Help With Shrinking a Circuit/Making it Fit Into a Tight Space

[Ozen]

TL;DR Summary

I came up with a circuit to drive an LED, but some of the components will make it too large to fit in the device.

Attached is the schematic for the circuit. It uses a TPS61042DRBT LED driver along with a PSoC 4000 8pin microcontroller to drive a 10mA LED with push button controls for brightness. The problem is some components, like the inductor and sense resistor is way too large (over 6mm long!). This is supposed to fit into a micro reflex sight and uses a CR2032 battery, which already has a diameter of 22mm. The footprint it is housed in has a length of 47.5mm and a width of 25.4mm. The adjustment mechanism also takes up ~14mm. That leaves around 11.5mm of space (not including wall thickness) to fit a PCB with all this on it.

How in the world can I possibly shrink this system? I looked at alternative drivers and even linear drivers, but they all loop back into the problem of it doesn't fit my need; like incorrect output current that varies drastically with low voltages. Or in the rare cases it doesn't have that problem, it still requires these large components that just won't fit. The only thing I could think of is to have it have expanded across multiple PCB's that are stacked on top of each other, but that seems significantly more costly and becomes a mess of wires connecting PCB to PCB.

Any insight on this is greatly appreciated! Feel free to critique the circuit as well, this is my first completed circuit so I am sure it is mediocre.

 

[hutchphd]

As a first question: why is the sense resistor so big?

 

[Ozen]

As a first question: why is the sense resistor so big?

This is the only one I could find meeting the Ohm requirement: https://www.mouser.com/ProductDetai...=sGAEpiMZZMvdGkrng054t2CwxgQNvPAT2YN1ukgPwp8=. I looked at cutting it in half over two resistors, but 13Ohm ones are the same size.

 

[hutchphd]

Why is it so important to have that exact resistor? You are just choosing illumination...how accurate and precise do you need to be. Have you written a specification?

Also a 1206 resistor is 0.12inx.06in = 3mmx1.5mm so I don't understand the spec. Perhaps you need someone more experienced on the project ?

 

[Ozen]

Why is it so important to have that exact resistor? You are just choosing illumination...how accurate and precise do you need to be. Have you written a specification?

Also a 1206 resistor is 0.12inx.06in = 3mmx1.5mm so I don't understand the spec. Perhaps you need someone more experienced on the project ?

That resistor was a calculated value to get the current to an acceptable level. The driver calls for it to be a sense resistor since it is critical in getting a constant current. I don't need that exact resistor, I just need a sense resistor that is small with at least 25.4 Ohms of resistance.

For the 1206, I found one in 33Ohm, sadly none closer to the desired resistance. But it will fit. I am not sure why when I was searching for sense resistors none of these even showed up...

I would love to have somebody more experienced, but I'm not wealthy in the slightest. The money I could put up people scoffed at, so its whatever. I designed my optical system with 0 experience in optics and got it to turn out just as good as the Noblex reflex sights'. With this though, it is a bit harder to understand, like there are sooo many parts to consider. And then there's the diode which i have no clue what to use there other than the recommended ones in the LED driver datasheet.

EDIT: I see what you mean now! I didn't realize resistors come in standard sizes, it makes perfect sense to have them standardized yet it flew right over my head. That 26.1 Ohm resistor is a 1206, which puts at at half the size I was thinking of.

Also putting what I learned from resistor to and applying it with the inductor, I did find one that is super small. Although everything still is going to be a very tight fit (if it even fits), is there a way to minimize what I have, or simplify it?

 

[Borek]

Resistors come in many values, I have here 10Ω, 15Ω, 22Ω and 33Ω (all SMD 1206) - bought in a set of 86 values, 20 pcs of each, for something like $10 (doesn't mean they have to be available in such sets in such price at your place, but it is worth checking). They don't take much space, so to get closer to the needed value you can always combine several is some parallel/series combination.

 

[hutchphd]

I appreciate your initiative, and learning by doing is absolutely a fine way to proceed, but do not underestimate the size of the task. You will need to build this circuit in several iterations, each one getting closer to final product form. For the first iteration do not worry about the form factor, just build a circuit that proves the basic design and allows you to test and refine the user interface. Then you can worry about board layout and sourcing for a form factor prototype.
There are professionals who do nothing but circuit board layout...the good ones are wizards. Multiple boards is always a bad idea. "if you need to use a connector, don't".

Do not be surprised by the learning curve, The most difficult part will be prioritizing the important factors and that is hard won knowledge. Good Luck.

 

[Tom.G]

That 26.1 Ohm resistor could be replaced with a 27 Ohm which is a standard value and therefor more readily available, and probably cheaper. That's only a 3% difference, which is much less than the typical component tolerance.

[EDIT]
For the size, can you put the low profile parts on one side of the board and the tall ones on the flip side?

The 4.7uF at the LED driver output can most likely be much smaller, perhaps 100nF, which means it can be ceramic capacitor rather than an electrolytic. This will save both space and cost. As long as the switching frequency of the driver is 150Hz or above, the user won't notice any difference. (see section 9.2 of the data sheet to calculate inductor and capacitor values that will stay within the frequency capabilities of the driver)

Cheers,
Tom

 

[Ozen]

That 26.1 Ohm resistor could be replaced with a 27 Ohm which is a standard value and therefor more readily available, and probably cheaper. That's only a 3% difference, which is much less than the typical component tolerance.

[EDIT]
For the size, can you put the low profile parts on one side of the board and the tall ones on the flip side?

The 4.7uF at the LED driver output can most likely be much smaller, perhaps 100nF, which means it can be ceramic capacitor rather than an electrolytic. This will save both space and cost. As long as the switching frequency of the driver is 150Hz or above, the user won't notice any difference. (see section 9.2 of the data sheet to calculate inductor and capacitor values that will stay within the frequency capabilities of the driver)

Cheers,
Tom

Thanks for the suggestion! I will get back to you around Thursday when I have some free time to try a different capacitor. The 26.1 ohm resistor was only selected based off meeting the closest current to the LED. I'm actually considering prototyping two boards, one with a resistor around that range and one substantially higher (the reason is because my competitor says on their brightness level 4, the battery last for up to 4 years giving it around 0.007mA on that setting, which I call BS on).

As for using both sides of the board, got to admit I didn't even consider it! Although doing so will limit me to only the front cavity of extra space. I will post a picture when I reply on Thursday of the tight area I am working in.

 

[Rive]

CR2032 battery, which already has a diameter of 22mm. The footprint it is housed in has a length of 47.5mm and a width of 25.4mm

That's a big one. Try to find https://hu.farnell.com/search?st=cr2032%20holder There are so many variations.
Also, you should consider the actual shape instead of the xx * yy size.

 

[Ozen]

That's a big one. Try to find https://hu.farnell.com/search?st=cr2032%20holder There are so many variations.
Also, you should consider the actual shape instead of the xx * yy size.

That is the industry standard battery. To my knowledge, all battery powered scopes and sights use these, with the rare exceptions in say a holographic sight (which demands high energy usage) where they use AA batteries or similar.

 

[hutchphd]

I agree that the CR2032 is a wiser choice for marketing. How are building the circuit (board type etc )?
You understand that LEDs are very happy and efficient at low current. The ratings are typically max current. You want to maximize lumens per watt (for the particular supplier) and optimal light pattern and size/shape of package. Just build something chunky to play with...you won't get it just right the first time..

 

[Ozen]

That 26.1 Ohm resistor could be replaced with a 27 Ohm which is a standard value and therefor more readily available, and probably cheaper. That's only a 3% difference, which is much less than the typical component tolerance.

[EDIT]
For the size, can you put the low profile parts on one side of the board and the tall ones on the flip side?

The 4.7uF at the LED driver output can most likely be much smaller, perhaps 100nF, which means it can be ceramic capacitor rather than an electrolytic. This will save both space and cost. As long as the switching frequency of the driver is 150Hz or above, the user won't notice any difference. (see section 9.2 of the data sheet to calculate inductor and capacitor values that will stay within the frequency capabilities of the driver)

Cheers,
Tom

So I looked into changing the 4.7uF capacitor, but it won't actually be necessary. I redid my parts search and was able to find 0402's for all the capacitors and found a 1kOhm resistor for 0201. The sense resistor sadly can't be reduced without changing the the value of it, and I don't know how bright this will be so I cannot make a determination on a better resistance value until the test board is built and tried; but it will likely be increased resistance. So, to sum up the sizes of the parts: the capacitors are all 1mm x 0.5mm, 4.7uH = 1.6mm x 0.8mm, 1kOhm = 0.6mm x 0.3mm, D1 = 1.8mm x 1.45mm, LED Driver = 3mm x 3mm, PSoC 4000 MC = 5mm x 4mm, and RSense = 3mm x 1.5mm. The push buttons will also takeup space on the right side of the device, but I am still trying to determine the right ones for this.

This is this the footprint I am working with. there is a 0.5mm gap under the adjustment mechanism that a board could fit into extend to the far right cavity. The walls of what will house all this are 3mm, which is called out on the head of the screw. But you can see, it is a extremely small space to work in. Every way I think of setting this up, it doesn't appear to fit. Like putting the LED Driver in the right cavity, but it requires all its components (except MC and LED) to be in very close proximity, but it looks like there isn't enough space for that when you include the mount sizes. The MC would barely fit, if at all. And both of them cannot fit on the left side.

I agree that the CR2032 is a wiser choice for marketing. How are building the circuit (board type etc )?
You understand that LEDs are very happy and efficient at low current. The ratings are typically max current. You want to maximize lumens per watt (for the particular supplier) and optimal light pattern and size/shape of package. Just build something chunky to play with...you won't get it just right the first time..

I do not know the board type yet, I have been focused on making sure this circuit hypothetically wokrs, what the components are, their sizes, and how in the world to jam all this into that small space. I did not know they were efficient at low currents, that is very interesting and partly explains Trijicons ability to get theirs to 4 years battery life. But I still need to know what the brightness will be to adjust the current to as low as possible while remaining visible during the day...I did just reach out to the manufacturer of the LED's, so hopefully they can shine some insight to that problem.

 

[John Kovach]

I was not able to read your circuit. What bothered me is the discussion about component size. Why don't you use surface mount components? I know etching a board and soldering surface mount components take a certain amount of skill that many of us lack, but it may help.

 

[Tom.G]

A trick for figuring out tight layouts is to draw a footprint of the available space at 5x or 10x scale. Then do a cutout of each component at the same scale. You then have a jigsaw puzzle to play with and figure out what goes where.

When you realize you need a double-sided board, the footprint can be on a clear plastic sheet, such as a notebook page protector. The drawback of double sided is you will need an adhesive on the components, something like a double-sided version of "Scotch" brand Post-It tape that is readily removable.

Cheers,
Tom

 

[John Kovach]

Why johnny can't read. Sorry.

 

[Ozen]

A trick for figuring out tight layouts is to draw a footprint of the available space at 5x or 10x scale. Then do a cutout of each component at the same scale. You then have a jigsaw puzzle to play with and figure out what goes where.

When you realize you need a double-sided board, the footprint can be on a clear plastic sheet, such as a notebook page protector. The drawback of double sided is you will need an adhesive on the components, something like a double-sided version of "Scotch" brand Post-It tape that is readily removable.

Cheers,
Tom

I appreciate all the help you, and everybody else has given me to get to this point! I finally got the PCB shape with the parts footprints into EAGLE.

It seems to fit at a rudimentary level, still have to connect everything. But before I can do that, I do have some last questions on connecting and doing things. So the LED is a bare die with a P and N contact plate that will be wire bonded to a separate PCB mounted on the "floater"; how do you create a footprint to have two separate but connected wires running between the PCB's? This also has to be done for the silicone rubber buttons on the left side of the board. Is there a specific term I can lookup to find a footprint done on SnapEDA? And lastly the battery will be connected with a contact strip similar to the one shown here: https://patents.google.com/patent/US6327806?oq=reflex+sight. Except instead of going all the way to the LED mount, I just need to connect it to the main PCB. Is there some sort of standard to do this?

 

[John Kovach]

I appreciate all the help you, and everybody else has given me to get to this point! I finally got the PCB shape with the parts footprints into EAGLE.

View attachment 272016

It seems to fit at a rudimentary level, still have to connect everything. But before I can do that, I do have some last questions on connecting and doing things. So the LED is a bare die with a P and N contact plate that will be wire bonded to a separate PCB mounted on the "floater"; how do you create a footprint to have two separate but connected wires running between the PCB's? This also has to be done for the silicone rubber buttons on the left side of the board. Is there a specific term I can lookup to find a footprint done on SnapEDA? And lastly the battery will be connected with a contact strip similar to the one shown here: https://patents.google.com/patent/US6327806?oq=reflex+sight. Except instead of going all the way to the LED mount, I just need to connect it to the main PCB. Is there some sort of standard to do this?

It is difficult to "see" from reading text. What struck me was the word wires doing the connection between two pcb's if I resd it correctly. I've been taking things apart since I was a kid to see how they are put together, most times to see if I can fix something that broke. Wires are always a problem if they are flexed either during construction or use. If you are connecting two boards, spring loaded contacts might be best approach. I've seem some tightly packaged devices use thin cardboard to isolate components from metal surfaces and each other. So if there is some danger to the rest of the board , you can cut slots in the cardboard for the clips. You should be able to find clips that mount to a PCB and go from there. I hope I understood the problem correctly.

 

[Ozen]

It is difficult to "see" from reading text. What struck me was the word wires doing the connection between two pcb's if I resd it correctly. I've been taking things apart since I was a kid to see how they are put together, most times to see if I can fix something that broke. Wires are always a problem if they are flexed either during construction or use. If you are connecting two boards, spring loaded contacts might be best approach. I've seem some tightly packaged devices use thin cardboard to isolate components from metal surfaces and each other. So if there is some danger to the rest of the board , you can cut slots in the cardboard for the clips. You should be able to find clips that mount to a PCB and go from there. I hope I understood the problem correctly.

The term wires was a rough term I used because I don't know exactly what it would be called. This is what I am thinking of using to connect them: https://www.google.com/search?clien...le&aqs=mobile-gws-lite..#imgrc=msOltaLryb9VIM. I don't know the name of these or if they have a standard footprint. But these seem ideal since it would be flexible to connect to the other boards while also being bonded between the boards.

As for the cardboard suggestion, the PCBs don't need to be compartmentalized, the problem is the main PCB in the picture is flat on the bottom of the device. Another PCB is perpendicular to it and mounted on the "floater" which moves. The other PCB is also perpendicular to the main PCB and the floater PCB. Both of those PCBs have to connect to the main PCB. And lastly the contact strip has to connect to the main PCB. That's the problem I'm struggling with; if I can just get over this hump I should have enough of an understanding to make the prototype, test it and refine it based off of the performance.

 

[berkeman]

I don't know exactly what it would be called. This is what I am thinking of using to connect them: https://www.google.com/search?clien...le&aqs=mobile-gws-lite..#imgrc=msOltaLryb9VIM. I don't know the name of these

Well, they are called what the picture that you linked to calls them...

 

[hutchphd]

The flex-cables are probably a good choice but why do you need three separate boards? If you need to put in a connector, don't. This thing puts an LED spot on a fixed location on some moveable optics. Why are there three moving pieces? Can you move a small plastic mirror rather than the LED? You need some more cleverness here I think, but I don't see the whole picture.

 

[Ozen]

The flex-cables are probably a good choice but why do you need three separate boards? If you need to put in a connector, don't. This thing puts an LED spot on a fixed location on some moveable optics. Why are there three moving pieces? Can you move a small plastic mirror rather than the LED? You need some more cleverness here I think, but I don't see the whole picture.

I need the three boards due to where critical pieces have to be. Like 3 push buttons will be on the side of the device, so perpendicular to the main PCB. The other one is mounted on the floater for the LED. You suggested using a mirror, but that would not achieve what has to happen. The source of the light has to move in incredibly small increments (1 MOA); having a flat mirror bounce the light at the correct angle would not allow the light source to move. That would require a significantly more complicated system then using a 3rd PCB. My competitors only use one but they have a very primitive electrical system, I don't even think they use an LED driver, yet alone a microcontroller.

 

[berkeman]

I need the three boards due to where critical pieces have to be.

You may have already considered it, but keep in mind that for buttons you can use extended push-rods to actuate them away from where the actual buttons are, and for LEDs, you can use plastic light pipes to present the illumination a distance away from where the actual LED is. Just something to keep in mind...

 

[Tom.G]

...you can use plastic light pipes to present the illumination a distance away from where the actual LED is.

Or some plastic fiber-optic cable to allow movement. Although mechanically that may be slightly more complicated, you eliminate the flexible wiring to the LED. Considering the high mechanical shock environment, a plastic fiber-optic cable would be much more reliable than any metal conductors.

...for buttons you can use extended push-rods to actuate them away from where the actual buttons are...

There are also right-angle mount pushbuttons available, their actuation direction is parallel to the board surface.

Note: Pushbuttons (and switch contacts in general) that are switching very low currents and/or voltages need to have Gold plated contacts. These are rated for "dry circuit" switching, meaning they are switching almost nothing.

This is because most metals acquire a thin oxide layer on the surface. Low voltages can not penetrate the oxide layer without a physical wiping action of the contacts. Currents higher than 10's of milliamps can supply enough energy to help burn away the oxide layer.

Cheers,
Tom

 

[John Kovach]

Or some plastic fiber-optic cable to allow movement. Although mechanically that may be slightly more complicated, you eliminate the flexible wiring to the LED. Considering the high mechanical shock environment, a plastic fiber-optic cable would be much more reliable than any metal conductors.

There are also right-angle mount pushbuttons available, their actuation direction is parallel to the board surface.

Note: Pushbuttons (and switch contacts in general) that are switching very low currents and/or voltages need to have Gold plated contacts. These are rated for "dry circuit" switching, meaning they are switching almost nothing.

This is because most metals acquire a thin oxide layer on the surface. Low voltages can not penetrate the oxide layer without a physical wiping action of the contacts. Currents higher than 10's of milliamps can supply enough energy to help burn away the oxide layer.

Cheers,
Tom

I like the comment about the thin oxide layer on contacts. Years ago I was writing assembly language software for a precursor to the PC, an intelligent terminal used for data entry. The engineers used to remove boards from the prototypes use erasers on the gold contacts and sometimes reseated the cards with their shoe. I still have trouble accepting that gold plating contacts can oxidize.

This website is truly AWESOME, not so much for the advice you get, but for the opportunity to communicate a problem you are solving. When I was working problems surfaced that took weeks to resolve. What I discovered that the best help you could give someone was to listen and keep the conversation going until the speaker resolved the problem. You may not fully understand the problem a coworker is having but listening and offering suggestions are asking questions always resulted in the problem solver finding a solution even though you had no clue what the solution might be.

So yes, you get excellent advice and information on this website, but its biggest benefit is being able to talk out the problem you are having until you can form a solution in your own mind.

 

[Ozen]

You may have already considered it, but keep in mind that for buttons you can use extended push-rods to actuate them away from where the actual buttons are, and for LEDs, you can use plastic light pipes to present the illumination a distance away from where the actual LED is. Just something to keep in mind...

Extending the push rods would go back into the issue of not enough space. And the buttons will be silicone rubber molded so I'm not sure about the stability for that either. Fiber optics like Tom suggested after you have the disadvantage of it changes the way the light emits from it. So with the selected LED, I know it produces a light beam that covers the entire lens so that no matter where your head is aligned you see the reticle where you should hit (parallax errors apply). Using the fiber optic likely changes the emission angle along with forcing me to move the adjustment mechanism back to get the fiber optic in the right position. And the nail in the coffin for that is I'm not a full fledged optical engineer, everything is self taught with discussions with my local university, I am not confident it would be worthwhile to invest another month looking into this when multiple PCB's is a clear solution. I appreciate the recommendations though!

Or some plastic fiber-optic cable to allow movement. Although mechanically that may be slightly more complicated, you eliminate the flexible wiring to the LED. Considering the high mechanical shock environment, a plastic fiber-optic cable would be much more reliable than any metal conductors.

There are also right-angle mount pushbuttons available, their actuation direction is parallel to the board surface.

Note: Pushbuttons (and switch contacts in general) that are switching very low currents and/or voltages need to have Gold plated contacts. These are rated for "dry circuit" switching, meaning they are switching almost nothing.

This is because most metals acquire a thin oxide layer on the surface. Low voltages can not penetrate the oxide layer without a physical wiping action of the contacts. Currents higher than 10's of milliamps can supply enough energy to help burn away the oxide layer.

Cheers,
Tom

The push buttons on the final product (hopefully prototype too!) will be silicone rubber molded, I don't think they offer them with right angle mounting? If they do, then that would certainly solve having to use a separate PCB for the buttons!

That is extremely interesting about the gold plating! I don't know whether they will need it; I have yet to calculate the current and voltages the button will experience, mostly because I am not quite sure on it lol. So the buttons tap power from the main bus, but only when pressed, then it is met with an intersection with a 1k resistor and an input pin. Getting the resistor current is easy, but what about the button and the input pin? My very basic understanding is telling me it will be ~3V (with slight drop from the power button and adjustment button), but current wise, no idea. But that's a different problem I'm researching!

The mechanical shock is another interesting point you bring up. Is there a way to epoxy the cables connection? Or perhaps a similar loose flat cable that can be soldered? Dismantling this device should only happen under my care and if their is a PCB issue, the easiest solution is to scrap it and put in a new one. Although I doubt an electrical issue will occur since I plan to personally test these (depends if ammo prices drop so it is economical!).

 

[Tom.G]

Using the fiber optic likely changes the emission angle along with forcing me to move the adjustment mechanism back to get the fiber optic in the right position.

The emission angle is reported as Numerical Aperture, or NA, of the fiber. It is the sin of the half-angle of the light cone. Note that this applies to both the emission angle and to the acceptance angle of light exiting (entering) a fiber. That is, any light outside that cone will not enter the fiber.

from: https://www.newport.com/t/fiber-optic-basics
--
Fiber Parameters
Numerical Aperture (NA)


Figure 5: Numerical Aperture

The Numerical Aperture (NA) of a fiber is defined as the sine of the largest angle an incident ray can have for total internal reflectance in the core. Rays launched outside the angle specified by a fiber's NA will excite radiation modes of the fiber. A higher core index, with respect to the cladding, means larger NA. However, increasing NA causes higher scattering loss from greater concentrations of dopant. A fiber's NA can be determined by measuring the divergence angle of the light cone it emits when all its modes are excited.
Qualitatively, NA is a measure of the light gathering ability of a fiber. It also indicates how easy it is to couple light into a fiber.
--

So the buttons tap power from the main bus, but only when pressed, then it is met with an intersection with a 1k resistor and an input pin. Getting the resistor current is easy, but what about the button and the input pin? My very basic understanding is telling me it will be ~3V (with slight drop from the power button and adjustment button), but current wise, no idea.

Since the input current to the MCU pin is quite small it can be ignored. Therefore the button current will be (BatteryVoltage)/1k, the button voltage will be (BatteryVoltage).

Is there a way to epoxy the cables connection? Or perhaps a similar loose flat cable that can be soldered?

Sure, you can epoxy the connectors after assembly. Though it's usually faster and easier to solder the cable to the boards. Ribbon cable is available in pre-cut lengths with the ends stripped and conductors formed into pins ready for board insertion and soldering.

BUT what about the unsupported cable between boards? That will move around with the shock and eventually fatigue and fail. You could pot the whole assembly but that is time consuming and labor intensive for small quantities (been there, done that), not recommended if at all avoidable.

Cheers,
Tom