TINKERERS POWER SUPPLY

[Tinkerer]

Thanks for the feedback Stefan,

porkluvr is doing a great job with the power supply.

Tinkerer

[porkluvr]

Glad to hear of your good results, Stefan. I came across some LT1171 and LT1172 as freebies, so I'm lucky they are so useful.

I am looking to use either LT1963 or LT1965 adjustable regulators (TO220-5 package) for the post regulator. Both have identical pinout. Their low noise may be over-kill, but it should be easy enough to go back and institute LM2940-15 should someone so desire.

It's embarassing to have to say, my power supply board layout is not finished. But it will be done, when it's done. I want it "right" and there were decisions to be made. Ponderously slow decision making processes were at work. (That's me.) The good news is that I'm done making decisions (yeah,sure ).

With Stefan's good luck using a protoboard, it makes me wonder "why bother with a PCB". But a PCB is almost a done deal. The hard part is over and it's downhill from here.

I have the schematic to post but the Geotech1 website is not accomodating me. The attachment process is so very SLOW. It was like that last time I posted something. Only now it's even slower.

I suppose that means I'll need to reload my PC operating system because it is mmmmMY problem. Tinkerer, I could email the schematic to you and then you could post it, OK?

Oh, joy. The last time I re-imaged I lost a bucketful of good music and movies I had downloaded.

"Preview Post" takes forever so I'll just pickle this off and hope it makes sense.

[Qiaozhi]

Glad to hear of your good results, Stefan. I came across some LT1171 and LT1172 as freebies, so I'm lucky they are so useful.

I am looking to use either LT1963 or LT1965 adjustable regulators (TO220-5 package) for the post regulator. Both have identical pinout. Their low noise may be over-kill, but it should be easy enough to go back and institute LM2940-15 should someone so desire.

It's embarassing to have to say, my power supply board layout is not finished. But it will be done, when it's done. I want it "right" and there were decisions to be made. Ponderously slow decision making processes were at work. (That's me.) The good news is that I'm done making decisions (yeah,sure ).

With Stefan's good luck using a protoboard, it makes me wonder "why bother with a PCB". But a PCB is almost a done deal. The hard part is over and it's downhill from here.

I have the schematic to post but the Geotech1 website is not accomodating me. The attachment process is so very SLOW. It was like that last time I posted something. Only now it's even slower.

I suppose that means I'll need to reload my PC operating system because it is mmmmMY problem. Tinkerer, I could email the schematic to you and then you could post it, OK?

Oh, joy. The last time I re-imaged I lost a bucketful of good music and movies I had downloaded.

"Preview Post" takes forever so I'll just pickle this off and hope it makes sense.

Perhaps it's time to move to a better OS? ... i.e. Ubuntu Linux.
You might be surprised to know that LTspice runs just fine under Ubuntu using WINE.
Say goodbye to slow bloatware, constant crashes, and the need for expensive antivirus protection. Say hello to opensource.

http://en.wikipedia.org/wiki/Ubuntu_linux

[Tinkerer]

Glad to hear of your good results, Stefan. I came across some LT1171 and LT1172 as freebies, so I'm lucky they are so useful.

I am looking to use either LT1963 or LT1965 adjustable regulators (TO220-5 package) for the post regulator. Both have identical pinout. Their low noise may be over-kill, but it should be easy enough to go back and institute LM2940-15 should someone so desire.

It's embarassing to have to say, my power supply board layout is not finished. But it will be done, when it's done. I want it "right" and there were decisions to be made. Ponderously slow decision making processes were at work. (That's me.) The good news is that I'm done making decisions (yeah,sure ).

With Stefan's good luck using a protoboard, it makes me wonder "why bother with a PCB". But a PCB is almost a done deal. The hard part is over and it's downhill from here.

I have the schematic to post but the Geotech1 website is not accomodating me. The attachment process is so very SLOW. It was like that last time I posted something. Only now it's even slower.

I suppose that means I'll need to reload my PC operating system because it is mmmmMY problem. Tinkerer, I could email the schematic to you and then you could post it, OK?

Oh, joy. The last time I re-imaged I lost a bucketful of good music and movies I had downloaded.

"Preview Post" takes forever so I'll just pickle this off and hope it makes sense.

Hi porkluvr,

thanks for the good work.
Sorry I am slow in answering, had problems on my own. No power for 2 days, then burned out the voltage regulator on the generator, then the power came back and the freezer with 200 pounds of meat in it broke down.... It's not been a good week.
Anyway, here is the schematic of the power supply.

all the best

Tinkerer

[Tinkerer]

I don't know much and am a bit lazy so I just ask some questions to knowledgeable people instead of searching for the answers in the datasheets.
So here it goes:
A nominal "14.4V Li-Ion" battery pack has a maximum peak charge voltage of 16.8V and a cutoff charge of 11V.
Is it OK to use the configuration above for an output of 15V?

If we were to take the input of the 5V regulators directly from the battery voltage. would there be a saving of power, instead of boosting and regulating and taking the input from the regulated 15V?

Thanks for all the help

Tinkerer

[porkluvr]

Mistake. Input voltage should be 11V. Good catch, Tinkerer.
You told me that 14.4V was full charge, 11V was nominal and 8.5V (I think) was time to recharge for the battery that you had selected. So, unless you want to change battery choice, the 14.4V is wrong and needs corrected.
But, please don't do it!

We can't use the higher voltage you just mentioned because above 15.4V the switching regulator would no longer be operating in the step-up mode. I don't think it would start. You would be running directly from the battery. (Hmmm, what happens then??) I would want to use the SEPIC topology if input exceeds 15V. Efficiency would suffer. What I have designed could also be used with a 12V gel-cel but won't work properly if input exceeds about 15V.

You are right about gaining efficiency by regulating directly from the battery but the gain would be relatively small because the transmitter is responsible for the vast majority of the current draw.

By using the +15V as the feed we know that the input voltage will be a constant 10V above the regulator input. Knowing that (and estimating the current draw through the regulator) allows an input resistor to be selected as part of an RC pi filter, and drop the excess voltage. If the input were allowed to vary 6V over the battery discharge cycle I would probably want to remove the 470Ω input resistors but no harm would be done.

It would be OK to jumper from the B+ to the "south" end of the 470Ω resistors (removing the resistors), or maybe I could redesign as you say, coming directly from the battery. It seemed to me a good idea to isolate the individual 5V regulators and the 15V output is a convenient, predictable input source, but I will admit a small efficiency advantage by regulating directly from the battery.

I've already placed and routed most of the components shown in the schematic so let me finish this. It will be easy enough to bypass the 15V regulator if you want, or I could rework the board. It might be especially good to run the "audio 5V" (which is my own addition) directly from the battery so I'll give this some more thought. (That's painful, and you know it takes a long time. ) Darn. I thought I was almost finished, but I would rather hear about this now than later.

Another thing that needs mentioned is that there will be more than one ground output. Be careful about not creating a ground loop. Only one ground wire should be run unless the ground areas on the main PCB are seperated.

I'm working from my backup partition so I can probably "Preview Post" AND post an image. Here goes...

observation: Oh yeah!!!! I need to re-image, soon.

edit: In the +5V audio circuit, the 470Ω resistor shown is probably way too high. The best value would depend on the audio current drawn (unknown to me) and it may be best to remove the R and run this regulator directly from the battery. I'll submit a change in a couple of days.

[porkluvr]

no LTC660 or MAX660

Here is another mistake or two. In trying to accomodate the widest range of CMOS voltage converter ICs possible I made a note on the schematic draft that if MAX660 or LTC660 were used, then the Zener diode at the input must be changed. I cannot guarantee that inserting a 1N757 will lower the 15V to below 6V "Absolute Maximum Input Voltage" for these devices, but if it does, then the output of those two devices will almost certainly not be high enough to use a 7905 regulator. National's 7905 datasheet states -2.3V is required overhead for the 7905 with a 100mA load, and -2V required at 40mA. Without getting into the fuzzy math I extrapolate that 7905 cannot be used with those two converters whose maximum output at no load would be -6V.

What's even worse is that I know of very few negative LDO regulators having the same pinout to solve the problem. There is probably one somewhere, and I will look, but the most comprehensive plan may be to exclude the MAX660 and LTC660 as possible choices for U2.

edit: I don't know of any -5V LDO regulators in either the TO92 or TO220 package. All I know are SMD. Anybody???

[porkluvr]

Perhaps it's time to move to a better OS? ... i.e. Ubuntu Linux.
You might be surprised to know that LTspice runs just fine under Ubuntu using WINE.
Say goodbye to slow bloatware, constant crashes, and the need for expensive antivirus protection. Say hello to opensource.

http://en.wikipedia.org/wiki/Ubuntu_linux

I'll try Ubuntu. I've tried using Linux in a dual boot system, and I even used two different store bought versions (Redhat and SuSe). I also tried Mandrake.

All versions I've tried so far did not support my video card. I could not use the dual monitor feature. Ackk!! That was a long time ago so I'll give it another try.

I'm using a top reated antivirus but I suspect that is what's slowing my sytem down, or is at least part of the problem.

I'm not so quick to fiddle with my OS as I was at one time. I've found that you can try and direct where a new OS or an old image will be installed, the indicated preference is often ignored or misinterpreted. WHAM!

[Qiaozhi]

I'll try Ubuntu. I've tried using Linux in a dual boot system, and I even used two different store bought versions (Redhat and SuSe). I also tried Mandrake.

All versions I've tried so far did not support my video card. I could not use the dual monitor feature. Ackk!! That was a long time ago so I'll give it another try.

I'm using a top reated antivirus but I suspect that is what's slowing my sytem down, or is at least part of the problem.

I'm not so quick to fiddle with my OS as I was at one time. I've found that you can try and direct where a new OS or an old image will be installed, the indicated preference is often ignored or misinterpreted. WHAM!

The current LTS (long time support) is Ubuntu 8.04 (hardy).
The best solution is to download the iso and burn a CD. You can test Ubuntu directly from the CD without removing Windoze. This way you can check if the video card is supported.

What is the make, model and spec of your machine?

If you're interested in giving this a go, I will move these OS posts to a different thread under Off Topic.
I have also used RedHat and several other distros, but Ubuntu is the best. That's why it's called "Linux for human beings."

[porkluvr]

The current LTS (long time support) is Ubuntu 8.04 (hardy).
If you're interested in giving this a go, I will move these OS posts to a different thread under Off Topic.
I have also used RedHat and several other distros, but Ubuntu is the best. That's why it's called "Linux for human beings."

I've gotten off to a bad start already because what was supposed to be a 698MB download ended at 85MB. I tried a second source and it hanged at about 84%. I'll try again tomorrow.

I have a home-built AMD XP2500+, and MSI KM4M MB. ATI Radeon 9000. Nothing cutting edge here. My HDs have been running for over two years so that will probably be my next big headache (that, or the laser in my OTHER DVD burner).

So, am I making a mistake downloading v9.04 instead of v8.04 or was that a typo? Maybe you should move this off topic stuff! I don't want to use this backup partition much longer or it'll get corrupted like the other one did, but I dread making major OS changes.

Thanks, Qiaozhi

[Qiaozhi]

Ubuntu Linux stuff has been copied to Off Topic.
We can continue that discussion there.

[porkluvr]

mo' trouble

Here's another something that needs correcting. For the LT1171 and LT1965 I have the voltage adjust resistors shown as the same values in the previous draft. (I won't call it a schematic because it still needs work.)

Those two devices have slightly different reference voltages, so although I calculate about 0.4V voltage differential using those resistor values applying the voltage output formulas - according to the specification sheets for the LT1171 and LT1965, simulation shows otherwise. The post regulator may be in dropout, and not what we want.

Funny thing, I get different results with those same components in different simulation files and I have yet to identify the source of variance. That's scary.

So, a simple catch-all adds a trimpot in either the switcher or the post regulator circuits. (Oh, crap. More decisions.) Just kidding, that's an easy one. It'll be whichever one is EASIER to ripup and re-route. (Yes, I'm lazy too, and behind schedule.)

In reality, I suspect that a LM2940-15 would work fine in this power supply so the obvious choice would be to make the switcher's output variable. That's what I'll do next time.

[Tinkerer]

Here's another something that needs correcting. For the LT1171 and LT1965 I have the voltage adjust resistors shown as the same values in the previous draft. (I won't call it a schematic because it still needs work.)

Those two devices have slightly different reference voltages, so although I calculate about 0.4V voltage differential using those resistor values applying the voltage output formulas - according to the specification sheets for the LT1171 and LT1965, simulation shows otherwise. The post regulator may be in dropout, and not what we want.

Funny thing, I get different results with those same components in different simulation files and I have yet to identify the source of variance. That's scary.

So, a simple catch-all adds a trimpot in either the switcher or the post regulator circuits. (Oh, crap. More decisions.) Just kidding, that's an easy one. It'll be whichever one is EASIER to ripup and re-route. (Yes, I'm lazy too, and behind schedule.)

In reality, I suspect that a LM2940-15 would work fine in this power supply so the obvious choice would be to make the switcher's output variable. That's what I'll do next time.

Yes it is a lot of work to do a good power supply. But it is an important part of the whole circuit. We appreciate your effort.

Earlier, you cautioned about different grounds. Could you expand on that a little?
Are you using a Star Ground or a Ground Plane?
How do we avoid getting noise on the ground?

Thanks for your help

Tinkerer

[porkluvr]

Star, what's that?

I am only making this a single sided board so far. All traces are on the bottom so it would be a simple matter to pour a copper plane onto the top. So far, the bottom trace tends to follow a star. I'll try to keep it that way. Also on the bottom I usually pour a copper ground polygon around the unrouted areas.

I am not a PCB layout expert so after I get stuff more completed I will submit it for criticism (which will hopefully be of the constructive sort). I will also eventually post the Eagle Lite (v416) schematic and layout files here, for anyone's use and possible modification.

So, somebody familiar with Eagle (simple stuff) would theoretically easily be able to convert the product to their own preference - whether you want to etch the board yourself with no ground plane, or with a plane on the top, or also with a bottom fill which can be connected to the top. Or somebody could legally send the board file to a PCB house. Just, have them make one for me too, will you?

Those are choices. Once the basic placement and routing is finished it is a very simple matter to pour either top or bottom or both. It takes only a few minutes to place vias if you want to add connections between top and bottom pours but there is nothing hard about it.

BUT: one possible problem for the home PCB maker is, how can you solder the top of the board underneath a large upright filter cap or underneath a Bournes 3296 style trimpot?

Possible solution: can a conductive pen (silver ink) be used to create plated thru holes? I also have some .072" eyelets that might work on uprights caps if I use large diameter pads (but I can't do eyelets here).

Tinkerer I guess the answer to your question is "star ground", and ground plane(s) is/are are optional (if you call the bottom fill a plane).
*******************
About the +5V regulator inputs; I would like to use input resistors (except possibly for the audio regulator). Those resistors could easily be removed and the B+ used as regulator inputs, but I have another issue that worries me.

I've been experimenting with the filter at the input to the FET and found that a smaller inductor and larger capacitor may give the lowest ripple on the +15V output. These tests take time, but I won't stress. L & C can be upgraded if something is found that works better. I'll make room.

Hopefully I'll have something concrete SOON. In Hours, not days, but I'll show it to Tinkerer before it comes here.

I'm doing some other things besides this just project but a lot of it is related. There is foundation work to do; library parts to build and modify; there were datasheets to consult. I'm also chasing down spice models for LM7905 and LM7805. I know where those are but I've neglected to build them into my simulation library until now. (I only hope they'll work.) If I could find any LM2940 model I would be almost happy.
*************************
In my previous post I mentioned inabilty to make the post regulator output the correct voltage in simulation. That problem is fixed. I had two nets with the same name on either side of the LT1965 so it was shorted across. I've done that same mistake too many times for it to have taken so long to fix, but it's fixed... . So, the resistor values should work as originally shown but I see where I can easily add a trimpot to the post regulator adjustment (or even better, at the switcher). Using the specified resistors, a trimpot would be extraneous and can probably be deleted if my math was not too fuzzy and the simulations are correct.

Hopefully we avoid noise on the ground because of how the parts are positioned and routed. We will see.

[porkluvr]

I finished the vast majority of design work. There are three toroid inductors that I need to specify in some greater detail besides just stating the inductance value.

I figure that the largest of the three (as far as physical size) will be at the switcher. It is also the most critical so far as physical composition. It will very likely be either FT-50-77 core, or even better, FT-63-77 core. I'll wind my own. When I get it. After I do the math.

I don't design switching power supplies every day so this is not second nature for me to design a toroid inductor to use at 100kHz and able to handle 4.5 watt. But, it's not rocket science so it will get done. I'm also working up the parts list details. This takes time. (And it's time for lunch.)
I never was good at keeping schedules.

I've already caught a few outright blunders on the initial draft I posted last week (Tinkerer pointed out one other) and there may be more.