I have 2 layer board that is working perfectly fine. I have created nearly identical 4 layer boards. On both 4 layer boards the power/signal layer distribution is done as follows:
signal/ground layer (outer layer, top)
+5V layer (inner layer)
-5V layer (inner layer)
signal/ground layer (outer layer, bottom)
I cannot physically move signal layer into inner layers due to the fact the signal must go through a few resistors. The difference between two 4 layer boards is that:
-board 1 has the entire layers dedicated to +5V and -5V (which are not close to the edges of the board, and have clearances for all vias)
-board 2 has traces about 3x normal trace with used for the power rails on the 2 layer boards.
I have made two vias in each place where the power needs drawn (see below). If the power is needed between components 1 and 2 such as component_1----------------component2 The do the following: component_1----via_1-------via_2----component2
The +5 and -5 voltages are being supplied to where they are supposed to be supplied (checked with multimeter). There is no shortcircuit on the 4 layer boards (checked with multimeter).
My application is: Transimpedance Amplifier (1st opamp) and Voltage gain (2nd opamp)
The problem: the opamps are not drawing any current and there is no signal output. On the power supply unit, I can see that my 2 layer board is drawing about 20mA.
Please suggest what can be causing this? Also, please let me know if I missed some important information required to answer this.
Time to get your trusty VOM out. Verify that there is continuety between the ground pins on the chips and the ground of the power supply. Now start measuring voltages. You might need a 'scope to see where you lose the signal.
Why? If it was working perfectly fine, why change it?
That's an unusual configuration. Most analog designs benefit more from a ground plane than from power planes without a ground plane.
Also, +5V and -5V are slightly unusual voltages for powering opamps. +/-15V, +/-12V, or even +/-10V are more commonly used. Are you using +5V so you can power some logic chips?
Is this a through-hole or surface mount design? If it is through-hole, which layer a trace is on has nothing to do with whether it goes to a resistor.
Here is how to troubleshoot a new analog board layout that supposedly has the same circuit (is built to the same schematic) as an existing board:
First, test them both and verify that they both work the same. If one of them is missing a signal starting at a particular stage, that's the place to look at first.
You have found that they don't work the same -- they draw 0mA and 20mA. Feed appropriate DC level into all the inputs, go through every circuit with a voltmeter, and make a list of everything else that is different.
There are only three possible reasons why the two would work differently when fed DC as described above:
[1] The circuits are not the same. Somewhere a trace is added, missing, or goes to the wrong place. [2] A component is bad or has the wrong value.
[3] (Rare) a circuit that didn't oscillate on the old board oscillates on the new board, or a circuit that did oscillate on the old board does not oscillate on the new board.
To troubleshoot [1] (The circuits are not the same): Get an assistant, have one of you use a DMM in continuity mode to see what each pin connects to, while the other highlights the connections on the schematic. It is important to do this single-blind. I do it like this:
Me: (looking at schematic) "Does U1 pin 3 go to U3 pin 6?"
Him: (looking at board with meter) "Yes."
I then use a yellow highlighter to color the connection on the schematic.
Me: "Good. Does it go anywhere else?"
Him: (looking at where the traces go on the board and checking the other pins of U1, U2, and any components close to them) "No."
Note that some meters won't beep if you aren't on the trace long enough. You want one that beeps when you slide the probe from pin-to-pin on a connector or IC and get a connection for a very short amount of time.
Me: "Good." ...and so on. To reduce the chances of errors, every few minutes I ask whether a pin that *doesn't* connect to to another connects, and say "good" when he says "No."
If he gives an unexpected answer, we look at the board, at the artwork, and at the schematic, and occasionally refer to the old board and its artwork.
To troubleshoot [2] (component is bad or has the wrong value), use an LCR meter and DMM in diode mode to measure every discrete component, a logic probe and pulser to measure every logic gate, and a signal generator and oscilloscope to test every opamp and transistor. If you do this a lot, get a Huntron Tracker (see
formatting link
and use that first.
To troubleshoot [3] (oscillations), start by making sure that troubleshooting method [2] didn't find anything, then start looking at signals with an oscilloscope. Check amplifier and oscillator circuits first.
Or, if you are doing this at work, hire a good electronics engineering technician from a temp agency. *Most* electronics engineers can do the same job, but there is an occasional one that graduated with no hands-on experience making things work.
Depends on what generation of parts you are using. There are a lot of hi-tech CMOS op-amps that are optimized for 1.8V, 2.5V, 3.3V operation and many of them are at or even beyond the upper limits of supply range for +/-5V operation.
Now, for old geezers like you and me maybe +/-18V rails on op-amps might seem more usual :-).
There are a lot of op-amps that can't take +/-3.3V... eg. OPA379 5.5V maximum total supply voltage, and even some that can barely handle
+/-1.8 (eg. TLV2781 3.6V maximum).
Best regards, Spehro Pefhany
--
"it\'s the network..." "The Journey is the reward"
speff@interlog.com Info for manufacturers: http://www.trexon.com
Embedded software/hardware/analog Info for designers: http://www.speff.com
Unless there is a mistake in the net then I cant see what is wrong. Have you run a "continuity check" and an "integrity check" on your layout ? I am assuming your software has these? if not dump it !
Hopefully the original poster lets those know if any suggestions worked or suspicions were correct.
Myself I am betting that his multi-layer board doesn't have connections to his internal layers. Note that one original comment below, where it states it has clearances for all vias. Well some of his vias should not have clearances to those internal layers if it is supposed to work. I figured that the simplest foul-up for someone starting a first multilayer design is not getting the connections to the internal layers connecting correctly.
--
Sincerely,
Brad Velander.
> On 30 Aug, 23:36, Vitaliy wrote:
>> Hello,
>>
Snip
>> The difference between two 4 layer boards is that:
>> -board 1 has the entire layers dedicated to +5V and -5V (which are not
>> close to the edges of the board, and have clearances for all vias)
Snip
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.