Forum - Electronics FAQ

I can't find any infos for the LEDs, for example the size, vendor, et al. Can anybody provide more?

Hi Violinzju,

The information is on our Master Parts List under 'Electrical Components' on the 'Head Mounted Scope' sheet. I can see how you might expect to find it located somewhere else on our site so I will go ahead and add the information to those pages as well.

Just to save you some time looking it up on our parts list, the excitation LED we currently use is a Luxeon Blue (P/N LXML-PB01-0030). Hope this helps

Hi, Tristanshuman!

About DAQ PCB board, I got a confusion in GP1 and GP2.

GP1-Ground plane GP2-Power plane Is it right matching to layer?

Hi Jys000415,

Yes this is correct and I just updated 'Miniscope_PCB_DAQ_Fab_Info.docx' on GitHub to clear up the confusion. So the layer stack should be:

• Top Layer (Component Layer)
• GP1 (Ground Plane Layer)
• GP2 (Power Plane Layer)
• Bottom Layer (Signal Layer)

Hope this helps and let me know if you have any other questions!

Hi, I can't find DAQ PCB components, U9, C80 and C81 in part list.

Could you let me know the information of those components? (Serial number or ordering sites)

Thank you.

Hi Jys000415,

U9, C80, and C81 should be left unpopulated.

Hello,

I was going through the datasheet for the CMOS sensor MT9V032 and I noticed that this sensor has can have a 10-bit serial output (LVDS). The figure on page 64 of this datashet also shows that using a deserializer DS92LV1212A up to 8 meters away can allow the transmission of signal through three cables.

However, the mini scope is equipped with a serializer (DS90UR913A) on the CMOS sensor and deserializer (DS90UR914A) on the DAQ board.

What is the advantage os implementing these additional chips? I could see that they could add more functionality, such as controlling the LED brightness, but is there any additional advantage?

Any input would be greatly appreciated.

thanks.

Hi Walter,

Great questions. Our first version of the Miniscope system did use the LVDS output along with the DS92LV1212A deserializer. While it performed adequately, it had the following issues:

1. We hand made the cable assembly that ran between the CMOS PCB and DAQ PCB which was a pain. I am sure we could have had a custom cable assembly made but it would have been expensive and required a very large order.
2. The LVDS signal seemed to be pretty temperamental and movement/shifting of the twisted wire pair carrying the LVDS data could easily cause the sync between the sensor and the deserializer to fail.
3. The cable assembly was 8 wires in total: 2 wires for the LVDS, 1 for power, 1 for ground, 2 for I2C, 2 for the excitation LED.
4. Only a few commercially available CMOS imaging sensors support LVDS (and usually they send data across multiple LVDS line pairs simultaneously). This severely restricts the future development of the Miniscope system. For example, the Inscopix microscope system relies on an Aptina CMOS sensor which outputs LVDS. Their cabling is much thicker (I think it is a 12 wire assembly), much more expensive, and less flexible than ours. Also, they are pretty much stuck using their current sensor or a few other sensors on the market.

Some of the advantages of using the serializer/deserializer pair in our system:

1. Extremely robust data stream. The serializer/deserializer dynamically adjust for attenuation in multiple frequency bands.
2. Data is sent over standard 50Ohm coax cable. This cable is super easy to find and relatively cheap.
3. System supports Power-Over-Coax which means we can send power and data on the same coax cable.
4. System serializes not only the pixel data but also I2C control lines and 4 GPO lines. Again, this means we can run everything off of a single coax cable and have access to 4 more output channels to the head mounted scope.
5. The power and data stream can be sent through a commutator (we use low torque commutators from DragonFly). When using our previous version with LVDS we were never able to recover the data signal though a commutator. This is also why Inscopix's system cannot be commutated.
6. We have successfully run our system on coax cables down to 0.3mm in diameter.
7. Since the serialization is being taken care of by a separate chip, our electronics support the vast majority of commercially available CMOS imaging sensors (the industry standard for CMOS pixel output is a parallel data bus with frame valid and line valid signals).

Thanks for the detailed response, it really brings to light a lot of aspects to take into account.

Since currently we are using the parallel output of the CMOS (MT9V032, area footprint 192.5 mm2), could we replace this sensor for something like the OVM7692 (footprint = 8.1 mm2)? of course, with the proper adjustment of the circuit. This sensor is not as good as the MT9V032, but it is significantly smaller, and could decrease the size of the top part of the microscope.

Is this something worth pursuing?

Hi Walter,

Yes, in theory you could replace the MT9V032 with the OVM7692. You would need to look into their control bus to see if it is compatible with I2C. My main concern would be its pixel size being only 1.75um x 1.75um. One reason we are using our current sensor is it has pretty large pixels, 6um x 6um (almost 12 times the area of the OVM7692), which improves SNR. At 1.75um pixels and 640 x 480 resolution, you would have a very small field a view without modifying the optics of the scope to decrease its magnification.

I plan at some point to switch over to an E2V CMOS sensor which has better performance, higher resolution, lower power requirements, and a smaller footprint than the MT9VV032.

If size and weight are a major concern then you may want to consider having the bare die of a CMOS sensor wire bonded to a PCB. If you take a look at almost any image sensor chip the majority of its size is due to the ceramic housing, not the actual IC. Wire bonding the bare die of the MT9V032 (or something similar) would save an incredible amount of size and weight (~1/4 the footprint and about .8 grams lighter). We are actually interested in pursing this ourselves so if you are interested let me know. Finding an assembler to wire bond the sensor is easy, sourcing the bare die and handling the PCB afterwards is a bit trickier.

I assume the Inscopix scopes use this approach to reach their stated mass of 2 grams.