Request for a small bit of help

129 views
Skip to first unread message

Matthew Daubney

unread,
Oct 19, 2012, 5:51:50 PM10/19/12
to reading-hackspace
Can someone take a quick look over the schematic for this and let me
know if it's anywhere near sane? I've not quite finished the board
layout yet! The schematic (and eagle files are here
http://solderpad.com/mattdaubney/mqtt-s-test-board/ )

Many thanks for any help/suggestions.

-Matt Daubney

Nigel Worsley

unread,
Oct 19, 2012, 8:21:57 PM10/19/12
to reading-...@googlegroups.com
Is the buffer chip between the processor and the LCD really neccesary? Most 77480 compatible
LCD controllers have TTL input thresholds which should work fine with 3.3V inputs.

The Vo pot will need a series resistor in the positive leg, Vo needs to be fairly close to 0V
in most cases and with some displays I have used in the past has had to be slightly negative.
If you want the pot to only use the bottom 10% of its range then leave it as it is.

Your power input is expecting the centre pin to be positive, but it is rather more common
for it to be the other way round. It may be worth including a diode to save the regulators
from being blown up in the event of the wrong wall wart being used.

Everything else that looks wrong is probably the 'not quite finished yet' that you mentioned.

Nigle



Jim Pugh

unread,
Oct 20, 2012, 2:56:46 AM10/20/12
to reading-...@googlegroups.com
>It may be worth including a diode to save the regulators
from being blown up in the event of the wrong wall wart being used<
Theres something to be tatooed on your arm...

> FIt may be worth including a diode to save the regulators
from being blown up in the event of the wrong wall wart being usedrom: nig...@googlemail.com
> To: reading-...@googlegroups.com
> Subject: Re: [RDG-Hack] Request for a small bit of help
> Date: Sat, 20 Oct 2012 01:21:57 +0100

graham

unread,
Oct 20, 2012, 4:24:59 AM10/20/12
to reading-...@googlegroups.com
On 20/10/2012 01:21, Nigel Worsley wrote:
> Is the buffer chip between the processor and the LCD really neccesary?
> Most 77480 compatible
> LCD controllers have TTL input thresholds which should work fine with
> 3.3V inputs.
Agreed...also watch out for another couple of "gotchas"...most
Hitachi/Densitron compatible character LCDs have quite low value (1K or
less) pull-downs on their inputs...this can spoil your day if you
use...for example..the ICSP pins to talk to yr LCD (yes...I've done
it!!) or multiplex them in some other way... leaving you in the position
where you can either have ICSP...or your LCD...making the board a pig to
debug. Clearly you can't sort that problem by fitting resistor
buffers...'cos everything then gets too low impedance for the ICSP to
operate...and UR screwed.
>
> The Vo pot will need a series resistor in the positive leg, Vo needs
> to be fairly close to 0V
> in most cases and with some displays I have used in the past has had
> to be slightly negative.
> If you want the pot to only use the bottom 10% of its range then leave
> it as it is.
Absolutely...a lot of designs use a 10K pot 'cos they begrudge the extra
cost of an ickle resistor....V false economy if you ask me. Second
gotcha....may need to fit 100nF between pot wiper and ground...some LCDs
suffer from noise feeding through the contrast pin and ending up as a
ghostly "snow" on the display.

PS if it helps I normally use a 2K2 pot with a 10K res to Vcc. works
for 96.8534% of LCDs.
>
> Your power input is expecting the centre pin to be positive, but it is
> rather more common
> for it to be the other way round. It may be worth including a diode to
> save the regulators
> from being blown up in the event of the wrong wall wart being used.
Too right...PS the sleeve-positive power supplies are a pain in the arse
because if you build a complex system all you have to do is drop the DC
connector onto a grounded bit of metal...instant short!

Other way to sort the problem is to bring back germanium transistors
(Vcc negative) LOL

graham

unread,
Oct 20, 2012, 4:31:17 AM10/20/12
to reading-...@googlegroups.com
Matt

Comes back as a "not found"???????

Can you send it to

g...@coastelect.com as an attachment??

Graham Coast

Ryan .

unread,
Oct 20, 2012, 5:36:52 AM10/20/12
to reading-...@googlegroups.com

Some reverse polarity protection and  maybe even a polyfuse never a bad thing! Also, there's plenty of dcdc converters at the space, make room for one and stick in a full bridge and it'll run on any wall wart that'll fit in the connector ;-)

Richard Ibbotson

unread,
Oct 20, 2012, 7:01:31 AM10/20/12
to reading-...@googlegroups.com
I would add a 100nf ceramic cap. as close as possible to the power pins of the processor.
Add pull up resistors to I2C bus.
Shorten track between ground of capacitor on crystal and processor ground.
 
Suggest you make other suggested changes and fully route, so we can take a proper look.

Matthew Daubney

unread,
Oct 20, 2012, 9:59:42 AM10/20/12
to reading-...@googlegroups.com
On 20 October 2012 01:21, Nigel Worsley <nig...@googlemail.com> wrote:
> Is the buffer chip between the processor and the LCD really neccesary? Most
> 77480 compatible
> LCD controllers have TTL input thresholds which should work fine with 3.3V
> inputs.

I tried it on an arduino and it kinda works... but the datasheet
suggested it shouldn't :) Felt it better safe than sorry.

> The Vo pot will need a series resistor in the positive leg, Vo needs to be
> fairly close to 0V
> in most cases and with some displays I have used in the past has had to be
> slightly negative.

I see the logic in this. Will have a look at it

> If you want the pot to only use the bottom 10% of its range then leave it as
> it is.
>
> Your power input is expecting the centre pin to be positive, but it is
> rather more common
> for it to be the other way round. It may be worth including a diode to save
> the regulators
> from being blown up in the event of the wrong wall wart being used.

I thought the solid line on the symbol for the power jack was the
center pin? This is the -ve connection on the diagram. Have I just got
confused on the symbol? Should the diode go on the negative side or
the positive side? (I'd assume +ve as most stuff seems to go that
side?)

> Everything else that looks wrong is probably the 'not quite finished yet'
> that you mentioned.

I'll keep updating it and putting it back here :) Thanks for having a look.

-Matt Daubney

Matthew Daubney

unread,
Oct 20, 2012, 10:02:27 AM10/20/12
to reading-...@googlegroups.com
On 20 October 2012 09:24, graham <g...@coastelect.com> wrote:
> On 20/10/2012 01:21, Nigel Worsley wrote:
>>
>> Is the buffer chip between the processor and the LCD really neccesary?
>> Most 77480 compatible
>> LCD controllers have TTL input thresholds which should work fine with 3.3V
>> inputs.
>
> Agreed...also watch out for another couple of "gotchas"...most
> Hitachi/Densitron compatible character LCDs have quite low value (1K or
> less) pull-downs on their inputs...this can spoil your day if you use...for
> example..the ICSP pins to talk to yr LCD (yes...I've done it!!) or
> multiplex them in some other way... leaving you in the position where you
> can either have ICSP...or your LCD...making the board a pig to debug.
> Clearly you can't sort that problem by fitting resistor buffers...'cos
> everything then gets too low impedance for the ICSP to operate...and UR
> screwed.

I'd read about that somewhere, hence avoiding the ICSP pins with the LCD :)

<snip>

> Absolutely...a lot of designs use a 10K pot 'cos they begrudge the extra
> cost of an ickle resistor....V false economy if you ask me. Second
> gotcha....may need to fit 100nF between pot wiper and ground...some LCDs
> suffer from noise feeding through the contrast pin and ending up as a
> ghostly "snow" on the display.

The pot is a 10k pot (it came with another LCD at some point), but I
can see the logic in using a smaller pot and a resistor. Will add the
cap.

<snip>

Thanks :)

-Matt Daubney

Matthew Daubney

unread,
Oct 20, 2012, 10:03:27 AM10/20/12
to reading-...@googlegroups.com
Which bit is throwing the "not found"? The whole url or the schematic diagram?

I've attached the schematic just in case
schematic.png

Matthew Daubney

unread,
Oct 20, 2012, 10:05:12 AM10/20/12
to reading-...@googlegroups.com
Are those switch mode DC-DC converters? Wikipedia isn't very enlightening.....

Matthew Daubney

unread,
Oct 20, 2012, 10:06:58 AM10/20/12
to reading-...@googlegroups.com
On 20 October 2012 12:01, Richard Ibbotson
<richard....@btinternet.com> wrote:
> I would add a 100nf ceramic cap. as close as possible to the power pins of
> the processor.

Thank you, meant to do that but obviously forgot!

> Add pull up resistors to I2C bus.

To both sda and scl?

> Shorten track between ground of capacitor on crystal and processor ground.

Check.

> Suggest you make other suggested changes and fully route, so we can take a
> proper look.

Will do, thanks for the help :)

-Matt Daubney

Ryan .

unread,
Oct 20, 2012, 12:18:20 PM10/20/12
to reading-...@googlegroups.com

Ill pull one out and get you a part number

Nigel Worsley

unread,
Oct 20, 2012, 12:23:48 PM10/20/12
to reading-...@googlegroups.com
I wrote:
>> Your power input is expecting the centre pin to be positive

But I meant to say negative, I should stick to my policy of not posting
anything to the internet after a night of heavy drinking!

> I thought the solid line on the symbol for the power jack was the center pin?

It is, sorry for the confusion!

> Should the diode go on the negative side or the positive side?

Positive is traditional, putting it there avoids problems with using a lab PSU with
earthed negative and then probing the circuit with an oscilloscope (which almost
always has a connection to mains earth as well).

Nigle


graham

unread,
Oct 20, 2012, 1:03:02 PM10/20/12
to reading-...@googlegroups.com
Matt

Connect LCD pins 7,8,9,10 to 0V.

I agree with others...no point to the TTL octal buffer chip. also agree
re reverse polarity protection etc around the input voltage regs.
Whether or not you fit a series diode (1n4001 would do) I would up the
value of C3...it's way too low...make it same as C4,C6 (10uF) to ensure
a suitably low source impedance for the regs (else you might get a very
nice c. 2 megs oscillator!!).
Also makes the BOM simpler.

Contrast pot=2k2 with a 10K series resistor to Vcc, and a 100nF between
wiper and 0V.
I agree..definitely need a local decoupler for the processor 10nF or
100nF close to power pins
Also I agree about keeping the crystal loop as small as poss....and keep
signal lines away from it (I haven't seen the PCB...but you will get
jitter if signals cross-couple to the clock.).

No idea what the XBEE thingy is...but does it need decoupling?? Are you
dead sure none of those other pins need to be either allocated or tied
hi or lo?? (just check)

I guess yr LCD pins 15,16 are the backlight....provide for a resistor in
series with the pin 15...some LCDs do not have an onboard current limit
resistor (pop !!)...remember that LEDs are like zener diodes...without
Ilim they will (worst case) try to clamp yr 5V line to 4.2V (ouch) and
you will have SUCH fun trying to find the problem.

I don't know where yr program comes from...but set the unused io pins as
outputs and drive them low.

I don't know what value you have for C1...but it needs to be somewhere
between 10K and 100K.

I assume marking C5 as 100pF is a typo...should be 100nFOn 20/10/2012
15:03, Matthew Daubney wrote:
> Which bit is throwing the "not found"? The whole url or the schematic diagram?
It wasn't the URL
>
> I've attached the schematic just in case

Coooolio

graham

unread,
Oct 20, 2012, 1:10:28 PM10/20/12
to reading-...@googlegroups.com
Matt

Also...check out this month's Elektor magazine...7805 and 7905 drop-in
switchers!!

GC

graham

unread,
Oct 20, 2012, 1:11:43 PM10/20/12
to reading-...@googlegroups.com
On 20/10/2012 15:06, Matthew Daubney wrote:
> On 20 October 2012 12:01, Richard Ibbotson
> <richard....@btinternet.com> wrote:
>> I would add a 100nf ceramic cap. as close as possible to the power pins of
>> the processor.
> Thank you, meant to do that but obviously forgot!
>
>> Add pull up resistors to I2C bus.
> To both sda and scl?


4K7 to each (2 off)

GC

Matthew Daubney

unread,
Oct 20, 2012, 1:49:55 PM10/20/12
to reading-...@googlegroups.com
On 20 October 2012 18:03, graham <g...@coastelect.com> wrote:
> Matt
>
> Connect LCD pins 7,8,9,10 to 0V.

check

> I agree with others...no point to the TTL octal buffer chip. also agree re
> reverse polarity protection etc around the input voltage regs.

Will double check the datasheet..... pretty sure it said it shouldn't
work with 3.3v signals. Then again it worked with the arduino with
3.3v signals.

> Whether or
> not you fit a series diode (1n4001 would do) I would up the value of
> C3...it's way too low...make it same as C4,C6 (10uF) to ensure a suitably
> low source impedance for the regs (else you might get a very nice c. 2 megs
> oscillator!!).
> Also makes the BOM simpler.

check

> Contrast pot=2k2 with a 10K series resistor to Vcc, and a 100nF between
> wiper and 0V.
> I agree..definitely need a local decoupler for the processor 10nF or 100nF
> close to power pins
> Also I agree about keeping the crystal loop as small as poss....and keep
> signal lines away from it (I haven't seen the PCB...but you will get jitter
> if signals cross-couple to the clock.).

I'm interested in why for this... can I just run it back to the GND on
the micro or will that cause similar issues? Is this just to reduce
noise?

> No idea what the XBEE thingy is...but does it need decoupling?? Are you dead
> sure none of those other pins need to be either allocated or tied hi or lo??
> (just check)

All of the other pins are optional. It's a micro in it's own right
really. Just doing zigbee things :) It might be worth decoupling it
anyway though I suppose.

> I guess yr LCD pins 15,16 are the backlight....provide for a resistor in
> series with the pin 15...some LCDs do not have an onboard current limit
> resistor (pop !!)...remember that LEDs are like zener diodes...without Ilim
> they will (worst case) try to clamp yr 5V line to 4.2V (ouch) and you will
> have SUCH fun trying to find the problem.

Check

> I don't know where yr program comes from...but set the unused io pins as
> outputs and drive them low.

I'm writing it, so will do.

> I don't know what value you have for C1...but it needs to be somewhere
> between 10K and 100K.

I had it as 22pF same as C2 (it's what the datasheet said, assuming
we're talking about the caps near the crystal?)

> I assume marking C5 as 100pF is a typo...should be 100nF

Yes, that's a typo :)

<snip>

Thanks again,

-Matt Daubney

Matthew Daubney

unread,
Oct 20, 2012, 1:50:30 PM10/20/12
to reading-...@googlegroups.com
I had never heard of this magazine.... now I'm considering a
subscription! Shall run up to Smiths on the bike tomorrow and grab a
copy.

Matthew Daubney

unread,
Oct 20, 2012, 1:50:55 PM10/20/12
to reading-...@googlegroups.com
On 20 October 2012 18:11, graham <g...@coastelect.com> wrote:
> On 20/10/2012 15:06, Matthew Daubney wrote:
>>
>> On 20 October 2012 12:01, Richard Ibbotson
>> <richard....@btinternet.com> wrote:
>>>
>>> I would add a 100nf ceramic cap. as close as possible to the power pins
>>> of
>>> the processor.
>>
>> Thank you, meant to do that but obviously forgot!
>>
>>> Add pull up resistors to I2C bus.
>>
>> To both sda and scl?
>
>
>
> 4K7 to each (2 off)

Ta :)

- Matt Daubney

graham

unread,
Oct 20, 2012, 2:02:57 PM10/20/12
to reading-...@googlegroups.com
Call me on 01202 244309
>
>> No idea what the XBEE thingy is...but does it need decoupling?? Are you dead
>> sure none of those other pins need to be either allocated or tied hi or lo??
>> (just check)
> All of the other pins are optional. It's a micro in it's own right
> really. Just doing zigbee things :) It might be worth decoupling it
> anyway though I suppose.
>
>> I guess yr LCD pins 15,16 are the backlight....provide for a resistor in
>> series with the pin 15...some LCDs do not have an onboard current limit
>> resistor (pop !!)...remember that LEDs are like zener diodes...without Ilim
>> they will (worst case) try to clamp yr 5V line to 4.2V (ouch) and you will
>> have SUCH fun trying to find the problem.
> Check
>
>> I don't know where yr program comes from...but set the unused io pins as
>> outputs and drive them low.
> I'm writing it, so will do.
>
>> I don't know what value you have for C1...but it needs to be somewhere
>> between 10K and 100K.
> I had it as 22pF same as C2 (it's what the datasheet said, assuming
> we're talking about the caps near the crystal?)

Shit...I meant R1..the pull-up for the reset line

Matthew Daubney

unread,
Oct 21, 2012, 5:10:41 PM10/21/12
to reading-...@googlegroups.com
I've sorted all (I think) of the suggested changes and finished the
routing. I know there are some Via's to come out on the right hand
side (after removing the transceiver).

Anything else I've missed?

http://solderpad.com/mattdaubney/mqtt-s-test-board/

-Matt Daubney

Matthew Daubney

unread,
Oct 21, 2012, 5:12:21 PM10/21/12
to reading-...@googlegroups.com
(Bad form replying to ones self, but thought I should add this)

I've not gone with DC-DC yet, as I'm still understanding the parts :)
Got some datasheets for some maxims ones I'm just flicking through.

-Matt Daubney

Ryan .

unread,
Oct 21, 2012, 5:15:22 PM10/21/12
to reading-...@googlegroups.com

Nono i mean a monolithic quarter brick package not a discrete thing fuck laying those out! So much leaky rf

Ryan .

unread,
Oct 21, 2012, 5:16:44 PM10/21/12
to reading-...@googlegroups.com

Come to the space tomorrow and I'll find you one

graham

unread,
Oct 22, 2012, 4:45:29 AM10/22/12
to reading-...@googlegroups.com
Matt

If yr PCB reflects the schematic...then.....

D1 is wrong way round
the cap next to it is way too small value

R5 should be between top of pot and +5 NOT from wiper to LCD

And the 2 resistors for sda/scl are shown in series with the signals...not as pull-ups to +5V

None of it will work as it is.

by the way...the contrast pot ideally want to go clockwise toward ground...if u want contrast to increase as u go clockwise.

Regards
Graham Coast

Matthew Daubney

unread,
Oct 22, 2012, 5:09:58 AM10/22/12
to reading-...@googlegroups.com
On 22 October 2012 09:45, graham <g...@coastelect.com> wrote:
> Matt
>
> If yr PCB reflects the schematic...then.....
>
> D1 is wrong way round
> the cap next to it is way too small value

Fixed and fixed

> R5 should be between top of pot and +5 NOT from wiper to LCD

Misunderstood that, sorry :) Fixed

> And the 2 resistors for sda/scl are shown in series with the signals...not
> as pull-ups to +5V

Again, misunderstood. Fixed

> None of it will work as it is.
>
> by the way...the contrast pot ideally want to go clockwise toward
> ground...if u want contrast to increase as u go clockwise.

Is that the "standard" way of doing it? Generally I'm happy if
something changes when I turn it :) I haven't turned this around, but
if it is the standard way of doing it then I will do (if it's a
preference thing I'm not so bothered).

Really do appreciate all the help with this! The only other PCB I had
made was a very simple thing with only one chip on the board, so this
is an extra level of complexity plus a bit. This is the bit where I
wish I'd done electrical engineering at Uni rather than 9/10's of a
physics course.

I've pushed the new version up to:

http://solderpad.com/mattdaubney/mqtt-s-test-board/

Many thanks,

-Matt Daubney

graham

unread,
Oct 22, 2012, 5:52:09 AM10/22/12
to reading-...@googlegroups.com
On 22/10/2012 10:09, Matthew Daubney wrote:
> On 22 October 2012 09:45, graham <g...@coastelect.com> wrote:
>> Matt
>>
>> If yr PCB reflects the schematic...then.....
>>
>> D1 is wrong way round
>> the cap next to it is way too small value
> Fixed and fixed
>
>> R5 should be between top of pot and +5 NOT from wiper to LCD
> Misunderstood that, sorry :) Fixed
>
>> And the 2 resistors for sda/scl are shown in series with the signals...not
>> as pull-ups to +5V
> Again, misunderstood. Fixed
>
>> None of it will work as it is.
>>
>> by the way...the contrast pot ideally want to go clockwise toward
>> ground...if u want contrast to increase as u go clockwise.
Getting pots the wrong way round is summat I regularly do!!!
UR only gonna set contrast once...so it's up "U whether it's worth any
effort
> Is that the "standard" way of doing it? Generally I'm happy if
> something changes when I turn it :) I haven't turned this around, but
> if it is the standard way of doing it then I will do (if it's a
> preference thing I'm not so bothered).
>
> Really do appreciate all the help with this!
Welcome to my world of pain!!
> The only other PCB I had
> made was a very simple thing with only one chip on the board, so this
> is an extra level of complexity plus a bit. This is the bit where I
> wish I'd done electrical engineering at Uni rather than 9/10's of a
> physics course.

I did combined science....3 sets of lecturers all used to teaching
single subject degrees...tactical error on my part. Likewise...it was
only when I got to Uni that I found out that what I really wanted to do
was called "engineering"....anyhoo I should have gone with plan A and
become a doctor.
>
> I've pushed the new version up to:
>
> http://solderpad.com/mattdaubney/mqtt-s-test-board/
I've looked at yr PCB...I note that the PCB has a 10uF after the DC
input diode...but yr schem still shows 0.1uF.

When you build it you may well find you don't need the big sod-off
TO-220 regulators...and scale 'em down to the TO92 equivs. I can't
believe that ickle board needs lots of welly.
> Many thanks,
>
> -Matt Daubney
>

Ginjaian

unread,
Oct 22, 2012, 5:58:25 AM10/22/12
to reading-...@googlegroups.com
I should have gone with plan A and become a doctor.

With your bedside manner? You would have been immediately struck off! :))
Ian

graham

unread,
Oct 22, 2012, 6:09:46 AM10/22/12
to reading-...@googlegroups.com
Wot....U mean I have the subtlety of a main battle tank doing 80 MPH???

A glowing endorsement of my attitude to life!!

Try makin an omelette WITHOUT breaking eggs

GC

graham

unread,
Oct 22, 2012, 6:11:04 AM10/22/12
to reading-...@googlegroups.com
Q...Doctor, I have a horrible pain in my belly

A...Feckin deal with it...build a bridge...get over it


No???





On 22/10/2012 10:58, Ginjaian wrote:

mikethebee

unread,
Oct 28, 2012, 3:20:07 PM10/28/12
to reading-...@googlegroups.com
I think one needs nerves of steel to build an electronic bridge of any sort at the Hackspace :) Well done Matt.!!!


On Monday, October 22, 2012 11:10:39 AM UTC+1, GeeSee wrote:
Q...Doctor, I have a horrible pain in my belly
A...Feckin deal with it...build a bridge...get over it
!

No???




On 22/10/2012 10:58, Ginjaian wrote:
Reply all
Reply to author
Forward
0 new messages