Using reverse biased zeners to correct for mosfett turnon voltagein a push pull stage

[legg]

Probably best to post a schematic here showing how you want to build it.
The risk with zeners like in the example you brought is that they charge
the gates fast but the discharge is slow. That does not make for a very
linear operation.

You don't want them in the actual drive path. Capacitive coupling
might help. As long as they don't DC-restore.....

The gate load, being capacitive, should return all charge as the
signal reverses polarity.RL

 

[mike]

I initially used BJTs as output instead of the mosfets, but as the
output load can spike to 6-10A easily at full speed and full swing i've
found that I achieve higher bandwidths with a fet output, at the cost of
some headroom and unfortunately also some crossover distortion.

Don't know if this is relevant. This month's edition talks about drivers.
http://www.ecnmag.com/search/site/piezomotors

I don't understand the objective. Is trajectory important? Or do you
just need
to get from one physical location to another fast and not worry about
the path from A to B? You can do some interesting things by
pre-calculating charge levels and stuffing in an approximate impulse
and let the loop take care of the details. Very messy...

Also, I don't think you have ANY room for a dead band. If the loop
opens, you're screwed.

 

[Fred Bartoli]

Joerg a écrit :

Probably best to post a schematic here showing how you want to build it.
The risk with zeners like in the example you brought is that they charge
the gates fast but the discharge is slow. That does not make for a very
linear operation.

Also, the threshold drifts with temperture and since the FETs are the
parts that will become the hottest that needs to be reckoned with.

Personally, I'd servo it. Since the piezo is a capacitive device it
cannot have a DC part in its current so it should be possible to lock
the quiescent current.

Only in class A operation, otherwise your average supply current depends
on the AC output current.

 

[Joerg]

The configuration i would try to build right now would be
https://www.circuitlab.com/circuit/trh76n/susprailamp/
with the zener voltages below the respective threshold voltages.

Hmm, that one is not very encouraging. It might work with capacitors
across the zeners but the opamp is not going to like to drive that.
You'll probably be better off completely rolling your own amp like Tim
suggested. If the Apex ones aren't fast enough that is pretty much the
end of the rope in terms of power/voltage/speed.

Another thing you could try since you already have an Apex amp is to try
to follow that with a pair of BJTs.

I don't exactly get what you mean by servo it, care to elaborate? Google
only turns up that servo seems to be the same as buffering.

Essentially an active circuit that sets the DC quiescent current. Since
temperature will make that drift away it needs a regular quiet interval
where the DC bias circuit measures the quiescent current and adjusts
itself. That would be the servo part. But Fred is right, DC can only be
measured while in full operation in a class A amplifier and this (or
shuold become) is class AB2. So you'd have to briefly stop the
microscope scan to adjust the DC bias.

 

[Joerg]

Joerg a écrit :
[...]

Also, the threshold drifts with temperture and since the FETs are the
parts that will become the hottest that needs to be reckoned with.

Personally, I'd servo it. Since the piezo is a capacitive device it
cannot have a DC part in its current so it should be possible to lock
the quiescent current.

Only in class A operation, otherwise your average supply current depends
on the AC output current.

Oui, j'ai eu tort. Was thinking of class A but this is class AB2 (or
should be). So that only leaves the option to regularly stop the
microscope scan. Probably not too bad since temperature drifts are slow.

Other options might be to gauge the DC from the input signal and bick a
low amplitude phase but that gets esoteric.

 

[Joerg]

It's funny, back in the 10MHz Tek days, I bet they would've loved having
some of the later TV tubes -- video output amps and such, huge
transconductance (some of them comparable with JFETs in current
capacity, gain and cutoff voltage, at a few times higher terminal
voltages). By my estimate, 7KY6 for example (which is usually pretty
cheap even these days) can do 5MHz bandwidth pretty easily, which means
they'd be able to do 10 or 20 with some effort.

But, the price for the cutting edge demands that one cannot wait, and
thus distributed amps and hybrid circuits were used.

But even those caught up, and the newest vertical deflection circuits
were quite involved indeed. Pull ups, pull downs, pulls up and down for
the pullups and pulldowns, etc. Double complementary double balanced
and who knows what else...

...Then they introduced distributed deflection plates and, with MCP,
pretty much finished the history of analog scopes.

One sad trend is that with the demise of the CRT the nice video driver
transistors also died away. Those would have been very useful in a
project like Adrian's.

 

[Joerg]

Uhh... i _really_ like the idea of using low frequency integrators to
correct dynamically for any threshold voltage. The two additional opamps
will introduce a tiny bit of noise, but really nothing to write home
about. This should get nearly completely rid of any THD. I'll try if i
can get this to run in my suspended rail configuration to make it swing
120V+.
Thanks a lot for the suggestion. If you're interested I'll post some
results and actual schematics when i get them.

It is a good idea but you have to make sure the capacitance to ground in
those isolated supplies is very minimal. Also, the slew rate in the
input signal must be limited because if the opamp-FET combos can't
follow fast enough, else ... PHUT ... *POP*

You could diode-clamp around the opamp to avoid excursions that would
exceed the abs max limits on its input. That should fix the startup
blow-outs and input pulse racing.

 

[Tim Williams]

One sad trend is that with the demise of the CRT the nice video driver
transistors also died away. Those would have been very useful in a
project like Adrian's.

I've got a couple pulls sitting around here, like the 20V 100mA 1GHz cascode
bottom device, or the 120V 100mA 1GHz device that sat atop it! Almost
always Sanyo parts, lovely specs.

Tim

 

[Joerg]

I've got a couple pulls sitting around here, like the 20V 100mA 1GHz
cascode bottom device, or the 120V 100mA 1GHz device that sat atop it!
Almost always Sanyo parts, lovely specs.

Probably all obsolete or "Not for new designs" :-(

But for Adrian that wouldn't matter since he only has to build one unit.

 

[Joerg]

John Larkin wrote:

[...]

The other piezo driver that I like is totem-pole high voltage
optoisolators, but that's too slow for this app. Unless....

This doesn't have to be an opto-isolator. You can also pump RF in the
GHz range into a tiny RF transformer, rectify on the other side and use
that to control the FET. Modulating the RF on the primary side is easy,
the signal into the modulator is your control signal. On the rectifier
side things must be fast and loaded so there will not be a slow decay.

The capacitance between the windings of such a transformer can be in the
low single-digit picofarads.

[...]

 

[Joerg]

Piezos tend to have high-Q resonances, though, and if the OP needs
speed, that generally means using a notch filter to keep the whole thing
from oscillating. Lower output impedance makes filters a lot easier to
build.

If you try to get high reaction speed out of a high-Q transducer that
usually ends with a shattered transducer and a puff of smoke wafting
away from the amplifier

Back in the palmy days, when I built the IBM SXM scanned probe
microscope workstation, I was using piezo bimorphs to excite cantilever
vibrations in the tip as well as doing the Z axis servo. That bimorph
had resonance at 30 kHz with a Q of about 30, meaning that with a 1-pole
rolloff, the loop BW had to be 1 kHz or less. Using a simple LC notch
got me a 10 kHz loop bandwidth, which was a big win.

How do they get fast movements if there isn't any stiff backing
material? The PZT transducers I dealt with (except for CW Doppler) all
had 6dB bandwidths between 30% and 60%. The only way to get there is to
have a lot of tungstens and stuff in the backing.

 

[Adrian Nievergelt]

Probably all obsolete or "Not for new designs" :-(

But for Adrian that wouldn't matter since he only has to build one unit.

Weell... make that two or three units at first. The circuit as proposed
by John Larkin seems to me to be the currently best option i have. I'll
be trying to combine that with the rest of the amp (voltage swing) and a
fast overall feedback and see if I can get a better amp. The power
handling capability of mosfets is after all quite stunning.

 

[Adrian Nievergelt]

It is a good idea but you have to make sure the capacitance to ground in
those isolated supplies is very minimal. Also, the slew rate in the
input signal must be limited because if the opamp-FET combos can't
follow fast enough, else ... PHUT ... *POP*

You could diode-clamp around the opamp to avoid excursions that would
exceed the abs max limits on its input. That should fix the startup
blow-outs and input pulse racing.

I am well aware of these problems as I already have a simpler version of
this physically running. Slew limiting i do simply with an RC lowpass at
the entrance. The startup i do using a relays to ground the input
initially and after the rails have been established (window comp)
together with a time delay (zener and RC) i turn on both rails
simultaneously. This already works reproducably.

 

[Adrian Nievergelt]

Piezos tend to have high-Q resonances, though, and if the OP needs
speed, that generally means using a notch filter to keep the whole thing
from oscillating. Lower output impedance makes filters a lot easier to
build.

Back in the palmy days, when I built the IBM SXM scanned probe
microscope workstation, I was using piezo bimorphs to excite cantilever
vibrations in the tip as well as doing the Z axis servo. That bimorph
had resonance at 30 kHz with a Q of about 30, meaning that with a 1-pole
rolloff, the loop BW had to be 1 kHz or less. Using a simple LC notch
got me a 10 kHz loop bandwidth, which was a big win.

Cheers

Phil Hobbs

Ah yes this is where the whole thing becomes increasingly interesting.
If you use dither piezos that are near MHz range with their resonances
you don't need much any drive circuitry, you need neither linearity nor
much power, as cantilever in air have a resonance Q of easy 300 or more.
I use cascaded PID loop control for the z feedback. The samples are
essentially mounted on a stack of piezos with increasingly higher
resonance frequencies and in turn decresing maximal deflections. This
scheme allows to boost z-feeback to well over 500kHz if done right, but
it takes a long time to set up and control for.
Also total z deflection measurements become a project of their own as
you'd have to sum up all single hysteresis-ridden deflections. The
current goal is using optical sensors instead.

 

[Jamie]

Weell... make that two or three units at first. The circuit as proposed
by John Larkin seems to me to be the currently best option i have. I'll
be trying to combine that with the rest of the amp (voltage swing) and a
fast overall feedback and see if I can get a better amp. The power
handling capability of mosfets is after all quite stunning.

But, don't get misled with those high amp ratings.. Heat is a killer
of MOSFETS.

Jamie

 

[Adrian Nievergelt]

But, don't get misled with those high amp ratings.. Heat is a killer
of MOSFETS.

Jamie

That's just the beauty of it all. Using piezos as fast nanopositioning
devices requires crazy currents of multiple amps at peak, but since you
don't have steadystate oscillations the actual power is limited. Granted
I will have to care for heat dissipation, but with the current amp
scheme i can scan around 100kHz feedback without the DPAK cased hexfets
going much above 30 maybe 35C even with deadbugging and without a heatsink.

 

[Joerg]

That's just the beauty of it all. Using piezos as fast nanopositioning
devices requires crazy currents of multiple amps at peak, but since you
don't have steadystate oscillations the actual power is limited. Granted
I will have to care for heat dissipation, but with the current amp
scheme i can scan around 100kHz feedback without the DPAK cased hexfets
going much above 30 maybe 35C even with deadbugging and without a heatsink.

Yeah ... but ... if it's swinging back and forth all the time you have
to make a proper time allowance in the SOA and then the power handling
capabilities are quite a bit reduced. Don't push it

 

[Adrian Nievergelt]

This deserves simulation with good part models. When one side swings
up, the opposite-side opamp may want to rail, so it may need some
diodes or fancier stuff. Ideally, one would keep some current flowing
in both fets all the time.

The opamps should *reduce* the noise of the fets!

That circuit is fun, but maybe this is better:

https://dl.dropbox.com/u/53724080/Circuits/Precision_Piezo_Driver.JPG

The impressive thing about this circuit is that neither fet ever turns
off, and the crossover behavior is essentially perfect.

This is a current source, and the piezo is a capacitor, so we have an
integrator that is the dominant pole of the control loop. That all
works out pretty nicely. This will be super-linear.

U1 needs to be fast and have low quiescent supply current, LT1217 sort
of thing. One can add a resistor+pot between the Q1-Q2 emitters to
trim fet quiescent bias. +-10 (or whatever) needs to be quiet.

Always design at least one circuit before breakfast.

I don't understand the circuit well enough yet, but isn't it a problem
attaching +-60V supplies this way with an opamp that can only take 30V?
The suspended rail configuration i drew earlier hardwires the rails to a
fraction of the supply range above and below the opamp. That one i've
already tested on board and it works really quite nicely. I might draw
the board flexible enough to try both schemes.
If i understand your first suggestion correctly, setting the dividers at
the input correctly will already keep each transistor always just
slightly conducting, eliminating any crossover distortion as well.

 

[Adrian Nievergelt]

In theory that looks good.
But..
Watch out for FET bias variations over temperature and the fact that
zener voltage variations over temperature go the other way..

From all i've gathered only for Vt > 5V, below zener diodes have a NTC
as have FETs.

 

[Jamie]

I don't understand the circuit well enough yet, but isn't it a problem
attaching +-60V supplies this way with an opamp that can only take 30V?
The suspended rail configuration i drew earlier hardwires the rails to a
fraction of the supply range above and below the opamp. That one i've
already tested on board and it works really quite nicely. I might draw
the board flexible enough to try both schemes.
If i understand your first suggestion correctly, setting the dividers at
the input correctly will already keep each transistor always just
slightly conducting, eliminating any crossover distortion as well.

The DC-DC converter is a isolated type and the common is being attached
to HV side output. This way the op-amp supplies will only see +/- 15
volts because they are basically riding ontop of the HV supply.. the
output will look like a common/ground to the op-amp no matter the level.

This poses a problem however, the common for your input signal..
Something just don't look right there.. SOmething tells me the inputs
also need to be isolated.. They do make isolated op-amps where the
common rail for the output side can attached in this manner and the
inputs can use a low voltage common.

Jamie