Author Topic: Fraser has two new patients inbound - E2V ARGUS 4 & Seek Reveal Pro FF  (Read 3790 times)

0 Members and 1 Guest are viewing this topic.

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
As requested, I have produced some higher resolution images of the PCB's with better lighting :)

The full 12MP camera image files are available from Filedropper (60MB total Zip File size)

http://www.filedropper.com/hrpics

I will upload smaller 2MB images to this thread below.

Fraser
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
1
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
2
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
3
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
4
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
5
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
I carried out some reverse engineering of the DC-DC converters and Microbolometer power supply today.

The main component involved with the production of the various power rails is the Texas Instruments TPS65217C PMIC. It is an all in one power management chip that takes external power supply, USB and Battery inputs and automatically connects the best, or available option to the System power output. The IC also provides the smart charger circuit for a Lithium Cell  :)

As if this were not enough, the TPS65217C provides multiple DC-DC converters and LDO's to provide for a microprocessor system and a string of LED's !

Outputs from the PMIC:

SYS = 5V nominal. Selected power input passed to this pin to power the system. Normally the DC-DC converters and LDO's are connected to this pin.

DCDC1 = 1.5V

DCDC2 = 1.1V

DCDC3 = 1.1V

LDO1 = 1.8V

LDO2 = 3.3V

LDO3 = 1.8V

LDO 4 = 3.3V


So which DCDC conveter or LDO powers the microbolometer ?

The answer is... none of them !

The SYS output from the PMIC provides a nominal 5V for use by the various DCDC converters and LDO's but it can also provide the main power rail for other IC's and their respective power supplies. This is what happens in the Reveal Pro. The SYS output rail is directly connected to the Microbolometer PCB connector and has no further filtering before doing so. It is therefore best considered a RAW power rail in need of some 'treatment' before being applied to the voltage and noise sensitive Microbolometer circuitry.  At this point it was clear to me that the Microbolometer PCB must host a low noise LDO voltage regulator of some sort. I knew exactly where to poke around !

Ever since looking at the Microbolometer PCB I have wondered about the function of the small 4 pin IC on the PCB edge, and its identity. It was now number one suspect that it was an LDO. Its positioning made sense as well as you do not want an LDO sat too close to a very temperature sensitive microbolometer die. Some quick continuity probing showed the component to indeed be some form of regulator. It received the power rail from the ribbon connector and supplies many of the components on the microbolometer PCB. The four pins are Vin, Vout, 0v and 'unknown'. I suspect the unknown pin is an enable input but this has yet to be confirmed. This LDO is nor marked with a number but there does appear to be a symbol on it. I thought it likely a Linear Technology device but I will need to investigate further.

So where does this lead us ? Well I find it unlikely that the PMIC three way MOSFET switching is creating a noisy output at the SYS pin unless something has gone very wrong with the control circuit. If such were the case I would expect the camera not to boot correctly as SYS powers all other circuits associated with the processor. I will check it though  ;) The little LDO regulator on the microbolometer PCB is currently the chief suspect in the investigation. Either it is in some form of failure or a support component that maintains its stability has failed. It has input and output capacitors that I will need to check. If the LDO is malfunctioning due to excess loading, I will need to investigate further down the line. I believe in keeping things simple though, so for the moment I will look at the small LDO circuit and see if it is producing excess noise due to self oscillation. I will also try to identify the component. Testing will reveal its output voltage and I can always fit another, compatible, part if such is required. We could just be dealing with a failed MLC.

I attach a picture of the chief suspect in the investigation..... guilty until proven otherwise  ;D

The reverse engineering was carried out visually and with continuity checks. No power was applied as that comes only after I know what I am 'poking'!

Fraser
« Last Edit: July 25, 2020, 09:38:34 pm by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
TPS65217 PMIC functional diagrams
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
The NCP140 LDO Regulator looks similar to that used on the Microbolometer PCB. The size of 1mm x 1mm matches. As I suspected, the fourth pin is very likely an “Enable” pin. It will be interesting to see what voltage appears at the output of the LDO. The NCP140 operates with, or without input and output capacitors so may be better than the LDO fitted in the Reveal Pro.

https://docs.rs-online.com/de84/0900766b81628526.pdf

Richtek make a similar device as well. It looks like I will have my choice of replacement parts if such is needed  :phew:

http://www.farnell.com/datasheets/2183318.pdf?_ga=2.73893776.30223683.1595718040-787969494.1589067057&_gac=1.23385160.1594899574.Cj0KCQjw9b_4BRCMARIsADMUIyr7bImqY48wb16DFj_Uoi3g3YNTRusSSK02WuPkx3PR13hb4sf-iNQaAsezEALw_wcB

Fraser

« Last Edit: July 25, 2020, 11:11:10 pm by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
I could not resist checking the voltage at the LDO.

Input voltage is a solid 3.8V

Output voltage is 2.8V but my trusty Fluke 87 III Bar graph display appears to show noise on the rail. A look with a scope will obviously be the next step but at least I now have an idea what voltage the LDO produces. Whether that voltage is correct and noise free we will have to investigate tomorrow :)

Fraser
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
As the PMIC is a TI part, I thought it possible that the LDO would be a TI part as well.

This part looks like a possibility. The TI TPS7A02 series 1mm x 1mm LDO. It is available in 2.8V output and normally has input and output capacitors. This is just a first look with Google so I am not done yet but it is a start.

https://www.ti.com/lit/ds/symlink/tps7a02.pdf

Fraser
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
I think I will be fitting high quality MuRata 0402 size 1uF, low ESR, MLC’s at the input and output of the LDO tomorrow   ;)

Fraser
« Last Edit: July 25, 2020, 11:55:01 pm by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
I woke up to a message from a friend in the USA today. He was sharing his knowledge of the Seek Pro dongle and its 2.8V Regulator that powers the microbolometer. In his testing, if you mess around with the value of the 2.8V regulator bypass capacitors it can create a rolling noise presence on the cameras display. This was very interesting to me for two reasons.

1. It adds further support to my theory that the 2.8V LDO in the faulty Reveal Pro is likely causing the issue that I am seeing.

2. A friend of mine with a Reveal Pro has a rolling noise issue on his unit similar to that seen if trying to record a CRT TV screen with a camcorder. It looks like a slow rolling shutter effect.

I will have a good poke around the LDO with a portable scope today to see what is present on the supply rails. I am beginning to wonder whether there is a stability issue with the 2.8V LDO used by Seek Thermal and it has a penchant to oscillate if it is not adequately bypassed with good quality capacitors if the correct values. The TI datasheet for an LDO that I posted last night suggests 1uF low ESR MLC’s for the bypass capacitors. I checked before going to bed and I have some very nice Wurth 1uF 10V low ESR 0402 MLC’s in my selection box :) It could be that the camera would benefit from the installation of a better LDO in that position but that can wait until I am clearer in what is happening in my ‘patient’. I like inherently stable regulator designs.

This all comes back to what I said earlier in this thread. When providing power to a microbolometer the designer needs to be very careful to ensure that the supply rail is very low noise. Any noise in the supply rail can act upon the analogue elements of the Microbolometer-ROIC combination and create noise on the output scene data. Separate digital and analogue regulated supplies are highly desirable, as found in many analogue/digital designs that require low noise levels in the analogue stages. In the Seek Thermal design I am seeing a budget solution for the Microbolometer supply rail. A RAW 3.7V to 5V SYS supply rail with very little filtering feeds a tiny LDO, that also has the minimum of filtering and noise reduction, and that LDO directly influences the noise performance of the Microbolometer. I would normally expect to see more filtering of the LDO output to remove as much noise from the supply rail as reasonably possible. Maybe I am wrong and Seek Thermal have selected an ultra low noise LDO that does not produce significant noise at its output ?

Interesting stuff to me as I wonder how much, if any, random noise is injected into the Microbolometer in a Seek Thermal camera that appears to be working OK. This Microbolometer supply design may deserve further investigation. Sadly there is also the issue of noise created within the Microbolometer ROIC and that is beyond our reach to address.

Fraser
« Last Edit: July 26, 2020, 11:15:50 am by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
Interesting article on reducing power supply noise :)

https://www.electronicdesign.com/power-management/power-supply/article/21808839/3-ways-to-reduce-powersupply-noise

It is not as simple as slapping a chunk of capacitance across a supply rail without first considering what you are trying to filter out of the supply and the appropriate specifications of the capacitor used. A combination of L and C filtration is what I would expect to see on the output of an common LDO supplying a Microbolometer. Some LDO’s are better than others when it comes to noise and stability. We could be seeing an unfortunate choice of LDO in my patients case.

Fraser
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
I forgot to mention this earlier but when looking at Mikes teardown of a prototype Seek Reveal Pro he highlighted the presence of a Cortex M0 processor on the Microbolometer PCB and the mystery of why it was needed. Well there does not appear to be a M0 processor on my production Seek Reveal Pro Microbolometer PCB. I am wondering whether the Microbolometer PCB used in the prototype was the Circa 2016 Seek Thermal J3 Microbolometer PCB that is normally mated with the “603” processor PCB. In the “J3-603” configuration there is SPi control of the FFC shutter and a Clock generator on the Microbolometer PCB.

Having an M0 processor mounted close to the Microbolometer would not be my choice of design so I am pleased that it appears to be absent in the Reveal Pro core. I wonder whether it has been moved/removed in the more recent Mosaic cores as well ?

Could it be that early Reveal Pro cameras used the J3 microbolometer PCB and later versions of camera were fitted with the Mosaic Microbalometer PCB ?

I note that the same LDO format and capacitor bypassing is present in Mikes core.

Mike discusses the M0 processor at time index 21.40 in his teardown video....

https://youtu.be/mgaBAm_sRnk

Fraser
« Last Edit: July 26, 2020, 12:07:41 pm by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
A picture of the standard Seek Thermal (non pro) Microbolometer PCB used in the seek Reveal that was shown in the Hackaday teardown. Compare it to my picture of the Pro Microbolometer PCB.

The same LDO is used..... is that a “Q” I can see on it ?

Fraser
« Last Edit: July 26, 2020, 12:26:48 pm by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
Yay  :-+ I think I found the identity of the LDO  :)

I believe it to be a Texas Instruments part.... LP5907A28YKMR as that is a 2.8V LDO with a marking of “Q”  :-+

It is also an Ultra Low Noise part designed for sensitive analogue applications .... that fits its use in the Seek Thermal core perfectly. So they were trying to keep noise injection into the microbolometer to a minimum  :-+ The datasheet states that no “noise bypass capacitor” is required. Maybe the problem lies with the output MLC as 1uF is specified for stability.

https://www.ti.com/lit/ds/symlink/lp5907.pdf?ts=1595767306668&ref_url=https%253A%252F%252Fwww.ti.com%252Fproduct%252FLP5907%252Fquality

Sorry to detail every step in this investigation, but I hope it helps others when carrying out their own investigations. Note how a I am not diving in with the probes and test kit. That comes only when I understand the circuit that I am probing and what I should expect to see. In this case I will be looking for a signal that should not be present on the Microbolometer supply rail so it’s pretty simple. By investigating the circuit and identifying the components used, I can now determine what is failing and the correct replacement parts.

Fraser
« Last Edit: July 26, 2020, 12:55:56 pm by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
Test complete on LDO output. Pictures attached.

Low Noise LDO ? Hmmmmmm ...... nope  :--

Signal is AC coupled to remove DC content. Pk-Pk measurement of the unwelcome signal is approx 200mV  :(

LDO specification added.

Sorry about the awful pictures. These were just 'working pictures' for my use and I thought I saved proper pictures to the USB stick...... I have not used this particular DSO for some time and must have done something wrong as files saved, but no files found on the USB Stick  :palm: I might repeat the exercise again later but other pressing tasks are tearing me away from this investigation.

Fraser
« Last Edit: July 26, 2020, 03:07:25 pm by Fraser »
 

Offline IwuzBornanerd

  • Frequent Contributor
  • **
  • Posts: 268
  • Country: us
Holy crap!  That's 7% of full scale; does it show up as 20-30C of image noise?

Perhaps that "Q" you see is actually a "C1"?  The only regulator I found that matches the pcb pad pattern on the dongle units is an On Semi NCP114.  Not sure if the marking on the part truly matches the specified marking, though.

[attach=1]
I am not opposed to exercise, unless it is an exercise in futility.
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
Hiya,

I cannot remember the temperature error on the Faulty Reveal Pro measurements, but it was significant !

Fraser
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
I just checked the pictures of the faulty Reveal Pro.

The measurements were jumping up and down due to the noise bars but as can be seen in the attached picture I was getting a reading of 70C when viewing an approx 20C ambient scene.

Fraser
 

Offline IwuzBornanerd

  • Frequent Contributor
  • **
  • Posts: 268
  • Country: us
I'd say that is in the ballpark.  :)

Also, after perusing your TI regulator spec., I see that the "C1" marking on the regulator in my old XR matches the marking specified for the LP5907SNX-2.8/NOPB 2.8V regulator, so it looks like you found the right part.  :-+
I am not opposed to exercise, unless it is an exercise in futility.
 
The following users thanked this post: Fraser

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
Sadly other commitments are taking higher priority than the Reveal Pro as they are outside jobs and we have good weather at the moment  ;)

It would be soooo easy to just replace the MLC’s and even the Regulator IC as well but I want to get my DSO setup on it again to record the noise signal with proper earth connections and recording to the DSO flash memory stick for better quality images. I do this as a record of what can occur in these camera cores. I currently believe that the Regulator and MLC bypass capacitor at its output are causing the noise issue with the camera but until I have a well configured probing setup and later resolve the issue, I cannot be certain. I am in no great hurry anyway but like to record details of failures in case they assist others in the future.

In the mean time, I found an interesting article reprint discussing the interaction of regulators and their associated bypass capacitors that is well worth a read. How many of us just deploy a regulator with a pair of bypass capacitors that we have always used, without considering the effects of this capacitors on the regulator ? I know I do ! I use LM78 series regulators with a couple of 0.1uF or 0.47uF polyester capacitors and think nothing more about it. What about paralleling bypass capacitors ? Is that a good idea ? Read on.........

Part 1

https://www.edn.com/why-bypass-caps-make-a-difference-part-1-how-a-regulator-and-its-output-capacitor-can-interact/

Part 2

https://www.edn.com/why-bypass-caps-make-a-difference-part-2-power-supply-excitation-and-ringing/

Fraser
« Last Edit: July 28, 2020, 03:07:00 pm by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
I set up the DSO and its probe to record the input and output signals of the LDO Regulator positioned on the Microbolometer board.

The results are interesting but differ to those I captured previously. The previous test was 'quick and dirty' so was open to error. Maybe my probes 0V connection was not perfect ?

Anyway, tonight I attached a nice solid 0V wire to the PCB for the probe to connect to and was able to capture decent snapshots of the noise on both sides of the LDO. I have yet to study the results but will upload them now for readers interest. Feel free to comment if you want to:) The noise on the output of the LDO is my greatest interest and and we need to consider the ROIC that is driven off of the supply rail and whether it is causing some of the supply rail signals that we are seeing. Anyway, here are the input and output signal captures. The probe was AC coupled and set to its X1 position for this test so all voltages are as declared in the images. The DSO software has a feature where it produces a list of measurements so I have included that for interest. It provides the Pk-Pk voltage reading. Do not get too involved in the other measurements as they are not really that applicable to the tests here. Time I was in bed so TTFN.

Each picture is named as a LDO input or output. The first 4 are all Input captures whilst the rest are output.

Fraser
« Last Edit: July 29, 2020, 03:04:15 am by Fraser »
 

Online Fraser

  • Super Contributor
  • ***
  • Posts: 9543
  • Country: gb
Now the Output captures.......
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf