100V common mode diff front end amp?

[Grant]

My (borrowed) Tek DPO doesn't do differential inputs very well (got a
math module a-b mode, too noisy plus crappy common mode range).

So I think of doing a differential frontend end amp to take out up to
say 100V common mode, for looking at switchmode signals on top of usually
24V (21V-30V) battery stuff. Sometimes need to see 100V signals on that,
or at least tolerate the overdriven inputs to see signals of interest,
So I expect to put in dynamic range limiting to stop the opamps saturating
and getting 'stuck' in saturated mode. You know how they can take some
time to come back out of saturation? Or is that something no longer happen
with newer opamps?


Equipment could be chargers, motor drives and so on, switchmode, dirty
signals. Quite noisy, and I'm usually stuck with the ground reference
on other side of the battery pack. Normally battery is run isolated from
mains ground, so noise pickup is a problem. Also the safety issue of
grounding wherever the CRO earth goes.



So the topology seems most useful is the inverting front end for each
input, feeding a diffamp. Ideas for one of these? I tend to use the
DPO on 20MHz bandwidth filter, so I guess I'm not expecting too much
speed from the thing, I have cheap 60MHz probes for a normal 1M input
impedance, good enough?

Ideas for which opamp would be nice. Is trying for 1M input too easy
for noise pickup? I suppose I could have a couple flying leads with
clips going into the box with this thing, a coax lead out into the DPO's
input set to 1M input. Or you might convince me to use the 50R input,
though I don't think I want to go there. Last time I mentioned doing
that somebody tried to tell me my observed voltages would not be halved?

Thanks,
Grant.

 

[Spehro Pefhany]

We have a TPS2024, with all four channels and the trigger input
isolated. It's amazing for probing power stuff. You can hook a probe
ground clip anywhere, like on one side of a 240 volt AC line, and just
not care.

John

Does yours make you sign a disclaimer that you won't put more than 30V
on the probes every time you power it up? And with silly yellow
stickers on the probes reminding you... (until they get cut off).


Best regards,
Spehro Pefhany

 

[mike]

My (borrowed) Tek DPO doesn't do differential inputs very well (got a
math module a-b mode, too noisy plus crappy common mode range).

So I think of doing a differential frontend end amp to take out up to
say 100V common mode, for looking at switchmode signals on top of usually
24V (21V-30V) battery stuff. Sometimes need to see 100V signals on that,
or at least tolerate the overdriven inputs to see signals of interest,
So I expect to put in dynamic range limiting to stop the opamps saturating
and getting 'stuck' in saturated mode. You know how they can take some
time to come back out of saturation? Or is that something no longer happen
with newer opamps?


Equipment could be chargers, motor drives and so on, switchmode, dirty
signals. Quite noisy, and I'm usually stuck with the ground reference
on other side of the battery pack. Normally battery is run isolated from
mains ground, so noise pickup is a problem. Also the safety issue of
grounding wherever the CRO earth goes.



So the topology seems most useful is the inverting front end for each
input, feeding a diffamp. Ideas for one of these? I tend to use the
DPO on 20MHz bandwidth filter, so I guess I'm not expecting too much
speed from the thing, I have cheap 60MHz probes for a normal 1M input
impedance, good enough?

Ideas for which opamp would be nice. Is trying for 1M input too easy
for noise pickup? I suppose I could have a couple flying leads with
clips going into the box with this thing, a coax lead out into the DPO's
input set to 1M input. Or you might convince me to use the 50R input,
though I don't think I want to go there. Last time I mentioned doing
that somebody tried to tell me my observed voltages would not be halved?

Thanks,
Grant.

This is another case of "trivial in concept, but the devil is in the
details."

Take a look at the TEK 7A13 vertical amplifier.
Download the operator and service manuals. There are some interesting
CMRR vs frequency graphs in the manual. This designed by people who
knew what they were doing.

And just getting the signals into the amp without messing up the CMRR
is a challenge.

Also check out the TEK A6902 isolation amplifier. It's more a dual-channel
isolation amp, so you can use one probe and stick the common lead
anywhere you want. I've found it very useful in debugging switching
power supplies safely. Never did find the service manual tho.
If anybody has a link to the schematic, I'd like to have one in the
archives.

I've never tried it, but for stuff that doesn't have to be that accurate,
you might be able to make something like a wideband LM3900.
I've always been intrigued by the simplicity of huge voltage compliance
you can get with that topology.

 

[Nico Coesel]

My (borrowed) Tek DPO doesn't do differential inputs very well (got a
math module a-b mode, too noisy plus crappy common mode range).

So I think of doing a differential frontend end amp to take out up to
say 100V common mode, for looking at switchmode signals on top of usually
24V (21V-30V) battery stuff. Sometimes need to see 100V signals on that,
or at least tolerate the overdriven inputs to see signals of interest,
So I expect to put in dynamic range limiting to stop the opamps saturating
and getting 'stuck' in saturated mode. You know how they can take some
time to come back out of saturation? Or is that something no longer happen
with newer opamps?

Divide down and use an instrumentation amp to amplify. You'll need
0.1% or better resistors for the divider. I've used this trick to get
100mV signals from several hundreds of volts common mode noise.

 

[Spehro Pefhany]

No, we must have an older, pre-lawyer model.

John

Damn, I should ask for a downgrade in the firmware.


Best regards,
Spehro Pefhany

 

[Grant]

In the mid '80's I had a Tektronix probe device that provided complete
isolation so I could look at off-line switchers. Surfing I can't
locate such an animal anymore. You might look on ebay.

I looked and looked for the comms module*, somebody in Hong Kong offered
me one for USD800 a few years ago, I hate to think what they'd want for
a real diff-amp input module. I sort of know what you mean, one job
we had a good HP (I think) with a plug-in diff module, was wonderful.

*Anybody got hardware interface spec and schematic for Tek TDS3034?

Only outside comms it has is a parallel printer port, yes I think about
emulating a printer to capture an image, for when the internal floppy
drive breaks, as they all do eventually in fan cooled equipment.

Grant.

 

[Grant]

....

We have a TPS2024, with all four channels and the trigger input
isolated. It's amazing for probing power stuff. You can hook a probe
ground clip anywhere, like on one side of a 240 volt AC line, and just
not care.

Is that when running off battery? Cos surely the mains earth would
change things? We got 240V mains, a little more exciting than yours

I'm wondering if I can power the DPO from a 12V SLA battery, instead
of the original NiCd battery pack. Seems ballpark voltage might be
okay, if I don't expect the DPO to charge the SLA battery.

Ah, re-reading what you write, you got isolated probes? I've got cheap
USD6 Chinese probes for my TDS3034. No budget for 'real' probes.

Grant.

 

[Grant]

....
Check out the INA117. It has 200Vcm input range.
Downside is the input impedance is not high Z.

I don't need hi Z, and yes, it the one I keep coming back to, $10 one off,
not bad. Just had another look at the spec +/- 500V CM and diff voltage
protection, so it's likely best I'll find? Just make a quiet* +/- 15V
supply for it, or run from a pair of 9V batteries?

* As in not add crap from the mains, be nice if I could find an
electrostatic shielded mains transformer from an old tape recorder
or something? Or wind my own earth shield plus secondaries onto a
mains toroidal xformer? Wonder how bad the turns/volt is?

Grant.

 

[Grant]

Grant wrote:

[ideas for a diff amp?]
....

This is another case of "trivial in concept, but the devil is in the
details."

Take a look at the TEK 7A13 vertical amplifier.
Download the operator and service manuals. There are some interesting
CMRR vs frequency graphs in the manual. This designed by people who
knew what they were doing.

And just getting the signals into the amp without messing up the CMRR
is a challenge.

That's one issue, a couple flying leads, with miniature clip? If
I could some really flexible coax I'd use that for the input leads.

But then you start into guard drivers and stuff, how far does one go?

I'm now tempted by the INA117 since it's been seconded by someone here.

Also check out the TEK A6902 isolation amplifier. It's more a dual-channel
isolation amp, so you can use one probe and stick the common lead
anywhere you want. I've found it very useful in debugging switching
power supplies safely. Never did find the service manual tho.
If anybody has a link to the schematic, I'd like to have one in the
archives.

I've never tried it, but for stuff that doesn't have to be that accurate,
you might be able to make something like a wideband LM3900.
I've always been intrigued by the simplicity of huge voltage compliance
you can get with that topology.

"Performance as a dc amplifier in a single-power-supply mode is not as
precise as a standard integrated-circuit operational amplifier operating
from dual supplies." --TI datasheet.

I'm surprised the things are still stocked. Never did find a use for
them, like you I've always thought they might be good for something.

But what?

Grant.

 

[Grant]

....

It wouldn't be hard to do a matchbox-sized battery-powered probe sort
of thing with fiberoptic output. That would work a million volts off
ground.

What diff amp plus ADC -> optic to a receiver, DAC to DPO? I'd have to
run everything through such little boxes to match up the times... or
somehow delay all other signals, or, or?

Nice, but it has issues..

 

[[email protected]]

What diff amp plus ADC -> optic to a receiver, DAC to DPO? I'd have to
run everything through such little boxes to match up the times... or
somehow delay all other signals, or, or?

At the differential output, run an other amplifier stage driving an
optoisolator. To linearize the opto output, put an identical
optoisiolator in series on the input side and use the output from it
as feedback to the second amplifier stage.

By the way, if you are measuring ripple voltages riding on some large
DC voltages, do you actually have to have a frequency response down to
DC ?

If frequency response down to DC is not needed, why not use some DC
blocking capacitors, either directly at inputs or after a floating,
battery powered differential amplifier and the scope ?

Of course, some resistors and voltage swing limiting diodes may be
needed to handle the capacitor charge/discharge currents during
startup and power down.
to drive the signal to equipment at ground potential. On the
optoisolator input side, put an other optoisolator in series and
return the second optoisiolator output to the floating

 

[Grant]

It doesn't need to be a diffamp if it's battery powered and floating.
Right.

Maybe you could do FM at some huge carrier frequency, with a tiny
commercial VCO as the modulator.

Nah, too hard, INA117 looking better Besides, Mr TI said he'd send
me some samples last night

Grant.

 

[Grant]

At the differential output, run an other amplifier stage driving an
optoisolator. To linearize the opto output, put an identical
optoisiolator in series on the input side and use the output from it
as feedback to the second amplifier stage.

By the way, if you are measuring ripple voltages riding on some large
DC voltages, do you actually have to have a frequency response down to
DC ?

Practically, yes. In the sense that subtracting two AC coupled signals
in the scope doesn't work very well, the DC is moving about, and I want
to see the small and large signals on top of that.

Older scopes I worked with had enormous input offset ranges so one could
amplify a pair of signals and have the boring bit miles off screen, this
modern DPO thing has 8 units of vertical, and barfs at only 5 units of
offset, putting up a signal out of range error.

If frequency response down to DC is not needed, why not use some DC
blocking capacitors, either directly at inputs or after a floating,
battery powered differential amplifier and the scope ?

Well, isn't that the same as switching to AC coupled on the scope?

Hmm, tick tick, tick, what I need is isolated ground, failing that, a CRO
input lead where the -ve connector is a local 'ground' for the wanted
signal on the +ve connector. With a fairly good frequency range, some MHz.

So I describe that as needing a diff amp front end with a believable
unity gain and very good CM rejection up to +/-100V.

Of course, some resistors and voltage swing limiting diodes may be
needed to handle the capacitor charge/discharge currents during
startup and power down.

Yes, and at some point one wonders if they seeing the signal of interest,
or artifacts of the signal collection circuitry. That's why I like to
be careful how I do this, the result needs to be trustworthy.

to drive the signal to equipment at ground potential. On the
optoisolator input side, put an other optoisolator in series and
return the second optoisiolator output to the floating

Seems like too much in the signal path, some unknown delay times involved.

Grant.

 

[Grant]

On Sun, 24 Apr 2011 07:09:07 +0300, [email protected] wrote:

[snip]

At the differential output, run an other amplifier stage driving an
optoisolator. To linearize the opto output, put an identical
optoisiolator in series on the input side and use the output from it
as feedback to the second amplifier stage.

[snip]

I did that with isolated sensing on switching power supplies in the
early '80's. I vaguely remember issues with linearity over a large
dynamic range.

Yeah, back then with slow signals I had better luck transferring the
analog value as a duty cycle though the optocoupler, transfer was very
accurate.

I suppose now you'd call the technique self-clocked delta-sigma, or
something like that? Signal into an integrator that forced output
duty cycle to match, therefore recovery on other side of opto was
very close to original signal.

I used CMOS inverters for rail-to-rail buffers to get the accuracy.

Grant.

 

[Grant]

If the requirement isn't for high frequency, there are isolation
amplifiers mass-produced for this purpose.
A DigiKey search for "isolation amplifier" brings up seven pages...

Sorry, it is for high frequency, well the sort of signals you see when
switching ion the 20MHz filter on the 'scope.

I still think the INA117 idea is winning here. Like John L. said upthread,
I want to get rid of a large DC offset to see the interesting part of the
signal, with spending heaps, and not get a significant delay. Hmm, unless
I build these for all four inputs? That'll stop ground loops for starters.

Probing around power stuff is hairy, and accidentally shorting out stuff with
the ground leads a constant worry for me. Even when working with isolated
battery bank. As soon as I clip scope ground to something, I have a
potential Bang! waiting to surprise me. And if it wasn't mains earth, it's
the common ground between probes, sometimes shorting current sense resistors.

So a diffamp for each probe seems like it would suit me, and all channels
suffer same delay, even better. Suit most of what I'm doing at the moment.

The INA117 claims +/- 500V input protection, so it looks very good for this
problem.

Grant.



Grant.

 

[Grant]

Divide down and use an instrumentation amp to amplify. You'll need
0.1% or better resistors for the divider. I've used this trick to get
100mV signals from several hundreds of volts common mode noise.

That's what makes the INA117 attractive, resistors onboard trimmed to
0.01%, +/-200V CM, and +/-500V peak input protection on chip. Unity
gain for the diff signal, that'll suit me, it would have to have
reasonably low noise as well? Datasheet says: 25uV <10Hz and
500nV/rootHz, max 1mV offset. Yeah, it'll do

Have to layout a PCB if I want four of them, 'cos I want a little stamp
size module with short signal leads to diff amp, INA117 says it stable
up to 1000pF output, drive coax okay, straight into 'scope. Might work.

Grant.

 

[Ronnie Jäger]

Am 23.04.2011 17:31, schrieb Grant:

My (borrowed) Tek DPO doesn't do differential inputs very well (got a
math module a-b mode, too noisy plus crappy common mode range).

So I think of doing a differential frontend end amp to take out up to
say 100V common mode, for looking at switchmode signals on top of usually
24V (21V-30V) battery stuff. Sometimes need to see 100V signals on that,
or at least tolerate the overdriven inputs to see signals of interest,
So I expect to put in dynamic range limiting to stop the opamps saturating
and getting 'stuck' in saturated mode. You know how they can take some
time to come back out of saturation? Or is that something no longer happen
with newer opamps?

Why not buy a commercially available differential probe? They start
about 200 Euros, for example from Testec. They usually have a wide
common mode range and a switchable differential mode range (eg. 10:1,
100:1).
I'd at least consider buying one of these before frying my scope's
frontend or even myself with a self-tinkered circuit...

 

[Ronnie Jäger]

Am 24.04.2011 02:12, schrieb Grant:

...

Is that when running off battery? Cos surely the mains earth would
change things? We got 240V mains, a little more exciting than yours

The TPS scopes run from integrated LI-ION batteries or from an isolated
bench power supply (which is so crappy that it will make the scope
display several millivolts of noise when it is connected).

Ronnie

 

[Ronnie Jäger]

Am 23.04.2011 22:19, schrieb John Larkin:

It wouldn't be hard to do a matchbox-sized battery-powered probe sort
of thing with fiberoptic output. That would work a million volts off
ground.

Good idea - the Yokogawa ScopeCorders have something like this built in.
Unfortunately the bandwidth is quite limited at these models...

Ronnie

 

[Nico Coesel]

That's what makes the INA117 attractive, resistors onboard trimmed to
0.01%, +/-200V CM, and +/-500V peak input protection on chip. Unity
gain for the diff signal, that'll suit me, it would have to have
reasonably low noise as well? Datasheet says: 25uV <10Hz and
500nV/rootHz, max 1mV offset. Yeah, it'll do

Looks nice but I don't think the (creapage) distance between the pins
is large enough to actually work reliably at 200V (or more). I also
like to add a capacitor between the inputs of the instrumentation amp
to filter unwanted signals.