Desired Specification for a new Driver?

[Immo1282]

Most people round here tend to use thermal adhesives to mount small drivers to heatsinks - which while compact, is far from ideal as it's really hard to take a driver off a heatsink if it's epoxied in place I'll definitely bear this in mind.

Yep - I totally agree with you on that solder-mask point We like green and red at work, Green for production for cost purposes, red for prototypes - It makes it easy, Nobody takes red boards outside of the building and none are ever sent to customers. Will be taking hints from my colleagues when i do lay it out, so I'll do my best to stick to sensible clearances and common sense layout. Still at a concept stage in my mind so that's some way off yet...

I've been pondering on what to do with this project for a few months, and whether to prototype something with digital power control, or whether to stick to a trusty analogue controller solution with a digital reference from a micro-controller. My goal with this is to build an all-in-one board that drives a diode and a little OLED display, with the ability to program in a current-to-power curve for each diode. Outcome for this would be a build with buttons to set the laser power, which would be displayed on the small screen. Ambitious - but as far as I'm aware nothing like this exists being sold right now

 

[rustynuts]

Most people round here tend to use thermal adhesives to mount small drivers to heatsinks - which while compact, is far from ideal as it's really hard to take a driver off a heatsink if it's epoxied in place I'll definitely bear this in mind.

Yep - I totally agree with you on that solder-mask point We like green and red at work, Green for production for cost purposes, red for prototypes - It makes it easy, Nobody takes red boards outside of the building and none are ever sent to customers. Will be taking hints from my colleagues when i do lay it out, so I'll do my best to stick to sensible clearances and common sense layout. Still at a concept stage in my mind so that's some way off yet...

I've been pondering on what to do with this project for a few months, and whether to prototype something with digital power control, or whether to stick to a trusty analogue controller solution with a digital reference from a micro-controller. My goal with this is to build an all-in-one board that drives a diode and a little OLED display, with the ability to program in a current-to-power curve for each diode. Outcome for this would be a build with buttons to set the laser power, which would be displayed on the small screen. Ambitious - but as far as I'm aware nothing like this exists being sold right now

if this is for a pointer then size is the priority. I use thermal conductive tape for my builds with 2-56 or 0-80 screws to keep the board, or high wattage leds on, to remove the board use alcohol to loosen the tape the board will fall off. As for my designs glue for me is a no no for high value items that will be reused. companies over use glue and plastic for the designs because it saves money and that is good if you want to maximise proffits. I just dont make compermises with personal items . dont forget about thin boards flexing just look at all the iphone failures caused by board flexing. I would like to see two versions, a driver for lab work and a portable small one . As one board wont fit both needs. would be nice to dial it in with knobs to control the software. look at Agilent Technologies oscilliscopes, they took away the knobs then after loosing alot of custoners the knbos were back for some reason.

 

[Immo1282]

I just dont make compermises with personal items . dont forget about thin boards flexing just look at all the iphone failures caused by board flexing. I would like to see two versions, a driver for lab work and a portable small one . As one board wont fit both needs. would be nice to dial it in with knobs to control the software.

I would like to eventually put it in a relatively compact pointer - but that's a pipe dream until I have some working hardware on my bench. Once I get past concept & design stages, will start out prototyping it as a larger board - I could always attempt to scale things down if there's demand. I don't see it ever fitting inside a standard Aixiz module (board size of 9x12mm or thereabouts), and yet remaining possible to hand assemble (i.e. no BGA or extremely fine-pitch SMD).

I should be able to keep the input options flexible enough with firmware variants - Analog knobs are nice - but seem a little redundant to me in this case. Maybe a nice quadrature encoder knob/button combo, but I absolutely could use a pot if there was a spare ADC on the chip I end up using.

 

[rustynuts]

What is your background? Are you a electronics engineer? what is your comfort level in electronics? can you take a breadboard and design and build using only spec sheets and no computer aid. It is hard for me to know if you are biased toward analog or to digital.
I am very new here. I have taken an interest with this project. Can you fab a good enclosure for the control panel or make a heat sink for your driver? or just electronics? do you use cad for the board layout?

I am a jack of all trades and a master of none. I grew up with analog and not with digital circuits. my only real experience is in repair for the digital stuff. The high failure rates keep me in business, so I dont mind that at all. This is why the analog route is usually my first choice. Also I use manual machines, and must consider that limitation for my designs I make sure boards are easy to install and remove.

Are you still wanting to build a ultra feature packed driver. or something with only a few well chosen adjustments? I dont think one board will cover both targets because , If the user needs a computer to program it this will limit your demographic user to be at high skill level.

I feel that for a ultra reliable driver is to be made economically, It should not need a computer to use it nor use a programmable chip,
It is often too tempting not to do the homework and to ignore the protection side of things especially when the project is near completion.

This is the most dangerous time for any build. It is when the job get old, and boring and no longer any fun. And you just want something working for your efforts. This is when corners are cut.
I know that that the programmable chips do have a high failure rate. and it is a matter of when and not if they act up. If the brain of a power management circuit acts up It will likely kill these ultra sensitive and sometimes expensive diode. which for me would limit its use to the cheap diodes.
on the flip side it definatly will be a awesome to have all of that flexibility. But if it is to be reliable there will be more to it than just the control interface chip, you will also need to at least incoperate a well thought out watchdog monitor protection to monitor the parameters of that driver. If you choose to incorperate proven and tested protection circuits, I suspect testing it will be the toughest challenge and will take most of the time spent.

 

[Immo1282]

Thanks for weighing in with such a detailed comment

I'm a recent Electronics graduate working in the field of mathematical modelling and Digital Control design for fairly high-power switchmode power supplies - The main reason I want to do this project is to actually practice the practical electronics design skills that my degree (and to a lesser extent, my job) never really covered. Best way of doing that is to dive straight in I reckon!

I have a highly digital bias to my current knowledge & skills, but I realise that Digital Control (i.e. a micro-controller directly driving the switching elements of any converter) is tricky to get right, especially when it comes round to reliability. I think the approach I want to take is an analog power design, with a digital interface (perhaps something like a digital potentiometer). This way, tried-and-tested analog designs can be used for the actual power base, and hopefully with some insight from some of my colleagues at work who've been doing analog electronics longer than I've been alive - the appropriate safety & protection circuitry can also be built into that.

I understand your concerns with reliability of micro-controllers - but properly used, with well written firmware, many modern uC's can be extremely reliable. Fault-tolerance is an important facet of any design like this - but designed right I don't believe this to be an impenetrable barrier to what I want to do.

I'm at the stage of the project where I pin down a requirements specification. I know in my head what I want from it but have seen projects at work that have been going on mostly unsuccessfully for 2-3 times as long as they should because of poorly specified systems.

I realise that there's a slim subset of the people looking for drivers that want something like this (might only be 1 person who cares hah...) but I feel that it's still something that I want to build. It's not going to be possible to make and sell this for comparable prices of most drivers that are sold - but it's the process and not the end result that I'm really looking forward to here

 

[rustynuts]

Thanks, for the great reply,
I never got past vocational school, I invent and build quite a bit, I feel the same as you I do this work to learn. knowing how is the end result, not the widget made. It is a great that You want to use the digital pot.
Yesterday I was asking if I could use aE3614a older hp power supply to directly control a 445 blue. The hp accepts analog programming 0-10v for current and amperage adj. I have use several of this type of power supplies with analog remote programming I want a super clean 0-10v voltage for controlling them. This will greatly add to their usefulness at my lab. maybe this can be a useful tool, it will add value to cheaper older power supplies. Is using one the ideal way to control that HP. maybe not directly the problem is the 0-10v the pot cant handle that voltage. due to the age i think the wire wound pots in the ps has too much noise. and may spike that diode.
any suggestions?

 

[Immo1282]

Nothing that I've used in the past unfortunately - my thoughts were either to use dedicated digital pot chip - or simply a DAC output on the microcontroller.

Your 0-10V reference might be fairly straightforward as it's a voltage - you can smooth out whatever you're giving as reference with a cap or a filter as it's a voltage reference controlling the current adjustment - so it's much easier to get a ramp between signals?

 

[rustynuts]

Thanks,
if you do get a digital pot and power one up, I would love know what you find out about them.
after your R&D is done how are you going to get a finished pcb made for your driver?
Just thinking about your project
The issue i see from the available hosts, thermal bridge issue is the orientation and poor access to the driver board and poor air circulation
gluing these small boards to hosts inside a round tube just not a best practice in my opinion I worked as a machinist for 8 years and we really are picky. If you want to do the electronics, I can design a practical host. it is not going to be fancy or shiny, or cut with wierd angles to make it look cool, first thing first it must have high thermal performance also making it easy to modify and easy to get to the electronics without screwing it apart.
also, I am designing a water cooled system for my first test block , I am going to use the radiator off a ice maker the magic is going to be how the block is designed. keep tuned in.

 

[Benm]

If you're doing the design from scratch, i'd still say go with the uC. There are some many changes you can do in firmware instead of re-wiring hardware that just makes it worth it.

Industry in general seems to have embraced this route too, in many products you'll find the "nameless 8 pin chip" that seems to be doing most of the work, and the vast majority of those are just uC's, but factory programmed and with the number removed or replaced by some internal code.

One reason to use them is their very low price too. In small quantities they cost $1 or so, but if you get a reel of a few thousands that cost drops to a tenth of that, even including factory programming.

I personally love to do the analog approach with opamps, transistors and mosfets, sense resistors and the whole load of passive's that come with that too. You can make excellent drivers doing so, but not to a cost or form factor that the market demands. For a lab style laser it's fine to have all that as analog circuitry on board, but for small handheld pointer it's not really feasible and the uC route just makes more sense.

 

[Immo1282]

if you do get a digital pot and power one up, I would love know what you find out about them.

No problem

how are you going to get a finished pcb made for your driver?

I'll probably get blank boards made up for all prototypes by someone like JLCPCB and hand assemble. I can't see this being something that's going to be in high enough demand to scale up to full automated assembly, but that expertise exists with people at work so if I get to that point I can ask for some advice.

If you want to do the electronics, I can design a practical host

That's an incredibly generous offer - I've no machining skills but I can 3d model serviceably and print in plastic... Thanks!

If you're doing the design from scratch, i'd still say go with the uC

Thanks Ben - that was what I was thinking, to try and keep the circuitry as simple as possible and to offload as much as I can to a uC. Actual digital control is still something I'm not sure is a good idea, but there's a good number of approaches that would work. Think I'll sit down and do some research once my family are sick of me next week

Thanks guys for both of your input - very helpful and insightful!

 

[rustynuts]

do you ever watch yoube?
I've learned alot from Watching the EEVblog videos..
if the host is made right, you can probably stack the boards if you cant fit everything on just one.

 

[Immo1282]

do you ever watch yoube?
I've learned alot from Watching the EEVblog videos..
if the host is made right, you can probably stack the boards if you cant fit everything on just one.

Yeah - i've been watching Dave Jones for a few years I'm going to see about making something on a large board for development reasons then shrinking it down to something that'll fit in a host. I'm pretty certain that multiple boards with some form of interconnect would be preferable later - but I'm less concerned about physical layout etc as I can always change the layout but retain a common design/schematic.

 

[rustynuts]

how much power do you want your driver to go up to? I guess it really depends on the size of the host. it should not be too hard to design it to go up to maybe 10amp. maybe have a power board and a control board. which can be interchanged or at least maybe make it easy to add higher amp components to beef it up if it is a single board also dont like it when chip numbers are scraped off. also i like the schematics to be provided. usually when I get older test gear, I go out of my way and find and will buy the paperwork for it with the schematics.

 

[Benm]

I like Dave's videos for sure - and there are several other channels like bigclive or julian illet amongs many others i'd recommend watching. I don't always agree with them, but their content is interesting to watch regardless.

Designing a laser driver has a lot of constraints: small envelope, limited thermal dissipation, but a wide range of voltages and currents selectable within those limitations. This already drives you into switching regulator design, and at that point you need to choose if it should be boost, buck, or capable of both (the latter making it much more complex).

As far as the actual switching element: the uC is not likely to be able to handle the current required for that, but you do get very compact mosfets that can handle a LOT of current for their size, and at pretty low cost. Just look at what goes into lithium battery protection circuits on a powerbank - typically only 2 tiny chips that can handle charge and discharge control at several amps - as well as do the under and over voltage cut-off for the cell, and they cost only a few cents in bulk.

 

[rustynuts]

Machining some test jigs now, Also rounding up materials for a small optical table. Have you started a layout yet?
I would like to test your first driver

I am starting / have a small business doing micro soldering and Electronics repair. I take jobs for the local shops when they are ill equipped. Also I am adding lab testing services, If you are still eager to go, And if you can round up all the parts for a first piece. I would be happy to help you get a first Driver going, I have bought diodes from 100mW to 7W and ran some tests, My linear driver is now covered . I have designed and built a few of them now.

I am looking into some extra laser power meters, Currently I am using a modified laser bee 5 watt . currently designing a new laser Power meter incorporating a optical reducer , thought I should be incorporating a Hershel wedge to get a higher power measuring capacity. I already have the wedge for solar observing. Any thoughts?

 

[Immo1282]

Have you started a layout yet?
I would like to test your first driver

Sorry about this - in all honesty I'm at least a few months away (probably more like 6) from the stage of laying anything out - Life suddenly got busy in the new year so I have to push this project to the bottom of the stack as other things have taken control a little more than I'd have liked.

I'll let you know when i've got something tangible - don't worry! I'm very grateful for your offer to test what I have - I've got some kit here to check things but it'd be nice to get a 2nd opinion