/var/www/restricted/ssh/stm32/www/stm32circle/ STM CircleOS forum / Primer2 analog ground bug

Username:     
Password:     
             

Forum

# 1   2009-04-09 22:18:11 Primer2 analog ground bug

kubark42
Member
Registered: 2009-02-13
Posts: 46

Primer2 analog ground bug

It seems that the Primer2 ties together the analog ground with the digital ground. This is quite unfortunate, as it dramatically reduces the usefulness of the ADC (and perhaps by extension the DAC). For instance, we've got some very sensitive analog accelerometers in the lab that are embedded with a GPS. Unfortunately, we can't use the Primer for this extension board as the GPS contaminates the accelerometer any time it sends serial data.

I'm not sure what can be done about this, except perhaps to fix it in a future revision. I've thought about cutting the traces where they tie into the digital ground, but I've got no experience with this and it might prove difficult. I'm open to any suggestions if anyone has ideas for how we can easily remedy this problem.

Offline

 

# 2   2009-04-09 23:16:52 Primer2 analog ground bug

nibj
New member
Registered: 2008-08-06
Posts: 1

Re: Primer2 analog ground bug

I think the grounds has to be tied together.

The realm problem is the ADC itself. Any SOC ADC's have noisy problems. Tricks can hide the problem in some applications, but for real sensitive signals the only choice is a good external ADC.
Look this article.

Offline

 

# 3   2009-04-10 06:25:09 Primer2 analog ground bug

kubark42
Member
Registered: 2009-02-13
Posts: 46

Re: Primer2 analog ground bug

nibj :

I think the grounds has to be tied together.

The realm problem is the ADC itself. Any SOC ADC's have noisy problems. Tricks can hide the problem in some applications, but for real sensitive signals the only choice is a good external ADC.
Look this article.

Thanks for the link. I learned quite a lot reading it.

I realize that one will never have the resolution in a uC as on a good external ADC. However, as I understand the ADC ground does NOT have to be tied together. Otherwise, there would be no point in the uC having separate analog ground and V_ref_minus pins. So even if the performance is nothing stellar, it seems we should still be able to get more out of the Primer2 with a separated analog ground.

In my opinion, they tied them together because otherwise they would have had to dedicate a header pin to it. Those pins were already in short supply, so from certain vantage points it seems like a good decision. Maybe what they could have / should have done was use a jumper on the link between analog ground and digital ground.

Cheers,
Kenn

Last edited by kubark42 (2009-04-10 06:26:37)

Offline

 

# 4   2009-04-10 08:10:16 Primer2 analog ground bug

Francis
Administrator
From: France-Grenoble
Registered: 2007-07-09
Posts: 890

Re: Primer2 analog ground bug

We are preparing the next production, with a modified PCB. Therefore, any suggestion is welcome. In anycase, the AGND pin is not on the extension connector, and we have to connect it to the digital ground (but perhaps we could add some filtering).

Offline

 

# 5   2009-04-10 08:23:10 Primer2 analog ground bug

kubark42
Member
Registered: 2009-02-13
Posts: 46

Re: Primer2 analog ground bug

Filtering might be a real good idea. Otherwise, you could bring the trace out somewhere on the extension connector side of the board, and 1) use a shorting block that we can unsolder; 2) put two blank pads just before the point where the AGND meets the DGND and then we can cut the trace with a knife, later soldering  a short across the pads if we want to relink AGND and DGND, or best, and space and budget permitting, a jumper.

Solution #2 is certainly the cheapest, as it requires no extra parts or manufacturing steps. However, it means lots of extra effort to get AGND running, which people less versed in soldering might be scared off.

Out of curiosity, at Embedded World Nuremberg, I was told that the first production run was for 5,000 units, and that future runs would be considered if the first run sold well. Am I to understand that you guys are seeing good sales data? I sure hope so!

Cheers,
Kenn

Offline

 

# 6   2009-04-10 16:42:43 Primer2 analog ground bug

jetcode
Member
Registered: 2009-04-03
Posts: 49

Re: Primer2 analog ground bug

I can't imagine the Primer2 as a sophisticated measurement system in it's current incarnation (demo development system) but hey I'm new to this product!

I am finishing up a measurement system project using a $1500 3 axis accelerometer for structural analysis on oil platforms. We designed the analog I/O board separate from the compute platform and provide power to the 16 bit 8 channel A/D (SPI), analog conditioning circuits, and the microcomputer. The power supply is essential for precise analog work and to date we have not had any noise issues. Wish I could say the same for dog slow 56k baud file transfer over a wireless radio!

Forgot to add that I'm interested in knowing what accelerometers you are using if it's convenient to post. I can't remember what accelerometer we are using but I do remember spending a solid week searching for it a couple of years back.

Joe

Last edited by jetcode (2009-04-10 16:47:21)

Offline

 

# 7   2009-04-11 19:28:10 Primer2 analog ground bug

kubark42
Member
Registered: 2009-02-13
Posts: 46

Re: Primer2 analog ground bug

jetcode :

I can't imagine the Primer2 as a sophisticated measurement system in it's current incarnation (demo development system) but hey I'm new to this product!

So true! However, supposing that the bug can be ironed out for (relatively) free, why not make the best product possible? Most won't use it, I'm betting, but those who will will really appreciate how everything on the Primer2 just... works.

jetcode :

Forgot to add that I'm interested in knowing what accelerometers you are using if it's convenient to post. I can't remember what accelerometer we are using but I do remember spending a solid week searching for it a couple of years back.

Nothing quite so expensive! It's just a dinky analog 2-D accelerometer that costs us about as much as the Primer2 did, if memory serves. I'd have to go dig up the board to remember the exact part number, but it was an Analog Devices, so nothing incredibly rare, either.

Cheers,
kenn

Offline

 

# 8   2009-04-12 21:33:42 Primer2 analog ground bug

jetcode
Member
Registered: 2009-04-03
Posts: 49

Re: Primer2 analog ground bug

I agree that separating the grounds and sources is a good idea. I work with embedded audio and design in separate supplies.

Offline

 

# 9   2009-04-13 13:04:45 Primer2 analog ground bug

stphilsalzmann
New member
Registered: 2009-03-30
Posts: 1

Re: Primer2 analog ground bug

Most often AGND (analog Ground) & DGND (digital ground) are seperated and tied together at only one point, often using a ferrite bead and maybe other filtering. The idea is that the AGND and DGND are well connected for low frequencies but not for high frequencies. In well designed Data Aquisition Instruments we can see the analog section of the design segragated and shielded in it's own area and the analog ground plane never overlaps the digital ground plane

Offline

 

Board footer