EVM430-FR6047: Cannot get the TIDA-01486 amplifier board working

Hi,

1. For no transmit signal, could you figure out is the transmit signal is coming out from the MSP430 MCU and breaks at somewhere in the external circuit or the MSP430 MCU does not even generate excitation pulses？

2. For the current consumption, have you checked the transducer spec? Can the transducer work at +-15V?

Best regards,

Cash Hao

Hello Cash. Thanks for your reply.

I just checked CH0 and it doesn't look like a transmit signal is even coming from the MSP430. Is it possible that I damaged the transmitter? My evm seems to still be functional in that it can connect to the USS software, so the MSP430 is still operational.

The transducers I'm using are Audiowell US0072. The datasheet says that the maximum drive voltage is 5 Vp-p, but Audiowell said that the absolute maximum drive voltage can be greater than 5 Vp-p but that the lifetime would be reduced. In this case, the absolute maximum drive voltage should be less than 24 Vp-p. I checked the operation of the transducers with a function generator and it looks like their are still working. I am re-confirming with Audiowell that the US0072 transducers can be driven upto 24 Vp-p.

Please disregard my last comment about not measuring the transmit signal out of the evm. I was using the wrong scope probe on my oscilloscope. I restored my evm to it's original form just to verify that I hadn't burnt out the transmit path and made the discovery about the scope probe then. I will reconnect the TIDA-01486 to my EVM and check the transmit signal again.

The comment about the transducers is still valid.

Hi,

Get it. Waiting for your updates on the board test. For checking the transducer, you can try measure impedance of transducer Check whether it still in a valid range or compare to a normal transducer. It can indicate it is damaged or not. 

Best regards,

Cash Hao

I was able to measure the transmit signal yesterday. It was odd, though. Let me explain...

I had the system powered up and the 2 op amps in the receive path were still getting really hot (too hot to touch) as was before. But there was no trasnmit signal. Then suddenly the transmit signal appeared and at the same time the current drawn from the external +/-15 V power supplies decreased from 150 mA to about 90 mA. In this case only 1 of the op amps in the receive path was still too hot to touch. I'm guessing that the other receive path op amp finally burnt out.

In summary, I can now measure the both CH0 and CH1 transmit signals, which are +/- 9 V (for a total of 18 Vp-p). this seems a little low, but I can measure it. I believe 1 receive path op amp burnt out while the 2nd is still hot. I expect it will burnt out soon, too.

Is there a proper power up sequence to use with external +/-15 V power supply?

I'm still looking for why the external power supply current is too high.

There's a typo in the THS3095 datasheet, by the way. Page 18, section 8.3.1, Equation 2 should be (where PD is supposed to have the NOT line above it):

PD >= REF + 2.0 V for enable

Hi,

Could you share the schematic of your board? 

And what is the receive signal level on the CHx_IN pin as the receive op amp input? If the receive op amp is hot, I would recommend to check the voltage level on all the op amp pins. Check if they exceed the op amp spec. 

Best regards,

Cash Hao

FYI, I confirmed with the my transducer manufacturer that I can drive my transducers up to 24 Vp-p.

I have attached the trace cuts I made and the jumper wires I added to drive the control signals to the appropriate voltage levels.

The trace cuts were necessary to disconnect the traces from the MSP430 IOs that drive those control signals. Then, I hardwired the control signals in accordance with Table 2 of the TIDA-01486 User Guide document (TIDUDO7) on page 6. For reference, the control signals are hardwired to use transducer pair 1 as follows:
S1 = 1 - connected to the 3.3 V power rail
S2 = 0 - connected to the GND power plane
S3 = 1 - connected to the 3.3 V power rail
S4 = 0 - connected to the GND power plane

PD1 = 1 - connected to the 3.3 V power rail
PD2 = 0 - connected to the GND power plane

The schematic I am using is TIDA-01486E1_Schematic.SchDoc, and is populated as shown in the schematic. That means all the op amps are set to a gain of 10.

I have a jumper plugged onto J6 of the TIDA-01486 to connect the VBIAS rail to the 3.3V rail from the MSP430-FR6047 evaluation module.

I can now measure the amplified 1 MHz ultrasonic drive signals from both op amps U1 and U2 in the transmit path.

The op amps U5 and U6 in the receive path are having problems. A few days ago both op amps were getting too hot to touch. Now, U5 is no longer getting hot - I think it has failed and I'll show why with some data below.

I believe U5 is burnt out because the node CH0_IN is measuring -15 Vdc, and the output of U5 measured at pin 6 is also -15 Vdc.

For U6, the input at node CH1_IN is -3.6 Vdc, and the output measured on pin 6 is -15 Vdc. This is why the op amp is getting hot - -3.6 V * 10 = -36 V and the op am output is being driven hard to the negative rail. But I can't figure out WHY CH1_IN is -3.6 V. Remember, VBIAS is set to 3.3 V so all the inputs to the receive path op amps after the VBIAS voltage divider should have a +0.658 V offset.

Before U5 burnt out, I suspect it was behaving the same way as U6. That is, node CH0_IN was probably at -3.6 Vdc.

Also, the REF pin of all the op amps is set to GND, and the PD pin of both U5 and U6 is set to 3.3 Vdc.

I am going to try replacing U5 with a new op amp.

Hi,

Do you have an oscilloscope to capture the signal on the CHx_IN? 

If the transducer can work under 24 Vp-p, it could be the receive signal is too strong and break the circuit. 

Best regards,

Cash Hao

The input levels in the receive path I mentioned before were stated as Vdc. The implication was that the insputs were a steady -15 V and -3.6 V on CH0_IN and CH1_IN, respectively.

That was for zero signal into the transducer. In this case, with no AC receive signal, the inputs to the op amps should be 0.658 V as set by the 3.3 V VBIAS rail through the voltage divider, right?

CH1_IN is at -15 Vdc because, I believe, the U5 op amp has failed. But why is CH1_IN, the input to U6, at -3.6 Vdc with no AC signal?

Here are the scope captures (CH0_IN on the left and CH1_IN on the right):

I think I have this system working. I am working on the bench right now, and transmitting the acoustic wave through a block of plastic. It's not the intended application but I think now I can see the transmit signal from one transducer and the resulting receive signal on the opposite transducer.

I must have damaged the receive path op amps when I was trying to find the correct power up sequence, which is to switch on the +/-15 V supply first, then plug the USB cable into my laptop to power up the EVM430 that supplies the 3.3 V power for the switching.

Anyway, I removed the 2 op amp chips in the receive path and put new chips on my board. Now the current consumption from the +/-15 V supply appears to be correct (40 mA), and I belive I can see the receive signals now.

Monday is the Canadian Thanksgiving holiday so Tuesday I'll put the transducers on the water pipes and see if I can measure  some water flow.

Hi,

What changes do you made and get the correct receive signal? By using a plastic instead of the transducers? Emm, interesting.

Best regards,

Cash Hao

I am using only 1 pair of transducers. I had to replace all the op amps on the TIDA-01486 with new op amp chips.So that's 2 opa mps in the transmit path and 2 op amps in the receive path.

To test to see if I was getting a receive signal, I attached the transducers to the opposite faces of a block of plastic so that T2 would capture the transmit signal from T1, and T1 would capture the transmit signal from T2. As shown in this picture:

The purpose of this was just get the TIDA-01486 working.

Today, I attached the EVM and the TIDA-01486 to transducers clamped in the V configuration to a 1 inch PVC pipe As shown here:

The gain in both the transmit and receive paths is 10. I know it's too large for the 1 inch PVC pipe, but I just want to get this thing working and I'll fine tune it later. This test should work, assuming my TIDA-01486 is working, because this setup works with no TIDA-01486. The signal I'm getting is this:

The signal I should be getting should look similar to this, which is Figure 21 from TIDUDO7

Clearly, I still have something wrong. Is it possible to over drive MSP430? Would would that look like in the ADC Capture window of the USS?

The TIDUDO7 docuement decribes some tests that were done with the TIDA-01486 on a 10-inch pipe. Can you tell me what material the pipe was made from? Was it PVC or metal?

Hi,

Okay, you are working on a clamp on meter. Clamp on meter needs special transducer and gel between the transducer and pipe surface. It would help you to get correct receive signal. 

With your current setups, I do not think it can work on the clamp on meter. When you run your test on your V configuration pipe, have you filled water inside the pipe? And you can modified your plastic into below one and check if you can get the correct receive signal. 

Best regards,

Cash Hao

The transducers I am using may not be optimal for a higher drive voltage, but the manufacturers confirm that they will work but the mean time to failure will be shorter. This is not an issue.

Yes, I know about the ultrasonic couplent between pipe and transducer. I am using industrial grease. This is not an issue.

Yes, I make sure the pipe is full of water and that all the air has been flushed out of the system before attempting to make ultrasonic measurements. This is not an issue.

There is something wrong with the TIDA-01486. Today, I am going to use a signal generator and an oscilloscope to check the transmit and receive paths.

Can you please tell me what the pipe material was in the experiments described in the TIDUDO7 document that were performed on the 10-inch pipe that used the TIDA-01486?

I spent the day in my lab with just the TIDA-01486 board. I'm using one power supply to power the 3.3 V rail for the AND gate and the multiplexers, and 2 more power supplies for the +/-15 V for the op amps. I'm using a signal generator set to 1 MHz and at various magnitudes which are all less than 800 mVp-p. I'm using a scope to make the measurements.

The summary is that both transmit and receive channels appear to behave the same. And each channel on its on is amplifying the input signal by a factor of 10 since all the gains on my TIDA-01486 are currently set to 10.

I am driving the receive path with a square wave from the signal generator. The receive path op amps output a square wave but larger in magnitude by a factor of 10. But the amplified receive signal is converted to a sawtooth wave by the clamping diodes. This sawtooth wave is what will be sent to the MSP430FR6047. Is the MSP430FR6047 expecting a square wave or can it work with a sawtooth?

Hi,

On the original TIDA-01486 is used for invasive large pipe. It does not tested for clamp on meter. We have another application note which is design for clamp on meter. https://www.ti.com/lit/pdf/slaa949 I believe you already aware of this. For a small clamp on pipe, it usually does not need 12V excitation voltage. So, in this app note, it directly uses EVM430-FR6047 board to achieve it. 

 MSP430FR6047 would expect a square wave. However, if this distortion on the receive is exactly the same for both UPS and DNS receive signals, then it still fine for the MCU to analyze the signal. Because our algorithm is based on calculating the phase difference between the UPS and DNS receive signal. As long as the external brings the same phase different on both UPS and DNS signals, it should be okay to process the signal. 

Another thought, since your pipe is small, and you have already use 12V to excitation signal. The receive signal should be strong enough and do not need to go through the receive op amp again. If you can measure the receive before the receive op amp for about Vpp 100mV. Then the strength of receive signal should be enough and can directly go into the MCU. It can avoid the influence bring by the receive op amp circuit. 

Best regards,

Cash Hao

I am working with the TIDA-01486 on a section of pipe that is 1-inch in diameter and made of PVC. I am using this section pipe to make sure that the TIDA-01486 is working properly because I was able to measure flow data with just the EVM430-FR6047 evaluation module (no TIDA-10486). But this is not my end application.

I tried the EVM430-FR4067 (no TIDA-01486) with metal pipes. I could see the receive signal but it wasn't strong enough for the EVM430-FR4067.

I need the amplification for clamp on transducers with metal pipes (stainless steel and copper).

You're right: I re-read the TIDUDO7 document where it clearly says, "A 10-cm diameter pipe with 4 in-line transducers was used for the measurements." "4 inl-ine transducers" being the key phrase.

Has TI done any clamp on transducer testing with metal pipes?

Apologies for the delay. I was on vacation last week.

I changed the TIDA-01486 board by setting the receive path gain to 2.I have transducers clamped on to the same 1-inch PVC pipe that I know is full of water (no air at all). I'm just trying to see the receive signals in the USS gui. The figure below is what the gui shows me:

I used a scope to look at the actual transducer signals. In the 2 pictures below the yellow trace is the CH1 transmit signal and the blue trace is the CH0 receive signal. The CH1 transmit signal (yellow) s measured at R36 (the impedance matching resistor for CH1) at connector J4. This is the signal that transmitted into the pipe. The CH) receive signal (blue) is measured at two places: the first is right out of the op amp U5, pin 6, and the second place is at the clamping diodes where the receive signal is fed into connector J3 and into the MSP430.

In the left picture, you can see the signal (yellow) on CH1 transmitted by one transducer which is received by the other transducer (blue) on CH0. The receive pulse is at an offset of just over 500 mV.

In the right picture, the signal is measured at the clamping diodes. You can see that somehow the received pulses are removed. I am assuming that this is why the gui doesn't show any signal at all - because it's not getting a signal.

What is happening to the blue received signal? Are the clamping diodes removing it? And if the MSP430 isn't getting any signal why are the traces in the ADC graph at -1400? Shouldn't they be at 0?

I am also don't know why the receive signal, measured at the clamping diodes (right picture) suddenly drops to 0 right about the time that the receive pulses arrive. Is this because the receive ADC is being enabled in the MSP430? Is there maybe a fault in the MSP430? A short circuit perhaps that simply cancels out the receive signl when the ADC is switch on?

Hi，

I think the issue is coming from the receive side op amp circuit. On the left picture, you are measuring from CH1_IN right? The VIBAS circuit should provide a DC offset on the receive signal. But I do not see it in the captures. 

On the right picture, you can remove the R40 resistor first. And check the output signal of the OPA. 

Best regards,

Cash Hao

I wasn't clear enough. Refer to the following picture for measurement locations:

The blue trace in the left picture was measured at Node B. The blue trace in the right picture was measured at Node C. 

I haven't sent any pictures measured at Node A. I have measured Node A before and I can measure a signal there along with the offset. My offset is 0.66 V, but the documentation says is should 0.75 V. My DC bias comes from 3.3 V voltage divided by R22 and R26.

The blue trace in the left and right pictures does have an offset. The blue trace zero marker is hidden under the yellow trace. The offset in the pictures is about 500 mV at the scale in the picture, but if I user a finer vertical resolution the offset is closer to the expected 0.66 V.

In summary, the blue traces are measured at the output of the op amp, not the input. Another question is why isn't the offset removed at the output of the op amp? The documentation says that the purpose fo C48  is to remove the offset in the output.

Today I tried removing the 2 switches in the receive paths that I am using on the TIDA-01486 (see the picture in this post). The switches U9 and U10 are the only thing in the receive path before the receive signals go onto the EVM430-FR6047 eval board. I added a jumper wire from pin 1 to pin 7 in place of each swtich.

This didn't help. I get the same operation is before: I can see the received pulse train at Node B but only the offset voltage as it goes to 0 V at Node C.

I am at a loss. I don't know why the received signal is not getting into the MSP430. I have double checked the wires I added to carry the receive signals from the TIDA-01486 amplifier board onto the EVM430-FR6047 board and all are correct and soldered properly. I don't know what else to do.

Is it worth having a conference call? Answering one question at a time through this forum is taking a long time.

Hi,

Have you tried also removed the clamping diodes? 

A meeting sounds okay. We can go through the hardware design again in the meeting. However, I am not familiar with this external circuit. I may not have enough information to provide through a meeting. 

Best regards,

Cash Hao

I will try removing the clamping diodes, but before I do that I will disconnect the jumper wires I installed yesterday when I removed the receive path switches. By removing the jumper wires I should be able to isolate the problem to the evaul board itself. And I will need to remove those wires anyway when I remove the diodes so that the transmit signal doesn't overwhelm the MSP430.

The following picture shows the two receive ptha op amps and the switches in the TIDA-01486 schemtic. I have removed the 2 switches to disconnect any influence from the MSP430 CHx_IN pins.

The next picture is for reference where I make measurements, namely at Node B and Node C.

The next 2 pictures are scope measurements with clamping diodes D2, D8, D9, and D10 still in the circuit. The left picture is measured at Node B (output of the op amp) and the right picture is measured at Node C (at the clamping diodes.

From the above 2 pictures it looks like the op amp hasn't removed the DC offset from the input. The clamping diodes are doing their job of clamping the voltage to safe levels for the 3.3 V switch and the MSP430, but the actual receive pules train gets clamped to to zero.

The next two pictures are scope measurements with clamping diodes D2, D8, D9, and D10 removed (the switch is also removed). The left picture is measured at Node B and the right picture is measured at Node C as before.

With no clamping diodes, these 2 pictures are essentially the same, which is expected.

My question is why isn't the DC offset being removed from the output of the op amp circuit. That's what the capactior C48 is supposed to do, right? And if the DC offset isn't being removed, then the clamping diodes just remove everything as they clamp to a low enough level for the 3.3 V switch that follows the op amp.

Please let me know how to arrange a call.

I misunderstood the receive path op amp amplifier circuit. C48 is supposed to allow the received AC pulse train signal to be amplified without amplifying the DC offset. So the output of the op amp should be an amplfied version of the AC input signal superimposed on the same 0.66 V DC offset at the input. This appears to be the case according to what I measure, as long as I measure at Node B. But somehow the AC pulse train signal disappears when I measure at Node C (at the clamping diodes). But the voltage levels of the received pulse train are not high enough for the clamping diodes to forward bias. Unless one of of the diodes is mounted on the circuit board backwards, probably D2....

But we build these boards in accordance with the TI documentation supplied with the design. I'll have to check this next...

I tried reversing D2, but that didn't work. D2 is on the right way. But the problem seems to be with D2 - when D2 is on the board and there is a positive voltage at Node C, D2 is conducting and clamping the signal at 0.2 V. This is weird because the forward voltage of D2 is about 0.2 to 0.3 V, but if there is a positive voltage on Node C then D2 is reverse biased and should be open circuit because it's a 30 V diode!

I even replaced the D2 diode with a new part in case the old one was damaged. No, same behaviour.

I checked that my bag of D2 parts is indeed the DFLS130L-7, and yes it's the correct diode in the bag.

Hi,

This is the capture without D2, right? I think this signal can goes into the MCU. The Chx_IN has C11. C12 which can block the DC offset on this signal. And the AC part should be good to goes inside the MCU. The Vpp of the AC signal is around 500mV according to the picture. It should be fine to be captured by MCU.

Yes, you are correct. The picture you showed is without all the clamping diodes. Are you suggesting that I just remove the clamping diodes? If so, I can't do that because each of the 2 transducers are used for both transmit and receive. The way this system works is that when a transducer transmits, the transmit signal still goes through the receive path even though the ADC is disabled at that time, and the clamping diodes limit the signal applied to the MCU to safe levels.

My next step is to try to verify that I'm using the correct part for D2. Maybe digikey sent me a different part in the DFLS130L-7 diodes.

I have verified that all the parts are correct.

I'm stumped. The D2 diode seems to be clamping when it shouldn't even be forward biased.

Can we arrange a conference call to go over this circuit?

Hi,

Our FAE should reach out to you to arrange a meeting discussing it. It looks difficult to find a time work for all of us.   

Best regards,

Cash Hao

Thank you for joing the conference call last night, Cash. Have you been able to talk to the author of the TIDA-01486 amplfier reference design for details on how to connect the TX and RX signals?

As I mentioned in the call last night, the reason my received AC signal did not exist when measured at the clamping diodes in the receive path was because the 10 ohm current limiting resistors, R39 and R40, had both failed. Their resistance had increased to about 15 Kohms. However, when I put new 10 ohm resistors in the R39 and R40 slots an over current problem that I thought I had fixed has returned. The over current is 150 mA drawn from the the external 15 V power supplies that I am using. The 2 receive path op amps were getting too hot to touch. This over current problem is in the receive path.

The TIDA user guide is not clear on how to connect the transmit and receive signals from the TIDA-01486 board to the EVM430-FR6047 board, so I had to guess. My guess is shown in the following picture and labelled Connection #1. In this connection, the impedance matching components on the EVM, C11, C12, R16, and R17, have been removed. Note that the amplified receive signals are fed directly in to the MSP430FR6047 microcontroller. The amplified receive signals are made up of the 1 MHz receive burst superimposed on a 0.6 V DC offset. I thought including the DC offset in the signal fed to the MSP430 was the cause of the high current problem.

In order to remove the DC offset from the amplified receive signals, I reinstalled the matching components on the EVM, C11, C12, R16, and R17, and cut the traces on the EVM between R16 and R17 and the connector. J8. This isolates the all 4 signals that still need to get to the MSP430, and capacitors C11 and C12 will remove the DC offset from the receive signals. The next diagram illustrates the Connection #2. Using this connection, the current drawn from the external 15 V power supplies has decreased to 90 mA. This is still too high, byt ajust over twice the expect current. The 2 receive path op amps still get too hot to touch, but it takes longer for them to get hot. And the USS is no longer reporting "No Signal" errors, and is actually not reporting any errors at all. Unfortunately, the signals looks terrible, but the system is at least measuring something.

I think I'm on the right track, but I'm still not there yet. I need to know exactly how to connect the transmit and receive signals between the EVM and the TIDA-01486 amplifier board. Can you please ask the author of the TIDA design for a detailed connection diagram?

Hi Anthony,

The two figures seem broken. Could you resend them again? 

I have checked with the author and the confirmation is that it do requires the cap and resistor connect between the two boards. 

In the application note, Figure 3 should shows the connection between them. According to this figure, CH0IN/OUT and CH1IN/OUT on the boosterpack board are not directly connect to the MCU pins. The Ouput pin needs to pass a register then connect to MCU pin and the Input pin needs to pass a cap to MCU.

Best regards,

Cash Hao

Here are the 2 pictures I posted last time. Connection #1 is how I first started, and according your note from the author, Cash, Connection #1 is wrong. Connection #2 is what I am using now, which should be correct. In my Connection #2, I am using the resistors and capacitors that were on the EVM, which are 200 ohms and 1000 pF.

But as I said in my last post, I still have an overcurrent problem: 90 mA is being drawn from my external +/- 15 V power supplies, and the receive path op amp are still getting too hot. Is it possible that I have damaged the MSP430 on my EVM be originally using Connection #1? This seems unlikely because the op amps on the TIDA-01486 are getting hot, not the MSP430 on the EVM. And I have removed the 2 switches, U9 and U10, on the TIDA-01486 and I am using jumper wires to bypass the switches (I don't need the switchies in my application), so those switches are not the cause of the over current situation.

Any insight would be very much appreciated.