Power Module Performance Comparisons

To break this out of the other thread http://www.tinyclr.com/forum/topic?id=8742 as requested.

The test setup is a Gadgeteer device which cycles power on/off, in this case 12 minutes on and then 3 minutes off for a board with a MakeGPS module v1.2, SD Module (micro v1.2) and a MulticolorLed (v1.1) to be used with a collection of power modules:

LoveElectronics.USBDCPower v1.2
GHIElectronics.UsbClientDP v1.3
Electronics.UsbClientSP v1.2

The idea is to compare the GPS fix times to see if the different power modules have any effect on performance

USB Wall Wart 5V 1A level 3
DC Wall Wart 9V 650ma

Love (USB Powered) 33 test cycles

Average 0:02:24

Min 0:00:46
Max 0:08:46
Dev 0:01:49

UsbClientDP (USB Powered) 51 test cycles

Average 0:03:16

Min 0:00:46
Max 0:10:40
Dev 0:02:27

UsbClientSP 54 test cycles

Average 0:01:36

Min 0:00:33
Max 0:04:27
Dev 0:00:54

UsbClientDP (DC Powered) 81 test cycles

Average 0:03:18

Min 0:00:35
Max 0:09:02
Dev 0:02:22

4 cycles unable to get a GPS fix.

Interesting as the UsbClientDP has pretty much no difference in performance between USB and DC connections.

Love USBDCPower (DC Powered) 107 test cycles

Average 0:03:28

Min 0:00:47
Max 0:10:17
Dev 0:02:16

Unable to get fix on 12 cycles so times are for 95 GPS fixes.

Currently testing SeeedGPS on Love power module (DC powered).

What mainboard did you run these tests on? It looks like a Cerberus from the picture, but I want to make sure…

And a Spider below.

The MakeGPS and power modules are being tested with a Cerberus. The device which is doing the power cycles is powered by a Spider.

Thanks for the update!

Would it be too much to ask for these numbers with the Seeed GPS module? I’m curious to see how the different power modules may be effecting it as well.

@ Duke Nukem - I’m glad you can get a fix at all with this combination of hardware. Did you say this testing was inside a building? Close to a window or not?

Skewworks’ Spider took many hours to get a fix (>10). On the other hand, HughB was able to get a fix powering his setup via a car lighter adapter while driving. There has not been enough consistency with field testing across users of the MakeGPS to discern a pattern to the (lack of) fix issue. It’s starting to really make me crazy, or crazier I guess. Maybe it’s a combination of hardware and locale factors? More observations from those who have one would really help.

I don’t have a self contained mobile solution (as of yet) - - so it on my car hood with an open sky above me.

  • With laptop + USB + Spider + DP + MakeGPS - sometimes never and sometimes it took a long time.
  • With a Domino + MakeGPS + USB + laptop - I repeatedly got very quick acquisition times.

When I get time to test again (work is really kicking my rear as of late) I will try to do more “scientific” testing.

Someone ought to put the DP on a scope, let’s see what it looks like.

I have, or should say, had someone else take a look with a “pro” scope and me hovering nearby. Both my DP modules are quite noisy. That’s why I put the linear regulator on the module - to remove (or dramatically reduce) the possibility it had something to do with the fix issue.

@ ransomhall - Being a hardware idiot so this is probably barking up several wrong tress but would a ferrite bead help? The reason i ask is i have been looking at some autopilot designs and they are using them on their GPS connectors.

There is a ferrite bead on the MakeGPS module.

@ godefroi - is it just the one on VCC? or on RX and TX as well?

Just on VCC. I don’t think you’d want one on TX and RX, but I could be incorrect about that (the datasheet certainly doesn’t recommend it). Even if you did, I doubt that would affect fix times (the datasheet certainly goes to great lengths to tell you clean power is necessary for proper functioning).

I have already did one side by side test of MakeGPS and SeeedGPS as the Cerberus has two U sockets:

After 48 test cycles of (12 min powered, 3 min non powered) the side by side comparison of the Seeed GPS to the new MakeGPS (I believe I was using the Love power module):

Number of cycles Seeed was able to get a fix, but the MakeGPS wasn’t able to: 8
Number of cycles MakeGPS was able to get a fix, but the Seeed wasn’t able to: 12
Number of cycles that neither was able to get a fix (within 12 minutes, remember I’m in my home office): 2

Seeed / MakeGPS
Average 0:07:00 / 0:02:47

Min 0:02:24 / 0:00:53
Max 0:10:36 / 0:07:07

Dev 0:02:15 / 0:01:32

I’ll do some Seeed GPS testing with the different modules after the current DC Love/MakeGPS test is done (I’m at 59 cycles now and might not be able to get back to it till this evening at which point I’ll end it, report the findings and then switch over to testing the SeeedGPS with the different power modules)

I’m in a home office (bungalow) and my desk looks out a window, so the GPS modules are 3 - 4 feet away from a window (in Calgary Alberta Canada). While GPS might suffer from temporarily poor statelite positioning, I don’t believe there are spots on the planet that suffer chronic GPS problems (except perhaps deep canyons or other similar local geological features).

@ godefroi - i have no idea only reason i asked is from looking at a design https://pixhawk.ethz.ch/px4/start as they have them on VCC TX and RX

So the MakeGPS is clearly better than the Seeed module, then.

Sure enough, they do. The bead they specify is for filtering noise in the GHz range. They must be picking up that kind of noise from somewhere.

@ Duke Nukem - Have you used this http://www.navcomtech.com/Support/Tools/satellitepredictor/main.cfm before?
Would be interesting to see if lower sat numbers makes any difference to your results.

@ godefroi - Ah ok thanks - slowly learning…

I use a clever combination of learning and making it up as I go. If it seems like I know what I’m talking about, it’s only because I managed to get the second part down :wink: