Forums

Hi, Stefan,

We are looking into this issue for you.

-d2

This does look like popcorn noise.  Popcorn noise is process related and is present in all bipolar processes.  We are constantly striving to make improvements in our processes to minimize this issue.  In many cases this issue is exacerbated by application decisions.  For example, applications with high gain and relatively narrow bandwidth are especially susceptible to this issue. 

http://www.en-genius.net/includes/files/avt_022508a.pdf

 Please note that this issue has not been seen in CMOS products.  The OPA1642 has a JFET input but is a Bipolar device.  Below is a link to an article that covers this subject in great detail.  Let me know if you have further questions.  I broke this answer into different posts to make it easer to read.  See next posts to answer your direct questions.

To address your specific questions:

 Is it possible that the OPA1642 is more susceptible to burst noise than the OPA2227?

Some devices are (by internal topology) prone to popcorn noise.  However, in general, it is a process variation issue.  Thus, for example it is possible that the OPA1642 may show no issues for years then have a wafer batch that has some popcorn noise.  I think the safest solution is to use CMOS for applications that are especially sensitive to popcorn noise. 

Is there a solution to minimize?

Some of the typical solutions may not apply to your application.  For example, decreasing input impedance helps if bias current is poping.  In your example the pop must be internal because the input stage is JFET.  Also, your device is running at full bandwidth.  Normally popcorn shows up when bandwidth is limited because the low frequency noise is more dominant.

 Is there the possibility that more devices show these effects depending on temperature or aging?

Popcorn noise may be affected by temperature.  The link to the article talks about this subject.  Also, popcorn noise is time dependent, so it is possible for the frequency that the popcorn noise appears to vary.  In some cases devices pop very infrequently and so it may be difficult to detect the pops.  In many cases I think that people mistakenly attribute a factor (e.g. temperature) to changing the popcorn frequency, but the actual change is just natural variation on the popcorn frequency.  I don’t know if ageing could affect popcorn noise.  Popcorn noise is caused by defects from fabrication.  Perhaps the defects could be aggravated (or even improved) by ageing.  I don’t think this is a major concern; the main problem is process variation.

My follow up questions:

 Would you please send me your schematic?  Also, have you confirmed that the OPA1642 is the device with popcorn noise?  In other words can you swap in and out an OPA1642 and see the popcorn appear and disappear?  Also can you send me the date code on the devices?  The date code is marked on top or the bottom of the device.  A picture or transcription of what is marked on the device will help us understand when the device was fabricated.  Any information regarding your application requirements will help me to recommend a CMOS device that may help you solve your issue.

Best regards,

Art

Hi Art,

Thank you for your detailed answer. Unfortunately I wasn’t able to reply earlier.

In relation to your questions:

We confirmed that the OPA1642 is the source of Popcorn Noise. We noticed that sometimes the gain stage has popcorn noise and sometimes one of the inverting amps. We tied the input of the OP to Gnd and the popcorn noise was still there.

When we replaced the OPA1642 by a new part the popcorn noise disappeared* in most cases.

*additional measurements let us draw conclusions that there is popcorn noise on a multiplicity of OPAMPS but it is so small that it is almost covered by other noise sources like 1/f etc. That’s not a problem for us.

Our approach is, that we try to detect the devices with high popcorn noise and to replace it. That’s costly in terms of time and we aren’t sure to find all affected devices.   

I found the following Datecodes on various PCBs:

05k EEK9, 9AK A3JE, 9BK CJTV, 12k DKLN, 17k DF4P, 17k DF4Q

It seems to me (but it is only a speculation) that especially the devices with Datecode 05k EEK9 and 17K DF4P show increased popcorn noise.

Devices with 9AK A3JE, 9BK CJTV, 12k DKLN we had on prototypes and advance prototypes. I think we haven’t had popcorn noise, but maybe we only haven’t detected.

Best regards,

Stefan

Hi Art,

here some further information. I did a few more measurements to validate, whether the popcorn depends on the gain of the amplifier stage. 

As you can see in the following pictures the popcorn noise varies with gain as expected.

Gain 1:

Gain 10:

Gain 50:

Kind regards,

Stefan 

Hi Art,

one more question.

Do you make tests during fabrication considering popcorn noise? Maybe for each batch?

Best regards,

Stefan 

Hi,

 here the second try to post the schematic. I think the first time it didn't work.

1440.Amplifier_schematics1.pdf

Stefan Gronloh

Hi,

 here the second try to post the schematic. I think the first time it didn't work.

1440.Amplifier_schematics1.pdf

Stefan,

We are implementing testing for popcorn on this device.  However, one alternative is to use a CMOS device with similar specifications.   We have not observed popcorn noise on CMOS processes.  The OPA1652 is a CMOS alternative that has similar specifications (see table below).  Let me know if this helps and if you think the OPA1652 is a suitable substitute.  

 Art

Hi Art,

thank you for your prompt reply.

I think the OPA1652 could be a good alternative even the Iq is 0.2mA higher than the OPA1642.

Having 24 devices of the OPA1642 on the PCB, that adds up to 9.6mA but I think it's practical.

I will discuss this with my colleaques.

In an optimal way we would assemble a hole board with the OPA1652 to verify the technical data, thermal behavior etc., but I'm afraid there aren't parts on the market.

Is it possible to get such a quantity of samples?

Best regards,

Stefan 

I received an e-mail from a colleague indicating that samples are being sent to you.  Let me know how the samples work out and if you have any other issues.

Art

Hi Art,

After a couple of weeks and a lot of measurements, here is the state of affairs:

It seems that there are 2 kinds of phenomena showing the same behavior respectively the same output: popcorn noise!

1^st: We notice popcorn noise, change the OPA1642 in gain stage and/or symmetric stage, and the popcorn noise disappears.

2^nd: We notice popcorn noise (symmetric stage), change the OPA1642 and the popcorn noise doesn’t disappear. Trying a second OPA1642 the popcorn noise is still there. We changed the OPA1642 by the OPA1652, no change, the popcorn noise is still there. Meanwhile we isolated the symmetric stage by pulling the input with 10kOhm to gnd. Even the output was isolated.   No Change, popcorn still there!!

 When we swap the OPA1642 with an OPA1642 of a neighbouring channel (not showing any popcorn noise), the pocorn noise is remaining, on the same channel!! The  neighbouring channel, now assembled with the former “Popcorning” OPA1642 doesn’t show any popcorn noise!

It seems that in this case the popcorn noise depends on the channel. We don’t see a correlation with the Layout because the affected Channel relating to the number of concerned PCBs is random.

We do some further tests:

-          powering the OPA with a battery (no change)

-          changing C403, C410 (no change)

Sometime by undoing these changes, the popcorn noise suddenly disappeared. We couldn’t activate it since then.

This 2nd effect we saw on a few PCBs. On all of them, the popcorn noise disappeared sometime doing changes and make them undone.

We are a little perplexed!!

Is it possible that this kind of “popcorn noise” depends on, or is effected by soldering!!!

Are there experiences of popcorn noise depending on cold soldered connections?

Best regards,

Stefan