Yes you are right about the BJT in different Beta groups. Take a look at parts like BC847 in A,B,C gain version. Higher gain => Lower linearity or "va".
But the linearity or currentgain modulation at constant collector current seems to be linear. So if you design with this in mind it should be okay.
Also BC847 from philips,fairchild,onsemi have different Spice models. So yes they produce them in different ways.
The thing about 2SC types are only produced by one vendor.. Not necesaryle true. Onsemi and philips does produce 2SC1815 but they call them 2PC1815 (philips) instead.
But i agree with you Jocko. Also the thing about how even the ib/ic is spread.
A note on this : 2SC2240 has wider area where the spreading is even than most transistors out there.
😉
Sonny
But the linearity or currentgain modulation at constant collector current seems to be linear. So if you design with this in mind it should be okay.
Also BC847 from philips,fairchild,onsemi have different Spice models. So yes they produce them in different ways.
The thing about 2SC types are only produced by one vendor.. Not necesaryle true. Onsemi and philips does produce 2SC1815 but they call them 2PC1815 (philips) instead.
But i agree with you Jocko. Also the thing about how even the ib/ic is spread.
A note on this : 2SC2240 has wider area where the spreading is even than most transistors out there.
😉
Sonny
on Bob Pease and engineering realities
Since we have gotten off topic, why not just keep it going....
I employ a significant number of engineers. Some are good, some are bad. Good marks in a good university does not make them a good engineer, but I can honestly say, that the best of them, generally did have good marks, and as it turns out, a good head on their shoulders. Good marks are often an indicator of hard work and knowing what is important. I also employ some very good "journeymen" designers. without degrees. Some a very good as well. It all comes down to knowing what interview questions to ask.
Is spice important? Of course. Ever tried to prototype a quarter micron CMOS IC. No, I thought not. Funny how some of those new op-amps sound pretty good. Maybe not great, but you can build a billion of them for about 35 cents cost each!
Does the university matter. Yes, some seem to turn out better engineers. Yes, it is generally ones with a better pedigree. Those "we teach you real things" universities and colleges can advertise all they want, but the reality is it, they don't attract the brightest minds. Do I need to teach some of these bright minds what decoupling caps are.. yes, but I don't need to teach them early voltage and how to calculate zeros and poles for feedback which is a heck of a lot harder.
Oh, and Bob Pease... I was not impressed at his last seminar. He showed some circuits that many of my new grads could come up with and espoused some wisdom that most of us learn in the first few years on the job.....
Since we have gotten off topic, why not just keep it going....
I employ a significant number of engineers. Some are good, some are bad. Good marks in a good university does not make them a good engineer, but I can honestly say, that the best of them, generally did have good marks, and as it turns out, a good head on their shoulders. Good marks are often an indicator of hard work and knowing what is important. I also employ some very good "journeymen" designers. without degrees. Some a very good as well. It all comes down to knowing what interview questions to ask.
Is spice important? Of course. Ever tried to prototype a quarter micron CMOS IC. No, I thought not. Funny how some of those new op-amps sound pretty good. Maybe not great, but you can build a billion of them for about 35 cents cost each!
Does the university matter. Yes, some seem to turn out better engineers. Yes, it is generally ones with a better pedigree. Those "we teach you real things" universities and colleges can advertise all they want, but the reality is it, they don't attract the brightest minds. Do I need to teach some of these bright minds what decoupling caps are.. yes, but I don't need to teach them early voltage and how to calculate zeros and poles for feedback which is a heck of a lot harder.
Oh, and Bob Pease... I was not impressed at his last seminar. He showed some circuits that many of my new grads could come up with and espoused some wisdom that most of us learn in the first few years on the job.....
Sorry you weren't impressed, or amused by Bob Pease. Maybe he held back his presentation for those of you north of the border. [joke]
But if your young engineers are that smart, let me know where I can steal some of them when I get back to a real job again.
Just make sure they don't come from Nortel! Not impessed with anything they put out.
Jocko
But if your young engineers are that smart, let me know where I can steal some of them when I get back to a real job again.
Just make sure they don't come from Nortel! Not impessed with anything they put out.
Jocko
Shure we do lack a lot experience and wisdom Jocko Homo wich you or other engineers have who have been working for a lot of years in the field!
... But some times new eyes have a different view... I learn this all the time.
😉 Did you all have a nice weekend..
Sonny
... But some times new eyes have a different view... I learn this all the time.
😉 Did you all have a nice weekend..
Sonny
Didn't make any money, if that is what you mean. That's life over 50 as an engineer.
(For those of you much younger.........this sort of stuff happens to all engineers eventually. And it is not unique to our field. I remember lots of friends from high school whose fathers were laid off about 1 year from retirement when USSteel closed their research down. Just thought you might like to know what your future holds.)
Cheers,
Jocko
(For those of you much younger.........this sort of stuff happens to all engineers eventually. And it is not unique to our field. I remember lots of friends from high school whose fathers were laid off about 1 year from retirement when USSteel closed their research down. Just thought you might like to know what your future holds.)
Cheers,
Jocko
Va of 1000V?
Sonny,
is this a hypothetical part or is it a device you have found in a data sheet but don't want to share the name of?
Greetings,
Eric
Sonny,
is this a hypothetical part or is it a device you have found in a data sheet but don't want to share the name of?
Greetings,
Eric
Look for zetex parts. Go to their site and start searching. I found a part after searching through data sheet for weeks.
"va" is not a holy grale but a important parameter in a voltagegain stage. It is helping to maximise openloop gain.
Sonny
"va" is not a holy grale but a important parameter in a voltagegain stage. It is helping to maximise openloop gain.
Sonny
Allright after some thoughts...
I am going to publish a circuit on my site where i am going to use the pair from zetex anyway... So here they are.
npn : ztx458 ; va = 1050.
pnp : ztx558 ; va = 349.
Sonny
I am going to publish a circuit on my site where i am going to use the pair from zetex anyway... So here they are.
npn : ztx458 ; va = 1050.
pnp : ztx558 ; va = 349.
Sonny
V_a and mystery parts
Hi Sonny,
the early voltage V_a is defined as the extrapolated point in the I_C vs. V_CE graph where I_C becomes zero if you extrapolate the linear region. This crossover point will depend on how much current gain you had in the first place.
If I understand your argument correctly, you are more interested in the slope d I_C/ d V_CE because this is what determines linearity. This, measurend in mA/V, is something like the reverse steepness g_r.
Anyway, I looked at a couple of Zetex data sheets, which can range from 1 to 5 pages. In none of them, V_a, g_r or a graph of I_C (V_CE) was given. Where did you find this information?
And while we are dealing with mysteries, does anybody have any idea what the S1D and S2D parts in the End Millenium amp are? I tried a reverse search with Zetex because the part marking should be mentioned in the data sheet I am looking for, but to no avail so far.
As I don't believe in purely local feedback, there is no danger that I will try to copy the end millenium amp...
Greetings to the north!
Eric (could do with a couple of weeks of vaction on the island of M/on...)
Hi Sonny,
the early voltage V_a is defined as the extrapolated point in the I_C vs. V_CE graph where I_C becomes zero if you extrapolate the linear region. This crossover point will depend on how much current gain you had in the first place.
If I understand your argument correctly, you are more interested in the slope d I_C/ d V_CE because this is what determines linearity. This, measurend in mA/V, is something like the reverse steepness g_r.
Anyway, I looked at a couple of Zetex data sheets, which can range from 1 to 5 pages. In none of them, V_a, g_r or a graph of I_C (V_CE) was given. Where did you find this information?
And while we are dealing with mysteries, does anybody have any idea what the S1D and S2D parts in the End Millenium amp are? I tried a reverse search with Zetex because the part marking should be mentioned in the data sheet I am looking for, but to no avail so far.
As I don't believe in purely local feedback, there is no danger that I will try to copy the end millenium amp...
Greetings to the north!
Eric (could do with a couple of weeks of vaction on the island of M/on...)
You have to look in the spice models to find the values.
They show a lot more like base,collector and emitter resistance, size of parasitic cap between be and bc etc...
If the va or early voltage is low a small change in Vce will have a great impact on the currentgain.
Like : MJL3281A
at 40V it has a hfe of ~240
and at 10V it has a hfe ~120.
So a ripple on the powersupply would also induce a ripple onto hfe!?
For S1D and S2D. They are a zetex marking and stands for SXTA42 and SXTA92
Sonny
They show a lot more like base,collector and emitter resistance, size of parasitic cap between be and bc etc...
If the va or early voltage is low a small change in Vce will have a great impact on the currentgain.
Like : MJL3281A
at 40V it has a hfe of ~240
and at 10V it has a hfe ~120.
So a ripple on the powersupply would also induce a ripple onto hfe!?
For S1D and S2D. They are a zetex marking and stands for SXTA42 and SXTA92
Sonny
thanks
Okay, I'm getting the point. You look at the spice models (which are not supplied for all parts).
Choice of HV transistors was not obvious to me as the used to have poor gain linearity and high base resistance. These parts are definitely superior even to my favorite BC850CW/BC860CW.
Have you simulated the influence of having a complementary pair with substantially different V_A?
By the way, have you been able to get a free text search from the Zetex site? When I enter things like "S1D" or "VAF=3" I should get a hit but I don't.
Eric
Okay, I'm getting the point. You look at the spice models (which are not supplied for all parts).
Choice of HV transistors was not obvious to me as the used to have poor gain linearity and high base resistance. These parts are definitely superior even to my favorite BC850CW/BC860CW.
Have you simulated the influence of having a complementary pair with substantially different V_A?
By the way, have you been able to get a free text search from the Zetex site? When I enter things like "S1D" or "VAF=3" I should get a hit but I don't.
Eric
Zetex part numbering system
Hi Sonny,
thanks for that information, too. You have obviously spent considerable time searching the Zetex data sheets.
Strange thing: I am beginning to consider through hole parts a nuissance. Take up more board space, more holes to drill, take more time to solder, thermal coupling difficult to achieve. So I look for SMD parts where possible.
I have a theory that I have confirmed only on a couple of zetex parts. The letters indicate the case and the digits indicate the die. That means that a ZTX123 and an FMMT123 will be identical except for thermal resistance and related parameters. Correct?
Eric
Hi Sonny,
thanks for that information, too. You have obviously spent considerable time searching the Zetex data sheets.
Strange thing: I am beginning to consider through hole parts a nuissance. Take up more board space, more holes to drill, take more time to solder, thermal coupling difficult to achieve. So I look for SMD parts where possible.
I have a theory that I have confirmed only on a couple of zetex parts. The letters indicate the case and the digits indicate the die. That means that a ZTX123 and an FMMT123 will be identical except for thermal resistance and related parameters. Correct?
Eric
Hi eric..
Yes a have spend some time reading spicemodel, datasheet and so on... But the books, is where i have a crossreference from... Nice old books 😉
You are right about their numbering system for BJT.
F*** is SMD in some way and ZTX is through hole.
When they have a A or B attached it is a another currentgain and maybe a lower breakdown voltage.
By the way your favorite BC850CW/BC860CW. Have you tried them in a B instead of C version ?? ... Lower gain => higher early voltage.
I think they are philipstypes!? Philips will have a spice models for both the B and C version.
You wrote : "Have you simulated the influence of having a complementary pair with substantially different V_A? "
Yes but i can not find them right know. Maybe later this week. I have some things to do first.. I have been sick for a week so i have some work to catch up with!!!
Sonny
Yes a have spend some time reading spicemodel, datasheet and so on... But the books, is where i have a crossreference from... Nice old books 😉
You are right about their numbering system for BJT.
F*** is SMD in some way and ZTX is through hole.
When they have a A or B attached it is a another currentgain and maybe a lower breakdown voltage.
By the way your favorite BC850CW/BC860CW. Have you tried them in a B instead of C version ?? ... Lower gain => higher early voltage.
I think they are philipstypes!? Philips will have a spice models for both the B and C version.
You wrote : "Have you simulated the influence of having a complementary pair with substantially different V_A? "
Yes but i can not find them right know. Maybe later this week. I have some things to do first.. I have been sick for a week so i have some work to catch up with!!!
Sonny
BC850CW...
Actually, I like to use the Siemens/Infineon parts because they have a lower thermal resistance then the BC850C (in spite of smaller case) and Philips C and CW parts. A Siemens app engineer once told me that was because Siemens used a copper lead frame on the CWs.
Siemens does not have Spice models but Philips does. You are right, C parts have 3x lower V_a than A parts. The same is reflected in the open circuit reverse voltage transfer ratio given in the Siemens data sheets, only that there we only have a factor of 2. I will have to look at the definitions...
Actually, I like to use the Siemens/Infineon parts because they have a lower thermal resistance then the BC850C (in spite of smaller case) and Philips C and CW parts. A Siemens app engineer once told me that was because Siemens used a copper lead frame on the CWs.
Siemens does not have Spice models but Philips does. You are right, C parts have 3x lower V_a than A parts. The same is reflected in the open circuit reverse voltage transfer ratio given in the Siemens data sheets, only that there we only have a factor of 2. I will have to look at the definitions...
Hi Eric
nice thing to know about infineon parts.
I didn't know that they have a lower thermal resistance.
To take advantage of this ... dont you need to connect collector to a large copper area?
Maybe i should have a deeper look at the BC850CW/BC860CW to.
Sonny
nice thing to know about infineon parts.
I didn't know that they have a lower thermal resistance.
To take advantage of this ... dont you need to connect collector to a large copper area?
Maybe i should have a deeper look at the BC850CW/BC860CW to.
Sonny
Yes, you need to use wide traces to really benefit from the package. I usually need the thermal resistance because I like to have the parts run at high voltage (40 V) and medium current (1-5 mA) to keep parasitics low. Neat thing too: you can get current mirrors (BCV61/62) and dual transistors (BC847S, 857S), the latter being two separate chips in one package.
Siemens does offer SPICE models (attached), they are just kind of hidden and they don't distinguish hFE groups.
In terms of hFe linearity and high VAF, the SMBTA06 looks nice, but its PNP counterpart 56 sucks. Similar VAF in a PNP part can be found in the SMBTA92 which is a completely different (higher voltage) part. Do you have any idea why the 92 PNP would have a higher VAF than its NPN counterpart 42 whereas it is the other way around for most other pairs?
Siemens does offer SPICE models (attached), they are just kind of hidden and they don't distinguish hFE groups.
In terms of hFe linearity and high VAF, the SMBTA06 looks nice, but its PNP counterpart 56 sucks. Similar VAF in a PNP part can be found in the SMBTA92 which is a completely different (higher voltage) part. Do you have any idea why the 92 PNP would have a higher VAF than its NPN counterpart 42 whereas it is the other way around for most other pairs?
stranger still...
... On Semi has the same part, callt MMBTA06 - VAF is 996.086 there. Unfortunately, they don't offer a model for the MMBTA56
... On Semi has the same part, callt MMBTA06 - VAF is 996.086 there. Unfortunately, they don't offer a model for the MMBTA56
As Geoffs told me. Sometimes the manufacture tweaks the spice model to get it up and running.
But it could very well be a difference in their process or design who gives the higher value from onsemi.
For the 42 versus 92 thing.. Again how have they done their chip and wich process are they using?
So try run a vce/ice sim on the model.. To see if it behaves/is as flat as indicated by the VAF value... It is dead easy to check in switchercad,pspice etc..
The models have been okay until know regarding VAF ... so try run a sim if it is okay then the value should be fine.
But a VAF in the area of 300 and above shure is good.
VAF is not everything but it tells you a bit about the linearity of the BJT.
The "hfe" curve could be fine even if it has a low VAF.
Still with a low VAF the hfe gain is more "vce" dependent than with a high VAF.
By the way one i find nice is BC847BPN from philips, i think Infineon has it to. Nearly matched complementary in a single SMD-box. Infineon does have wide program of devices.
Sonny
But it could very well be a difference in their process or design who gives the higher value from onsemi.
For the 42 versus 92 thing.. Again how have they done their chip and wich process are they using?
So try run a vce/ice sim on the model.. To see if it behaves/is as flat as indicated by the VAF value... It is dead easy to check in switchercad,pspice etc..
The models have been okay until know regarding VAF ... so try run a sim if it is okay then the value should be fine.
But a VAF in the area of 300 and above shure is good.
VAF is not everything but it tells you a bit about the linearity of the BJT.
The "hfe" curve could be fine even if it has a low VAF.
Still with a low VAF the hfe gain is more "vce" dependent than with a high VAF.
By the way one i find nice is BC847BPN from philips, i think Infineon has it to. Nearly matched complementary in a single SMD-box. Infineon does have wide program of devices.
Sonny
Found an exception to the new law 😱
My speakers are highly magnetic and they sound really good, they improve with fine wines too😛 maybe we should look in to the wine / speaker magnetism theory
My speakers are highly magnetic and they sound really good, they improve with fine wines too😛 maybe we should look in to the wine / speaker magnetism theory
- Status
- Not open for further replies.