-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Sensitivity of opamps to air coupled EM fields, especially of the LM4562/LME497X0 family
- Thread starter pma
- Start date
Pavel, sorry but as mine and @SIY information has demonstrated this is an avoidable issue. It's all down to appropriate design, which will vary from component to component.
All of your plots show very high levels of mains pick up that should not be experienced, and are not experienced in normal audio designs with those various op amps. I highlighted the 5532 simply because we know that thousands of audio designers have decades of experience of that part without any issue.
As such it should have been obvious to you that investigation and review of your test methods was required as there is clearly something wrong. You have created pathological conditions which are not relevant to the real world. As such your info is misleading and unhelpful.
Last edited:
Any references to this you could send my way?
Unfortunately, no. I was given this tidbit from a very skilled former TI engineer.
miero
Active Member
- Joined
- Aug 1, 2018
- Messages
- 241
- Likes
- 292
Here are guesses of Bob Pease (IC designer in NSC) written by Mark Brasfield (LME49713) why metal cans could be better sounding:
- https://www.diyaudio.com/forums/chip-amps/154106-sounding-audio-integrated-opamps-post3699846.html
KSTR
Major Contributor
Working around a design flaw is not the point here. The point is that this chip has this design shortcoming that actually does appear in real-world, non-pathological applications (compared to other, comparable opamps) and the sad part is that the problem is undocumented by the maker.
solderdude
Grand Contributor
The op-amp rolling crowd may get into trouble when rolling it in a circuit with a non suited PCB design/decoupling.
In all honesty I have been using the LM4562 a lot and never heard anything nasty but then again I do not 'measure' as deep and usually steel metal enclosures mostly. Maybe I've been lucky or didn't catch it yet.
In all honesty I have been using the LM4562 a lot and never heard anything nasty but then again I do not 'measure' as deep and usually steel metal enclosures mostly. Maybe I've been lucky or didn't catch it yet.
Well I think we need to perform more representative testing to characterise this to see how much of a real world problem it actually is. Different op amps have different characteristics that need to be appropriately designed for. Many op amps can have problems with EM/RF. What SIY said about this and the issues of op amp rolling is absolutely correct.
KSTR
Major Contributor
Agreed, but for this the characteristics have to be known and anything "non-typical" should be stated in the datasheet. Good example is the AD797 with its need of 100Ohms//22pF in the feedback, and preferably 100Ohm on the non-inverting input as well, for unity gain stability.
Also full ack on the imbecility of "op-amp rolling".
There are probaby a lot of "real world" implementation issues that are not covered in data sheets.
KSTR
Major Contributor
Depends, I'd say. Some makers care more than others. Let's take the venerable 5532, the only competent datasheet that is available is the one for the NJM5532, with lot's of hints how to avoid pitfalls.
Now the 4562/49720 spec sheet is bloated with useless distortion graphs but some really important stuff is missing.
Circuit designer's time and effort is a premium and some companies provide extensive information even for most simple and cheapest consumer parts, some don't.
I just discovered this discussion due to a reference from another forum.
I can confirm that the LME49720 (family) is very sensitive to EMI. I have seen that on several occasions. Especially with unshielded test boards.
I think it would be beneficial to look at the sources of the EMI. This can help explain why some people experince the EMI problems and others don't, or at least to a lesser degree.
The 100 Hz noise clearly comes from a DECT system (I have worked with DECT development for almost 30 years, so I have seen this a lot!).
The 10 Hz is most likely from a Wi-Fi router.
In the first measurement in post #39 both can be seen. The DECT interference with intervals of exactly 10 ms and the Wi-Fi intereference with an interval just over 100 ms.
A quick test you could do: If you hook off the DECT phone and hold the handset close to your work bench, you will probably get a 200 Hz signal instead of the 100 Hz you have now. The peak level from the handset will generally be different from the peak level from the base station, due to the difference in distance etc.
The LME49720 can show good performance, but to avoid problems shielding is definitely recommended. A good PCB layout will help, but it cannot solve the problem in all cases.
I would recommend that you move the DECT base station and Wi-Fi router away from your test area. I used to have my DECT base station in the same room as my test setup, but I have moved it to a different room in the house.
Be aware that even when no phone connection is active the DECT base station will normally transmit a short slot every 10 ms. And, in Europe, it will normally transmit at a power level of 250 mW (in the US it is only 100 mW).
I have used a couple of LME49720's in the RTX6001 Audio Analyzer that I designed. With the lid removed I see some noise issues from nearby DECT equipment (we have a lot of that at work). But with the lid in place the noise disappears.
I can confirm that the LME49720 (family) is very sensitive to EMI. I have seen that on several occasions. Especially with unshielded test boards.
I think it would be beneficial to look at the sources of the EMI. This can help explain why some people experince the EMI problems and others don't, or at least to a lesser degree.
The 100 Hz noise clearly comes from a DECT system (I have worked with DECT development for almost 30 years, so I have seen this a lot!).
The 10 Hz is most likely from a Wi-Fi router.
In the first measurement in post #39 both can be seen. The DECT interference with intervals of exactly 10 ms and the Wi-Fi intereference with an interval just over 100 ms.
A quick test you could do: If you hook off the DECT phone and hold the handset close to your work bench, you will probably get a 200 Hz signal instead of the 100 Hz you have now. The peak level from the handset will generally be different from the peak level from the base station, due to the difference in distance etc.
The LME49720 can show good performance, but to avoid problems shielding is definitely recommended. A good PCB layout will help, but it cannot solve the problem in all cases.
I would recommend that you move the DECT base station and Wi-Fi router away from your test area. I used to have my DECT base station in the same room as my test setup, but I have moved it to a different room in the house.
Be aware that even when no phone connection is active the DECT base station will normally transmit a short slot every 10 ms. And, in Europe, it will normally transmit at a power level of 250 mW (in the US it is only 100 mW).
I have used a couple of LME49720's in the RTX6001 Audio Analyzer that I designed. With the lid removed I see some noise issues from nearby DECT equipment (we have a lot of that at work). But with the lid in place the noise disappears.
- Thread Starter
- #53
Thank you for your input, @JensH . As always personal experience and results are needed to make an opinion, armchair design approach is of little value.
I would like to add that I want for my circuits to work my designs under real world conditions flawlessly. So I will not create special conditions like moving modem away. If OPA2134 works flawlessly and LME49720 has big issues, then the latter part will never take place in my designs.
I would like to add that I want for my circuits to work my designs under real world conditions flawlessly. So I will not create special conditions like moving modem away. If OPA2134 works flawlessly and LME49720 has big issues, then the latter part will never take place in my designs.
Last edited:
- Thread Starter
- #55
Now you may consider looking at 1.9 GHz interferers around your lab.
Nevertheless specific locations will need much more care in terms of electronic shielding, i.e: Zizkov TV broadcast tower Prague.
Thanks.
Yeah, you are right, that's why it is important to me to select immune parts and not to bother with sensitive parts. Once again, no problems with any of the JFET opamps, no problems with AD797, ADA4898 and no problems with LT1028. Some issues with NE5532 and OPA211. For the reason that AD797 is much better part in audio band than LME49710, I stick with using AD797 if I need a low noise bipolar opamp.
More on opamps in audio band by Samuel Groner, the guy whose job I appreciate very highly.
http://www.nanovolt.ch/resources/ic_opamps/pdf/opamp_distortion.pdf
And that of @scott wurcer , former Analog Devices designer and the father of AD797. From the newer bipolar opamps, ADA4898 is approaching to AD797, but is not that good.
This is not an armchair approach. This is precisely why I previously recommended to the DIY section not to use the Hypex Lm4562 eval board as it is not a "complete" design and will almost certainly have issues.
I showed you a hypex amp with an LM4562 buffer not showing a massive spread of 50 Hz harmonics (sat in a hostile RF environment and with a wifi router less than 1 m away).
NWav guys LM4562 plots were impeccable.
SIY showed you an impeccable LM4562 plot.
Nothing "armchair" about this. These are real world results.
There is nothing new about RF susceptibility in op amps. The solution is about appropriate design.
Your "personal experience and results" failed to get a 5532 working properly which should tell you there is something wrong with your test set-up.
Last edited:
What in the DECT phone causes the 100Hz? I just did a quick test with my Panasonic KX-TG9541B phone...
https://shop.panasonic.com/cordless-corded-telephones/cordless-telephones/KX-TG9541B.html
Using my QA401. From what I've seen he's got an opamp (1612) and a diff amp (1632) -see these photos:
You'll see the OPA1612 appears to come right off of the atten relay. The 1632 appears to be the driver for the ADC. I know the mfg is on this site - be interesting for him to clarify this.
Here's a crappy Youboob vid of the test setup. There's a set of open leads on the QA401 left input. I use both ATTEN and non-ATTEN (you can see the switching).
All I see is typical SMPS and 60 cycle noise.
This is with the lid open.
The large spike is the 1.9GHz DECT from the handset... The second spike to the right is the WiFi router upstairs from my bench - about 3 meters away thru a wood floor...
Near the end of the vid I touch the clip probe - you can see the typical rise in noise...
Last edited:
There's nothing wrong with Pavel's test setup. It's just what's being tested and what conclusions are drawn from it that may be the source of disagreement. Remember, the test circuit (which he was very explicit about) is not at all representative of anything that would be used in an audio gain stage. So the results are correct for that circuit, but are irrelevant to circuits used in engineered audio equipment.
BTW - if anyone's interested in what was in that DECT spike on the siglent - here's a close up:
Just a typical freq hop... A technique invented by Hedy Lamarr and her piano-playing friend back in 1941... sort of based on how player pianos work... whihc is also the basis of term Mechanical Royalties in the music biz...
Patent : https://patents.google.com/patent/US2292387A/en
It was filed under her married last name...
The story behind Hedy Lamarr’s game-changing flash of genius
https://www.iam-media.com/patents/hedy-lamarr-genius
Just a typical freq hop... A technique invented by Hedy Lamarr and her piano-playing friend back in 1941... sort of based on how player pianos work... whihc is also the basis of term Mechanical Royalties in the music biz...
Patent : https://patents.google.com/patent/US2292387A/en
It was filed under her married last name...
The story behind Hedy Lamarr’s game-changing flash of genius
https://www.iam-media.com/patents/hedy-lamarr-genius
