the 400-500 range. This is the reality of doing a project like this.
Those who got in on the parts group buy should consider themselves lucky even if you need to buy more transistors. The group buy still saved you hours of precious time.
I purchased my own parts as I was too late for the parts group buy. It took HOURS to research parts for correct value, size, type etc. Even then I screwed up on the trimmers which were side adjust instead of top adjust. I had to re-order trimmers at a cost of $25.
For the transistors I bought Fairchild in ammo pack. Almost all the NPN’s measured between 500-550 Hfe. Almost all the PNP’s measured between 350-400 Hfe. I had to re-order more transistors myself and measured 1250 transistors to get enough NPN and PNP to match in the 400-500 range. This is the reality of doing a project like this.
Those who got in on the parts group buy should consider themselves lucky even if you need to buy more transistors. The group buy still saved you hours of precious time.
THAT is extremely unfair. Someone handled a gigantically large group buy as a single person to his best knowledge and information available at the point of time where he worked on it. Looking at the large amount of questions which came up recently regarding matching, I wonder how he could have done things better?
MAYBE he could have given one or two quick updates on the Subject, but I can fully understand that you might loose interest in doing a lot of communication when you suddenly realize what you have done to yourself and are busy sorting out the mess...
And mistakes like the size of the caps, well thats human.
A little less stronger wording would have suited you well, sorry!
Mike
Hi Simon
Sorry to say I had same problem with 1000 of each and those where from Fairchild direct before I even got the boards for the Beta build
99% of the NPN are above 500 and 99% of the PNP are above 540
It is just the way the transistors are made
I have not jet opened the GB goodies bag butas soon as I have a chance I will mesure those up and send you anythink you need
By the way tanks forsending the LRC Capacitors I will mesurethem up and return what I don't need
LOOKING FOR VOLUNTEERS
Dear all, after a lot of thinking, trying and playing I think I might have a solution to the potential oscillation issue. Unfortunately, my own boards do not oscillate....
So I am looking for a volunteer that has built his / her Paradise R3 boards and knows they are oscillating, to ship them to me for testing. I will cover shipment cost, and promise to return them either unchanged or no more oscillating (hopefully!)
Please PM me in that case - I think I can take 1 or 2 but not 300, just in case..... But there shouldnt be a ton of issues anyway, after all that betabuilding....
Dear all, after a lot of thinking, trying and playing I think I might have a solution to the potential oscillation issue. Unfortunately, my own boards do not oscillate....
So I am looking for a volunteer that has built his / her Paradise R3 boards and knows they are oscillating, to ship them to me for testing. I will cover shipment cost, and promise to return them either unchanged or no more oscillating (hopefully!)
Please PM me in that case - I think I can take 1 or 2 but not 300, just in case..... But there shouldnt be a ton of issues anyway, after all that betabuilding....
Agree with you although frustration of not been able to make new toy for Crimbo is understandable.
I suggested few times that people post what they have in goodies bag and swap whit others
SO I would really like to have same of yours Hfe 350 to 400 PNP and Swap for my PNP that measure above 540 as my batch is opposite of yours.
Could you help please?
Another thing just in case is forgotten Joachim recommended 10 %match between transistors so a 500 Hfe can be considered matched to either a 450 or a 550 I did build the beta following this and it works fine so if you can do better great but no big deal
If you are not happy with this, build as you can and swap transistors later the boards can take a lot of punishment
Hi there,
recently ive made my day on studying some specsheets.
About 13 each 327/337 40
Almost every one told me that hfe should be between
250 and 625 with a statistic center @ 400.
So in terms of datasheets the Gauss Bellcurve has to appear
somewhere.
After measuring 1.600 of those creeps im going bonkers....
Aint got enough pairs for just one pair of boards.
Btw its not the fault of those honorable inventors of
Paradise !
Funny,
according to those sheets, Motorola seems not to offer
BC327-40 !
But, anyways, ill go further in it !
Just ordered 500/each more...
recently ive made my day on studying some specsheets.
About 13 each 327/337 40
Almost every one told me that hfe should be between
250 and 625 with a statistic center @ 400.
So in terms of datasheets the Gauss Bellcurve has to appear
somewhere.
After measuring 1.600 of those creeps im going bonkers....
Aint got enough pairs for just one pair of boards.
Btw its not the fault of those honorable inventors of
Paradise !
Funny,
according to those sheets, Motorola seems not to offer
BC327-40 !
But, anyways, ill go further in it !
Just ordered 500/each more...
Hi there,
recently ive made my day on studying some specsheets.
About 13 each 327/337 40
Almost every one told me that hfe should be between
250 and 625 with a statistic center @ 400.
So in terms of datasheets the Gauss Bellcurve has to appear
somewhere.
After measuring 1.600 of those creeps im going bonkers....
Aint got enough pairs for just one pair of boards.
Btw its not the fault of those honorable inventors of
Paradise !
Funny,
according to those sheets, Motorola seems not to offer
BC327-40 !
But, anyways, ill go further in it !
Just ordered 500/each more...
Unless one get different wafers about (20K transistors) Hfe will be around the same
So one need try different supliers and different brands
Any one tried ON semy ?
Any Idea how much for big order from manufacturer asking with close Hfe matching.
I purchased Fairchild npn and pnp devices from Mouser, as well as ON Semi npn and pnp devices from Digikey. The Fairchild npns and pnps don't match together, nor do the On Semi npns and pnps.
But ... it turns out that there are some good matches between the Fairchild pnps and the ON Semi npns, as well as between the Fairchild npns and the ON Semi pnps. So trying different sources, date codes, or manufacturers seems to pay off. I already posted measurements of about 750 devices, and will post in coming weeks measurements of another 750.
The distribution of HFE values I've measured (within the Fairchild population and the ON Semi population) looks like a bimodal. It does not appear to be normal/standard.
Matching and trading is part of the DIY experience IMO.
Pierre
Looks like i was pretty lucky with matching, but i am still a bit confused on what to match to what as i don't get to perfect matching of 8 pnp's to 8 npn's.
Let's first look at the input stage. Say, we have the pcb's, board A and B. each pcb has one pnp bank (4 transistors) and one npn bank (4 transistors). Let's call those bank 1 and 2. So we have 4 banks: A1,A2,B1,B2.
In what priority will we be matching?
- first within each bank?
- secondly between bank 1 and 2 of each pcb?
- thirdly between bank A1 and B1, and between A2 and B2?
Alternatively????
- first between bank 1 and 2 of a single pcb? (this could mean for example bank 1 (npn) could look like 450/475/450/475 and bank 2 (pnp) like 450/475/450/475
- second between pcb's?
- third within a single bank?
Essentially the question is : what are the effects of mismatches, and what is worse
- mismatches within one bank (pnp bank or npn bank)?
- mismatches between the pnp bank and the npn bank?
- mismatches between channels?
Talking servos and regulators, i imagine we are only looking at matches between single pairs of npn and pnp, as these need to be adjusted using trimpots.
Where else is matching crucial?
Let's first look at the input stage. Say, we have the pcb's, board A and B. each pcb has one pnp bank (4 transistors) and one npn bank (4 transistors). Let's call those bank 1 and 2. So we have 4 banks: A1,A2,B1,B2.
In what priority will we be matching?
- first within each bank?
- secondly between bank 1 and 2 of each pcb?
- thirdly between bank A1 and B1, and between A2 and B2?
Alternatively????
- first between bank 1 and 2 of a single pcb? (this could mean for example bank 1 (npn) could look like 450/475/450/475 and bank 2 (pnp) like 450/475/450/475
- second between pcb's?
- third within a single bank?
Essentially the question is : what are the effects of mismatches, and what is worse
- mismatches within one bank (pnp bank or npn bank)?
- mismatches between the pnp bank and the npn bank?
- mismatches between channels?
Talking servos and regulators, i imagine we are only looking at matches between single pairs of npn and pnp, as these need to be adjusted using trimpots.
Where else is matching crucial?
Mine are not accusations, they are questions.
I only want to build this once and I want it to be right..the best I can make it.
So I triple check the points that I am not sure about and the matching ( or complete absence of matching ) of the transistors is a critical point in my build and my build is stopped at this point.
I would like the designers to write down in simple terms what values I should aim for and what should match what.
None of this is a complaint or implies any criticism .
I am very grateful for what everyone involved has done so far.