IC capable of driving 30MHz 10 Vp sin into 300pF

[Frank Mikkelsen]

Hi group
Can someone recoment a IC witch has the capability to drive 30MHz 10V p sin
into 300pF cap. load.

Thanks in advance

Frank

 

[Pooh Bear]

Hi group
Can someone recoment a IC witch has the capability to drive 30MHz 10V p sin
into 300pF cap. load.

An ic ?

That's a substantial current. Have you checked those numbers ?

Graham

 

[Frank Mikkelsen]

An ic ?

That's a substantial current. Have you checked those numbers ?

Xc for 300pF at 30MHz is app 17r give a peak current at 570mA.

My first ides was to use the EL2008 buffer from élantec. (BW 55MHz, 1A) but
it is obsolete, and I can't find any other types driving > 570mA.

Had found two other types that can be taken into consideration:

-BUF634T, but it is only capable of driving 250mA maybe two in parallel.

-AD815 but then I must use both amps in one case and a transformer to make
the output single ended.

Frank

 

[Winfield Hill]

Frank Mikkelsen wrote...

Pooh Bear skrev ...

Xc for 300pF at 30MHz is app 17r give a peak current at 570mA.

And you need to check the slew rate, 10Vp at 30MHz is 1885V/us.

BUF634T, but it is only capable of driving 250mA maybe two
in parallel.

Or three or four. Generally the slew rate capability will be
degraded when an amp is used near its maximum output-current
capability. Furthermore, when driving a capacitive load the
current maximum occurs as the output voltage goes through zero,
a situation that maximizes each amplifier's power dissipation.

Also, be sure to add a ballast resistor for each paralleled
output, say 0.5 to 1.0 ohm.

Even so, the 2kV/us slew rate is typical, not a minimum spec,
and furthermore is uncomfortably-close to your requirement.

AD815 but then I must use both amps in one case and a transformer
to make the output single ended.

In bridge form each amp only needs 942V/us, but that's still a
bit past the AD9815's specs. You can solve this with some turns
step-up in the transformer, and give yourself some good slew-rate
safety margin. Again, if you use two AD815s (four amps) there'll
be a better safety margin. Use separate feedback resistors for
each amp's gain setting, and also ballast the paralleled amps.
This will also help to prevent the AD815 IC from oscillating.

 

[Fred Bartoli]

Frank Mikkelsen wrote...

And you need to check the slew rate, 10Vp at 30MHz is 1885V/us.


Or three or four. Generally the slew rate capability will be
degraded when an amp is used near its maximum output-current
capability. Furthermore, when driving a capacitive load the
current maximum occurs as the output voltage goes through zero,
a situation that maximizes each amplifier's power dissipation.

Also, be sure to add a ballast resistor for each paralleled
output, say 0.5 to 1.0 ohm.

Even so, the 2kV/us slew rate is typical, not a minimum spec,
and furthermore is uncomfortably-close to your requirement.


In bridge form each amp only needs 942V/us, but that's still a
bit past the AD9815's specs. You can solve this with some turns
step-up in the transformer, and give yourself some good slew-rate
safety margin. Again, if you use two AD815s (four amps) there'll
be a better safety margin. Use separate feedback resistors for
each amp's gain setting, and also ballast the paralleled amps.
This will also help to prevent the AD815 IC from oscillating.

Hmm, is the 300p really 300p?

At 30MHz the OP can tune the reactive part out with about 94nH serial
inductance and adjust the voltage gain (Q) with a series resistor.

Say we want +/-10V excursion on the amplifier, that translates to Q=1,
R=17.7R and Io=563mA.
He can lower the voltage excursion requirements by setting a higher Q.
Say Q about 3 / R=5.6R then this translates to Vo=3.2V and Io unchanged.

 

[Winfield Hill]

Fred Bartoli wrote...

He can lower the voltage excursion requirements by setting a higher Q.
Say Q about 3 / R=5.6R then this translates to Vo=3.2V and Io unchanged.

Yes, and he can lower both the voltage and current requirements
by connecting his added inductor to ground, and driving a tap.
If he can employ resonance, it's a whole new ball game.

 

[John Larkin]

Hi group
Can someone recoment a IC witch has the capability to drive 30MHz 10V p sin
into 300pF cap. load.

Thanks in advance

Frank

That's a lot of current. Is it a fixed frequency? If so, resonate the
capacitive load with an inductor and that will reduce your drive
current requirement by maybe 10:1.

John

 

[Greg]

How about a CRT driver amp?
They can drive high currents at high frequencies.
Greg

 

[Frank Mikkelsen]

That's a lot of current. Is it a fixed frequency? If so, resonate the
capacitive load with an inductor and that will reduce your drive
current requirement by maybe 10:1.

Unfortunately I had to scan the frequency from 1 to 30MHz so there no way
out of driving the large current into the cap. load.

Frank

 

[[email protected]]

Yipes, that's quite a load. You're talking several watts of power to
be sure.

How's about a middling op amp with a couple RF power transistors tacked
on?

 

[Klaus Bahner]

Hi group
Can someone recoment a IC witch has the capability to drive 30MHz 10V p sin
into 300pF cap. load.

Difficult spec:

THS6182 has enough output current (1A), small signal bandwidth of 80 MHz
(not too much for my taste when operating at 30MHz), and slew rate is
too low, i.e. 450V/us. Maybe bridged?

OPA2677 offers 500mA output current and has a slew rate of 2000 V/us,
but has a maximum supply voltage of +/- 6.5 volt, allowing for an output
swing of a bit below +/- 5 volts, so 10 volts peak peak are probably not
possible.

Anyway, power dissipation is anyway a serious issue with these specs, so
you probably have anyway to come up with a bridged/parallelled solution.

HTH
Klaus

 

[John Larkin]

Unfortunately I had to scan the frequency from 1 to 30MHz so there no way
out of driving the large current into the cap. load.

Frank

OK, how about a bunch of medium-power fast opamps in parallel? Use one
to get the basic voltage swing, then a bunch of paralleled followers,
each connected to the load rail through smallish resistors, so they
don't fight one another too much.

TI has about the only fast, high voltage linear process around...

THS3061

THS3112

Both are available in the SO-8 PowerPad package, so can be heat sunk
pretty well. They *will* get hot.

John

 

[Ian]

OK, how about a bunch of medium-power fast opamps in parallel? Use one
to get the basic voltage swing, then a bunch of paralleled followers,
each connected to the load rail through smallish resistors, so they
don't fight one another too much.

TI has about the only fast, high voltage linear process around...

THS3061

THS3112

Both are available in the SO-8 PowerPad package, so can be heat sunk
pretty well. They *will* get hot.

John

Another candidate would be the THS3001, with 2 paralleled pairs of
2N4401/2N4403 type (or better) emitter followers helping the output current.
Connect a resistor from the opamp output to the load, connect the transistor
bases to the opamp output (possibly add stopper resistors in series), put
small resistors from each emitter to the load to equalise currents (and help
with stability). You'll need a better package than a TO92, at least an SO-89
or similar. A quick search throws up PXT4401, I don't know who makes those.

Regards
Ian

 

[Winfield Hill]

Frank Mikkelsen wrote...

Unfortunately I had to scan the frequency from 1 to 30MHz so
there no way out of driving the large current into the cap load.

What are you working on? Is it a sinusoidal output whose
frequency you scan, or some other kind of waveform? How
precise must the output be?

BTW, you could still take advantage of output resonance,
peaking-coil style, by placing a parallel inductor with
a series R, making a resonance at the highest frequency,
that's low-Q, but still helpful in reducing the current
and shifting its phase. The Q-spoiling resistor becomes
a simple resistive load at low frequencies. A series cap
can be used to remove or reduce its effect at DC and the
lowest 1MHz frequency region.

 

[Frank Mikkelsen]

Frank Mikkelsen wrote...

What are you working on? Is it a sinusoidal output whose
frequency you scan, or some other kind of waveform? How
precise must the output be?

I am working on a sweep generator to exercise an ion trap, used on a atom
physics experiments on cern (http://public.web.cern.ch/Public/Welcome.html).
The ion trap consists of 6 electrodes each having a capacity to ground of
300 pf. The design of the ion trap can't be changed and it is placed in a
vacuum chamber and the temperature must be close to 0 Kelvin -> no power
dissipation allowed in the chamber. The signal paths to each electrode are
precisely matched. The generator consist of six outputs producing the same
frequencies to each electrodes with a constant phase offset during the
frequency sweep, I need therefore a close phase match f(frq) for each output
stages, and therefore I attempt have as smoothly phase change as possible up
to 30 MHz.

The frequencies with constant phase offset during the sweep are generated
with six synchronize AD9954.

The requirement to phase match between the electrodes over a 1MHz - 30 MHz
sweep are not defined but the scientist wished around 1 deg which I am not
sure I can meet.



Frank

 

[Frank Mikkelsen]

Another candidate would be the THS3001,

I had missed these TI amps, they are fast! I take a a closer look - Thanks

with 2 paralleled pairs of

2N4401/2N4403 type (or better) emitter followers helping the output
current.
Connect a resistor from the opamp output to the load, connect the
transistor
bases to the opamp output (possibly add stopper resistors in series), put
small resistors from each emitter to the load to equalise currents (and
help
with stability). You'll need a better package than a TO92, at least an
SO-89
or similar. A quick search throws up PXT4401, I don't know who makes
those.

Is it possible to make the feedback loop stable when adding emitter
followers in the loop (when using op-amps with bandwidth that high) ?
Or is the output emitter follower uotside the loop ?

Frank

 

[Rich Grise]

I am working on a sweep generator to exercise an ion trap, used on a atom
physics experiments on cern (http://public.web.cern.ch/Public/Welcome.html).
The ion trap consists of 6 electrodes each having a capacity to ground of
300 pf. The design of the ion trap can't be changed and it is placed in a
vacuum chamber and the temperature must be close to 0 Kelvin -> no power
dissipation allowed in the chamber. The signal paths to each electrode are
precisely matched. The generator consist of six outputs producing the same
frequencies to each electrodes with a constant phase offset during the
frequency sweep, I need therefore a close phase match f(frq) for each output
stages, and therefore I attempt have as smoothly phase change as possible up
to 30 MHz.

The frequencies with constant phase offset during the sweep are generated
with six synchronize AD9954.

The requirement to phase match between the electrodes over a 1MHz - 30 MHz
sweep are not defined but the scientist wished around 1 deg which I am not
sure I can meet.

Look into rigid coax - I don't know what they call it these days, but you
should be able to find some - military equipment uses it all over the
place. If you have the six outputs, you could run rigid coax to each of
your electrodes, and, much like the way car buffs do exhaust headers, see
to it that they're all the same overall length.

Good Luck!
Rich

 

[artie]

Look into rigid coax - I don't know what they call it these days, but you
should be able to find some - military equipment uses it all over the
place. If you have the six outputs, you could run rigid coax to each of
your electrodes, and, much like the way car buffs do exhaust headers, see
to it that they're all the same overall length.

Good Luck!
Rich

141 semirigid line, terminated with SMA connectors, readily available
(eBay is a good source), but matching line lengths is a small part of
this prolblem.

The burr-brown buf634 will do the 2000v/us slew rate and 30 MHz bw to
about 250 mA. It's a pretty bullet-proof part.

The old LH0002H from natsemi is (only) 200 v/us, but can do 250-400 mA
and up to 30 MHz.

You might also check out the power op amps from Apex
(http://www.apexmicrotech.com) Apex have lots of app notes and sample
circuits. As an example, they have a nice app note on driving
capcitive loads (app note 25), going beyond slew rate and max current
calculations to looking at stability, power dissipation, and the like.

 

[Winfield Hill]

Frank Mikkelsen wrote...

Ian wrote...

Is it possible to make the feedback loop stable when adding emitter
followers in the loop (when using op-amps with bandwidth that high)?
Or is the output emitter follower uotside the loop?

I wouldn't waste time attempting it inside the loop, certainly not
with those parts. Outside the loop is a non-starter, unless you're
considering a class-A circuit.

I'd start by taking bench measurements on the current capability of
BUF634T buffer ICs at their full slew rate, to evaluate paralleled
opamps. BTW, Burr Brown (see AB-101) says series resistors aren't
needed because of the amplifier's internal 10-ohm output resistance,
but you may find some added external resistance may help it to deal
with a capacitive load. You may also find an R+C network to ground
helpful in reducing ringing.

I previously suggested evaluating inductor peaking of the capacitive
load near 30MHz to shift the phase of the current, or to reduce the
voltage swing at 30MHz, either of which can reduce the stress on the
amplifiers. Something along those lines may be part of your solution,
but a quick calculation reveals you may also have serious problems
involving inductance. For example, if you have a high-capacitance
electrode at the end of some coax, what the electrode sees at 30MHz
won't be what you're sending at the start of the coax. Also, simple
wiring may be a problem. Consider, even a modest 250nH of wiring
inductance resonates with 300pF at only 18MHz. Sheesh! One wonders
how your 300pF is constituted, and what inductances are involved.
Your first task should be to evaluate these issues. Often a spice
simulation can be helpful in evaluating a circuit with many complex
elements.

 

[colin]

How about a CRT driver amp?
They can drive high currents at high frequencies.
Greg

I think a similar aplication apeared here some time ago I sugested crt
driver, but voltage was much higher .
they are designed to drive lower capacitances but at much higher frequencies
and voltages so the current
should be within its capabilities, also you get 3 in a high power package
wich you can parallel or use
one per electrode etc...

Colin =^.^=