Netburner (MOD54415) with other components in a PCB

[nicobari]

Hi,
This is a very general PCB question and may not belong here, so my apologies before hand. Since it involves Netburner I thought I should try asking here anyways. So I have a PCB which has a MOD54415 and an Novatel OEMV1-GPS receiver board connected to it. When I don't connect the MOD54415 and power up the PCB the GPS gets a position fix. However when I hook up the MOD54415 to the PCB and power up the PCB I never get a position fix. I checked for shorts, its drawing around 600 mA of current (whole PCB). Also when I take off the MOD54415 from the PCB I get a position fix again. During all this test the GPS status signal never showed any problem with its voltages or for that matter of fact the voltage always stayed around 3.3V. I was assuming it might be some AC noise and I have capacitors to take care of that. I also used jumper cables and hooked up just the Vcc and Gnd pin of the MOD54415, this way I ruled out any other MOD54415 pins creating any problems for the GPS. I know it is very user specific problem but I was wondering if anyone has any suggestions. Also everything is connected in parallel to the power pins.

Regards,
TM

[gavinm]

Certainly sounds like a power supply problem.
- If the whole pcb is drawing 600mA, what's the current draw with the MOD54415 unplugged? (The MOD54415 should take about 400mA).
- Use an oscilloscope and check for noise on the psu with / without the MOD54415 connected. You do get a bit of processor frequency superimposed on the 3.3v so you do need adequate decoupling.
- I don't know what the actual data connection is between the 54415 and GPS module, but I assume it's going to be serial - TTL or CAN or SPI or similar, an obvious point I know, but have you got correct ground connection between the devices (no ground loops, no extra impedance in series etc).
- You could always try earthing the board zero volts (but again watch out for ground loops).

I've just had a look at the OEMV1 datasheet, when you say the 3.3v "stayed around 3.3v", in fact the requirement is 3.3v +5%/-3%.
Putting this into absolute figures, this is quite tight 3.201 to 3.465 especially on the minimum side. Are you sure the ripple on the 3.3v is not dipping below 3.2v?

My best recommendation is to get a scope on the pcb supply so you can see the effect of connecting the module.

All the best with your investigations ...

[rnixon]

Just want to second Gavin's comments. I'm not a power supply expert, but I see inductors as well as capacitors used on the supply to gps components. I don't think capacitors alone will do the job. The picture of the novatel part does not have a metal shield. Does the actual physical part have a shield? If not there could also be a radiated noise concern.

I looked for a data sheet and noticed the part you are using seems to have went end of life in 2012, but maybe I have the wrong part number? Here's the link: http://www.novatel.com/support/info/doc ... 8#expanded

NovAtel Announces End of Life for the complete OEMV-1, OEMV-1G, OEMV-1GR, and OEMV-2 family of receivers, including all hardware variants. This extends to the following enclosure product: FlexPak-G2TM containing OEMV-1, OEMV-1G and OEMV-2.

These products will be available for order until:

October 31, 2012 (or until inventory is depleted)
- See more at: http://www.novatel.com/support/info/doc ... 8#expanded

[nicobari]

Hi,
Thanks for the reply. I hooked up the PSU to oscilloscope perviously and didn't see any oscillations except for small ripples (which dies out) when MOD54415 powers up. The MOD54415 is taking approximately 400 mA. Also GPS is hooked up to MOD54415 via TTL and OEMV1 that I am using was bought before they went out of service but it is similar to OEM615 (which is supposed to be a replacement for the OEMV1) and it is not inside a metal shield but the whole PCB is going to go inside a metal box in the future. Is there any way to solve ground loop problem without designing a new PCB? But right now I am hooking up Oscilloscope to PSU to check anything I might have missed before.

Regards,
TM.

[gavinm]

When looking at the 3.3 on the scope, remember you're interested in just a 0.1v deviation.

rnixon's comment about inductors for psu smoothing is perfectly valid, but be careful, they must be used with adequate damping capacitance otherwise the stored energy in the inductor can easily cause quite high voltage oscillations (I've seen over 90v in a circuit powered by 24v recently because the ripple current through the 10mH inductor caused it to burst into oscillation (it should have been 10uH!!).

You want a good (very low impedance) connection between the GPS zero volts and the MOD54415 zero volts. Watch out for the screens on any connecting cables being the source of ground loop currents (if you have any of these then try only connecting the screen at one end of the cable).

If you're certain it's not the power supply the next obvious thing is that when you connect the MOD54415, it is sending something via the serial interface that is preventing the GPS from working, or at least holding some of the interface signals in a state that prevents it from working.

Have you tried (just as a suggestion), downloading software to the 54415 and then connecting the GPS. Something simple that doesn't use the pins that you use to drive the GPS - perhaps like the Factory Demo app?

I'm thinking that maybe when you connect the 54415 your software is driving a signal that should be an input and causing some over current condition in the GPS (and 54415).

Good Luck - when you find your problem it will be blindingly obvious!

[roland.ames]

alternative to downloading the factory app to the MOD54415 is to short the link (TP1) that forces it into boot monitor mode.

This will still use the selected monitor serial port, so you should set the monitor serial port to one that is not connected to the GPS if that is possible.

then again

I also used jumper cables and hooked up just the Vcc and Gnd pin of the MOD54415, this way I ruled out any other MOD54415 pins creating any problems for the GPS.

so what runs in the MOD54415 is really only relevant as far as it affects power consumption ??

[dciliske]

I'll weigh in on this a little bit, considering we've been talking about GPS designs lately (a custom NTP server and other stuff that's being discussed). So, to begin with, GPS is a RF signal. Therefore, you need an antenna to receive it (which you have, undoubtedly). It's also an incredibly weak RF signal, meaning it needs a lot of amplification and hardware to lock onto the carrier wave. As it's a very weak signal and there's a lot of amplification going on, noise on the power will have a large impact on the receiver's ability to find the signal. As a result, you want as low of noise as possible on the power rail.

This is the reason for the suggestion of inductors on the power rail for the receiver. You may also consider a fully separate power supply for the receiver (you can probably just whack a linear down and feed it raw input if you don't have 5 volts available elsewhere).

The second thing to consider is what is your antenna situation. What kind of antenna are you using? Where is it in relation to the MOD54415? Is the MOD54415 shielded? (this doesn't sound to be the case currently) Unfortunately, the MOD54415 has some significant radiated emissions on harmonics of 125 MHz, as everything in the world is synchronous with it (CPU, RAM, Ethernet, CPU peripherals). Another thing to look at is whether there is an ethernet cable connected and whether it is shielded or unshielded. An unshielded ethernet cable becomes an antenna radiating on a 125MHz frequency due to it's symbol rate at 100Mb.

Look for traces that might be prone to capacitive or inductive coupling. If you have an ethernet cable connected, does it cross over the board at all? Is it near your antenna?

This is about all I can think of for things to try for the time being. And I used to think I was a software guy...

-Dan

[nicobari]

Thanks Dan,
The GPS does have an Antenna. From my debugging I figured the issue most likely is noise on power rail and ground loops. After some trace cutting to remove the suspected loop, I removed the MOD54415 got a GPS fix and then plugged in the MOD54415 back in and the GPS fix stays in place. This makes me think that along with ground loop RF noise is also an issue. We are designing a new board where GPS receiver board has its own separate power supply. What would be an easier way to shield MOD54415 without increasing too much weight and can used easily on existing PCB designs.

However before we design a new PCB I want to make sure that we understand the problem properly, I am still working with the existing design and cutting traces here and there to remove suspected ground loops and couplings. Thanks for the help.

TM

[nicobari]

So I finally solved the issue and I really want to thank Dan for mentioning EMI shielding. After I shielded my GPS PCB (as it was much smaller than MOD54415 and hence easy to shield) I always get a GPS fix. Thanks again for the help.

Regards,
TM

[rnixon]

Thanks for the update. Its always interesting to know how those things work out. Did your GPS module have a metal shield at all, or was it just a pcb with exposed components?