Decided to try a new idea, where the complementary driver transistors only see < 10 volts.
There is usually a trade-off between BVce and Ft, so I can optimize for speed and maybe save some phase shift to allow improved compensation.
Previous, conventional circuit used 2SC3503 / 2SA1381 but for this application there should be better parts.
Current requirement is modest, few 100 mA at most so power is also fairly low, maybe a watt.
Anyone have any recommendations for stand-out fast drivers?
David
There is usually a trade-off between BVce and Ft, so I can optimize for speed and maybe save some phase shift to allow improved compensation.
Previous, conventional circuit used 2SC3503 / 2SA1381 but for this application there should be better parts.
Current requirement is modest, few 100 mA at most so power is also fairly low, maybe a watt.
Anyone have any recommendations for stand-out fast drivers?
David
DSL CFA op amps? - are built on GHz processes, op amp with feedback can be faster than most discrete driver Q candidates
There are many "wideband" transistors intended for CATV or instruments applications that fall into those power requirements; since the NPN member is not going to pose problems, I'll give just PNP examples: 2N5160, 2N5583.
They are obsolete TO5 devices (which shows my age), but there are more modern examples, in more modern cases, BFRxxx, BFQxxx in SOT89 or "capstan" cases for the fastest.
The NPN is normally not difficult to find: 2N3866 (a real ancestor), 2N5109 (a bit more modern), BFW16.
In HP/Agilent/Keysight talk, they would be called 1853-201 and 1854-247: that's the complementary pair that equips all of their historic generators, be they PDH/SDH, video, function, etc
They are obsolete TO5 devices (which shows my age), but there are more modern examples, in more modern cases, BFRxxx, BFQxxx in SOT89 or "capstan" cases for the fastest.
The NPN is normally not difficult to find: 2N3866 (a real ancestor), 2N5109 (a bit more modern), BFW16.
In HP/Agilent/Keysight talk, they would be called 1853-201 and 1854-247: that's the complementary pair that equips all of their historic generators, be they PDH/SDH, video, function, etc
It would be helpful if you were more specific.
That is an idea that I had not considered, clever, but it's a whole new area to research, maybe for version 3.
Thank you for the recommendations. These are faster than I need, optimized for RF.
They have quite low hFE by audio transistor standards, which reduces the feedback I have available to reduce distortion.
I look for hFe at least 100, typical audio transistor value, just faster.
The 3503/1381 are already around 150 MHz so Ft around 400 MHz looks achievable after trade-off of lower BVce.
Problem is that the sellers don't usually have any way to search only for transistors with complements so it's a bit tedious.
Best wishes
David
Some of the discontinued Sanyo CRT drivers are 400 MHz,
Edmond Stuart reported that the lower phase shift was helpful in the Poppa/Stuart ultra-low distortion amp.
Anyone know a source?
Last edited:
At what currents are you comparing?
The typical 50V small signal devices would be a reasonable choice, probably the better ones from Zetex.
You could also try some switching 'driver' transistors for size, if the capacitance isn't an issue (e.g. if it lives under a cascode). Sort of like MCH6564 or CPH5541. But since they are not intended for signal use some curve tracing and characterisation time may be in order.
The typical 50V small signal devices would be a reasonable choice, probably the better ones from Zetex.
You could also try some switching 'driver' transistors for size, if the capacitance isn't an issue (e.g. if it lives under a cascode). Sort of like MCH6564 or CPH5541. But since they are not intended for signal use some curve tracing and characterisation time may be in order.
2N3904/3906 ... lol too old for you?
By driver are you saying an EF?
Does Early voltage matter in your application?
By low voltage are you looking for good saturation characteristics, or does it not matter?
By driver are you saying an EF?
Does Early voltage matter in your application?
By low voltage are you looking for good saturation characteristics, or does it not matter?
Last edited:
IMZ2A is ~400MHz, +/-50V, Hfe >200...
And dual in one package.
http://www.mouser.com/ds/2/348/emz2-210654.pdf
And dual in one package.
http://www.mouser.com/ds/2/348/emz2-210654.pdf
It's a clean sheet so few constraints, typically I run the current close to the maximum of the Ft curve, so doesn't have to be the same for each candidate.
Well, yes, that's precisely why I asked, in case someone had already spent the time to search the datasheets and check the contenders to find a winner.
Or at least a few top picks.
The specific application was to drive a chunky pair of IXYS FETs in Common Source. so the "driver" sees the Vgs of the FET, only around 4 V to maybe 10V.
So saturation should not matter, or Early effect, as far as I can see at this point when it is still a new circuit for me.
But such devices would also be useful for bootstrapped drivers, as in Bob Cordell's book for instance.
So I don't want to be too restrictive, at the moment.
Best wishes
David
2SA1407/2SC3601 FT=400MHz http://rtellason.com/transdata/2sc3601.pdf
I don't think they manufacture these anymore but, since you are interested in simulation rather than physically building something, you can succesfully use them, and for free. I haven't seen any decent models for these, though.
I've seen once a chart with these devices process, they are 8 mask with polysilicon emitter, very expensive for discretes, no wonder Onsemi dropped them right away after acquiring Sanyo.
I don't think they manufacture these anymore but, since you are interested in simulation rather than physically building something, you can succesfully use them, and for free. I haven't seen any decent models for these, though.
I've seen once a chart with these devices process, they are 8 mask with polysilicon emitter, very expensive for discretes, no wonder Onsemi dropped them right away after acquiring Sanyo.
Dual is often nice but I don't think I need it here and it makes the power dissipation a bit limited, only 150 mW per device.
Looks useful for certain application however.
Yes, I already mentioned that just a few posts back, in #6.
Do you have any spares to sell?
Best wishes
David
I found some 2SA1403E / 2SC3597E with minimum hFE of 100.
800 MHz Ft so phase shift should not be a problem😉
Local oscillation maybe...
David
800 MHz Ft so phase shift should not be a problem😉
Local oscillation maybe...
David
I bought 150 of each when DigiKey was selling off their Sanyo inventory. However peak fT is at 70mA for the 2SA1407 and 50mA for the 2SC3601. If you run them from ±45 volt rails, they'll be dissipating 2.7 watts so you need a generous heatsink.
However the Sanyo datasheet doesn't give a Theta-JC. If we assume it's 18 deg C/W as for other TO-126 devices, then we'll need a heatsink whose temperature rise is less than 4.2 deg C/W (!) That's for the worst case scenario where chassis internal air is 50 deg C and BJT junction temperature is 110 deg C, thus delta-T is 60 deg C. {math: 2.7W * (18 deg/W + 4.2 deg/W) = 60deg}
That's quite a big heatsink: example. So I expect that most people will choose to operate their A1407 and C3601 at considerably lower current than peak-fT, and get considerably less than 400 MHz in practice.
However the Sanyo datasheet doesn't give a Theta-JC. If we assume it's 18 deg C/W as for other TO-126 devices, then we'll need a heatsink whose temperature rise is less than 4.2 deg C/W (!) That's for the worst case scenario where chassis internal air is 50 deg C and BJT junction temperature is 110 deg C, thus delta-T is 60 deg C. {math: 2.7W * (18 deg/W + 4.2 deg/W) = 60deg}
That's quite a big heatsink: example. So I expect that most people will choose to operate their A1407 and C3601 at considerably lower current than peak-fT, and get considerably less than 400 MHz in practice.
No, Dave I'm not selling any.
Mark, his question was for 10V and I was going to mention that it gives him the option to
bias any small transistor near the peak of the Ft curve.
Mark, his question was for 10V and I was going to mention that it gives him the option to
bias any small transistor near the peak of the Ft curve.
- Status
- Not open for further replies.