Using a 40 dB, 50 Ohm attenuator on my Scope

[manytoolsmike]

Hello EEVryone,

I am using an acoustic emissions sensor to measure ultrasound (10 to 100 kHz). The sensor is a piezo-electric transducer, I attached the datasheet below. The sensor connects directly to my scope's input via BNC. My problem is that the signal can be too high for my scope to handle and there is no switch (10x / 1x) like you would find on a probe. I'm guessing the voltage would peak at 15 V and the power is very low. I think 10x (20dB?) would be perfect.

I was surprised to find nothing about choosing and using attenuators on Youtube or EEV blog. (except for high power RF measurement). I am looking at the Rigol RA5040K 40dB Attenuator. I assume this would mean the signal would be attenuated 100x, which is more than I would like. I often have millivolt and volt level signal in the same measurement, the millivolt signals might get lost in the noise.

I am also not sure what the effect of the 50 Ohm impedance would be. Because the signal has such low power I am concerned the 50 Ohms will essentially act as a short and kill the signal. I'm assuming the impedance of a normal probe is in the Mega ohms even when the 10x attenuation is selected.

One "solution" is to cut the BNC off a probe that has the attenuation switch in the BNC side and connect that to my sensor. So how do I attenuate the signal without chopping the BNC connector off my expensive Acoustic Emissions sensor? Is there a high impedance 20 dB (10x) BNC attenuator?

Thanks for helping,

Mike

[nctnico]

Just use a standard 1:10 probe with a BNC connector conversion adapter. AFAIK probes from Testec come with the BNC adapter.

[manytoolsmike]

Just use a standard 1:10 probe with a BNC connector conversion adapter. AFAIK probes from Testec come with the BNC adapter.

You mean a BNC to Banana jack? I use this setup every day so I am looking for a cleaner solution. If it was a one time measurement I would use a probe on a BNC adapter.

I have found this, it appears to be what I want. I am surprised this is so hard to find.

[tggzzz]

I am looking at the Rigol RA5040K 40dB Attenuator. I assume this would mean the signal would be attenuated 100x, which is more than I would like.

It will, if and only if the source and load impedances are 50ohms. If your scope's input was 50ohms, you would know about it

To find out whether the source's voltage would damage the scope, look at the scope's spec. A standard *10 probe would give you an extra factor of 10 (duh!), but be careful of the probe's voltage vs frequency spec. Using a *10/*1 switchable probe will, sooner or later, produce "disappointing" results.

[nctnico]

Just use a standard 1:10 probe with a BNC connector conversion adapter. AFAIK probes from Testec come with the BNC adapter.

You mean a BNC to Banana jack? I use this setup every day so I am looking for a cleaner solution. If it was a one time measurement I would use a probe on a BNC adapter.

I have found this, it appears to be what I want. I am surprised this is so hard to find.
(Attachment Link)

That TA197 seems to do what you want but it has a 20MHz bandwidth.

No, I didn't mean BNC to banana; the silver adapter in this picture:

[DL2XY]

A (piezo) ultrasonic transducer is a charge shifting device. It's gain and frequency response depends purely on (complex) termination impedance.

So changing from direct scope input ( 1Meg/100pF) to 10:1 Probe (10Meg/10pF) will have little effect, since the transducer output voltage will be about 10 times as high (depending mostly on the actual capacitive load).

The proper way (in terms of calibration and precision) would be using an appropiate signalconditioner (charge mode preamplifier).

   

[justme1968]

nctnico meant something like this: https://www.ebay.com/itm/184090742816

[manytoolsmike]

A (piezo) ultrasonic transducer is a charge shifting device. It's gain and frequency response depends purely on (complex) termination impedance.

So changing from direct scope input ( 1Meg/100pF) to 10:1 Probe (10Meg/10pF) will have little effect, since the transducer output voltage will be about 10 times as high (depending mostly on the actual capacitive load).

The proper way (in terms of calibration and precision) would be using an appropiate signalconditioner (charge mode preamplifier).

I know the usual high impedance probes would not effect the quality of the signal, it is the 50 ohm attenuators that I am concerned about. It seems that 50 ohm attenuators are the norm for scopes. The only high impedance attenuator I have found is marketed for automotive use. I think I will end up making a solution by chopping off a suitable probe. Although the probe adapter shown by others is a good solution as well.

[DL2XY]

A (piezo) ultrasonic transducer is a charge shifting device. It's gain and frequency response depends purely on (complex) termination impedance.

So changing from direct scope input ( 1Meg/100pF) to 10:1 Probe (10Meg/10pF) will have little effect, since the transducer output voltage will be about 10 times as high (depending mostly on the actual capacitive load).

The proper way (in terms of calibration and precision) would be using an appropiate signalconditioner (charge mode preamplifier).

I know the usual high impedance probes would not effect the quality of the signal, it is the 50 ohm attenuators that I am concerned about. It seems that 50 ohm attenuators are the norm for scopes. The only high impedance attenuator I have found is marketed for automotive use. I think I will end up making a solution by chopping off a suitable probe. Although the probe adapter shown by others is a good solution as well.

A high impedance attenuation probe will not solve your problem. 

Please understand  https://en.wikipedia.org/wiki/Piezoelectric_sensor and for solution https://en.wikipedia.org/wiki/Charge_amplifier.

The transducer is not a voltage source but a charge source, so its output voltage is proportional to load resistance.
The 10x Probe has 10 time the resistance, so voltage is 10 times as high, divided by probe factor 10 would be  the same scope input voltage as without 10x probe!

You may try a 50 passthrough terminator, but this would give you only a view mV  with first order highpass response.

A quick solution would be a capacitor in parallel to transducer with a capacitance of about 10 times the transducer capacitance. That would reduce voltage by a factor of about 10 at reasonable flat frequency response.  There are still drawbacks in frequency response due to scope input resistance, so don't trust the datasheet gain chart too much. 

[xmo]

Maybe this?

[xmo]

The sensor manufacturer has this:

[manytoolsmike]

The sensor manufacturer has this:

I already have their preamp. when the signal is very strong it overwhelms the preamp. For my application the signals are so strong that preamps are pointless. The signals are between 50 mV and 15 Vac with very high dynamic range.

[tggzzz]

The sensor manufacturer has this:

I already have their preamp. when the signal is very strong it overwhelms the preamp. For my application the signals are so strong that preamps are pointless. The signals are between 50 mV and 15 Vac with very high dynamic range.

So, 300:1. You may need to be aware of a digitising scope's ENOB (effective number of bits) and how that affects your measurements.

[DL2XY]

The sensor manufacturer has this:

I already have their preamp. when the signal is very strong it overwhelms the preamp. For my application the signals are so strong that preamps are pointless. The signals are between 50 mV and 15 Vac with very high dynamic range.

The most straighforward solution is to use a shunt capacitor and direct connection to scope. Choose the capacitance N times higher than Csource of tranducer gives an attenuation by N. Measure Csource at center of band.

Second possibility is to modify the modify the existent preamplifier. If it is a charge mode amp modify Cfeedback.
But you may run into restrictions:
1. The used OPAMP may not be able to deliver the transducers peak current at hight levels.
2. Dynamic range may be insufficient  due to OPAMP noise figure.

Third way: Build your own signal conditioner according to your needs.

hth Walter

[manytoolsmike]

The sensor manufacturer has this:

I already have their preamp. when the signal is very strong it overwhelms the preamp. For my application the signals are so strong that preamps are pointless. The signals are between 50 mV and 15 Vac with very high dynamic range.

So, 300:1. You may need to be aware of a digitising scope's ENOB (effective number of bits) and how that affects your measurements.

Im pretty good at using my scope to capture these complex signals and Im aware of how the scope's ADC and sample rate limit what I can analyse. Im probably pushing the limits of the Rigol DS1000 series in ways few have before. I routinely capture 6 second long signals with dozens of frequencies and a wide dynamic range. attached are a few screen shots of what Im measuring. The first and second screen shots are of the same capture. Im analysing deep into the 6 second capture to view discreet wave forms.  In the third shot Im running a sweep on with my Signal generator to pinpoint the resonant frequency of a transducer.

I am looking into signal conditioners, but it looks like they will only cause me more headaches. I will likely change the physical setup of my test jig to attenuate the signal. This problem occurred when I switched to a more sensitive Transducer. Im operating the Transducer way out of its intended purpose, but it is giving reliable and repeatable measurements.