Maker Pro
Maker Pro

Need replacement for BC847S better suited for HF

E

Erik Baigar

Jan 1, 1970
0
Hi,

in a project I am using the dual transistor BC847S in the
SOT363 package. This transistor has a maximum frequency of
200MHz according to the datasheet.
Does anyone know about a pin compatible replacement with
a higher frequency (lets say 500MHz). Gain and Uce are not
so important for me and maximum Ic of 50mA would be OK, too.

The project is that in my russan oscilloscope (C1-122)
there are hybrid switches used for chopping channel
and trigger signals. These switches (04KH009 or
04KN009) tend to die rather often and so I designed a
replacement using the BC847S. In principle it works
fine but the bandwith of the oscilloscope is severely
reduced since the BC847S is not well suited for HF
applications. But using transistors in such a small
package was mandatory to obtain th small form factor
to insert the replacement circuit in the oscilloscope.
This project is also relevant for the following russian
oscilloscopes used by lots of hobbyists: C1-112, C1-112A.

Any hints are welcome,

Erik.


P.S. If someone is interested I'd be able to share my
information and documentation. I have some PCBs
left, too.
 
J

Joerg

Jan 1, 1970
0
Hello Erik,
in a project I am using the dual transistor BC847S in the
SOT363 package. This transistor has a maximum frequency of
200MHz according to the datasheet.
Does anyone know about a pin compatible replacement with
a higher frequency (lets say 500MHz). Gain and Uce are not
so important for me and maximum Ic of 50mA would be OK, too.

Check Toshiba's transistor pairs. IIRC Digikey has them.

Else, how about using a couple RF transistors and glue their bodies
together for better thermal coupling? Just keep in mind that their beta
will usually be quite low and so will their breakdown voltage.

Regards, Joerg
 
E

Erik Baigar

Jan 1, 1970
0
Joerg said:
Hello Erik,
Check Toshiba's transistor pairs. IIRC Digikey has them.

Hi Joerg,

thanks for you hint. I checked toshibas transistor pairs:
Regarding their parameters they would be very fine but none
of them has the pinout I need:

BC847:

C1 B2 E2
+--------+
! !
!1 !
+--------+
E1 B1 C2



Toshiba's transistors all have the same pinout:


B1 B2 E2
+--------+
! !
!1 !
+--------+
C1 E1 C2

Applying a patch with wires is not really fun here
because of the size of the parts and because of
the bad RF properties of such wire-patches :-(
Else, how about using a couple RF transistors and glue their bodies
together for better thermal coupling?
Size of such a dual transistor is below 2mm and I have
got three of them in 1.5cm^2. I think applying such a patch would
be really challenging, too. But maybe it is worth a try before
redesigning the PCB...

What would be your choice for the transistor? Maybe in SOT23
package with teh following pinout?

C
+
!\
! \
! \
+---+
E B
Just keep in mind that their beta
will usually be quite low and so will their breakdown voltage.
Thats OK. The total gain of the stages is below 5 and breakdown
voltage of 5V would be fine.

Thanks for your hints,

best regards,

Erik.
 
J

Joerg

Jan 1, 1970
0
Hello Erik,

What would be your choice for the transistor? Maybe in SOT23
package with teh following pinout?

C
+
!\
! \
! \
+---+
E B



Thats OK. The total gain of the stages is below 5 and breakdown
voltage of 5V would be fine.

In that case maybe one of my old favorites, the BFS17. But your desired
pin-out is non-standard. There used to be a reversed version called
BFS17-R. I don't know why they came out with it, maybe because in the
early days lots of SMT layouts went wrong (and they still do...). I
haven't seen "-R" versions in a long time. So you may have to take a
regular one and carefully bend the leads backwards. Or use a non-SMT
transistor like the 2N2369.

But keep in mind, these are hot rods. Transistors with an fT in the GHz
range can easily become oscillators. Thing is, there ain't much around
in the "medium speed" range. Most popular transistors are either below
300MHz or above 1GHz.

Regards, Joerg
 
E

Erik Baigar

Jan 1, 1970
0
Joerg said:
Hello Erik,


In that case maybe one of my old favorites, the BFS17. But your desired
pin-out is non-standard. There used to be a reversed version called
BFS17-R. I don't know why they came out with it, maybe because in the
early days lots of SMT layouts went wrong (and they still do...). I
haven't seen "-R" versions in a long time. So you may have to take a
regular one and carefully bend the leads backwards. Or use a non-SMT
transistor like the 2N2369.

But keep in mind, these are hot rods. Transistors with an fT in the GHz
range can easily become oscillators. Thing is, there ain't much around
in the "medium speed" range. Most popular transistors are either below
300MHz or above 1GHz.

Regards, Joerg
OK, thanks a lot. I will order some BFR17 and give it a try!
I will keep you up to date!
 
R

Rich Grise

Jan 1, 1970
0
P.S. If someone is interested I'd be able to share my
information and documentation. I have some PCBs
left, too.

Interested??!?!?!!! Around here, that's _required_! ;-)

Cheers!
Rich
 
J

Joerg

Jan 1, 1970
0
Hello Erik,
OK, thanks a lot. I will order some BFR17 and give it a try!
I will keep you up to date!


It's BFS17, not BFR17.

Regards, Joerg
 
E

Erik Baigar

Jan 1, 1970
0
Joerg said:
Hello Erik,


In that case maybe one of my old favorites, the BFS17. But your desired
pin-out is non-standard. There used to be a reversed version called
BFS17-R. I don't know why they came out with it, maybe because in the
early days lots of SMT layouts went wrong (and they still do...). I
haven't seen "-R" versions in a long time. So you may have to take a
regular one and carefully bend the leads backwards. Or use a non-SMT
transistor like the 2N2369.

But keep in mind, these are hot rods. Transistors with an fT in the GHz
range can easily become oscillators. Thing is, there ain't much around
in the "medium speed" range. Most popular transistors are either below
300MHz or above 1GHz.

Thanks for the Tip - I patched one of my boards with two BFS17 instead
of
one BC847S. Additionally I replaced the 1N4148 with a half BAT68-06
which
has much lower capacity. Analysis with the network analyzer shows that
it is now nearly as good as the original 04KH009:

On http://www.baigar.de/electronics/04kh009-1.0a.pdf you can see the
plot.
In this plot the patched version (V2) is shown against the original
04KH009
and the unpatched, BC847-based version (V1). The original-Version has
been
shifted by around +2dBm. Transmission is sufficient now and much of the
difference at higher frequencies may be attributed to the missing
shielding
of the small replacement boards compared to the original 04KH009.

My plan now is to redesign the board for using transistors from the
BFS480/481/482 series since they are readily available. As well I will
put in the BAT68-6 since the board can be even smaller than allowing
me to apply some kind of shielding...

Thanks again,

Erik.



BTW: Applying the patches (i.e. 8 BFS17 transistors) was real pain since
SOT363 is very tiny and SOT23 is still small - but huge compared to
SOT363
;-)
 
J

Joerg

Jan 1, 1970
0
Hello Erik,


Looking pretty good there but you have some droop towards 500MHz. You
might need an even hotter transistor with a higher fT. Since you are in
Germany I'd look into Infineon. In the US I found it nearly impossible
to obtain samples so I don't use them anymore but that might be easier
in Europe. Right now they data sheet link don't work (happens a lot with
them, they seem to use some stupid java script...) but check the BFR
series. There are some above 10GHz fT. And really, really small.

Regards, Joerg
 
E

Erik Baigar

Jan 1, 1970
0
Joerg said:
Hello Erik,


Looking pretty good there but you have some droop towards 500MHz. You
might need an even hotter transistor with a higher fT. Since you are in
Germany I'd look into Infineon. In the US I found it nearly impossible
to obtain samples so I don't use them anymore but that might be easier
in Europe. Right now they data sheet link don't work (happens a lot with
them, they seem to use some stupid java script...) but check the BFR
series. There are some above 10GHz fT. And really, really small.

Regards, Joerg

I think the BFS480 are "hotter" than the BFS17 - at least a factor
of two. The oscilloscope under investigation is a C1-122 and the
other componentes have -3dB at around 150MHz so a drop at 750MHz
(as I expect in using the BFS48x) might be OK. In my opinon the SOT363
package is the smallest I want to solder by hand ;-)

If their site works again I will be happy to check out the BFR
series - maybe I will make two different PCBs to give them a try, too...

Regards,

Erik.
 
F

Fred Bloggs

Jan 1, 1970
0
in a project I am using the dual transistor BC847S in the
SOT363 package. This transistor has a maximum frequency of
200MHz according to the datasheet.
Does anyone know about a pin compatible replacement with
a higher frequency (lets say 500MHz). Gain and Uce are not
so important for me and maximum Ic of 50mA would be OK, too.

The project is that in my russan oscilloscope (C1-122)
there are hybrid switches used for chopping channel
and trigger signals. These switches (04KH009 or
04KN009) tend to die rather often and so I designed a
replacement using the BC847S. In principle it works
fine but the bandwith of the oscilloscope is severely
reduced since the BC847S is not well suited for HF
applications. But using transistors in such a small
package was mandatory to obtain th small form factor
to insert the replacement circuit in the oscilloscope.
This project is also relevant for the following russian
oscilloscopes used by lots of hobbyists: C1-112, C1-112A.

Any hints are welcome,

Erik.


The Russian stuff is probably using an inferior and corrupt design to
begin with. Rip their dumbass chopper and replace with cheap GaAsFET
logic controlled switches and possibly a video summing amplifier if you
need gain.
 
E

Erik Baigar

Jan 1, 1970
0
Fred said:
The Russian stuff is probably using an inferior and corrupt design to

The design is old (maybe between 1970 and 1980) but I'd not
say that it is corrupt. It was state of the art this time and
is found in contemporary tek oscilloscopes as well.
begin with. Rip their dumbass chopper and replace with cheap GaAsFET
logic controlled switches and possibly a video summing amplifier if you
need gain.
There is no gain needed in this application but the
impedance of the chopper is of importance since
directly after the chopper there is the circuitry
responsible for bandwidth selection. Using CMOS-Switches
would be difficult, too: The 04KH009 is designed fo differential
signals and shifts the level of both lines by several volts
from input to output side. I.e. replacing the 04KH009 by any
type of modern switch would require severe patches to the
vertical amplifier plugin I think. But the idea of using a
video summing amp for such purposes is a interesting hint
one should keep in mind. Do you have a type number, so I
can look at a datasheet?

Thanks for your ideas,

Erik.
 
J

Joerg

Jan 1, 1970
0
Hello Erik,
I think the BFS480 are "hotter" than the BFS17 - at least a factor
of two. The oscilloscope under investigation is a C1-122 and the
other componentes have -3dB at around 150MHz so a drop at 750MHz
(as I expect in using the BFS48x) might be OK. In my opinon the SOT363
package is the smallest I want to solder by hand ;-)

Just be careful with these. A GHz transistor can oscillate without you
even seeing it because none of the equipment is fast enough for that.

Regards, Joerg
 
E

Erik Baigar

Jan 1, 1970
0
Joerg said:
Hello Erik,


Just be careful with these. A GHz transistor can oscillate without you
even seeing it because none of the equipment is fast enough for that.

Do you really think, it will be a problem in my configuration?
The overall gain of my circuit is unity and at the 10mA of
DC current used, ft is 2G in tha case of the BFS17 and max.
8 in the case of the BFS482. And gain of both tranistors is
with around 13dB comparable, too?

I could design two versiobs of the board (one with BFS482s and
a second with lots of BFS17) but I think temperature drift is much
worser with the BFS17-solution. And this drift of DC-level
during warm up is the only "feature", which is inconvenient with
the russian oscilloscope.

Best regards,

Erik.
 
J

Joerg

Jan 1, 1970
0
Hello Erik,
Do you really think, it will be a problem in my configuration?
The overall gain of my circuit is unity and at the 10mA of
DC current used, ft is 2G in tha case of the BFS17 and max.
8 in the case of the BFS482. And gain of both tranistors is
with around 13dB comparable, too?

I have seen unity gain buffers oscillate. That was because some traces
weren't really shorts at high frequencies and the buffer developed some
'gain'.

I could design two versiobs of the board (one with BFS482s and
a second with lots of BFS17) but I think temperature drift is much
worser with the BFS17-solution. And this drift of DC-level
during warm up is the only "feature", which is inconvenient with
the russian oscilloscope.

I'd go with the BFS482 if it is better. We have a hard time getting this
stuff in the US. EU manufacturers aren't that great with engineering
samples. Usually you can't get them and so we often don't use their
parts. Temp drift can often be fixed with a tracking diode.

Regards, Joerg
 
Top