Best way to match impedance between an amplifier and transducers

  • Thread starter rwooduk
  • Start date
743
55

Main Question or Discussion Point

I am trying to figure out the best way to match impedance values between an amplifier (50 Ohm) which powers a number of transducers (initially 50 Ohm but could change). For one transducer I have previously used a step up transformer (SUT) device between the amplifier and transducer. Now I have multiple transducers.

The signal from the amplifier (40 kHz, 2000 W, 50 Ohm) will be split by a power splitter which feeds 4 transducers. Would it be best to place an SUT between the amplifier and power splitter to match all of the transducers at once, or have some sort of variable capacitor for each transducer (after the power splitter)?

BrJSnx1.png


I will also have to use some device to measure the impedance values, I like the idea of the smith chart...

0FLl29c.png


But the instruments are quite costly. Are there any other instruments that would do something similar at 40 kHz?

Thanks for any advice.
 

Answers and Replies

berkeman
Mentor
56,131
6,156
between an amplifier (50 Ohm) which powers a number of transducers (initially 50 Ohm but could change)
Why would the input impedance of the transducers change? Are you worried about aging? Or maybe interactions between the transducers if they are spaced closely in a medium? Are these ultrasonic transducers, or some other transducer modality? Can you link to their datasheet? Do you know how closely they are matched, in case you want to drive them in series to increase efficiency?
 
367
245
Are there any other instruments that would do something similar at 40 kHz?
Since you are operating at high power (high S/N ratio) and at low frequency, you should be able to measure impedance with an oscilloscope and a current probe or with the addition of a known impedance into the circuit.
There are a lot of other ways using other instruments too.
 
743
55
Thank you for the replies.

Why would the input impedance of the transducers change? Are you worried about aging? Or maybe interactions between the transducers if they are spaced closely in a medium? Are these ultrasonic transducers, or some other transducer modality? Can you link to their datasheet? Do you know how closely they are matched, in case you want to drive them in series to increase efficiency?
I've worked with a single transducer before, they are notoriously difficult to match because their vibration / operation determines the impedance value. In air at low power they are great, but as soon as you place them in contact with solution / solid surface (think ultrasonic welding) the impedance changes. This is because the frequency of vibration is dampened or the internals heat.

In my previous work we were lucky enough to have an amplifier that gave a reflected power (RP) value, this shows the degree of mismatch between amp and transducer. We would then tweak the frequency to enable us to match the impedence. But, now I have no amplifier with RP value.

Yes, the changes in vibration of the surface (in this case a metal sheet) due to multiple transducers in contact will cause the transducers' impedance to change. I will be measuring the surface properties of the sheet, but envisage poor matching which may damage the amplifier (and reduce efficiency).

3uFUHGy.png


They are ultrasonic transducers. Actually they are closer to 20 Ohm, here is the data sheet:

https://www.allendale-ultrasonics.co.uk/media/downloads/ultrasonic-transducer-40-technical-sheet-v1.pdf
We are hoping to get some 50 Ohm transducers, but without being able to first match to the amp whilst in operation, we expect very poor matching.

Since you are operating at high power (high S/N ratio) and at low frequency, you should be able to measure impedance with an oscilloscope and a current probe or with the addition of a known impedance into the circuit.
There are a lot of other ways using other instruments too.
Thanks for the suggestion. I am not an electrical engineer though, please could you tell me where I would take the measurement and what I would be looking for?

edit I need to measure the impedance of the transducer during operation, not just its impedance which is already given.

Thanks again.
 
Last edited:
berkeman
Mentor
56,131
6,156
Wow, interesting problem. Given the situation, I don't think it's practical to try to series-connect the transducers to improve the efficiency. It seems like you will need an impedance matching network/module for each transducer, driven by your power splitter (like in your 2nd diagram in your OP).

Do you know if any manufacturer makes power matching modules that can sense the mismatch and adjust the impedance transformation dynamically? It seems like that's the kind of building block you need in order to feed power to each of the transducers. If nobody already makes such a module, you may need to figure out the best way to design and build your own. What do other people in the industry do in this situation? Just try their best to come up with a compromise matching network for the anticipated power level?
 
743
55
Wow, interesting problem. Given the situation, I don't think it's practical to try to series-connect the transducers to improve the efficiency. It seems like you will need an impedance matching network/module for each transducer, driven by your power splitter (like in your 2nd diagram in your OP).

Do you know if any manufacturer makes power matching modules that can sense the mismatch and adjust the impedance transformation dynamically? It seems like that's the kind of building block you need in order to feed power to each of the transducers. If nobody already makes such a module, you may need to figure out the best way to design and build your own. What do other people in the industry do in this situation? Just try their best to come up with a compromise matching network for the anticipated power level?
Indeed. This is why I wanted to match each individual transducer first by a method yet to be determined, then also have some integrated impedance matching device incase the impedance changes with operation.

The matching device I used previously was an SUT:

https://www.tcpowerconversion.com/rf-impedance-transformers
Which worked well for matching a single transducer to the amplifier, but it only worked when the amp gave an RP value. This way the frequency of the amplifier could be altered slightly to reduce the impedance and then the SUT could be switched to its most effective impedance selection value.

Other people in industry,as I understand, use either a computerised smith chart, filter network and / or tune their circuits.

I don't have the funds to buy an SUT for each transducer. I just need a way to measure the reflected power (RP) from a single transducer, then I can develop a matching circuit for it, repeat for each transducer, then use the SUT to match the sum of the lower impedance values.

But no idea how I would measure the RP whilst in operation (which gives some idea of impedance mismatch).

Thanks.
 
berkeman
Mentor
56,131
6,156
But no idea how I would measure the RP whilst in operation (which gives some idea of impedance mismatch).
In HAM radio, we use in-line SWR meters to ensure that we are getting a good match between our amp and feedline and antenna:

1574186664826.png


Maybe something like that can work for you. You'd want to check the bandwidth of the SWR meter to be sure you were getting one (well, several) that covers down to your 40kHz Ultrasonic frequency band.

You might also do some reading about HAM radio antenna tuner devices, since that is their purpose to match the output impedance of the amplifier to the feedline and antenna (the antenna may be well off of its natural resonance, depending on its size and configuration). I don't know much about them, but they may help you think of ways you can set up your system...

https://radiotransmitter.wordpress.com/2018/07/28/reviewing-and-improving-the-semi-automatic-antenna-tuner/
1574186932732.png
 
743
55
In HAM radio, we use in-line SWR meters to ensure that we are getting a good match between our amp and feedline and antenna:

View attachment 253062

Maybe something like that can work for you. You'd want to check the bandwidth of the SWR meter to be sure you were getting one (well, several) that covers down to your 40kHz Ultrasonic frequency band.

You might also do some reading about HAM radio antenna tuner devices, since that is their purpose to match the output impedance of the amplifier to the feedline and antenna (the antenna may be well off of its natural resonance, depending on its size and configuration). I don't know much about them, but they may help you think of ways you can set up your system...

https://radiotransmitter.wordpress.com/2018/07/28/reviewing-and-improving-the-semi-automatic-antenna-tuner/
View attachment 253064
Very interesting berkeman I will look further into this idea, thank you.
 
743
55
Btw for those who may find this in the future, I found a really good paper on impedance matching

inbisyd.png


I will be using their "Single Resonant Frequency Based Design" (see 3.1.) for matching except with a variable capacitor:

zdnFuhj.png
 
tech99
Gold Member
1,720
575
I am trying to figure out the best way to match impedance values between an amplifier (50 Ohm) which powers a number of transducers (initially 50 Ohm but could change). For one transducer I have previously used a step up transformer (SUT) device between the amplifier and transducer. Now I have multiple transducers.

The signal from the amplifier (40 kHz, 2000 W, 50 Ohm) will be split by a power splitter which feeds 4 transducers. Would it be best to place an SUT between the amplifier and power splitter to match all of the transducers at once, or have some sort of variable capacitor for each transducer (after the power splitter)?

View attachment 253003

I will also have to use some device to measure the impedance values, I like the idea of the smith chart...

View attachment 253004

But the instruments are quite costly. Are there any other instruments that would do something similar at 40 kHz?

Thanks for any advice.
What type of power splitter do you use? It may be absorbing reflected power.
 
Baluncore
Science Advisor
7,060
2,218
What are you doing with the 2kW of ultrasonic power?
How do you combine the four generated waves in the unspecified medium?

Do the four transducers need to operate in phase at the same frequency?
Or is it one frequency because they now share one 2 kW amplifier?

Do you modulate the power, or just run them flat out? in which case a dirt cheap H-bridge could do the job of the amplifier.

By adjusting individual frequencies they would be more immune to neighbours and they could be matched by changing frequency.
The sum of four different frequencies will eliminate deep nulls between the side lobes of the beam.

If they all drive one plate, as an array, at one frequency, matching the individual impedances will phase shift the signals, which will distort the combined beam and affect the tuning of the other transducers.

You may have to tune them mechanically by trimming their individual masses.
 
743
55
What type of power splitter do you use? It may be absorbing reflected power.
Just started working on that aspect today. In contact with this company...

http://uk.pasternack.com/power-dividers.html
The power splitters they provide are 50 Ohm, which is what I need, but I had not considered that they may adsorb the reflected power! I could potentially remove the power splitter and just tune each transducer direct to the amplifier, is that what you would suggest?

What are you doing with the 2kW of ultrasonic power?
How do you combine the four generated waves in the unspecified medium?

Do the four transducers need to operate in phase at the same frequency?
Or is it one frequency because they now share one 2 kW amplifier?

Do you modulate the power, or just run them flat out? in which case a dirt cheap H-bridge could do the job of the amplifier.

By adjusting individual frequencies they would be more immune to neighbours and they could be matched by changing frequency.
The sum of four different frequencies will eliminate deep nulls between the side lobes of the beam.

If they all drive one plate, as an array, at one frequency, matching the individual impedances will phase shift the signals, which will distort the combined beam and affect the tuning of the other transducers.

You may have to tune them mechanically by trimming their individual masses.
The 2 kW is split so it feeds multiple transducers, for example 50 W+ power dissipation per transducer. The high power allows us to add more transducers if we require. The four transducers must be at the same frequency because the frequency is given by the signal from the amp, which then split. In phase, I'm not sure exactly of the context of that question, but ideally the transducers would vibrate at the same time to get the metal sheet to resonate efficiently.

Thank you for the H-bridge suggestion, do you mean use a H-bridge as a splitter? We need a broad bandwidth, high power, RF power source.

Adjusting the frequency per transducer is a very good idea and something we are working on!

Do you think that having a variable capacitor circuit for each transducer would not work very well?

Thank you for your suggestions.
 
Baluncore
Science Advisor
7,060
2,218
The 2 kW is split so it feeds multiple transducers, for example 50 W+ power dissipation per transducer.
50 W * 4 = 200 W.
You will not identify why the plate must resonate, or the environment the plate is in, or what you use the energy for. Your understanding of the technical situation is incomplete, and I can't fill in the gaps due to insufficient information. I'm probably wasting my time here.
 
743
55
50 W * 4 = 200 W.
You will not identify why the plate must resonate, or the environment the plate is in, or what you use the energy for. Your understanding of the technical situation is incomplete, and I can't fill in the gaps due to insufficient information. I'm probably wasting my time here.
Apologies for any confusion. I agree my technical knowledge when it comes to electronics is limited, that's why I posted the question, to learn more. When I say 50+ W this could be greater than 50 W and will be dependant upon the power capability of the transducers. Also, as mentioned, we may wish to add more transducers to the plate. This is why I stated a value of 2 kW (the maximum power my amplifier has to offer).

The plate must resonate because I am studying the surface vibration behaviour. This will be done in air using an instrument such as a Micro epsilon optonCDT (https://www.micro-epsilon.co.uk/service/glossar/optoNCDT.html) for displacement of the plate in the micron range and an eddy current probe (http://www.monitran.com/products/eddy-current-probes-and-drivers) for mm range displacement. These are point measurements. I would like to go with 3D mapping, such as with a 3D Laser Scanning Vibrometer, but the equipment is very expensive.

Hope that helps.
 
Baluncore
Science Advisor
7,060
2,218
The plate must resonate because I am studying the surface vibration behaviour.
Then you will not require very much power at all.

The point that you select to excite the plate will be critical to the modes of resonance excited.

Larger amplitude vibrations will cause problems when the flexing plate vibrates against the solid flat face of the transducer. You may have to use a mechanical tapered transformer to reduce the footprint and focus the transducer onto a smaller patch.

Driving the plate with two transducers will probably be impossible to arrange as you will be unable to identify two points that do not result in interference when two or more transducers are mounted and tuned.

Matching the amplifier to the transducer will shift the phase of the ultrasonic signal. If you try to match two transducers, driven by the same amplifier, exciting the same plate, you will get weird interference patterns due to phase differences between the two matched transducers.

Some transducers have a bolt that sets the pre-compression of the piezo elements and holds it all together. By adjusting that bolt tension you can adjust the resistive component of the impedance at different frequencies. By loading the transducer with different masses you can trim the reactance at different frequencies. Those weights may be a metal shim disc placed on the back piezo element.
 
pbuk
Gold Member
1,159
231
...
If ever an Advisor's user id was better suited to a thread title I would be surprised!
 
743
55
Then you will not require very much power at all.
The plate is stainless steel around 1 inch thick and the aim is to cause substantial displacement. The term "sheet" in the diagram was misleading, I apologise.

The point that you select to excite the plate will be critical to the modes of resonance excited.
I agree

Larger amplitude vibrations will cause problems when the flexing plate vibrates against the solid flat face of the transducer. You may have to use a mechanical tapered transformer to reduce the footprint and focus the transducer onto a smaller patch.
The transducer will be pinned to the metal surface with a pushing force, similar to the principle used in ultrasonic welding.

Driving the plate with two transducers will probably be impossible to arrange as you will be unable to identify two points that do not result in interference when two or more transducers are mounted and tuned.
Matching the amplifier to the transducer will shift the phase of the ultrasonic signal. If you try to match two transducers, driven by the same amplifier, exciting the same plate, you will get weird interference patterns due to phase differences between the two matched transducers.
I agree, it will be difficult. I have the option to increase the distance between the transducer such that interference occurs less readily.

The impedance of the transducers will change during operation. This will be a difficult aspect to address.

Some transducers have a bolt that sets the pre-compression of the piezo elements and holds it all together. By adjusting that bolt tension you can adjust the resistive component of the impedance at different frequencies. By loading the transducer with different masses you can trim the reactance at different frequencies. Those weights may be a metal shim disc placed on the back piezo element.
This is very helpful. I had intended to use the centre bolt to fix the transducer to the surface and, as mentioned, to apply a pushing force to the transducer. I had not considered that by adjusting these forces the impedance could be better matched. Thank you!
 
sophiecentaur
Science Advisor
Gold Member
23,883
4,091
@rwooduk : I'm still a bit unsure about your experiment but, if you are looking at transverse waves on the plate and if you need a large amplitude as possible, did you consider putting them in a square configuration? There could be a lot less phase variation (1/3?) over the smaller array footprint.

As mentioned previously above, the wavelength of vibrations along the plates is relevant to the optimum spacing of the transducers. This has a lot in common with multi-element radio transmitting arrays in which mutual impedance between elements can have significant effects on the impedances and array pattern.

As you have a lot of power from just one transducer, why not use a more sensitive receiving system?
 
Baluncore
Science Advisor
7,060
2,218
Stainless Steel has a P-wave velocity of about 5790 m/s and a S-wave velocity of 3100 m/s.

At 40 kHz, λ = 5790 / 40000 = 144.75 mm. λ/2 = 72.4 mm. λ/4 = 36.2 mm.

The diameter of the transducer contact should be smaller than λ/10 = 14.5 mm.

The plate will act like a waveguide with a depth of 25.4 / 144.75 = 0.175 λ

The difference between the P and S wave velocity will make energy injection into the thick plate an interesting challenge.
 
sophiecentaur
Science Advisor
Gold Member
23,883
4,091
Stainless Steel has a P-wave velocity of about 5790 m/s and a S-wave velocity of 3100 m/s.
That's interesting but how thick does the plate have to be before there is no straightforward transverse (not bulk) wave - like a thick diaphragm? That would be a much lower wave velocity. Perhaps the energy coupled into that mode is only small for a comparatively light transducer? But the OP does mention "flexing of the plate"

To be honest, I think we need to know a lot more about the context and details of the experiment. Which of the three modes is relevant?
 
Baluncore
Science Advisor
7,060
2,218
To be honest, I think we need to know a lot more about the context and details of the experiment.
I agree. Secrets can be expensive when they obstruct the return flow of information.

The unspecified shape and dimensions of the plate outline will be critical. It may contain simple corner reflectors, or long path delays, like the original PAL ceramic plate delay lines.
 
sophiecentaur
Science Advisor
Gold Member
23,883
4,091
the original PAL ceramic plate delay lines.
Iirc, they used P mode which was launched into the edge of the plate. The OP's arrangement seems to be exciting a different mode.
Mention of PAL took me back and mention of delay lines made me recall the FIELD delay that was used in the old analogue 50Hz / 60Hz Field Store Standards Converters. They used a polygonal quartz crystal, about 20mm thick and perhaps 300mm across. There were multiple reflections to give a long enough delay. I handled one once and (before serious optical fibres were common place) it was amazingly optically clear to look through. But. again, the mode was 'obvious'. [Edit: ?]
 
Last edited:
743
55
Thank you for the replies.

I'm still a bit unsure about your experiment but, if you are looking at transverse waves on the plate and if you need a large amplitude as possible, did you consider putting them in a square configuration? There could be a lot less phase variation (1/3?) over the smaller array footprint.
Yes, we have different configurations to test, but thank you for this idea.

As mentioned previously above, the wavelength of vibrations along the plates is relevant to the optimum spacing of the transducers. This has a lot in common with multi-element radio transmitting arrays in which mutual impedance between elements can have significant effects on the impedances and array pattern.
Indeed, it was interesting for me to see the similarities! Especially if one of the transducers hits resonance then this will act as a short and the other transducers would no longer receive the signal! I am working on this aspect.

As you have a lot of power from just one transducer, why not use a more sensitive receiving system?
Not sure what this means, there is no receiver.

The difference between the P and S wave velocity will make energy injection into the thick plate an interesting challenge.
I agree, thanks again for your interpretation.

To be honest, I think we need to know a lot more about the context and details of the experiment. Which of the three modes is relevant?
I agree. Secrets can be expensive when they obstruct the return flow of information.
The unspecified shape and dimensions of the plate outline will be critical. It may contain simple corner reflectors, or long path delays, like the original PAL ceramic plate delay lines.
The test sheet / plate will initially be 1m by 1m and around 1 inch thick, it will be square in shape.
 
sophiecentaur
Science Advisor
Gold Member
23,883
4,091
The test sheet / plate will initially be 1m by 1m and around 1 inch thick, it will be square in shape.
OK, thanks. And what mode will you be studying?
there is no receiver.
The equivalent to a receiver will be your measuring system. It will have a noise figure, a frequency response and some gain - all the same features.
 

Related Threads for: Best way to match impedance between an amplifier and transducers

Replies
16
Views
2K
  • Last Post
Replies
8
Views
816
  • Last Post
Replies
3
Views
1K
  • Last Post
Replies
2
Views
1K
  • Last Post
Replies
4
Views
9K
  • Last Post
Replies
3
Views
3K
Replies
12
Views
1K
Top