| Author |
Message |
b-funk
Joined: Jul 17, 2007
Posts: 193
Location: Berlin
|
 Posted: Wed Jan 23, 2008 4:08 am Post subject:
Question about transistor pairs - split off |
  |
|
hey guys,
as i don´t have "too much" experience with transistor-pair stuff, this is very interesting for me...
maybe it´s worth a "real" thread? ups and downs of matched transistor pairs?
best wishes,
tobias.
[Blue Hell : split this off from : http://electro-music.com/forum/post-165312.html] |
|
|
Back to top
|
|
 |
frijitz
Joined: May 04, 2007
Posts: 1734
Location: NM USA
Audio files: 54
|
| Posted: Wed Jan 23, 2008 8:48 am Post subject:
|
|
|
| b-funk wrote: | as i don´t have "too much" experience with transistor-pair stuff, this is very interesting for me...
maybe it´s worth a "real" thread? ups and downs of matched transistor pairs? |
Yeah, good idea. Unfortunately, we don't seem to have much around by way of moderators. Krunkus just got married and Stites? Haven't seen him anywhere lately ... hope he's OK.
Ian |
|
|
Back to top
|
|
|
frijitz
Joined: May 04, 2007
Posts: 1734
Location: NM USA
Audio files: 54
|
| Posted: Thu Jan 24, 2008 5:36 pm Post subject:
|
|
|
Hey great, a separate thread for this discussion. Thanks Jan!
This thread stems from the following statement:
| Etaion wrote: | | The LM394 is "low end" really. Most people achieve better VCO tracking with the SSM2210. |
Can the LM394 really be considered "low end", especially in view of the statement in National's app notes (Super Matched Bipolar Transistor Pair Sets New Standards for Drift and Noise):
"The relation between emitter-base voltage and collector current, for instance, is perfectly logarithmic over an extremely wide range of collector currents, deviating in the pA range because of leakage currents and above several mA due to the finite 0.4 Ohm emitter resistance. This gives the LM194 a distinct advantage in non-linear designs where true logarithmic behavior is essential to circuit accuracy."
That's >15 octaves, folks!
A careful study of the SSM2210 vs LM194/394 data sheets shows that the SSM2210 is a slightly better device than the LM194. For example, the emitter resistance is 0.3 Ohms (vs 0.4). And of course the LM394 is not quite as good as the LM194. But these are rather small differences, in the big scheme of things. For example the emitter resistance of the popular CA3046 array is 5 Ohms, and my favorite, the CA3083, weighs in at 1.3 Ohms. I have made a number of VCO's with perfect tracking up to 20 kHz or even 40 kHz with the 3083 -- it requires a bit more HF tracking feedback, but this correction for emitter resistance is quite accurate.
So based on just a couple of anecdotes about poor performance of the LM319, I don't think anyone should avoid this chip, especially given the excellent price from Futurlec.
It probably wouldn't be too hard to track down the problem with the 394 in the RW VCO. There could only be a couple of explanations, such as bad chips, chips damaged in handling and soldering (very easy to do, in my experience), or improper circuit design. For example, has anyone checked carefully for HF oscillations in the converter's servo amp? That's actually the most common cause of tracking problems, and in my experience the 100 pF stabilizing cap in the RW design is marginal. This part of the circuit might very well work better with one device vs the other.
Two years ago I breadboarded a hot-rodded version of the RW design to see how good the core could do with high-performance opamps and a decent switching FET. I was able to get tracking to within 0.1% up to 15 kHz or so. This was using CA3083 transistors in the converter. The RW core is seriously limited by its use of a TL082 chip for the integrator and for the comparator. I was using an OPA 2134, which is significantly better for both applications. But really one should use an LM311 comparator (as in the original Terry Michaels design) here.
Oh, and I had to use 180 pF for the servo stabilizer. Isn't that funny, an amp specifically designed to be stable in servo operation requires twice as much capacitance as the jelly-bean device in the RW design?
My updated version of Terry's design may be found on my website. It uses the LM394 and tracks perfectly (as close as I could measure at the time) up to 40 kHz.
Anyway, the point is that you folks should all snap up some of these from Futurlec. I got a dozen, and just from measuring the hFE on my multimeter they seem OK.
Ian |
|
|
Back to top
|
|
|
Uncle Krunkus
Moderator
Joined: Jul 11, 2005
Posts: 4761
Location: Sydney, Australia
Audio files: 52
G2 patch files: 1
|
| Posted: Thu Jan 24, 2008 6:52 pm Post subject:
|
|
|
| frijitz wrote: | Yeah, good idea. Unfortunately, we don't seem to have much around by way of moderators. Krunkus just got married and Stites? Haven't seen him anywhere lately ... hope he's OK.
Ian |
I actually don't even know how to split a thread.  I s'pose I'm really only an honourary moderator, as I don't know a lot of stuff I should. I wonder sometimes how the other moderators do these things, but there doesn't seem to be a "handbook" of moderator jobs.
And it's true, I've been quite pre-occupied with other stuff lately.
_________________
What makes a space ours, is what we put there, and what we do there. |
|
|
Back to top
|
|
|
frijitz
Joined: May 04, 2007
Posts: 1734
Location: NM USA
Audio files: 54
|
| Posted: Thu Jan 24, 2008 7:25 pm Post subject:
|
|
|
| Uncle Krunkus wrote: |
I actually don't even know how to split a thread. I s'pose I'm really only an honourary moderator, as I don't know a lot of stuff I should. I wonder sometimes how the other moderators do these things, but there doesn't seem to be a "handbook" of moderator jobs.
And it's true, I've been quite pre-occupied with other stuff lately. |
Ooops, hope you weren't right in the middle of anything.
Ian |
|
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007
Posts: 1536
Location: nyc
Audio files: 41
|
| Posted: Thu Jan 24, 2008 9:39 pm Post subject:
|
|
|
i am using the lm394 from futurlec in all my circuits that call for a matched pair...so far (at least with my limited knowledge) it seems fine.
and they are so cheap!!
i also got a bunch of the ca3046
so Ian you recommend switching out the opamps in the RW vco? thats my main vco right now.
and you think the 100pf cap is marginal? do you mean youd recommend a different value?
btw---my vca is feeding through much less now that i isolated the wiring and also since i put the lfo on its own bus...seems it was also causing interference on the vca
_________________
-------------------------------------------- check out various dan music at: http://www.myspace.com/lossnyc
http://www.myspace.com/snazelle
http://www.soundclick.com/lossnyc.htm http://www.indie911.com/dan-snazelle |
|
|
Back to top
|
|
|
andrewF
Joined: Dec 29, 2006
Posts: 1176
Location: australia
Audio files: 4
|
| Posted: Thu Jan 24, 2008 9:46 pm Post subject:
|
|
|
| frijitz wrote: |
Ooops, hope you weren't right in the middle of anything.
Ian |
I must confess I have never been too fussed about perfect tracking/tuning, guess that is a reflection of my minimal (= none) muscial talent.
Nevertheless I usually use LM394 or 2SC1583 in VCOs.
2SC1583 is no longer produced, I searched Akihabara from top-to bottom in December, one little hole-the-wall shop had just one left in stock. 
how does the 2SC1583 compare? |
|
|
Back to top
|
|
|
frijitz
Joined: May 04, 2007
Posts: 1734
Location: NM USA
Audio files: 54
|
| Posted: Thu Jan 24, 2008 10:24 pm Post subject:
|
|
|
| loss1234 wrote: | i am using the lm394 from futurlec in all my circuits that call for a matched pair...so far (at least with my limited knowledge) it seems fine...
so Ian you recommend switching out the opamps in the RW vco? thats my main vco right now. |
Ah ... so have you characterized the tracking? Are you getting more than 3-4 good octaves? I just can't understand having a problem with the LM394, unless there is something messed up.
If you want to try some mods on the RW VCO, I have some ideas -- not all tested, but based on my adventures trying to design high-performance oscillators. To start, replace U2 (TL082) with an OPA2134. (Buy a buch of those if you can; you can use them all over the place.) Then replace Q1 (MPF102) with a 2N4391 (or PN4391) This is the best commonly available JFET for this application. And believe me, I hunted through a pile of data sheets to find it. And Terry Michaels agrees that it is a good replacement for the one he used originally.
U1 is OK, I guess, but I usually use an OPA2227 there. It may be a bit of an overkill with this design, though. I'd increase C3 to 330 pF just to be sure, and also add a 1 nF cap in parallel with the R24 tempco -- these sometimes cause spurious HF oscillations, depending on their construction. It won't hurt anything if the oscillations are not a problem.
| Quote: | btw---my vca is feeding through much less now that i isolated the wiring and also since i put the lfo on its own bus...seems it was also causing interference on the vca
|
Good work! It can take hours to hunt down all the little problems that degrade your S/N. Careful power and signal routing and lead dressing are a big part of this.
Ian |
|
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005
Posts: 761
Location: Utrecht, NL
|
| Posted: Fri Jan 25, 2008 12:35 am Post subject:
|
|
|
| frijitz wrote: | | Then replace Q1 (MPF102) with a 2N4391 (or PN4391) |
Would that be generic? I.e. would that also be valid for other VCO's that use the MPF102 or equivalents?
Haven't found a source for the 2N4391 yet, other than some very expensive ones on eBay, but if it's worth the trouble I might get some.
_________________
http://www.casia.org/modular/ |
|
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005
Posts: 761
Location: Utrecht, NL
|
| Posted: Fri Jan 25, 2008 1:42 am Post subject:
|
|
|
Ian, I replaced the TL082 by an OPA2134 and used the two caps you mentioned. Haven't replaced the FET yet as I don't have one.
As a test, I removed the SSM2210 and put a LM394CH back in (a can).
From a first quick test it appears to track better right away even without recalibrating. I haven't done any accurate tests, I just ran it next to my CEM3340 based VCO and it is much closer to that than it was.
_________________
http://www.casia.org/modular/ |
|
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007
Posts: 1536
Location: nyc
Audio files: 41
|
|
|
Back to top
|
|
|
frijitz
Joined: May 04, 2007
Posts: 1734
Location: NM USA
Audio files: 54
|
|
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005
Posts: 761
Location: Utrecht, NL
|
| Posted: Fri Jan 25, 2008 10:09 am Post subject:
Re: 2N4391 |
|
|
| frijitz wrote: | | for twelve cents |
And $57 shipping. So pretty expensive 
_________________
http://www.casia.org/modular/ |
|
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005
Posts: 761
Location: Utrecht, NL
|
| Posted: Fri Jan 25, 2008 10:16 am Post subject:
|
|
|
No, I'm telling a lie. It's $35 shipping to The Netherlands.
But I'm still open to alternative source suggestions
[edit]
found a local source!
At 5.45 euro each (about $7). Hmmm...
[editedit]
Down to 2.61 each...
_________________
http://www.casia.org/modular/ |
|
|
Back to top
|
|
|
frijitz
Joined: May 04, 2007
Posts: 1734
Location: NM USA
Audio files: 54
|
| Posted: Fri Jan 25, 2008 10:40 am Post subject:
|
|
|
| Etaoin wrote: | | frijitz wrote: | | Then replace Q1 (MPF102) with a 2N4391 (or PN4391) |
Would that be generic? I.e. would that also be valid for other VCO's that use the MPF102 or equivalents? |
Yes, absolutely! The MPF102 is a small-signal RF transistor. I'm totally baffled as to why anyone would use it for a high-current switching application. It has a high series resistance and a low current rating.
If you have trouble finding the 4391, you could look for the J109. Some folks swear by this one. For me, it would be a second choice because of its larger off-state leakage current. But it's a pretty good discharge switch.
There might be a European device that would work well, but I didn't look through those.
| Etaoin wrote: | | As a test, I removed the SSM2210 and put a LM394CH back in (a can). From a first quick test it appears to track better right away even without recalibrating |
Glad you were able to get on this so quickly and that my ideas are working! 
I would be interested in knowing whether you still see a difference between the two pairs. My guess is that your original troubles were probably due to HF oscillations, as those could possibly be different for the two devices.
Ian |
|
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005
Posts: 761
Location: Utrecht, NL
|
| Posted: Fri Jan 25, 2008 10:44 am Post subject:
|
|
|
[quote="frijitzThere might be a European device that would work well, but I didn't look through those.
[/quote]
What specs would I be looking for specifically? I might check a number of commonly available ones over here.
_________________
http://www.casia.org/modular/ |
|
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007
Posts: 1536
Location: nyc
Audio files: 41
|
|
|
Back to top
|
|
|
frijitz
Joined: May 04, 2007
Posts: 1734
Location: NM USA
Audio files: 54
|
| Posted: Fri Jan 25, 2008 11:59 am Post subject:
|
|
|
| Etaoin wrote: | | frijitz wrote: | There might be a European device that would work well, but I didn't look through those.
|
What specs would I be looking for specifically? I might check a number of commonly available ones over here. |
Look for:
type: N-channel JFET for switching
current handling: Idss >= 40 mA
on resistance: Rds <= 40 Ohm
capacitance: Ciss <= 20 pF
GL!
Ian |
|
|
Back to top
|
|
|
bubblechamber
Joined: Nov 04, 2006
Posts: 280
Location: NYC
Audio files: 1
|
| Posted: Fri Jan 25, 2008 12:08 pm Post subject:
|
|
|
I would be happy to order a bunch of PN4391 with my next mouser order(in about 2 weeks) and ship them to Europe if you can't find a cheap local source. I'm sure it'll be a bit cheaper to have me ship them to you.
And thanks a lot Ian for that tutorial. it cleared some stuff up for me. I really don't care for the AD chips(too expensive) and I've been looking at the LM394, glad for the independent confirmation.
david |
|
|
Back to top
|
|
|
b-funk
Joined: Jul 17, 2007
Posts: 193
Location: Berlin
|
| Posted: Fri Jan 25, 2008 4:19 pm Post subject:
|
|
|
wow... thank you ian for the explanations. now stuff looks a bit clearer.
i have to experiment a little with different transistor pairs to verify if i understood what you´re saying
@andrewf:
i have a couple of 2sc1583 left. do you need some?
best wishes,
tobias. |
|
|
Back to top
|
|
|
frijitz
Joined: May 04, 2007
Posts: 1734
Location: NM USA
Audio files: 54
|
| Posted: Fri Jan 25, 2008 6:12 pm Post subject:
|
|
|
| andrewF wrote: | 2SC1583 is no longer produced, I searched Akihabara from top-to bottom in December, one little hole-the-wall shop had just one left in stock.
how does the 2SC1583 compare? |
MAT Electronics has them listed for <$2. matelectronics.com or
http://www.matelectronics.com/acatalog/Mat_Electronics_2SC__Semiconductors_191.html
I have a bag I bought when all the Japanese pairs started getting hard to find. But I haven't tried one in a VCO yet, and I haven't measured the Re or the matching.
But Rene likes them! That's something to go by.
Ian |
|
|
Back to top
|
|
|
loss1234
Joined: Jul 24, 2007
Posts: 1536
Location: nyc
Audio files: 41
|
|
|
Back to top
|
|
|
Tallhobbit
Joined: Aug 10, 2007
Posts: 15
Location: Kirkland, WA USA
|
| Posted: Fri Jan 25, 2008 11:19 pm Post subject:
|
|
|
| Quote: | what about circuits that call for NPN/PNP pairs??
|
I bought 50 of these: BC847BPN They're a nasty little SM part with a .65mm/.025" lead pitch but the data sheet looked as if the critter just may work. I haven't tried any of them yet and don't think I'll get around to it for several months, at best. If someone wants to experiment with a couple and is willing to solder little tiny wires onto even smaller leads, I'll part with a few as free samples. The catch is that you have to actually try to see if they're useable and report back. Just PM me. I think I paid about $.02 each for them.
Here's a link to the datasheet. http://www.nxp.com/acrobat_download/datasheets/BC847BPN_3.pdf
James |
|
|
Back to top
|
|
|
andrewF
Joined: Dec 29, 2006
Posts: 1176
Location: australia
Audio files: 4
|
| Posted: Fri Jan 25, 2008 11:44 pm Post subject:
|
|
|
| b-funk wrote: |
@andrewf:
i have a couple of 2sc1583 left. do you need some?
|
cheers, but I should be fine - have a few in the cupboard and can still get LM394 from Futurlec, if I need more (their $4 delivery fee can't be beat). Sometimes I find them in old lab gear, I recycle, too!
it was just disappointing to scour Akihabara and only find one dual NPN (also NO LM3080 anywhere too.....plenty of SN76477 tho...oh and a bag of 65 sub-mini dual triodes for free  )
Ian - thanks for the link. I have used 2SC1583 in CGS and farm (Takeda) VCOs, have also tried LM394 in CGS VCOs. I cannot tell the difference once they are up and running. |
|
|
Back to top
|
|
|
etaoin
Joined: Jun 30, 2005
Posts: 761
Location: Utrecht, NL
|
| Posted: Sat Jan 26, 2008 3:14 am Post subject:
|
|
|
| frijitz wrote: |
type: N-channel JFET for switching
current handling: Idss >= 40 mA
on resistance: Rds <= 40 Ohm
capacitance: Ciss <= 20 pF
|
Ian, you mentioned the J109 above. Futurlec sell the J108 which appears to be the similar but with a larger Idcc (80mA for the J108) and lower Rds.
Both have a Ciss of 85pF though.
@bubblechamber: I might take your offer. I'll let you know in time.
_________________
http://www.casia.org/modular/ |
|
|
Back to top
|
|
|
|
Forum index » DIY Hardware and Software
FAQ
Calendar
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
Chat Room
