I buy the ARRL "Radio Amateur's Handbook" every three years or so. For some reason I use the 1964 version most frequently!
Has anything changed over the years. Isn't it pretty much all the same with something like that?
On the other hand, I don't see it as any big issue to use NLA parts unless the examples
are construction projects. I don't remember but I did not think that was the case in your
book Bob. So what if they are NLA? People can look at the specs and hope to find
something similar. You are just explaining concepts as I understand it. If it was a
book on construction projects then of course that is different. I wouldn't go so far
back as to use MJ802/4502 there is no reason to at all and you didn't do this - just
an example.
If you need more room with the book getting too large, I'd rather see intro and
advanced analog in one book and another book on digital. The two are not really
related much and I think there are many people who would pick one or the other.
are construction projects. I don't remember but I did not think that was the case in your
book Bob. So what if they are NLA? People can look at the specs and hope to find
something similar. You are just explaining concepts as I understand it. If it was a
book on construction projects then of course that is different. I wouldn't go so far
back as to use MJ802/4502 there is no reason to at all and you didn't do this - just
an example.
If you need more room with the book getting too large, I'd rather see intro and
advanced analog in one book and another book on digital. The two are not really
related much and I think there are many people who would pick one or the other.
Forgot to mention that I also would not use MJE243/253 today since they have quite
bad SOA. Last time I used them was 1979. I also understand not wanting to redo the
many figures in your book.
Wondering if you've looked at Sanken 2SC2837 and comp to use as something
better than the MJE15032/33 for big output stages. Their specs seem to be all
around better and there are many in stock at Digikey:
http://www.semicon.sanken-ele.co.jp/sk_content/2sc2837_ds_en.pdf
bad SOA. Last time I used them was 1979. I also understand not wanting to redo the
many figures in your book.
Wondering if you've looked at Sanken 2SC2837 and comp to use as something
better than the MJE15032/33 for big output stages. Their specs seem to be all
around better and there are many in stock at Digikey:
http://www.semicon.sanken-ele.co.jp/sk_content/2sc2837_ds_en.pdf
That Sanken transistor looks okay, although it's no faster than the transistors it would be driving. The Hfe is tested at 4Vce but as you can see in the Vce/Ic graph, the transistor is in quasi-saturation. At higher Vce, it probably has better Hfe at high currents.
What do you think, Kean, better or worse than the MJE15032/33 as drivers?
Did you notice that hfe is ranked and the 50 is min for the lowest grade.
Best grade is 90 min.
Also consider that these are twice the power of the MJE15032/33 for even more
current/power headroom for amps that might drive 1 ohm loads or without
output protection.
Just noticed that Ft peaks at 2 amps and drops off fast below, might actually
be worst than the MJE15032/33.
Did you notice that hfe is ranked and the 50 is min for the lowest grade.
Best grade is 90 min.
Also consider that these are twice the power of the MJE15032/33 for even more
current/power headroom for amps that might drive 1 ohm loads or without
output protection.
Just noticed that Ft peaks at 2 amps and drops off fast below, might actually
be worst than the MJE15032/33.
Last edited:
They look better to me, although we don't know what Early voltage for the Sankens is. The MJE15032/33 are actually slower than the MJL3281/MJL1302. That doesn't stop people from making good amps with them, but generally when you get a weaker transistor you want to make up for it in the other parameters.
The Sankens are LAPT, you can tell by the way the Ft rises quickly and then within a few octaves of current drops like a bomb. They are probably faster than the MJEs at all operating points, especially at 2A due to the LAPT process.
Whether you like LAPT might depend on how you design amplifiers. Some designs might prefer Ft being relatively flat across the current range as in the MJL21194/21193, but this comes at the cost of very low overall Ft.
The Sankens are LAPT, you can tell by the way the Ft rises quickly and then within a few octaves of current drops like a bomb. They are probably faster than the MJEs at all operating points, especially at 2A due to the LAPT process.
Whether you like LAPT might depend on how you design amplifiers. Some designs might prefer Ft being relatively flat across the current range as in the MJL21194/21193, but this comes at the cost of very low overall Ft.
might be worth looking at the newer mje15034/5
Slightly faster and slightly higher gain
The reduction in maximum Ic does not have any effect on ability to drive a 2pair output stage.
Slightly faster and slightly higher gain
The reduction in maximum Ic does not have any effect on ability to drive a 2pair output stage.
I use 2SC6145/2SA2223, and I like their sound, over the Onsemis. The drivers are 2SC4883a/2SA1859a, which are quite fast, and looks reliable up to +/-60V.
My favourite driver was the 2SC5171/2SA1930 pair, but they oscillate sometimes, if the supply was over 50V.
Sajti
My favourite driver was the 2SC5171/2SA1930 pair, but they oscillate sometimes, if the supply was over 50V.
Sajti
The C4883A/A1859A are one of my favorite pairs, I made SPICE models for them, I even made a small speaker amp using them as outputs. Available from Digikey.
Some readers may not be aware that Winfield Hill (co author of The Art Of Electronics) used the 2SC4883 / 2SA1859 devices as the output stage of his
http://www.diyaudio.com/forums/solid-state/287023-winfields-100w-dc-10mhz-1000v-us-amplifier.html
at Harvard. He generously shared the schematics and his design spreadsheets, here on diyAudio. Take a look, it might be fun.
_
http://www.diyaudio.com/forums/solid-state/287023-winfields-100w-dc-10mhz-1000v-us-amplifier.html
at Harvard. He generously shared the schematics and his design spreadsheets, here on diyAudio. Take a look, it might be fun.
_
Just checked the book, no mention of FOM (I may have missed it, so correct me if I am wrong). Reading the 'Winfield's 100W ...' thread makes me wonder, should this FOM be explained in the book?
Thanks everyone for all the interesting suggestions, I've not noticed many of these.
As Kean mentions the Ft of the 2SC6145/2SA2223 drops like a rock at high Ic and I
wonder what this does in practice with a very low Z load - would it tend to go unstable?
I would probably lean toward a compensation scheme that bypasses the output devices
at HF, something I've preferred for many years.
Also interesting, the 2SC6145/2SA2223 data sheet suggests the 2sc4382 and comp as
drivers, they are slow at 15MHz but with fairly low Cob and they have the best SOA at
50V of nearly any 2A/20W driver that I've seen.
The 2SC4883A/A1859A also have far better than most SOA but also with high Ft - I like
these very much.
Sajti I've looked at the 2SC5171/2SA1930 pair for years but their SOA is so bad that I
won't be using them as drivers in any demanding designs. Perhaps as a high bias VAS.
I like the 2sa1837/2sc4793 pair far better even with a max IC of only 1A their SOA at 50
and 100V is double the 2SC5171/2SA1930.
The KSA1220A/KSC2690A pair have about the same SOA as 2sa1837/2sc4793, slightly
higher 1.2A max Ic, and are readily available. I'd put these in parallel when I need
greater SOA.
As Kean mentions the Ft of the 2SC6145/2SA2223 drops like a rock at high Ic and I
wonder what this does in practice with a very low Z load - would it tend to go unstable?
I would probably lean toward a compensation scheme that bypasses the output devices
at HF, something I've preferred for many years.
Also interesting, the 2SC6145/2SA2223 data sheet suggests the 2sc4382 and comp as
drivers, they are slow at 15MHz but with fairly low Cob and they have the best SOA at
50V of nearly any 2A/20W driver that I've seen.
The 2SC4883A/A1859A also have far better than most SOA but also with high Ft - I like
these very much.
Sajti I've looked at the 2SC5171/2SA1930 pair for years but their SOA is so bad that I
won't be using them as drivers in any demanding designs. Perhaps as a high bias VAS.
I like the 2sa1837/2sc4793 pair far better even with a max IC of only 1A their SOA at 50
and 100V is double the 2SC5171/2SA1930.
The KSA1220A/KSC2690A pair have about the same SOA as 2sa1837/2sc4793, slightly
higher 1.2A max Ic, and are readily available. I'd put these in parallel when I need
greater SOA.
Some knowledge about quasi-saturation is helpful when interpreting these datasheets.
In quasi-saturation Ft drops very fast. In most datasheets Ft is measured at low Vce where the end of the range runs into quasi-saturation. So actually at normal operating voltages high current Ft and Hfe could be much higher.
IIRC this is the case for the 2SC4793/A1837, which makes them even more competitive with the 2SC5171/A1943.
Since this is the case for so many datasheets, we really don't know how these transistors behave at normal operating points, but it's likely that Hfe and Ft drop much more gradually. The datasheet charts are mainly applicable to near-clipping conditions.
With the LAPT process, Ft is maintained right up to the quasi-saturation limit. For other transistors, different Ft limits tend to kick in before then.
In quasi-saturation Ft drops very fast. In most datasheets Ft is measured at low Vce where the end of the range runs into quasi-saturation. So actually at normal operating voltages high current Ft and Hfe could be much higher.
IIRC this is the case for the 2SC4793/A1837, which makes them even more competitive with the 2SC5171/A1943.
Since this is the case for so many datasheets, we really don't know how these transistors behave at normal operating points, but it's likely that Hfe and Ft drop much more gradually. The datasheet charts are mainly applicable to near-clipping conditions.
With the LAPT process, Ft is maintained right up to the quasi-saturation limit. For other transistors, different Ft limits tend to kick in before then.
I've also been curious about these and wonder how different they really are as compared to
the lower voltage versions. The thermal resistances are the same which suggests that the
die sizes are the same. I have to wonder if the half current rating is just the bonding wire
limit and if the lower voltage parts are double bonded or use larger wire? Odd that the
mje15034/5 data sheets provide Ft curves, the peak is much higher than the rated Ft but
also drops like a rock above roughly 1A.