Could someone check my design?

Am 07.10.2015 um 14:02 schrieb Jacob Thastrup:

Hi Guys

I'm totally new to this. I'm trying to build a 4 channel opamp (boost my

signal about 100x)

I'm hoping someone has the time to quickly check my design and see if it

makes sense and perhaps let me know how to improve it.

 

Thanks

Jacob

 

--

To view any images and attachments in this post, visit:

http://www.element14.com/community/message/162669

 

Attachments:

opamp.zip

 

No capacitors?

--

Mit freundlichen Grüßen / With best regards

Joern Paschedag

You mean to filter the input?

The opamp is meant to amplify a 7-50mV signal around 100 times so I can reliably read it with a National Instruments DAQ.

I'm sorry I'm such a newbee, any and all help is really appreciated! I'll try and add some capacitors to the diagram and see if I can work out how to add a GND plane :-)

I'll be back ;-)

Thanks

Jacob

Hmmm not sure i got the capcitors right. ;-)

How's this version?

Thanks again

Jacob

Am 08.10.2015 um 14:43 schrieb Jacob Thastrup:

Hmmm not sure i got the capcitors right.

 

How's this version?

 

Thanks again

Jacob

 

--

To view any images and attachments in this post, visit:

http://www.element14.com/community/message/162771

 

Attachments:

test4.zip

 

Where is your attachment???

Anyway I didn't like the many junctions and the separated connectors in

the schematic,

so I messed around in your design (as an example of course). Same went

for the board

I also changed the width of the tracks, because I suppose you want to

etch yourself...

--

Mit freundlichen Grüßen / With best regards

Joern Paschedag

Hi Joern

Maybe I'm getting it wrong. But I don't see any files from you?

Thanks

-Jacob

Hmmm not sure i got the capcitors right. ;-)

Partially...

For the 5µF capacitor: it's connected correctly in schematic, but I'm not sure you will found a value of  5µF, but a value of 4,7µF, 10µF, 22µF, 47µF... are standardized values . Did you check if you use the good package ? often this king of capacitor have a cylinder shape, and it polarized (+ shoufd be connected to +24V, and - to GND)

A good advice would be to found a capacitor (buy it or at least found were you will buy it), so you be sure of the value, the dimensions, the shape, if it's polarized or not....

The physical position on the board doesn't really mater for that capacitor.

For the 2 others, It's NOT OK: you need only one capacitor, and it should be in parallel not in series with LM324 :

     -> in schematic : you connect the +V pin (pad number 4) of the LM324 directly to +24V, and do the same with the GND pin (pad number 11), connect directly to GND. Then you add your 100nF capacitor and connect one pin to +24V, the other one to GND

     ->  in board, you place physically this capacitor near the LM324, the nearest is the best.

The role of these capacitor is :

- filter noise coming from your supply.

- provide a local reserve of energy for you amp : when the amp will need to swing is output, it will sink current, and will need fast, and if there is no capacitor around, the current will flow from your supply through your cable and the copper of your board. With ideal wire no problem. but in real life, they have a little resistance and inductance: both will slow down current flow, and the voltage of the supply near the LM324 will goes down....

That why a good practice is to add :

- 1 huge capacitor (10µ - 100µF) for the board: it's provide energy for 'slow' consumption raises, and filter low frequency noise from the power supply...

- 1 little capacitor for each integrated circuit (chip), near the chip,  for fast consumption variations, and filter for higher frequency noise...

This is only a very very simplified explanation and complex design need more than this little cook book, but in you case that could be a good start !

That also why you need to have larger wire for supply than for the other signal.

Guillaume.

Am 08.10.2015 um 17:38 schrieb Jacob Thastrup:

Hi Joern

Maybe I'm getting it wrong. But I don't see any files from you?

Thanks

-Jacob

 

--

To view any images and attachments in this post, visit:

http://www.element14.com/community/message/162826

 

I attached it but I suppose it is the glorious interconnection between

the newsgroup and element 14 that never works since years...

Next try now...

--

Mit freundlichen Grüßen / With best regards

Joern Paschedag

Am 09.10.2015 um 10:42 schrieb Joern Paschedag:

Am 08.10.2015 um 17:38 schrieb Jacob Thastrup:

Hi Joern

Maybe I'm getting it wrong. But I don't see any files from you?

Thanks

-Jacob

 

--

To view any images and attachments in this post, visit:

http://www.element14.com/community/message/162826

 

 

I attached it but I suppose it is the glorious interconnection between

the newsgroup and element 14 that never works since years...

Next try now...

 

No, sorry it  didn't go throu, maybe because I'm not registered at

element14 .

And I'm not interested to register there.

--

Mit freundlichen Grüßen / With best regards

Joern Paschedag

On 09.10.2015 11:12, Joern Paschedag wrote:

Am 09.10.2015 um 10:42 schrieb Joern Paschedag:

Am 08.10.2015 um 17:38 schrieb Jacob Thastrup:

Hi Joern

Maybe I'm getting it wrong. But I don't see any files from you?

Thanks

-Jacob

 

--

To view any images and attachments in this post, visit:

http://www.element14.com/community/message/162826

 

 

I attached it but I suppose it is the glorious interconnection between

the newsgroup and element 14 that never works since years...

Next try now...

 

 

No, sorry it  didn't go throu, maybe because I'm not registered at

element14 .

And I'm not interested to register there.

 

You can find them here

http://www.eaglecentral.ca/forums/index.php/t/51848/5abd43c14367c14cffef482390e897d4/

Once again thank you guys!!

Guillaume, thank you for the explanation it really helps!

I've added a ground plane to Joern's design... does that make sense?

I hope you can see the attached file..otherwise I'll go register at eaglecentral as well.

-Jacob

Hi,

That last design is better, the ground plane is OK.. You remove the big capacitor  (your 5µF) I suggested to check the value and the package, but not t remove it ? If it was one of my design I would  put a 10µF electrolytic capacitor between GND and +24V, just right after the supply connector...

Nice work ....

Hi,

That last design is better, the ground plane is OK.. You remove the big capacitor  (your 5µF) I suggested to check the value and the package, but not t remove it ? If it was one of my design I would  put a 10µF electrolytic capacitor between GND and +24V, just right after the supply connector...

Nice work ....

Guillaume barrey[1

wrote on Sat, 10 October 2015 05:53]Hi,

 

That last design is better, the ground plane is OK.. You remove the

big

capacitor  (your 5µF) I suggested to check the value and the

package,

but not t remove it ? If it was one of my design I would  put a 10µF

electrolytic capacitor between GND and +24V, just right after the

supply

connector...

 

Nice work ....

Ah ok hehe, I've added the 10uF capasitor now.

Quote:

I think there should be a ground connection for each channel. In other

words, you will need 3-pin instead of 2-pin connectors. Otherwise your

only ground connection is through the power supply connector. The ground

pin should be between each input and output pin.

I'm not sure I understand this bit.

The signal I'm trying to amplify is coming from small oxygen cells

(galvanic cells). If I understand what you are suggesting, is that I use

the GND from each cell to connect to their respective inverting input of

the opamp? My original idea was to tie all the GND lines from the cells

into the same GND connector (the one I also connect the power supply to).

Potentially a stupid question.. will that work? I don't need to ground

everything "together"?

Quote:

If you swap the ends of R5 and R6, then you can have a single trace for

OUT3, instead of 2 traces. You can do the same thing for the other

channels.

Cool, that does look better

Once again thanks for all the help!!

-Jacob

--

Web access to CadSoft support forums at www.eaglecentral.ca.  Where the CadSoft EAGLE community meets.

Hi,

The sensor cell will produce a voltage, between the 2 signals :

- the signal you will connect to your amplifier input

- and the GND.

Only the voltage between these 2 signals (the difference) is important. The current that will flow out from your cell into the input of the amplifier is exactly the same than the current flowing from the GND of your board into the GND pin of your cell.

So, the role played  by the 2 wires of the oxygen cell (signal and GND)  is equivalent, and there is no reason to have a different path (except if GND is also used to supply the sensor see (*) ).

On the amplifier side : it amplifies the difference between the input signal and the GND plane, the GND plane is a REFERENCE.

If instead of connecting directly the GND of sensor to your board, you connect it to a point away from the plane (name it STAR point), with all the other sensor GND and supply GND, then you connect this start point to your board with a single wire. On the paper this will work.... but....

In this last GND segment (wire from the GND plane and star point)  you will have the sum of all the currents of all sensors, plus the current drained by the amplifier it self (and this last one is far more important than the others).

Where is the problem ?

The problem is that the copper is a good conductor, but not an ideal one !

You will have a little voltage difference between your GND plane and the start point, that will be dependent of that current, and the length of the wire, and so amplifier REFERENCE will not be exactly the reference of the sensor.

In other word, each signal will be polluted with the signal of the other cell and the consumption of the amplifier...

There is also other advantages of using same path for your sensor: you can use twisted pair wire, that way you reduce noise, specially if your sensor are far away from the borad.

Just one thing (*) : I don't know your sensor: is this sensor need a supply ? How this supply is wired? is the GND of sensor is only a output reference or also used for the return path of the supply ? because in that case you have to figure how best can the GND be wired.... there is no magic solution, and we can help you unless you give us more info...

Sorry with this very quick fast ultra simplified explanation... and if you want to have more fun... you can also question yourself about the output of your amplifier : How the GND path mcan be optimised and wired, on the output side (I suppose you will connect it to an acquisition device...)

Guillaume

Guillaume barrey[1

wrote on Sun, 11 October 2015 08:04]Hi,

 

The sensor cell will produce a voltage, between the 2 signals :

- the signal you will connect to your amplifier input

- and the GND.

Only the voltage between these 2 signals (the difference) is

important.

The current that will flow out from your cell into the input of the

amplifier is exactly the same than the current flowing from the GND of

your board into the GND pin of your cell.

So, the role played  by the 2 wires of the oxygen cell (signal and

GND) 

is equivalent, and there is no reason to have a different path (except

if GND is also used to supply the sensor see (*) ).

 

On the amplifier side : it amplifies the difference between the

input

signal and the GND plane, the GND plane is a REFERENCE.

 

If instead of connecting directly the GND of sensor to your board, you

connect it to a point away from the plane (name it STAR point), with

all

the other sensor GND and supply GND, then you connect this start point

to your board with a single wire. On the paper this will work....

but....

In this last GND segment (wire from the GND plane and star point) 

you

will have the sum of all the currents of all sensors, plus the current

drained by the amplifier it self (and this last one is far more

important than the others).

Where is the problem ?

The problem is that the copper is a good conductor, but not an ideal

one

!

You will have a little voltage difference between your GND plane and

the

start point, that will be dependent of that current, and the length of

the wire, and so amplifier REFERENCE will not be exactly the reference

of the sensor.

 

In other word, each signal will be polluted with the signal of the

other

cell and the consumption of the amplifier...

 

There is also other advantages of using same path for your sensor: you

can use twisted pair wire, that way you reduce noise, specially if

your

sensor are far away from the borad.

 

Just one thing (*) : I don't know your sensor: is this sensor need a

supply ? How this supply is wired? is the GND of sensor is only a

output

reference or also used for the return path of the supply ? because in

that case you have to figure how best can the GND be wired.... there

is

no magic solution, and we can help you unless you give us more info...

 

Sorry with this very quick fast ultra simplified explanation... and if

you want to have more fun... you can also question yourself about the

output of your amplifier : How the GND path mcan be optimised and

wired,

on the output side (I suppose you will connect it to an acquisition

device...)

 

Guillaume

 

--

To view any images and attachments in this post, visit:

http://www.element14.com/community/message/162998

Hi Guillaume

Thank you again for your reply.

Just to make sure I understand. I should still connect all the GND signals

together, just at different points on the GND plane?

The Sensor does not require any power, it generates it itself though a

chemical reaction (dependent on the oxygen fraction).

The amplified signal will be going to an acquisition device, should I also

wire their GND "away" from the power supply GND? I guess I could wire them

in the same terminal as the GND from each sensor?

Thanks

Jacob

--

Web access to CadSoft support forums at www.eaglecentral.ca.  Where the CadSoft EAGLE community meets.

Hi,

Yes you understand very well, all GND will be connected together. Here we only speak about where physically it's better to do it. You can  do it every where, it will work, but some location can reduce noise and measurement error, and other will be introduce more noise.

Since your cell need no power, the best will be to use a twisted pair wire for each cell, and  connect it directly to the board, so for each channel you need to add an extra connection for the GND of the cell.

For the acquisition device, it's dependent on if it have an extra pin for GND or not.... If yes, you use a separate twisted pair wire for each channel and use this extra pin.

if not, then you can also use twisted pair wire but on the acquisition device side you connect all the GND to the acquisition device GND pin.

In both case on your amplifier board, you choose a 3 pins connector for each channel: input, GND and output. On the GND pin, you connect the GND wire from your cell and the GND wire from the acquisition device together.

Guillaume

Guillaume barrey[1

wrote on Mon, 12 October 2015 05:20]Hi,

Yes you understand very well, all GND will be connected together. Here

we only speak about where physically it's better to do it. You can 

do

it every where, it will work, but some location can reduce noise and

measurement error, and other will be introduce more noise.

 

Since your cell need no power, the best will be to use a *+twisted

pair

wire+* for each cell, and  connect it directly to the board, so for

each

channel you need to add an extra connection for the GND of the cell.

 

For the acquisition device, it's dependent on if it have an extra pin

for GND or not.... If yes, you use a separate twisted pair wire

for

each channel and use this extra pin.

if not, then you can also use twisted pair wire but on the

acquisition device side you connect all the GND to the acquisition

device GND pin.

 

In both case on your amplifier board, you choose a 3 pins connector

for

each channel: input, GND and output. On the GND pin, you connect the

GND

wire from your cell and the GND wire from the acquisition device

together.

 

Guillaume

 

--

To view any images and attachments in this post, visit:

http://www.element14.com/community/message/163048

OK, cool thanks.

So version god knows what is attached

(I must admit I think the traces on this version are a lot cleaner than

some of my previous ones)

With regard to twisted pair I guess I could use some CAT 5e Ethernet cable

I have lying around (somewhere).

Thanks once again!

-Jacob

--

Web access to CadSoft support forums at www.eaglecentral.ca.  Where the CadSoft EAGLE community meets.

Hi,

That's Great !

Just a last little advice for the GND plane.

When my board is almost finish, I do some little adjustment to be sure that my GND plane will be good.... It's hard to explain with my poor English....(sorry to be a French frog)

I hide all layer except the dimension, via, pads and the copper layer on which I work, and I run a ratsnest command to process the polygons

Then you use the show command and select the polygon (in your case you only have GND). The polygon is highlighted so you can see it well.

Now you try to imagine how the current can flow from here and there...

Sometime you will see that for connecting to  point that are not really far from each other, the current need to do a huge travel because the plane is interrupted by some other tracks. Sometime it's easy to reduce drastically the effective distance just by moving a little bit a track or a part.

In your example, the +24V track going from C1 to IC1 completely cut he polygon. If you move down a little this track, the GND plane will be not cut any more...

That way, I check all the little improvement I can do, and some time it's not enough, so I reduce the isolation or the wire of the polygon.

I do that for all polygons, layer by layer (copper layers of course).

But any way good job !

Guillaume.

Guillaume barrey[1

wrote on Wed, 14 October 2015 13:10]Hi,

 

That's Great !

Just a last little advice for the GND plane.

When my board is almost finish, I do some little adjustment to be sure

that my GND plane will be good.... It's hard to explain with my poor

English....(sorry to be a French frog:p)

I hide all layer except the dimension, via, pads and the copper layer

on

which I work, and I run a ratsnest command to process the polygons

Then you use the show command and select the polygon (in your case

you

only have GND). The polygon is highlighted so you can see it well.

Now you try to imagine how the current can flow from here and there...

Sometime you will see that for connecting to  point that are not

really

far from each other, the current need to do a huge travel because the

plane is interrupted by some other tracks. Sometime it's easy to

reduce

drastically the effective distance just by moving a little bit a track

or a part.

 

In your example, the +24V track going from C1 to IC1 completely cut he

polygon. If you move down a little this track, the GND plane will be

not

cut any more...

 

That way, I check all the little improvement I can do, and some time

it's not enough, so I reduce the isolation or the wire of the polygon.

 

I do that for all polygons, layer by layer (copper layers of course).

 

But any way good job !

 

Guillaume.

 

--

To view any images and attachments in this post, visit:

http://www.element14.com/community/message/163345

Hi Guys

Sorry for the slow response.

Thank you all for all the help. I feel I have learned a lot already!

Best wishes

Jacob

--

Web access to CadSoft support forums at www.eaglecentral.ca.  Where the CadSoft EAGLE community meets.