Luck must be part of the succes in matching.
A number of NPNs devices can be from the same batch with low deviation, same goes for a batch of PNPs.
But the average parameters of the NPNs may differ much from the average of the PNPs.
With Jappon devices it's like trying to find a match between an O series NPN and a Y series PNP.
Transistor production will show a Bell curve as well.
Although matching small signal devices seems to be a lot easier, buying different batches of both types can make all the difference.
A real audio nut basket shouldn't mind spending 25 bucks on a suitcase full of TO92s that lasts the family for the next 5 to 10 generations.
I gathered that there are plenty on this club who have thousands of different variety little ones. I still have plenty of GE/RCA devices left which i'm not likely to use again.
Nowadays i'll buy plenty, look for good matches and then dump the rest.
(real problem is tripping over the carpet and mixing the lot by dropping the assorted component box
)
Last year i bought another 10k of 0.6 watt 2% metalfilms for Jack Sheet, spent several weeks sorting and measuring each and everyone in 5 different classes.
The story goes that Mark Levinson used to have his (many) girlfriends do a lot of matching in the old/early days for his Cello gear.
Swiss audio makers seem to miss the good old mechanical watch days, so they spend awfull hours matching components.
Unless someone is willing to cough up enough for high precision Rikens, Caddocks or isabellenhüte power resistors, i'm clueless where to buy 1% accurate emitter resistors ?
Only trick i know is to put plenty of 0.6/1W metalfilms in parallel, trademark of Birdy Pavel (and P-A).
A number of NPNs devices can be from the same batch with low deviation, same goes for a batch of PNPs.
But the average parameters of the NPNs may differ much from the average of the PNPs.
With Jappon devices it's like trying to find a match between an O series NPN and a Y series PNP.
Transistor production will show a Bell curve as well.
Although matching small signal devices seems to be a lot easier, buying different batches of both types can make all the difference.
A real audio nut basket shouldn't mind spending 25 bucks on a suitcase full of TO92s that lasts the family for the next 5 to 10 generations.
I gathered that there are plenty on this club who have thousands of different variety little ones. I still have plenty of GE/RCA devices left which i'm not likely to use again.
Nowadays i'll buy plenty, look for good matches and then dump the rest.
(real problem is tripping over the carpet and mixing the lot by dropping the assorted component box
)Last year i bought another 10k of 0.6 watt 2% metalfilms for Jack Sheet, spent several weeks sorting and measuring each and everyone in 5 different classes.
The story goes that Mark Levinson used to have his (many) girlfriends do a lot of matching in the old/early days for his Cello gear.
Swiss audio makers seem to miss the good old mechanical watch days, so they spend awfull hours matching components.
Unless someone is willing to cough up enough for high precision Rikens, Caddocks or isabellenhüte power resistors, i'm clueless where to buy 1% accurate emitter resistors ?
Only trick i know is to put plenty of 0.6/1W metalfilms in parallel, trademark of Birdy Pavel (and P-A).
I bought about 500 .47 ohm 5 watt 3% Dale resistors and have had no problem matching them to 1% for the KSA-50 Klone. I use a Racal 5-1/2 digit bench meter to do the job... Its good at reading low resistances accurately.
If I just had the time to learn to use the curve tracer.... anyone have any tips for me on using it just starting out?
Thanks,
Mark
If I just had the time to learn to use the curve tracer.... anyone have any tips for me on using it just starting out?
Thanks,
Mark
When i match resistors (very selden) i take a little psu about 5-6v and makes a voltage divider with 1 permanent resistor close to the wanted value, then i change the other resistor as long as ther are some in stock and measures the voltage across them.
First time i let the resistors get "hot".
Each of them gets numbers and i write the values into an excelsheet wich automatic calculates the difference in % to the permanent one.
And yes i know that it might change value as matching goes on but the if the value measured at the "loose" resistors are close that is good enough matching for me.
First time i let the resistors get "hot".
Each of them gets numbers and i write the values into an excelsheet wich automatic calculates the difference in % to the permanent one.
And yes i know that it might change value as matching goes on but the if the value measured at the "loose" resistors are close that is good enough matching for me.
Mark A. Gulbrandsen said:
It's not like:
-Reset.
-Switch NPN or PNP
-Choose peak voltage and load resistor, depending on the device and whether heatsinked or not.
-Choose base resistor and current step size for the function generator.
-Adjust horizontal div for the voltage range you picked
-Adjust vertical div for the max voltage/load resistor
-Press go ?
(practically the same like doing a static test on a piece of perf board ?)
So are we looking for the curves to be identical? How does one know what voltages and load resistors to select... Is the selection based on the operating points of the circuit the device will ultimately be used in? Or just some random settings for a particuluar device to confirm it's printed specs?
Thanks,
Mark
Thanks,
Mark
Most logical would be the operating range, no ?
Otherwise you'd have to be able to store a lot of curves.
I'm currently (evenings) matching the K170/J74 i got from James in Singapore. That tube Tek would be a bless.
Was thinking of hacking a Steber tracer together, with the option of feeding it into the PC:
www.arrl.org/qst/2006/07/steber.pdf
www.arrl.org/qex/2006/07/qx7steber.pdf
Good thing i have a DSO, even found the RS232 interface software on the web.
I've only been able to fool around with a Tek tracer a long time ago for matching some 550Cs, one of the later models (576?)
The electronics faculty had those expensive machines, maybe Loek has some Tek curve tracer mileage, he works there.
Otherwise you'd have to be able to store a lot of curves.
I'm currently (evenings) matching the K170/J74 i got from James in Singapore. That tube Tek would be a bless.
Was thinking of hacking a Steber tracer together, with the option of feeding it into the PC:
www.arrl.org/qst/2006/07/steber.pdf
www.arrl.org/qex/2006/07/qx7steber.pdf
Good thing i have a DSO, even found the RS232 interface software on the web.
I've only been able to fool around with a Tek tracer a long time ago for matching some 550Cs, one of the later models (576?)
The electronics faculty had those expensive machines, maybe Loek has some Tek curve tracer mileage, he works there.
What I might do is start a whole new thread on using the curve tracer... perhaps you guys that know how to use one might teach a few of us that don't a thing or two. I paid $30.00 plus $48.00 shipping for my TEK curve tracer off of E-Bay... MIssing graticule and bezel and three bad tubes... and clean all the switches is all it needed to get it functioning 100%... CRT has a small burn spot but it doesn't interfer at all. I've looked at some PNP and NPN devices on it but I'm not 100% sure what I'm looking at....
Mark
Mark
Mark A. Gulbrandsen said:
Wow !
is S.L.C Tek heaven or is that the bonus of being in a mobile business, Mark ?
I thought Tek curve tracers were good for several hundreds of dollars, US alone ?
(for stuff like that i might as well shack up with ZenoModo, or whatever he calls himself this week)
Harry3 said:Also FYI
I tried matching one hundred MPSA42 (NPN) to MPSA92 (PNP) from Onsemi and could not get a close match.
I don't think you ever will. Usually MPSA PNPs have far higher fe (in order of twice that of NPNs).
BUT ! You really only need to match transistors of the same type (in fact the only critical ones are those forming the differential pair) - there is no need to match upper and lower pairs. The only possible problem is that Tr's with higher gain will draw less base current (if everything else is the same) and you might get a slight DC offset at input. Either blocked by a capacitor or trimmed out by adjusting idle current of the input diff pairs.
There's the fun part of Jappin devices.
Japanese PNPs also have a higher average hFe than NPNs.
But they come in different hFe classes, and their hFE bell curves overlap.
A lower hFe class GR type batch of PNPs and a batch of BL type NPNs have a higher hFe matching rate than trying to find matching pairs from NPN and PNP devices of the same hFE class.
So, those who go for Japanese and who are not into hFe matching are better off buying same hFe class PNP/NPNs. The hFe matcher buys the highest hFe grade NPNs, and the lower hFE grade PNPs.
Japanese PNPs also have a higher average hFe than NPNs.
But they come in different hFe classes, and their hFE bell curves overlap.
A lower hFe class GR type batch of PNPs and a batch of BL type NPNs have a higher hFe matching rate than trying to find matching pairs from NPN and PNP devices of the same hFE class.
So, those who go for Japanese and who are not into hFe matching are better off buying same hFe class PNP/NPNs. The hFe matcher buys the highest hFe grade NPNs, and the lower hFE grade PNPs.
I have to agree with the NPN and PNP's not being close enough for matching to one another with the MPS and BC stuff. I didn't experience much better figures when I matched my clone's 2SA/2SC's though, and I bought a huge bag full. NPN-NPN and PNP-PNP was easy. But as jacco said, it's possible I didn't get some of the overlapped hFe classes.
However, for the clone at least matching of the NPN-PNP's isn't really important, at least not compared to matching of each other.
Mark: have you got any feedback on the boards yet?
However, for the clone at least matching of the NPN-PNP's isn't really important, at least not compared to matching of each other.
Mark: have you got any feedback on the boards yet?
