Why The Latest Changes To Spin 2 Don't Worry Jared

Marcos' unmaintained, but still in-use, Puma for FreeEMS circuit board/hardware design!
User avatar
jharvey
1N4001 - Signed up
Posts: 1607
Joined: Tue Jun 10, 2008 5:17 pm

Why The Latest Changes To Spin 2 Don't Worry Jared

Post by jharvey »

EDIT: Source of quotes in this post, and material to which it refers: viewtopic.php?f=53&t=1164

Hello Fred. Perceptions are certainly important to consider as a design spec. That's why I choose to reply to your post. I hope you didn't really expect to post such negativity with out some kind of inquiry, or reaction to your post. You imply some very strong things that I feel are simply untrue, so I'd like to request some clarification of your post(s). I'm not sure your post is really all that clear to most people.
(18:38:56) fredcooke: 1) added extra switching regulator without removing others, simply to reduce heat into the others, but adding board realestate and adding switching noise, probably, from a design that is highly noise critical.
Last I recall the board LxW have not changed, so I don't understand what you mean by increased real estate.

The SMP circuit has been prototyped and is being tested now under a variety of loads and conditions. If we can measure noise, then this circuit will not be used. However if the measured ripple is reasonable, then it will likely be included as an option. Last I recall the circuit that has been proto'ed has unacceptable ripple, but also hasn't really been looked at, so it's not complete yet. Could have a wrong part installed for all we know now. Is there a particular load condition you are concerned about? We can likely simulate that load to measure the noise replacing perception with measurable tangible results.
(18:40:30) fredcooke: 2) moving all fuel to220 devices to SMD devices, moving resistive p&h control to PWM (noisy) p&h control in the name of vibration resistance and meeting some arbitrary test standards mil spec or similar
Last I recall, the plan is to include the SMT option, and TO220 option. So one can choose which one they want. The SMT can be an OVP IGBT or MOSFET and driven directly, with out modification. The TO220 is driven by the LM chip and requires jumper wires to use with OVP devices. If you use the LM data sheets recommended transistor, you can use the same old circuit, but you require thru hole which has assembly costs associated. So as far as I've seen, you can put in what you want. It would appear your concerned that a DIYer can put in devices you don't approve. I don't understand the concern. You appear to be concerned that it's offering to many options and to much flexibility.

About the mill spec, that's a physical thing, not an electrical thing. The push for an SMT option is to meet vibration and physical stress design specs. The thru hole leads can't sustain those stresses. If that transistor could be found in SMT, I'm sure that transistor would be used as SMT.

About P&H PWM, there isn't any backing software for such an option, so I don't know what the switching frequency is, and therefore don't know why the concern of RF/noise. If one uses an over sized OVP IGBT or MOSFET, then writes some code to PWM it, they might get some weird results. But for now, the SMT would be for HighZ applications, and would allow one to experiment with the potential of using it as LowZ, but that's a long ways off. It again implies your concerned we are planning to offer to much flexibility.
(18:41:05) fredcooke: 1 is silly because there is very little heat anyway, a better solution would be to ditch the fancy smd regulator and move it onto the heatsink bus as a to220 (with appropriate leg layout)
You seem to indicate you have a lack of understanding of thermodynamics. Testing that regulator circuit is fairly straight forward, predict the maximum load, put on that load, then toss it in an oven set for around 110F (43C) and watch the noise floor rise because the LDO's internal compensation kicks in. The linear as design now, has a limited scope, and an optional SMP will widen the scope of applications. If you don't want that option, don't use it. Again it appears your concerned with the flexibility.

I ran out of time to finish my reply. Several of your above comments appear to be out of context to me, and I don't follow them. The design specs we have set forth on Puma certainly include perceptions, and negative comments will likely cause a reaction from those of use involved in the design. The goal is to make a quality product both in reality, and in perceptions.
Last edited by Fred on Thu Jul 07, 2011 11:30 am, edited 1 time in total.
Reason: Added link to thread referred to at the top for ease of finding if they separate over time.
User avatar
Fred
Moderator
Posts: 15431
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by Fred »

EDIT: The result of my concerns about the Puma project direction: viewtopic.php?f=9&t=1225

You're entitled to your opinion, and this is the place for it.

I wasn't aware that the to220 pads were being left on the board along with the SMD pads. Did you state that the non-p&h drive would be done with the SMD components and the P&H drive would be done with to220 parts? This even makes a bit of sense, but not as much as using the same to220 slot for either autofet or darlington+p&h chip. I'm not a fan of this extra complexity and double parts count stuff, personally. That is my opinion, yours differs, good for you.

By real estate, I mean board are used, of course. If less is used, there is more flexibility for better routing of other traces and more robust footprints, etc etc.
You appear to be concerned that it's offering to many options and to much flexibility.
It again implies your concerned we are planning to offer to much flexibility.
Again it appears your concerned with the flexibility.
Is what you wrote.
You appear to be concerned that it's offering to many options and to much complexity.
It again implies your concerned we are planning to offer to much complexity.
Again it appears your concerned with the complexity.
Fixed.

Fred.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
User avatar
jharvey
1N4001 - Signed up
Posts: 1607
Joined: Tue Jun 10, 2008 5:17 pm

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by jharvey »

Following the thread topic, I feel the changes we are experimenting with in the next spin require the prototypes and such analysis as the current regulator design is expected to exhibit noise issues under normal automotive environments.

Based on hands on experience with a switch mode power supplie build with TO220 bolted to the case (very similar to how it was done with PUMA spin 1) vibration and time will cause the TO220 leads to crack and fail. The power supplies I have fist hand experience with, had a 5% failure rate in the first year of service. Some things we recommended to the MFG, were to use spring clips instead of bolts, and some other mounting stress relief mechanism, like mounting the case with rubber grommets to prevent shocks from being well coupled to the case. This got the failure rate down to about 1% per year, and I'm left feeling it's simply bad mechanical design.

Now for a reply to the above and the other thread this was moved from.

I don't understand your complexity notes. My understanding of the plan, is to offer it as pre-assembled, removing the complexity of the standard setup. However if one want's to modify, or DIY from scratch, yes that is more complex then buying it. So because as a normal user wouldn't see any of that stuff, I say flexible, but I guess you're concerned with complexity of the optional items.

I understand there would be two standard potential builds. One with LowZ, and no claims of mill specs (even though it will be as strong as possible). The other build would be HighZ with SMT and mill spec's. The SMT would be over built, to help minimize heat, which technically will allow for a potential LowZ PWM P&H, but I haven't heard of any real plans to back it with software at this point. However if someone wanted to, why not try and see what happens.

Also locking that other thread where you try to create false impressions that PUMA is of poor design, is bad form man. Your not playing nice, and your not playing with logic. Your comment are purely emotion and it make me concerned about the quality of your code. If your code is as squishy and vague as your hardware notes, we've got problems.

Any how, I hear your comments as trash talking PUMA, even though it hasn't been drafted yet. The concepts above are only concepts, and we won't know what will happen until Marco gets a chance to draft something. Some concepts have been tossed out there, and you feel a need to create some negativity around it for some reason. Not much I can say, it's pure emotion. They are negative nonconstructive emotions and I hope those that read it lump it in the appropriate trash bin.
slacker.cam
QFP80 - Contributor
Posts: 69
Joined: Sun Jan 27, 2008 10:25 pm

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by slacker.cam »

Jared. I'm really coming in to all this part way through so I have minimal understanding of what's going on however I do have some thoughts regarding what you're discussing. I'm an electronics design engineer working for a company who specialises in power electronics so I may have something to add here.

Firstly, I don't think it matters whatsoever if the board is designed with all SMD, all through hole or a mix of both. This is from a reliability/vibration stand point ONLY, obviously target markets, automated assembly, etc, etc will change what's most attractive. But for simply getting some electronics together into a reliable circuit I honestly don't think it matters. My opinion is based on a number of observations. Auto manufacturers built countless ECUs using 100% through hole components including power devices mounted to the case for years before SMD became cheap enough to produce. All of the early Toyota ECUs use a basic single sided circuit board with to220 (for example) leaded devices bolted against the case for voltage regulators, fuel switching and whatnot. You very rarely hear of one of these old ECUs dying due to cracked pads, dry joints or broken component legs. 99% of the failures are due to the old electros going dry and/or leaking and corroding the board. I think if the technology is good enough for 30 years of service then it's good enough for us. A double sided PTH board provides even more support for the component legs and gives greater soldered area. The second observation is that a huge majority of power electronics these days still use through hole components bolted up to some form of heatsink purely because of the practicalities of thermal transfer to the heatsink. It's not like since SMD has come along that everyone has suddenly dropped though hole components because they fail all the time.

I know personally I have a habit of designing boards where I use all SMD components even when there really isn't any need. Probably because it makes me feel like I'm a cooler engineer when I get to break out the solder paste and hot air. Is there really any actual need for going to SMD for power components? Sure it makes sense for the small stuff as automated assembly will bang that out in no time. Another point to make is if you design in SMD parts and plan to have them assembled by a machine then reels of every component will need to be purchased. This could prove to be very costly for whoever is backing the production. How many boards would need to be run at a time?

On the topic of SMPS vs linear. What are the current issues with the linear option? It certainly seems like the easier route to take in my eyes.

I don't want to step on anyone's toes here. Just throwing my 2c out there :)
User avatar
Fred
Moderator
Posts: 15431
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by Fred »

jharvey wrote:Based on hands on experience with a switch mode power supplie build with TO220 bolted to the case (very similar to how it was done with PUMA spin 1) vibration and time will cause the TO220 leads to crack and fail. The power supplies I have fist hand experience with, had a 5% failure rate in the first year of service. Some things we recommended to the MFG, were to use spring clips instead of bolts, and some other mounting stress relief mechanism, like mounting the case with rubber grommets to prevent shocks from being well coupled to the case.
Good feedback!
I don't understand your complexity notes.
More components, more traces = more complexity. Just like more lines of code (traces) and more branches (components) make the cyclomatic complexity of code worse.
My understanding of the plan, is to offer it as pre-assembled, removing the complexity of the standard setup.
This reduces complexity to the user, not of the system. This may or may not be a good thing. Less exposed complexity where the complexity still exists by necessity is effectively keeping the user blind. If they want to remain blind, and get paid support, there are companies such as Motec who are more than happy to provide such services.
I understand there would be two standard potential builds. One with LowZ, and no claims of mill specs (even though it will be as strong as possible). The other build would be HighZ with SMT and mill spec's.
The SMT would be over built, to help minimize heat, which technically will allow for a potential LowZ PWM P&H
ONLY if there are SMD darlingtons and if the traces route back to the p&h chip and if the p&h chip can be configured for both linear and pwm operation on the same layout.
but I haven't heard of any real plans to back it with software at this point. However if someone wanted to, why not try and see what happens.
It will never be backed from software in the main MCU, and there are many many fundamentally good reasons for this. It is not a function that belongs on the main MCU, not at all.
Also locking that other thread where you try to create false impressions that PUMA is of poor design, is bad form man.
Look, first of all, I didn't lock it until after you disrespected my wishes. Secondly, I'm NOT trying to create any impression, false or otherwise, of Puma being of poor design. I'm planning to base my design from Puma, which entirely contradicts your theory. I really like the Puma design, overall. I just feel that the direction it has taken recently is not conservative enough to use as a default and reference design for the project and the project's image. That is all. I had subtly mentioned this in several places where it fitted, and a number of people had said "what do you mean, fred" and I got sick of typing it. Now I can just link it. The two threads have virtually the same title and thus will be found together and this thread, where you get your feelings off of your chest, un-editted by me, will serve as a comments thread for the other.
Your not playing nice, and your not playing with logic.
Yes I am, you just don't see it, or didn't, hopefully.
Your comment are purely emotion
No, they are not, they are founded purely on engineering and marketing logic. No emotions, what-so-ever. I assure you.
and it make me concerned about the quality of your code. If your code is as squishy and vague as your hardware notes, we've got problems.
LOL, now now, Jared, don't let your temper make you type things that don't make any sense. You have a lot of good stuff to add here, and everywhere on the forum. You detract from it all with bullshit little comments like this, though. Try to keep your emotions in check. If you're having troubles with that, let's skype and talk about it before making yet another forum mess that I need to either clean up or watch slowly fade.
Any how, I hear your comments as trash talking PUMA
Not at all, I like the original Puma concept. I like my Puma despite its many many flaws. I dislike some specific choices that have been made recently. I will fork, rename, and polish it in a way that I feel is beneficial TO THE PROJECT, not anyone's eccentric tastes, or specific needs or egos.
Not much I can say, it's pure emotion. They are negative nonconstructive emotions and I hope those that read it lump it in the appropriate trash bin.
It's not emotion, Jared. It's opinion, and justified opinion, for my own reasons, which are all about growing this project as rapidly and in-control as possible. It belongs on the table, just as much as any of your comments.

I removed 3 posts, 2 from our resident shit stirrer, 1 of yours, in response to him, bait firmly taken, hook set, fish on the fucking scales! ;-) They were entirely worthless.'

As for Cam's comments, for once, I agree entirely with them. I'd just like to add and clarify that it DOES matter whether you choose SMD if there is heat involved as that heat raises the temperature of the board, and all components on it, reducing reliability in the process. Sometimes this is OK, sometimes it's not. The heat has to be got out somehow, it can be, with bus bars and so on, but that takes up a lot of realestate. Or through radiation and convection, but this is inefficient and creates a small oven from the enclosure. Other than the thermal concerns, and manufacturing concerns and style concerns and so on, sure, it doesn't matter at all.

Fred.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
User avatar
jharvey
1N4001 - Signed up
Posts: 1607
Joined: Tue Jun 10, 2008 5:17 pm

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by jharvey »

slacker.cam wrote:I'm an electronics design engineer working for a company who specialises in power electronics so I may have something to add here.
Great and thanks for adding your thoughts.
slacker.cam wrote:Firstly, I don't think it matters whatsoever if the board is designed with all SMD, all through hole or a mix of both.
In this case the SMD drive is purely for automation assembly needs. AAPCB (and several local assembly houses in Argentina where Marco lives) have quoted pricing that is attractive. AAPCB does not require purchasing a real of components, they don't even require the parts to be on a real. You can have loose part, and it make no difference to them. It's one of the things that makes them a very attractive short run assembly house. I personally find SMT easier to hand assemble, but I understand that many folks prefer thru hole. That's why I drew up DFH with hybrid foot prints, where I put 0805 pads between the thru hole vias. It allows extreme flexibility. However PUMA limited the scope to a specific range of engine applications, and Marco's long term plan is to offer them for sale, so he's got a stronger focus on MFG.
slacker.cam wrote:This is from a reliability/vibration stand point ONLY, obviously target markets, automated assembly, etc, etc will change what's most attractive.
Marco has engineering ties to the Argentina military. I believe this is part of why he's interested in maintaining mill specs, and why part of his market might be focused on mill specs. I understand they have lots of equipment that is mechanically good and could stand an upgrade. The Argentina military is primarily a good will organization not a defensive or expansive organization. They work on bridges, and maintain the peace.
slacker.cam wrote:You very rarely hear of one of these old ECUs dying due to cracked pads, dry joints or broken component legs. 99% of the failures are due to the old electros going dry and/or leaking and corroding the board.
Correct, they once did it, but now they don't. They were driven primarily to keep costs down, while maintaining a high quality and reliability in the design. All of the older case designs I can think of typically involved cast alum cases that were typically .125 inch thick or thicker. They also typically had shock mounting mechanism to dampen the shocks absorbed by the case, and they mounted the ECU in the firewall where the shocks are minimized to begin with.

Our cases don't offer such strength, and is much thinner and would be costly to offer a stronger case. We don't know where it will be mountable, or if it will even have a firewall. So it's hard to ensure it's mounted in an area with minimal vibration.
slacker.cam wrote:I think if the technology is good enough for 30 years of service then it's good enough for us.
NASA has used and currently uses wire wrap for things that have sustained much harsher environments, for longer periods of time. The down side is size, cost, and assembly time. So I don't suggest we use wire wrap. Which are the same reason why I suggest some of the design constraints I've suggested.
slacker.cam wrote:A double sided PTH board provides even more support for the component legs and gives greater soldered area. The second observation is that a huge majority of power electronics these days still use through hole components bolted up to some form of heatsink purely because of the practicalities of thermal transfer to the heatsink.
I certainly agree that there are times when thru hole is required, or offers better performance. Try to find a 50 watt 0805 resistor. Not going to happen. I also agree that thru hole can be made to work, but requires additional expense and increased physical size.
slacker.cam wrote:I know personally I have a habit of designing boards where I use all SMD components even when there really isn't any need.
For me that's not the case. I'm tend to learn a variety of technologies and choose the one that best fits the application or specifications. In that order as well. if the specification doesn't match the application, I change the design specs. I've actually had applications where I successfully suggested wire wrap, as it was technologically superior.
slacker.cam wrote:Another point to make is if you design in SMD parts and plan to have them assembled by a machine then reels of every component will need to be purchased. This could prove to be very costly for whoever is backing the production. How many boards would need to be run at a time?
AAPCB charges around $50 in qty 1, and $30 in qty 5 to PNP assemble. They require a little overage for in case a part fails to assemble correctly. For resistors and caps that's typically 20% and MCU's and such often require no extra stock. You get the extra unused stock with your order. The cost is quite low compared to me spending a day or more obtaining and assembling the parts. The also offer electrical testing if you provide a test fixture. One thing they do that I haven't found from another fab house is to use paper solder stencils for the proto run. Then if you do a significant run, they do the typical steel stencil.
slacker.cam wrote:On the topic of SMPS vs linear. What are the current issues with the linear option? It certainly seems like the easier route to take in my eyes.
The issue is that it's predicted to draw on the order of .25 amps to .5 amps under some fairly common built setups. So the regulator needs to drop (18v-5v).5A = 6.5 watts and we have been planning for 1 amp as extra room for growth isn't bad, so it could have to dissipate around 13 watts. With an ambient temp of 110F(43C) and LowZ injector drives in the case (estimated at 9 watts each for 8 cyl's), we are talking of a total case dissipation in the range of 78 watts to 85 watts for something the size of your hand. The internal component temperatures would likely be near the top range of their limits. So the LDO's temp compensation, and abilities to provide a 5V supply vs a 4.8V supply are likely degraded.

Fred feels we don't have a problem because his runs cold. He's using HighZ injectors with very high drive impedance, so his board is likely drawing less than 50mA not .5A. His HighZ are dumping in far less than 1 watt into the case, and he's typically doing 4 cyl, not 8 cyl. I understand he did one 8 cyl, but I don't know how he did it. It was likely with 4 injectors, but I don't know the specs of what he did there. The point being he needs to dump less than 1 watt, but the design spec is on the order of 50 watts to 100 watts.
slacker.cam wrote:I don't want to step on anyone's toes here. Just throwing my 2c out there :)
I'm not going to hold back because of fear of conflict, nor will I seek it for the sake of seeking it out.

EDIT: Content that breaches the LONG established rules was removed from here. READ THE (very reasonable) RULES viewtopic.php?f=17&t=2

As a bit of a side note, Marco's first language is Spanish I believe, English is secondary. This is one reason why I'm doing a lot of commenting semi on his part. It takes him longer to type, and his goals are to make PUMA, and less to socialize in the forums. So chatting with me about design details will likely steer PUMA's design even though it's actually being laid out by Marco.

As another side note, I have a BS in Electro-Mechanical Engineering. I tend to jump electrical and mechanical bounds on a daily basis. It's common that electrical's will design something, then the mechanicals will make it work. I tend to look at the entire picture a bit more than most typical design builds. It's likely that mechanical issues or obstacle are semi hidden from the electrical designer. In a typical design cycle.

Thanks for the 2C.

EDIT: Content that breaches the LONG established rules was removed from here. READ THE (very reasonable) RULES viewtopic.php?f=17&t=2
Last edited by Fred on Wed May 25, 2011 1:14 pm, edited 2 times in total.
Reason: Remove content that broke the rules and offered NO value.
User avatar
nitrousnrg
LQFP144 - On Top Of The Game
Posts: 468
Joined: Tue Jun 24, 2008 5:31 pm

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by nitrousnrg »

I think I can't thank Jared enough for his backup.

I also feel that Fred is being dictatorial and unsupportive, at least with me and the people that doesn't think his way (Its a fu**ing feeling, don't lose time commenting or deleting this).

SMD stuff is mostly because of the reliability against vibrations, then because assembly ease, and lastly because modern industrial equipment have plenty of SMT stuff.

I'm 23, in the last semester of Electronics Engineering, and my family (me, my dad, his dad, a couple of uncles, my mum, his dad) have been working for the Argentine Military since 1950 or so. I actually live at 150 meters of a military base, and we did lots of designs for the army (mine were only 2 so far, the rest were from my parents and uncles). That's why I care about complying with the MIL-STD.

Also, I'm in close contact with people that designs or validates aerospacial stuff, in particular rad hard or FPGA related designs.

I totally agree about the reliability of TH components. I have in this table a TH board that worked since 1960 in a ship (yes, in the sea, and driving big motors), I can't say its unreliable, but modern equipment we see is SMT, the only difference is that we're not in 2070 to say for how long the thing was working.

So back, to the tech stuff: my Puma, almost completely populated had one of its 5v reg getting fairly hot, at an ambient temperature of less than 30°C. That is not good for me, so I've chosen to give the SMPS a try. I assembled a test board with it, and I'm getting non acceptable ripple measurements, although I didn't put any work on getting it work properly, I only had the time to assemble it and see how it performs with the oscilloscope and a bunch of low value resistors.

About the injector drivers, they don't need to be big for high z. They need to be huge if they're going to control the current in a linear way. I gave a mate a bit of help with the LM1949 some months ago and he is almost ready to produce a P&H expansion board for some commercial ECU. He shown me some oscilloscope measurements of both linear and PWM, and told me he "can't touch the darlington in linear mode". That is mostly why I'm going to try the pwm path. Its only a 2 resistor difference. I also saw another different P&H pwm design, and most of the companies out there use pwm, and Puma has a pretty isolated signal/power circuitry. By the way, in Argentina P&H is must, others may find it unnecesary, but it can be unpupulated (just a different text file for the MFG).

Ok, now lunch, and I'll see if I post the pics I have in the SMPS section, or start working to get that ripple down at least for a couple of hours
Marcos
User avatar
jharvey
1N4001 - Signed up
Posts: 1607
Joined: Tue Jun 10, 2008 5:17 pm

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by jharvey »

Fred wrote:You're entitled to your opinion, and this is the place for it.
Apparently not Fred, apparently not.
User avatar
Fred
Moderator
Posts: 15431
Joined: Tue Jan 15, 2008 2:31 pm
Location: Home sweet home!
Contact:

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by Fred »

jharvey wrote:
Fred wrote:You're entitled to your opinion on Puma Spin 2 changes, and this is the place for it.
Apparently Fred, apparently.
Fixed. See the rules that Motofab earned us. Commenting on moderation action in the technical forums, or any place outside the closed admin forums is not acceptable. Drop it. Further posts with content not about Puma Spin 2 changes will be removed or cropped until they contain only appropriate material. There will be no exceptions to this.
DIYEFI.org - where Open Source means Open Source, and Free means Freedom
FreeEMS.org - the open source engine management system
FreeEMS dev diary and its comments thread and my turbo truck!
n00bs, do NOT PM or email tech questions! Use the forum!
The ever growing list of FreeEMS success stories!
slacker.cam
QFP80 - Contributor
Posts: 69
Joined: Sun Jan 27, 2008 10:25 pm

Re: Why The Latest Changes To Spin 2 Don't Worry Jared

Post by slacker.cam »

jharvey wrote:In this case the SMD drive is purely for automation assembly needs. AAPCB (and several local assembly houses in Argentina where Marco lives) have quoted pricing that is attractive. AAPCB does not require purchasing a real of components, they don't even require the parts to be on a real. You can have loose part, and it make no difference to them. It's one of the things that makes them a very attractive short run assembly house.
And to potentially comply with some military standard by the sounds of it. I'm sure that you could still comply with the mil spec and still have through hole components. Some of the military equipment I've seen is extremely primitive. This is just speculation though. Fair enough to push towards as much SMD as possible for automated assembly. Although one avenue that I can see to make the board more flexible would be to not populate the injector drive circuitry and let it be user populated. By leaving it through hole you could allow anyone with a soldering iron to populate either the LM1949 and darlington option or a simple autofet. This would allow the highZ people to save a few bucks on the component cost. Maybe it's more attractive from Marcos's perspective to just populate them all with with lowZ hardware but that cant be automatically assembled and it wont be mil spec. Bah. I don't know :) It would be nice to keep it as simple as possible from a users perspective rather than having a huge number of build time options like MS.
jharvey wrote:I personally find SMT easier to hand assemble
Me too.
jharvey wrote:Marco has engineering ties to the Argentina military. I believe this is part of why he's interested in maintaining mil specs, and why part of his market might be focused on mill specs.
Fair enough. If he's the one designing the board then he's free to determine what it's going to look and feel like. Just one comment that's related to what Fred had to say the other day about at this stage of the project 'we' should probably be doing whats best for the community rather than what's best for ourselves as what's really needed at this point is engines running on the platform. It's easy to put together another board design that fully complies to a standard down the line once we have mature code to run on it.
jharvey wrote:All of the older case designs I can think of typically involved cast alum cases that were typically .125 inch thick or thicker. They also typically had shock mounting mechanism to dampen the shocks absorbed by the case, and they mounted the ECU in the firewall where the shocks are minimized to begin with.
Yes, I agree. The cases I'm thinking of are cast alum that are reasonably thick. I'm not sure that they had any shock absorbing mechanism though. And yes, obviously all the OEMs put the ECU in the cabin to keep it out of the engine bay environment. I agree that it's unrealistic to assume that the environments where a Puma board will be installed will be as nice as the cabin of a car - good point. I think without actually quantifying what vibration scenarios we expect our hardware to deal with it's really just an argument about preference though. Sure, a SMD design will be MORE resistant to vibration but that's not to say that a through hole design wouldn't be sufficient in all of the usage scenarios that the board is likely to experience.
jharvey wrote:Our cases don't offer such strength, and is much thinner and would be costly to offer a stronger case.
What case is the Puma board designed for?
jharvey wrote:NASA has used and currently uses wire wrap for things that have sustained much harsher environments, for longer periods of time. The down side is size, cost, and assembly time.
Yeah, interesting isn't it. I always thought that the main reason for NASA extensively using wire wrap was to prevent oxidation of the joints however. I read that the forces of the copper biting into the wire wrap leg were very high and forced out all oxygen that could potentially corrode the joint. They then potted everything in epoxy to provide vibration resistance I believe.
jharvey wrote:I also agree that thru hole can be made to work, but requires additional expense and increased physical size.
Don't mean to nit pick but in our application both SMD and through hole options will be included so it will actually increase size, not decrease it.
jharvey wrote:AAPCB charges around $50 in qty 1, and $30 in qty 5 to PNP assemble. They require a little overage for in case a part fails to assemble correctly. For resistors and caps that's typically 20% and MCU's and such often require no extra stock. You get the extra unused stock with your order. The cost is quite low compared to me spending a day or more obtaining and assembling the parts. The also offer electrical testing if you provide a test fixture. One thing they do that I haven't found from another fab house is to use paper solder stencils for the proto run. Then if you do a significant run, they do the typical steel stencil.
Wow, sounds like quite an impressive outfit. I don't think we have anywhere near as cost effective or that provides the range of services here in NZ!
Post Reply