MPY634: a strange frequency responce

Hello Oleg,

It appears that you are testing the MPY634 in a frequency doubling configuration with a very low frequency passive low-pass filter (-3 dB @ 16 Hz) following the output. I suggest testing the MPY634 by itself, without the input amplifier and the low-pass at the output. You may try ac coupling the output as datasheet Fig. 10 shows to remove the dc level - should that be an issue. Then, if the multiplier appears to be functioning correctly add the input amplifier and check the operation. Finally, add the low-pass to the output. The datasheet shows the MPY634 driving a resistive load, the resistive load with the capacitor peaking, or a RC high-pass. There may be a drive issue driving the RC low-pass load. Make sure there aren't any signs of oscillation at the op amp, or multiplier outputs.

Also, the 47 uF capacitors are not effective power supply decoupling for a wideband analog circuit. They will have too much ESR and ESL to be effective at frequencies in the Megahertz. Datasheet Pg. 6 states, "As with any wide bandwidth circuit, the power supplies should be bypassed with high frequency ceramic capacitors. These capacitors should be located as near as practical to the power supply connections of the MPY634. Improper bypassing can lead to instability, overshoot, and ringing in the output." I would suggest 10 nF to 100 nF, right from the device +Vs and -Vs pins to ground. Be sure the op amp has correct bypassing too.

Regards, Thomas

Precision Amplifiers Applications Engineering

Dear Thomas
I found that this is not a first complain on the circuit assembled according to fig 6

e2e.ti.com/.../528477

We rechecked it and still see the same problem.

Do you have a recent lab data report over full frequency range?

Is it possible that the devices can be similarly distorted by static voltage and this leaded to such behavior?

Best
Oleg

Hello Oleg,

The MPY634 is a legacy product and it is unlikely that any recent characterization work have been performed. Nevertheless, I am interested in coming to an understanding as to why you, and the other gentleman, are observing the gain peaking behvior across frequency.

One possible explanation may be the actual bandwidth definition for the multiplier. Bandwidth, in the Electrical Specifications table is specified as Small Signal BW, VOUT = 0.1 VRMS. That is what the 6 MHz minimum and 10 MHz typical are referenced to. You indicate that the A - Amplitude, applied to the X1 and Y1 inputs is 1.55 V. Is that volts peak, peak-to-peak, or RMS? The actual amplitude level applied to the X1 and Y1 inputs may be affecting the bandwidth behavior.

I have ordered some MPY634 devices from our TI distributor so that I can get a sample of the product that we are currently supplying customers. They will take a day or two to arrive here. A bench test is being assembled and we plan to check the over-frequency behavior of the devices to see if we can replicate the results that you are receiving. I'll let you know what we find.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi, Thomas

The amplitude is  A=Peak_to_Peak/2

By the way, Alex got very similar result:   the response increases 10 times in the frequency range 100 to 2000 kHz,  while the  input amplitude was 3,5 times less.

Best

Oleg

Hello Oleg,

I am working with Thomas on this issue. We have received the part and I have tested and replicated your issue with the following setup and data, but I have questions:

The "Out" signal is just a DC level offset (no frequency component), which increases with the frequency of the input "A" signal (this is in line with what was seen in the E2E you referenced https://e2e.ti.com/support/amplifiers/precision_amplifiers/f/14/t/528477 ).

Does your design call for reading the output as a DC level?

Did you re-test your circuit using the below configuration from the datasheet that Thomas suggested earlier? This includes the DC blocing capacitor, C. You should get an output signal (without a DC offset) that rolls off with frequency. If you're still seeing the problem with this configuration and with a properly traced board, then we can continue to test this further.

Please let me know if this answers your question.

-Peter Iliya

Hi Peter

Thanks for reproducing the measurement we did.

Yes, we need to know DC level as function of multiplication. We have designed a board with two multipliers on it and found some unreasonable responce from them, then we bought two more and  built a circuit   for testing the multiplier.

We attempted to design a wideband lock-in amplifier using MPY634  referring to the "Normal Connection" of the data sheet.

and

Unfortunately, our and your measurement does not corresponding to the Data Sheet and currently  we are looking for another solution.

Best

Oleg