LM2621 + ZXCT1009 Current mode 5W Boost regulator

Starting a new post just for this design.

It is nearing completion and I have finished RUN503 (RUN5XX) Where XX is the actual configuration of collected data.

My design notes can be found on my web page at:
LM2621 Desgin Notes.

Here is the run data in case you don't want to look at the web site.


NOTES: CVTR = DC/DC Power dissipated by Converter board (ICs, diode, inductor etc).
Blink = IC hit die thermal limit and shuts down, cools, turns on, heats up and cycles this over and over at several times per second blinking the LS.


Possible Specifications:

Vin-min ~4V
Vin-max ~6V
Output Current adjustment range: 100mA - 750mA

Form factor - Target 17mm circular. (123 battery diameter)
Components: All on one side.
Board Construction: All traces on top side:
Bottom side: Solid Copper (GND).

Component cost: Total <$10.00
Board Cost: ~1.00 in low volume.
First run: Double sided plated thru holes, no silkscreen, no solder mask, tin plated both sides.

Hi again Wayne,

I didnt see the new waveforms for the
circuit with the extra diodes added
for the 5w LS simulator. I can read
your spreadsheet data, as im sure a lot
of people cant either. If you post the
spreadsheet data in tabular text form,
i can download and view it, or else
update the waveforms.

Did you notice any difference in the
internal switch Vsat with the higher
BOOT voltage? This was one of the tests
we were going to look at.


Good luck with it,
Al

Hello MrAl,

I didn't grab any waveforms for this run. The scope waveforms are all over the place. Very similar to the one shown. Only the amplitude changes or the OFF cycle duration.

I didn't get too much time today and opted to get as many changes made and spot checks made to start honing in on a valid component selection.

The spreadsheet is available on the web. I just boink on it on my web site and it popped up. There are many sheets in the Excel spreadsheet. You might need to select a different one to view the entered data.

Increasing the boot voltage made no difference in the switch on resistance.

As for the actual diagram of the changes I will post someone more definitive after some more changes I want to try tomorrow and over the weekend. So, the circuit may get altered again.

One thing that is of major concern is the part is always dissipating near or over the rated 500mW. If I didn't have it thermally bonded to the circuit board with ASIII it would almost always be in thermal shutdown mode. At best this is on or over the hairy edige of stability.

Congrats so far Wayne !

Klaus

Hello again Wayne,

Looks like the output mosfet isnt nearly
as good as the spec sheet seems to imply.
If the conduction didnt improve at higher
BOOT voltages, then i guess it never will.
Now im wondering if they made an improvement
in this transistor for the newer version,
the LM2623 chip. If so, you might get
better results with that chip. I have my
doubts because they spec the same Rds, but
maybe it's still better.

Also, since this package has 0.0256 pitch
leads, how are you soldering them to the
pc board? I was just trying some smaller
chips which only have 6 pins (not 8) and
even they are pretty hard to do. The
problem i see is that it's very hard
to inspect the finished solder work after
it's done. Any ideas here?

Thanks and good luck with your circuit,
Al

Hello all,

I added RUN503 Scope waveforms on my web site and run504.

Run504 I added another diode in series with R3 the output of the high side sense IC. This has the effect of subtracting 1.1V from the feedback node. This lowers the current sense feedback gain. You can see the scope waveforms that the green trace (feedback signal) is much lower than in the 503 run.

Also I increased the sense resistor to 0.22 ohms to increase the sense output. Now the bandwidth of the sense IC tracks the the sense voltage. Before it had too much lag.

MrAl,

I think the 2621 RDS on doesn't cut the mustard for 5W. My gut feeling is that the 2623 will have the same issues or same FET design.

As I noted above the run time data collected for 503 was still exceeding the device power specification and if it wasn't for the addition of ASIII and an external fan blowing air across the board this little IC would have shut down.

Soldering these things requires some form of magnifying glasses. At work I use a stereo microscope 7X-35X. I almost always use the 7X setting. At home I use a tweezers with a built in magnifier. I saw one at RS the other day. Tweezers and magnifier built it. I use a headband magifier in combination with the tweezers to inspect my toner transfer boards for copper shorts.

Soldering is made much easier if you use a flux pen. Flux the pads before placing the IC down. The clean the tip of the soldering iron add a little solder to the tip. Then just touch one of the leads and the excess solder will transfer to the lead and pad with the help of the flux on the board. Once one lead is down. You can re-heat it if you need to adjust it around. Then continue the same process moving from pad to pad.

For this type of IC a good small tip soldering iron and small diameter solder plus the flux pen should do the trick.

To remove excess solder I use solder wick and always load up extra flux on the solder wick to aid in transfering heat.

The tiny pads are not forgiving, so heating the pad too long and excessive pressure are sure ways to remove the copper land pattern.

Rene might be able to add his thoughts. I think he uses similar techniques.

See also the post in this section titled.
Max1674/75 on a SurfBoard by jeff1500. I think in there both Rene and I offered soldering tips and suggestions.

I just purchased a small stereo microscope from ebay for home use. If I was not clear let me emphasis I do not do this type of soldering at home. I do it at my work soldering station with the stereo microscope. I use the tweezers magnifier only for inspection of the etch process. When I get my stereo microscope I will be setting up my home to solder the assemblies at home.

(Well, I'm back off to work to assemble some more MADMAX boards and to see if I can do some more experiments on the LM2621.) I won't get more time till WED.

This design as it stands I think works really comfortably when putting out 500mA-600mA. Putting out 700mA-750mA is I think asking for too much from this IC.

Group:

Please take a look at the scope waveforms and if you have any suggestions or observations, please feel free to toss them into the pool here.

The top purple trace is the inductor current waveform (1A/div). the yellow (middle trace) is the output waveform before the sense resistor and the green waveform (bottom trace) is the feedback signal from the output of the current sense amplifer.

Wayne

It's a pity the LM2621 does not yet work as supposed because of the internal MOSFET ..

MrAl:
Concerning soldering these tiny little pins I don't have to add much to what Wayne said as it seems we do it in a similar way. But looking at the MADMAX1674 he has soldered I can only say: Perfectly! I can't do it as good (and for sure not as fast)!

There is one big difference: Wayne is as usual much better equiped. I don't have a high sophisticated soldering iron. Mine is an almost twenty years old Weller with a magnetic switch. And I also don't have a stereo microscope but instead this I use such a 'helping hand' with an integrated magnifying glass. Here you will find a picture of it: Helping Hand. I paid about ten dollars for it. I don't really use the clips to hold the pc board as it is too complicated to turn and twist the little pc board into a position where one can easily solder it. But the magnifying glass (something between 5x to 8x) is a real good thing and it is easy to work with it as it is not too big so one can easy work under the glass!

And last but not least one of the most important things: Flux, flux, flux .... This is what my 'teacher' (our production manager in the company I work) teached me most! And after all I only can admit: Flux, flux, flux ...

And then: Give it a try! That's the only way to learn it! For the first 10uMAX chip I used almost an hour to solder it correctly. For me it still is a hard job to solder these chips. After doing it you for sure will enjoy to solder some SOT23 parts ...

Hi again Wayne and Rene,

Thanks very much for the tips,
im having a problem doing the 0.0256
pitch parts and havent even attempted
doing the 0.02 pitch parts yet!
I think i see what you mean about the
flux, i think that is what im missing,
mainly because i noticed that the solder
likes to 'blob' up rather then flow nicely
under the pin like the larger packages.

So my next question is:
What kind of flux do you use, like, what
is the brand name of it and where can i get
some to try?

Also, have you ever tried the solder paste
(not flux) used with SM parts?
I was thinking of trying this too.
If i ever want to try the Max1698 chip,i
have to be able to do a 0.02 pitch part.


Take care,
Al

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:

Thanks very much for the tips, im having a problem doing the 0.0256 pitch parts and havent even attempted doing the 0.02 pitch parts yet!
<HR></BLOCKQUOTE>

As I already wrote: Just give it a try and you will see you also can do it! If the first try doesn't work the second or third one will do it for sure.

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:

I think i see what you mean about the flux, i think that is what im missing, mainly because i noticed that the solder likes to 'blob' up rather then flow nicely under the pin like the larger packages.
<HR></BLOCKQUOTE>

This is exactly what happens if you don't use the flux ...

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:

So my next question is: What kind of flux do you use, like, what is the brand name of it and where can i get some to try?

Also, have you ever tried the solder paste (not flux) used with SM parts? I was thinking of trying this too. If i ever want to try the Max1698 chip,i have to be able to do a 0.02 pitch part.
<HR></BLOCKQUOTE>

I use a flux I've got from my 'soldering teacher'. I can't give you any information about it as I have only an part no on the little bottle. And its the part no of a Swiss distributor so I think this doesn't help you ... I also didn't try the solder paste you mentioned. I'm sure Wayne or someone else can give you some better help!

<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by MrAl:
Also, have you ever tried the solder paste
(not flux) used with SM parts?
I was thinking of trying this too.
If i ever want to try the Max1698 chip,i
have to be able to do a 0.02 pitch part.<HR></BLOCKQUOTE>

I think someone else posted about the SMT paste with a strong recommendation ... lemme look around ...

Ah, here it is. It was dat2zip in this thread:

Max1674/75 on a SurfBoard

-XCN-

Hello dat2zip:

1.Do you think Sanyo OS-Con can help you reduce "Run503"'s output ripple Voltage?

20SA100M, 20V 100?F, DxH 10x10mm,
Allowable Ripple current: 3210(mArms)

I can sent 2 for you.(Free include mail fee)

2.Can you help me buy components which use in your circuit?
(The Inductor make me pain)

I am also strongly in favor of the use of solder past for surface mount devices. I use it to solder the LT1932 devices, which use the same 0.5mm pitch (roughly 0.02 inch). Solder past is essentially solder powder is _lots_ of flux.

The great thing about using the solder paste is that you _don't_ need to deal with individual pins. You just bleb down the solder paste on the land areas of the PCB, and then put the part down, and then heat up all the pins at once. You can use a fat tip soldering iron, or you can put the whole PCB in a toaster oven. If you put down too much solder paste, then a quick brush with the solder wicking braid fixes everything.

-Jon

Hello again,

Thanks to all of you for all the great
soldering ideas.
Im hoping to be able to try some of
these ideas soon, and im sure my results
will be much better now.

Sincerely,
Al