can some body point to me any schematic or circuit that will help me match a pair of 2N5401 or BC556 for an ltp stage .....???
unfortunatelly i own a fluke that features a hfe meter but its damaged beyond repair ....
any other options ????
unfortunatelly i own a fluke that features a hfe meter but its damaged beyond repair ....
any other options ????
http://www.diyaudio.com/forums/showthread.php?postid=1245546#post1245546
http://www.diyaudio.com/forums/attachment.php?s=&postid=1250046&stamp=1183734588
I find the near matching by using one of the LTP as a fixed reference.
Then compare a batch to the reference, noting the Ic of the DUT (V+ to Tp1b) and noting the base current (common to Tp2b).
label each DUT with these two readings and pierce them into an A4sheet.
Now find sets that are very close for both these readings. Place a selected pair in the jig and note the difference in collector voltage (Tp1a to Tp1b). and note difference in base voltage (Tp2a to Tp2b). If both these differences are zero then you have an exact match for both hFE and Vbe at that current. Now see how far you can change the current both up and down and for the differences to remain near zero. A 2k0 pot across the 120r header pins allows you to swing the current and see the deltaIc and deltaIb change or stay almost fixed.
If both DUTs face each other and are clipped together and pass the same Ic then they must both be at the same temperature. Temperature is the biggest problem for us amateurs to overcome!
To check a matched pair more precisely, a 4pin header or dpst switch to short out the two 10k base resistors allows Ic to be compared when Vbe is exactly equal.
Good organisation and a bit of patience and you will find many nearly matched pairs and a few super matched pairs. If you're very lucky you might find an ultra matched pair. Keep it for something really special.
http://www.diyaudio.com/forums/attachment.php?s=&postid=1250046&stamp=1183734588
I find the near matching by using one of the LTP as a fixed reference.
Then compare a batch to the reference, noting the Ic of the DUT (V+ to Tp1b) and noting the base current (common to Tp2b).
label each DUT with these two readings and pierce them into an A4sheet.
Now find sets that are very close for both these readings. Place a selected pair in the jig and note the difference in collector voltage (Tp1a to Tp1b). and note difference in base voltage (Tp2a to Tp2b). If both these differences are zero then you have an exact match for both hFE and Vbe at that current. Now see how far you can change the current both up and down and for the differences to remain near zero. A 2k0 pot across the 120r header pins allows you to swing the current and see the deltaIc and deltaIb change or stay almost fixed.
If both DUTs face each other and are clipped together and pass the same Ic then they must both be at the same temperature. Temperature is the biggest problem for us amateurs to overcome!
To check a matched pair more precisely, a 4pin header or dpst switch to short out the two 10k base resistors allows Ic to be compared when Vbe is exactly equal.
Good organisation and a bit of patience and you will find many nearly matched pairs and a few super matched pairs. If you're very lucky you might find an ultra matched pair. Keep it for something really special.
very nice
and actually very sophisticated andrew !!!!!
thank you very much ( i will also check the links )
thanks
and actually very sophisticated andrew !!!!!
thank you very much ( i will also check the links )
thanks
Comparison is a very powerful tool.
We do not have the resources to do absolute measuring.
Think about what we do with a measuring bridge. It's comparison again.
We do not have the resources to do absolute measuring.
Think about what we do with a measuring bridge. It's comparison again.
check this out .....
since and as you said this is a painfull job to do i thought about starting with something very simple like a chinese multimeter that features hfe function ....
i was actually very surprized with what i found in my BC 557 storage for example .....
normally i buy these from a good company like sgs or philips and the all thing is like a big mess
as about BC 559 hfe in the same make and family vary from 120 to 400 ( ????? ) amazing ....
the average though between my BC 559 is arround 300 ....do you think that figures of 400 are realistic ????
AndrewT said:
since and as you said this is a painfull job to do i thought about starting with something very simple like a chinese multimeter that features hfe function ....
i was actually very surprized with what i found in my BC 557 storage for example .....
normally i buy these from a good company like sgs or philips and the all thing is like a big mess
as about BC 559 hfe in the same make and family vary from 120 to 400 ( ????? ) amazing ....
the average though between my BC 559 is arround 300 ....do you think that figures of 400 are realistic ????
the BC range go from 110 to 800 hFE @ Vce=5V and Ic=1mA and Ta=25degC.
If you use a multimeter you have no idea what the test conditions are and worse they will be different for every transistor because most use near constant Ib and monitor Ic. this means each transistor runs at different Ic if hFE varies and thus run at a different temperature and when the current varies one would probably find that the Vce has changed as well. Forget using the DMM transistor function for anything other than an indication, mostly it will be wrong!!!
If you use a multimeter you have no idea what the test conditions are and worse they will be different for every transistor because most use near constant Ib and monitor Ic. this means each transistor runs at different Ic if hFE varies and thus run at a different temperature and when the current varies one would probably find that the Vce has changed as well. Forget using the DMM transistor function for anything other than an indication, mostly it will be wrong!!!
Dear Sakis,
Andrew is very right in what he is saying. You can try this with measuring Hfe on the meter, if you hold the transistor between your fingers, just the heat will change Hfe. You will find the same transistor measures different depending on the temperature.
Now the bridge method that Andrew describes, I think is a good way because the two transistors are compared under identical ambient conditions.
I tried playing a little with the idea yesterday and believe me it is a very tedious and boring exercise.
I am skeptical that there is any audible difference doing this, it would mean measuring every component that you use.
Nico
Andrew is very right in what he is saying. You can try this with measuring Hfe on the meter, if you hold the transistor between your fingers, just the heat will change Hfe. You will find the same transistor measures different depending on the temperature.
Now the bridge method that Andrew describes, I think is a good way because the two transistors are compared under identical ambient conditions.
I tried playing a little with the idea yesterday and believe me it is a very tedious and boring exercise.
I am skeptical that there is any audible difference doing this, it would mean measuring every component that you use.
Nico
hello.
if i want to do it quickly,i use the hfe-test on the dmm.is better than nothing..............and it always worked well.
regards..............
if i want to do it quickly,i use the hfe-test on the dmm.is better than nothing..............and it always worked well.
regards..............
sakis said:
Personally I dont think matching the LTP transistors makes any discernible difference.
You are more likely to find problems from delays around the loop than in the LTP stage.
well i am sorry to say
for me to understand matching and the bennefits of it has to be done first to a very simple circuit like a P3 or P3A to see what is going on .....
previously on a P3 there was siemens bc 559 never troubled to match them then removed them and checked first the basic testing with a multimeter and found that one had hfe of 400 and the other 800 ( ???? ) at this point also amplifier presented about 75 mv offset that might be inside specs as rod says up to 100 mv is ok for this amp .....
then replaced with "multimeter matched pair" of bc 559 from philips and offset droped to almost 20 mv .......
as about listening impresions i thng that the matched pair presented a bit less bass but for sure much more tight ..... then on the high area couldn't hear any serious diference but there was a thingy there telling that high was cleaner without too much crisp ....
also in my P3 boards ltp transistors are in thermal junction
for me to understand matching and the bennefits of it has to be done first to a very simple circuit like a P3 or P3A to see what is going on .....
previously on a P3 there was siemens bc 559 never troubled to match them then removed them and checked first the basic testing with a multimeter and found that one had hfe of 400 and the other 800 ( ???? ) at this point also amplifier presented about 75 mv offset that might be inside specs as rod says up to 100 mv is ok for this amp .....
then replaced with "multimeter matched pair" of bc 559 from philips and offset droped to almost 20 mv .......
as about listening impresions i thng that the matched pair presented a bit less bass but for sure much more tight ..... then on the high area couldn't hear any serious diference but there was a thingy there telling that high was cleaner without too much crisp ....
also in my P3 boards ltp transistors are in thermal junction
hello.
now you can see the multimeter-hfe-test can help.20mv offset is well done.and your listeningtest too (i think there are some designer that use to missmatch the transistors for more volume in the bass,e.g. musicamps or so..............)
best regards..................
now you can see the multimeter-hfe-test can help.20mv offset is well done.and your listeningtest too (i think there are some designer that use to missmatch the transistors for more volume in the bass,e.g. musicamps or so..............)
best regards..................
I have two identical amps, build about 14 weeks apart. These amps use complimentary symmetry topology throughout the design. They are high-end in every respect. They use dual regulated +-50V/20A high standard 300 kHz SMPS that our company makes for telecommunication applications.
When I purchased the transistors I got 2000 of each small signal type MPSA06/56, 500 each BD140-16/139-16 from Philips and 100 each MJ15003/4 from On-Semi.
However on a quick inspection I found Hfe mismatch between NPN and PNP of about 200 - 400 and between the same polarity Hfe of 100 - 200.
Obviously I was very disappointing. The power transistors from On-Semi came in matched pairs as I ordered and the Hfe was spot on, but cost me about 40% more than if I ordered a batch of each.
Not knowing at the time that I would do a second amp, the original amp I painstakingly matched every one of the 26 transistors with its opposite and complimentary mate to within an Hfe of 1.
I made a PCB with transistor sockets and constant current sources using voltage references and a clamp to thermally couple the devices under test in my air conditioned office that is maintained at 18 deg centigrade.
To find the transistors and the matching criteria I strive for took me five evenings after work and was incredibly boring.
I was very proud of the outcome, all the transistors I matched the Hfe exactly.
I then completed the amp and at work I measured it to design/simulated specifications and found that it measured perfectly.
I andseveral other of my musician friends auditioned the amp during beer drinking and chatting over the next 8 or so weeks and I was very happy with my system.
On night surfing on e-Bay I found two pairs of KEF B139 (old style German built) that I have been searching for for over 25 years. I eventually outbid the competition and bought the four drivers from two different auctions for an average price of about US$ 350 each.
I then proceeded having transmission line cabinets made that I designed years ago using two B139s in isobaric configuration and two KEF SP1024 (what was used in the Coda), and two Scanspeak 2008 tweeters. I then decided to build a matching stereo amplifier to drive the B139s at a cross over frequency of 70 Hz.
My 51st birthday was coming up two weeks later and my wife planned a gathering of my musician friends and I had to complete the second amplifier within about ten days from nothing, no metal work and no PCBs, but I had all the components.
I decided to build the amp not matching a single semiconductor and thought that I could always do this later as my amps are made in such a way that servicing is a walk in the park.
The interesting part of this exercise, the second amp did not measure any worse than the first one nor was there any difference in sound - no even in the finest detail, they were identical.
The matched amp had an off-set (I do not have off-set adjustment) of +1.3 mV while the unmatched amp offset was -2.7 mV. (42 deg stable on the heat sink after one hours)
In my opinion matching did not make any difference in the perceived sound, since many musically trained ears have listened over the years without any conclusion and I never bothered to match the second amp.
The semiconductors in my pre/line amp at perfectly matched, but my reasoning was if there is any imperfection in the pre-amp, it will be pumping out of the speakers at +30 dB.
The pre-amp measured offset of only a few micro volt, distortion products one could hardly see above the noise floor off the HP 8510 network analyser and phase deviation of -0.3 degrees from DC to 100 kHz. Whether the matching was partly responsible for the results, I am not totally sure of.
What I can say is this, and probably the most important characteristic of an amplifier, is its phase response, this parameter alters the characteristic of the sound that you hear.
You can have zero distortion using pure single frequencies, and a bandwidth from dc to daylight, slew rates of kilo volts/uS are fine, but if the amplifier phase response varies more than 2 - 3 degrees over the audio band, it adds it own recognizable character to the sound and believe me this is not subjective.
This is what makes two well designed low distortion amps sound different from another. If you know your instruments or the recording well then you will recognize the inadequacy of one perfect amp over another.
This, however is not that revealing with loudspeakers in a typical listening environment since there are many time related interfering boundaries, but with a good set of headphones it becomes very noticeable indeed.
Anyway I am drifting off the topic Sakis, so I will refrain from further comment.
Kindest regards Sakis and may you have an exceptional year ahead.
Nico
When I purchased the transistors I got 2000 of each small signal type MPSA06/56, 500 each BD140-16/139-16 from Philips and 100 each MJ15003/4 from On-Semi.
However on a quick inspection I found Hfe mismatch between NPN and PNP of about 200 - 400 and between the same polarity Hfe of 100 - 200.
Obviously I was very disappointing. The power transistors from On-Semi came in matched pairs as I ordered and the Hfe was spot on, but cost me about 40% more than if I ordered a batch of each.
Not knowing at the time that I would do a second amp, the original amp I painstakingly matched every one of the 26 transistors with its opposite and complimentary mate to within an Hfe of 1.
I made a PCB with transistor sockets and constant current sources using voltage references and a clamp to thermally couple the devices under test in my air conditioned office that is maintained at 18 deg centigrade.
To find the transistors and the matching criteria I strive for took me five evenings after work and was incredibly boring.
I was very proud of the outcome, all the transistors I matched the Hfe exactly.
I then completed the amp and at work I measured it to design/simulated specifications and found that it measured perfectly.
I andseveral other of my musician friends auditioned the amp during beer drinking and chatting over the next 8 or so weeks and I was very happy with my system.
On night surfing on e-Bay I found two pairs of KEF B139 (old style German built) that I have been searching for for over 25 years. I eventually outbid the competition and bought the four drivers from two different auctions for an average price of about US$ 350 each.
I then proceeded having transmission line cabinets made that I designed years ago using two B139s in isobaric configuration and two KEF SP1024 (what was used in the Coda), and two Scanspeak 2008 tweeters. I then decided to build a matching stereo amplifier to drive the B139s at a cross over frequency of 70 Hz.
My 51st birthday was coming up two weeks later and my wife planned a gathering of my musician friends and I had to complete the second amplifier within about ten days from nothing, no metal work and no PCBs, but I had all the components.
I decided to build the amp not matching a single semiconductor and thought that I could always do this later as my amps are made in such a way that servicing is a walk in the park.
The interesting part of this exercise, the second amp did not measure any worse than the first one nor was there any difference in sound - no even in the finest detail, they were identical.
The matched amp had an off-set (I do not have off-set adjustment) of +1.3 mV while the unmatched amp offset was -2.7 mV. (42 deg stable on the heat sink after one hours)
In my opinion matching did not make any difference in the perceived sound, since many musically trained ears have listened over the years without any conclusion and I never bothered to match the second amp.
The semiconductors in my pre/line amp at perfectly matched, but my reasoning was if there is any imperfection in the pre-amp, it will be pumping out of the speakers at +30 dB.
The pre-amp measured offset of only a few micro volt, distortion products one could hardly see above the noise floor off the HP 8510 network analyser and phase deviation of -0.3 degrees from DC to 100 kHz. Whether the matching was partly responsible for the results, I am not totally sure of.
What I can say is this, and probably the most important characteristic of an amplifier, is its phase response, this parameter alters the characteristic of the sound that you hear.
You can have zero distortion using pure single frequencies, and a bandwidth from dc to daylight, slew rates of kilo volts/uS are fine, but if the amplifier phase response varies more than 2 - 3 degrees over the audio band, it adds it own recognizable character to the sound and believe me this is not subjective.
This is what makes two well designed low distortion amps sound different from another. If you know your instruments or the recording well then you will recognize the inadequacy of one perfect amp over another.
This, however is not that revealing with loudspeakers in a typical listening environment since there are many time related interfering boundaries, but with a good set of headphones it becomes very noticeable indeed.
Anyway I am drifting off the topic Sakis, so I will refrain from further comment.
Kindest regards Sakis and may you have an exceptional year ahead.
Nico
Attachments
thanks for join in nico
i can tell you only this
my approach about matching ( i also have a thingy about ltp thermal junction ) probably i have more than one or two or three thingyes ha ha ha thats my personal vice
it is possible that you noticed these results in a sophisticated amp including CCS or curent mirrors and so on
i thing that in circuit like the one mentioned above a bootstrap CFP , there is a grate possibility that matching is very very importand
to my ears made obvious diference since i listen to this amp for years now
i think it made amp more detailed more sharp and precise
( i will dare to say that this actually killed the sonics of the low end a bit but finally i thing it sounds more correct than coloured )
i can tell you only this
my approach about matching ( i also have a thingy about ltp thermal junction ) probably i have more than one or two or three thingyes ha ha ha thats my personal vice
it is possible that you noticed these results in a sophisticated amp including CCS or curent mirrors and so on
i thing that in circuit like the one mentioned above a bootstrap CFP , there is a grate possibility that matching is very very importand
to my ears made obvious diference since i listen to this amp for years now
i think it made amp more detailed more sharp and precise
( i will dare to say that this actually killed the sonics of the low end a bit but finally i thing it sounds more correct than coloured )
hello.
perhaps it is a question of low feedback or not.........can higher feedback in an amp compensate missmatching.........?
greetings...........
perhaps it is a question of low feedback or not.........can higher feedback in an amp compensate missmatching.........?
greetings...........
mjf said:
Yes most definitely. I design my amps all with high OLG and use a lot of negative feed-back, mainly because I want little phase distortion, that is how like to design my amps.
However, I have made several that has low OLG and very little NFB and to this end careful matching will make a great deal of difference as you have no NFB to fix things. I guess if NFB is less that 15 dB, matching would be a prerequisite.
There is a definite difference in sound between amps with low OLG and little NFB. I will not say which I think sound better, but I have lived with both types.
Nico
- Status
- Not open for further replies.