4W/2W Converter and 2175Hz Notch Filter

[ronsimpson]

When the -5V comes from a good source the noise is gone.
In the green circle is the RC filter. If the ESR of the cap on the right is large the filter will not function well. What Tony said.
I do not see capacitors on the power supply of the op-amps.

 

[Les Jones]

Hi Bruce,
If you found that the voltage drop across the 100 ohm resister value that I suggested in the filter too great you could try an inductor. That way you would have a high reactance at 80 Khz but a low resistance at DC. A 1mH inductor would have a reactance of about 500 ohms at 80Khz

Les

 

[Bruce Glazier]

Thanks Ron, I didn't set up the RC properly as I used the existing CAP and didn't add a new one. So to understand what I need here, something like this?

https://www.vishay.com/docs/40005/593d.pdf

If I used your suggested values of 10Ω resistor and 10uF capacitor for the RC, the cutoff would be at 1.5KHz, which I assume since this is suppose to be -5VDC, the lower the cutoff the better?

 

[Bruce Glazier]

Thanks Les,
I will look into both options right now.

 

[Tony Stewart]

Improved dual supply charge pumps have integrated LDO's for both polarities.
https://www.linear.com/product/LTC3260


If you used a 1mH choke instead of 10 OHms it would appear as 250 Ohms at 40kHz and suppress noise /25 or double at 80kHz
**broken link removed**

 

[ronsimpson]

If I used your suggested values of 10Ω resistor and 10uF capacitor for the RC, the cutoff would be at 1.5KHz, which I assume since this is suppose to be -5VDC, the lower the cutoff th

I don't know what cap you are using but from the data sheet I think it has 2 ohm of ESR.
So if you added a 1 ohm resistor and are using a 2 ohm cap then at high frequencies you have a 1 to 2 ohm divider. So the noise will not be down much. (2/3) So you need to make the 1 ohm bigger and/or reduce the ESR to get a 10: reduction or 100:1 reduction in noise.

 

[Bruce Glazier]

Thanks Tony, that supply looks like it would be much better for this purpose and could replace both of the regulators I have on-board. I will look into using this for Rev B of the board which I need to get into production ASAP.

Ron, I actually used **broken link removed** for c48, the one I just linked in the previous post was what I found when I searched for low ESR capacitors. I really have no idea how to determine a capacitors ESR if it is not spelled out in the datasheet (which it doesn't appear to be in this case). I understand what your saying about the ratio, I really need to know my CAPS ESR to determine if it is even in an acceptable range so I don't have to bump the resistor value too high. I think I'll go check to see if I can dig up an inductor @ 1mH as recommended by Les, yourself and Tony and see if that makes a difference.

I think the real solution moving forward is to move away from the MAX660 altogether as suggested multiple times throughout this thread but I need to first fix these as quickly as possible. I will re-post with any updates using the inductor.

Thanks again

 

[ronsimpson]

The inductor should make a difference.
At DC you will have some resistance. But at 40khz and above it should open up and help disconnect you noisy supply.

 

[Les Jones]

I have just realised that there was a flaw in the way I suggested doing the tests yesterday to see if the noise was finding its way to the op amps via the +5 or -5V rails. When I suggested re connecting the +5 supply to the input of the max660 I should have also said to load the output of the MAX660 to the same current as that taken by the negative supply to the op amps. With load on its output the pulse current on the input will be larger so it will probably put more noise on the +5V rail. So it is still possible some noise from the +5 V rail is causing the problem when the MAX660 output is loaded.

Les.

 

[Tony Stewart]

D.F. is related to ESR and given in specs. Low ESR types will be clearly specified.
Usually D.F. is rated at 120Hz for large C(>10uF) and at 1kHz for small C

D.F. = ESR/X(f)

 

[Bruce Glazier]

Thanks again guys,
I've ordered the recommended inductor to try that tomorrow and will report back.

Les, that makes sense, I'll determine the current draw on the -5V rail and attempt to draw the same amount off the MAX660 and recheck for ripple.

 

[Bruce Glazier]

FYI, I didn't order them in time so I won't have them until tomorrow. Will report back tomorrow with results.

 

[Bruce Glazier]

So results are in, I added the inductor capacitor filter between the output of the MAX660 and the -5V plane and the 40KHz signal has been neutralized. I still have the 500KHz signal present but it is minimal in comparison to the noise present before. I could always add the LC filter to each line if I wanted to refine THIS design more.

I actually am looking into using the Active Twin-T Filter as suggested by Tony Stewart and I am going to use +12V and GND biasing the op-amps at half that, I already have the parts to breadboard and test. Thanks for that suggestion, you were right, the passband is so much flatter and the notch is very fine. .

At this point, I am going to add the LC filter to the existing and redesign the next revision using the Twin-T Notch, so this discussion can be closed by a Moderator as far as I am concerned.

I want to personally thank you all for your expertise and valuable input as you've done an excellent job in helping this newbie.
-Bruce

 

[ronsimpson]

The 500khz is not good. Do you want to fix that to?
If you change the filter design then it does not matter. It might be nice to know for experiance.

 

[Tony Stewart]

Op Amp may be higher output impedance than 10.
Load is required for depth of notch and tuning C.

It could be due to lack of gain margin in unity gain Op Amps with 500kHz =GBW

 

[Bruce Glazier]

Update: I am doing a second revision of this board and I've cleaned up a lot of mistakes from the first. I am going to use the low ESR capacitors as recommended from the MAX660 datasheet like you all have mentioned this entire time. I'm not sure why I didn't notice how important the ESR of the caps was before as it is spelled out clearly in the datasheet,

"To reduce the output ripple caused by the charge pump, increase the reservoir capacitor C2 and/or reduce its ESR. Also, the reservoir capacitor must have low ESR if filtering high-frequency noise at the output is important."

Furthermore I also didn't realize the datasheet gives a list of manufacturer's of recommended capacitors. I will report back with the results in a couple weeks. I am actually not in a bad position as high frequency noise is filtered by our product anyway so the noise will not persist at the end result whether I fix this or not.