Where have all the video transistors gone? (needed for wideband amplifier)

[Laurent Lamesch]

The application: a discrete 50 Ohm wideband amplifier with
f-3dB >= 100MHz and max 10Vpkpk into 50 Ohm, 20Vpkpk into open circuit.

Topology is a current feedback amplifier with small signal wideband
transistors in the input stage, and the idea is to use fast CRT video
driver transistors for the I to U stage and for the push-pull output driver.

Sanyo had some very intresting devices in their 'Ultra-high Quality
Output Devices' brochure, like the 2SA1403/2SC3597 NPN/PNP pair with
Vceomax=60V, Icmax=500mA, fT=800MHz typ, or Philips had the
BFQ232/BFQ252 pair with Vcermax=95V, Icmax=300mA, fT>1GHz.

Unfortunately, it seems to me that the potential manufacturers of
video transistors which I have looked at (Sanyo, Philips, ST, OnSemi,
Fairchild, Rohm, Panasonic, NEC, Toshiba, Samsung) do not make any
suitable devices any more, if they have ever done so.

Do any CRT driver transistors exist which have some probability to be
available in the next few years?

Another option: Philips still makes some UHF NPN and PNP transistors
with fT=5GHz, Icmax=100mA, however Vceo is limited to 15 or 20V. I
tried to stack two of each of these to get a higher Vcemax, but the
frequency response of the stack isn't sufficient any more.

Still another option: In most datasheets, only Vceo is given. As the
transistors are not operated with open base in the application, how much
can Vceo be exceeded?

Thank you.
Laurent Lamesch

 

[Ancient_Hacker]

The application: a discrete 50 Ohm wideband amplifier with

Still another option: In most datasheets, only Vceo is given. As the
transistors are not operated with open base in the application, how much
can Vceo be exceeded?

Good surmise! I recently measured some humdrum MPSA42's, which IIRC
are rated at 300 volts max. Well with 1K ohms between base and
emitter, to bleed off leakage current, they ran just FINE with 800
volts on the plates, er, collectors. Every one I measured out of a
batch of 100 worked fine (AT ROOM TEMPERATURE) at 800 volts, most
broke down around 830 to 860 volts. I wouldnt recommend doing this as
a regular thing, for mil-spec job, or for a nuke-bomb trigger, but you
might be just fine pushing the ratings to say 30% below the measured
breakdown voltage.

Also fainter in my memory was my measures of some high-speed
transistors to replace those whiz-bang ones Tektronix used as vertical
output transistors. The specs were around 25 to 30 volts, but with a
small drive resistor, none broke down below 75 volts.

Note that this isnt necessarily a poor practice: for example, A LOT
of transistors in Tek scopes are "selected" more humdrum types, being
pushed waay beyond their rated limits of voltage, leakage and speed.

For example, the CRT grid blanking driver transistors in the 4xx
series of scopes are asked to switch about 120 volts in a very few
nsec-- if you look up the ratings for the 2Nxxxx part, the rating is I
faintly recall, 65 volts and 15nsec. They must have culled a LOT of
these transistors to find a few golden ones, or maybe they found a
batch that were way better than spec-sheet specs.

 

[Joerg]

Hello Laurent,

The application: a discrete 50 Ohm wideband amplifier with
f-3dB >= 100MHz and max 10Vpkpk into 50 Ohm, 20Vpkpk into open circuit.

Topology is a current feedback amplifier with small signal wideband
transistors in the input stage, and the idea is to use fast CRT video
driver transistors for the I to U stage and for the push-pull output
driver.

Sanyo had some very intresting devices in their 'Ultra-high Quality
Output Devices' brochure, like the 2SA1403/2SC3597 NPN/PNP pair with
Vceomax=60V, Icmax=500mA, fT=800MHz typ, or Philips had the
BFQ232/BFQ252 pair with Vcermax=95V, Icmax=300mA, fT>1GHz.

Unfortunately, it seems to me that the potential manufacturers of
video transistors which I have looked at (Sanyo, Philips, ST, OnSemi,
Fairchild, Rohm, Panasonic, NEC, Toshiba, Samsung) do not make any
suitable devices any more, if they have ever done so.

Do any CRT driver transistors exist which have some probability to be
available in the next few years?

With the probability of most CRT being available in the next few years
diminishing I guess the same will happen to video transistors. I used to
design around a BF type, forgot which one because it's been long ago. It
was in almost all European color TVs but now these are migrating to LCD.

Another option: Philips still makes some UHF NPN and PNP transistors
with fT=5GHz, Icmax=100mA, however Vceo is limited to 15 or 20V. I
tried to stack two of each of these to get a higher Vcemax, but the
frequency response of the stack isn't sufficient any more.

Still another option: In most datasheets, only Vceo is given. As the
transistors are not operated with open base in the application, how much
can Vceo be exceeded?

You might want to try FETs in common gate configuration. That's what I
did a few times for fast amps that needed to generate a few volts out
into a load. Mostly because there was less of a chance of them going
obsolete on my client since they were popular in switchers.

 

[Winfield Hill]

Laurent Lamesch wrote...

The application: a discrete 50 Ohm wideband amplifier with
f-3dB >= 100MHz and max 10Vpkpk into 50 Ohm, 20Vpkpk into open circuit.

Topology is a current feedback amplifier with small signal wideband
transistors in the input stage, and the idea is to use fast CRT video
driver transistors for the I to U stage and for the push-pull output driver.

Sanyo had some very intresting devices in their 'Ultra-high Quality
Output Devices' brochure, like the 2SA1403/2SC3597 NPN/PNP pair with
Vceomax=60V, Icmax=500mA, fT=800MHz typ,

These are still in production by Sanyo, and MCM will deliver
them with 1 to 2 month leadtime. IMHO, it's not so much the
high f_T that makes these transistors valuable, it's the low
output capacitance. These two go hand-in-hand, of course,
but if you concentrate on the capacitance and load impedance
issues, you may find other satisfactory parts.

 

[Laurent Lamesch]

Good surmise! I recently measured some humdrum MPSA42's, which IIRC
are rated at 300 volts max. Well with 1K ohms between base and
emitter, to bleed off leakage current, they ran just FINE with 800
volts on the plates, er, collectors. Every one I measured out of a
batch of 100 worked fine (AT ROOM TEMPERATURE) at 800 volts, most
broke down around 830 to 860 volts. I wouldnt recommend doing this as
a regular thing, for mil-spec job, or for a nuke-bomb trigger, but you
might be just fine pushing the ratings to say 30% below the measured
breakdown voltage.

Wow. Pretty impressive.

Also fainter in my memory was my measures of some high-speed
transistors to replace those whiz-bang ones Tektronix used as vertical
output transistors. The specs were around 25 to 30 volts, but with a
small drive resistor, none broke down below 75 volts.

Note that this isnt necessarily a poor practice: for example, A LOT
of transistors in Tek scopes are "selected" more humdrum types, being
pushed waay beyond their rated limits of voltage, leakage and speed.

Yes, all the part numbers starting with 153 IIRC. If Tek got away with
this method, why shouldn't I try...

For example, the CRT grid blanking driver transistors in the 4xx
series of scopes are asked to switch about 120 volts in a very few
nsec-- if you look up the ratings for the 2Nxxxx part, the rating is I
faintly recall, 65 volts and 15nsec. They must have culled a LOT of
these transistors to find a few golden ones, or maybe they found a
batch that were way better than spec-sheet specs.

Right, this may have been a very expensive process, apart from the fact
that you can only hope to find enough good parts in the future to keep
your production running.

Thanks a lot,
Laurent.

 

[Jim Thompson]

Wow. Pretty impressive.

[snip]

Not really. Look up BVCER.

...Jim Thompson

 

[Laurent Lamesch]

Hello Laurent,
....

With the probability of most CRT being available in the next few years
diminishing I guess the same will happen to video transistors. I used to
design around a BF type, forgot which one because it's been long ago. It
was in almost all European color TVs but now these are migrating to LCD.

This might have been something like BF869 or 871. These have of course
also disappeared.

....

You might want to try FETs in common gate configuration. That's what I
did a few times for fast amps that needed to generate a few volts out
into a load. Mostly because there was less of a chance of them going
obsolete on my client since they were popular in switchers.

Hmm, even small signal FETs like BSS138 (N channel) and BSS84 (P) seem
to have a somewhat large Coss (13pF for the BSS138, 10pF for the BSS84)
for this application. I will have to try this out.

Thanks a lot,
Laurent.

 

[Laurent Lamesch]

Laurent Lamesch wrote... ....


These are still in production by Sanyo, and MCM will deliver
them with 1 to 2 month leadtime. IMHO, it's not so much the
high f_T that makes these transistors valuable, it's the low
output capacitance. These two go hand-in-hand, of course,
but if you concentrate on the capacitance and load impedance
issues, you may find other satisfactory parts.

Thank you very much for the MCM hint, this is very valuable!

Of course I have been looking on www.sanyo.com and couldn't find
any video transistors. After some searching I have now found
http://www.semiconductor-sanyo.com/discrete/index.htm which seems
to be the correct address. Under 'High frequency device', there is a
'Transistors', then 'Video output transistors' category, however trying
to access it gives me currently a 'Server Application Error' message.
Oh well.

Thanks a lot,
Laurent.